diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,140042 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1315.7894736842106,
+  "eval_steps": 500,
+  "global_step": 200000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.06578947368421052,
+      "grad_norm": 165.546875,
+      "learning_rate": 0.0001,
+      "loss": 21.3446,
+      "step": 10
+    },
+    {
+      "epoch": 0.13157894736842105,
+      "grad_norm": 92.7155990600586,
+      "learning_rate": 0.0001,
+      "loss": 4.7673,
+      "step": 20
+    },
+    {
+      "epoch": 0.19736842105263158,
+      "grad_norm": 64.37399291992188,
+      "learning_rate": 0.0001,
+      "loss": 3.561,
+      "step": 30
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 55.39424514770508,
+      "learning_rate": 0.0001,
+      "loss": 3.1393,
+      "step": 40
+    },
+    {
+      "epoch": 0.32894736842105265,
+      "grad_norm": 55.13695526123047,
+      "learning_rate": 0.0001,
+      "loss": 2.8728,
+      "step": 50
+    },
+    {
+      "epoch": 0.39473684210526316,
+      "grad_norm": 51.248966217041016,
+      "learning_rate": 0.0001,
+      "loss": 2.6172,
+      "step": 60
+    },
+    {
+      "epoch": 0.4605263157894737,
+      "grad_norm": 39.10142135620117,
+      "learning_rate": 0.0001,
+      "loss": 2.415,
+      "step": 70
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 47.69247055053711,
+      "learning_rate": 0.0001,
+      "loss": 2.2343,
+      "step": 80
+    },
+    {
+      "epoch": 0.5921052631578947,
+      "grad_norm": 49.21741485595703,
+      "learning_rate": 0.0001,
+      "loss": 2.0767,
+      "step": 90
+    },
+    {
+      "epoch": 0.6578947368421053,
+      "grad_norm": 44.604793548583984,
+      "learning_rate": 0.0001,
+      "loss": 2.0309,
+      "step": 100
+    },
+    {
+      "epoch": 0.7236842105263158,
+      "grad_norm": 51.46554183959961,
+      "learning_rate": 0.0001,
+      "loss": 1.8886,
+      "step": 110
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 46.2479133605957,
+      "learning_rate": 0.0001,
+      "loss": 1.7757,
+      "step": 120
+    },
+    {
+      "epoch": 0.8552631578947368,
+      "grad_norm": 49.21684646606445,
+      "learning_rate": 0.0001,
+      "loss": 1.6854,
+      "step": 130
+    },
+    {
+      "epoch": 0.9210526315789473,
+      "grad_norm": 49.37424087524414,
+      "learning_rate": 0.0001,
+      "loss": 1.614,
+      "step": 140
+    },
+    {
+      "epoch": 0.9868421052631579,
+      "grad_norm": 40.79842758178711,
+      "learning_rate": 0.0001,
+      "loss": 1.494,
+      "step": 150
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 43.07915115356445,
+      "learning_rate": 0.0001,
+      "loss": 1.4474,
+      "step": 160
+    },
+    {
+      "epoch": 1.118421052631579,
+      "grad_norm": 37.799560546875,
+      "learning_rate": 0.0001,
+      "loss": 1.3618,
+      "step": 170
+    },
+    {
+      "epoch": 1.1842105263157894,
+      "grad_norm": 37.14927291870117,
+      "learning_rate": 0.0001,
+      "loss": 1.3081,
+      "step": 180
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 35.22480010986328,
+      "learning_rate": 0.0001,
+      "loss": 1.2341,
+      "step": 190
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 33.31498718261719,
+      "learning_rate": 0.0001,
+      "loss": 1.169,
+      "step": 200
+    },
+    {
+      "epoch": 1.381578947368421,
+      "grad_norm": 27.80243492126465,
+      "learning_rate": 0.0001,
+      "loss": 1.0758,
+      "step": 210
+    },
+    {
+      "epoch": 1.4473684210526316,
+      "grad_norm": 35.85008239746094,
+      "learning_rate": 0.0001,
+      "loss": 1.0106,
+      "step": 220
+    },
+    {
+      "epoch": 1.513157894736842,
+      "grad_norm": 29.738868713378906,
+      "learning_rate": 0.0001,
+      "loss": 0.9601,
+      "step": 230
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 29.869279861450195,
+      "learning_rate": 0.0001,
+      "loss": 0.9334,
+      "step": 240
+    },
+    {
+      "epoch": 1.6447368421052633,
+      "grad_norm": 23.091772079467773,
+      "learning_rate": 0.0001,
+      "loss": 0.8842,
+      "step": 250
+    },
+    {
+      "epoch": 1.7105263157894737,
+      "grad_norm": 31.321971893310547,
+      "learning_rate": 0.0001,
+      "loss": 0.8304,
+      "step": 260
+    },
+    {
+      "epoch": 1.776315789473684,
+      "grad_norm": 31.007795333862305,
+      "learning_rate": 0.0001,
+      "loss": 0.7835,
+      "step": 270
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 23.761444091796875,
+      "learning_rate": 0.0001,
+      "loss": 0.741,
+      "step": 280
+    },
+    {
+      "epoch": 1.9078947368421053,
+      "grad_norm": 29.322738647460938,
+      "learning_rate": 0.0001,
+      "loss": 0.7034,
+      "step": 290
+    },
+    {
+      "epoch": 1.973684210526316,
+      "grad_norm": 24.82952308654785,
+      "learning_rate": 0.0001,
+      "loss": 0.6807,
+      "step": 300
+    },
+    {
+      "epoch": 2.039473684210526,
+      "grad_norm": 20.884836196899414,
+      "learning_rate": 0.0001,
+      "loss": 0.6234,
+      "step": 310
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 22.192235946655273,
+      "learning_rate": 0.0001,
+      "loss": 0.5987,
+      "step": 320
+    },
+    {
+      "epoch": 2.1710526315789473,
+      "grad_norm": 22.435510635375977,
+      "learning_rate": 0.0001,
+      "loss": 0.5711,
+      "step": 330
+    },
+    {
+      "epoch": 2.236842105263158,
+      "grad_norm": 20.451364517211914,
+      "learning_rate": 0.0001,
+      "loss": 0.5399,
+      "step": 340
+    },
+    {
+      "epoch": 2.3026315789473686,
+      "grad_norm": 22.708984375,
+      "learning_rate": 0.0001,
+      "loss": 0.5199,
+      "step": 350
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 20.40320587158203,
+      "learning_rate": 0.0001,
+      "loss": 0.4973,
+      "step": 360
+    },
+    {
+      "epoch": 2.4342105263157894,
+      "grad_norm": 17.256328582763672,
+      "learning_rate": 0.0001,
+      "loss": 0.4806,
+      "step": 370
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 19.842121124267578,
+      "learning_rate": 0.0001,
+      "loss": 0.4759,
+      "step": 380
+    },
+    {
+      "epoch": 2.5657894736842106,
+      "grad_norm": 17.68121337890625,
+      "learning_rate": 0.0001,
+      "loss": 0.4528,
+      "step": 390
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 19.98098373413086,
+      "learning_rate": 0.0001,
+      "loss": 0.4489,
+      "step": 400
+    },
+    {
+      "epoch": 2.6973684210526314,
+      "grad_norm": 18.062952041625977,
+      "learning_rate": 0.0001,
+      "loss": 0.4187,
+      "step": 410
+    },
+    {
+      "epoch": 2.763157894736842,
+      "grad_norm": 15.641585350036621,
+      "learning_rate": 0.0001,
+      "loss": 0.4068,
+      "step": 420
+    },
+    {
+      "epoch": 2.8289473684210527,
+      "grad_norm": 16.796161651611328,
+      "learning_rate": 0.0001,
+      "loss": 0.3884,
+      "step": 430
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 15.872922897338867,
+      "learning_rate": 0.0001,
+      "loss": 0.3997,
+      "step": 440
+    },
+    {
+      "epoch": 2.9605263157894735,
+      "grad_norm": 15.037050247192383,
+      "learning_rate": 0.0001,
+      "loss": 0.3687,
+      "step": 450
+    },
+    {
+      "epoch": 3.026315789473684,
+      "grad_norm": 18.485353469848633,
+      "learning_rate": 0.0001,
+      "loss": 0.3606,
+      "step": 460
+    },
+    {
+      "epoch": 3.0921052631578947,
+      "grad_norm": 17.94466209411621,
+      "learning_rate": 0.0001,
+      "loss": 0.3568,
+      "step": 470
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 15.609881401062012,
+      "learning_rate": 0.0001,
+      "loss": 0.3631,
+      "step": 480
+    },
+    {
+      "epoch": 3.223684210526316,
+      "grad_norm": 15.842222213745117,
+      "learning_rate": 0.0001,
+      "loss": 0.3499,
+      "step": 490
+    },
+    {
+      "epoch": 3.2894736842105265,
+      "grad_norm": 14.633220672607422,
+      "learning_rate": 0.0001,
+      "loss": 0.3442,
+      "step": 500
+    },
+    {
+      "epoch": 3.3552631578947367,
+      "grad_norm": 16.64904022216797,
+      "learning_rate": 0.0001,
+      "loss": 0.3311,
+      "step": 510
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 15.60223388671875,
+      "learning_rate": 0.0001,
+      "loss": 0.3356,
+      "step": 520
+    },
+    {
+      "epoch": 3.486842105263158,
+      "grad_norm": 14.84737491607666,
+      "learning_rate": 0.0001,
+      "loss": 0.3172,
+      "step": 530
+    },
+    {
+      "epoch": 3.5526315789473686,
+      "grad_norm": 15.952341079711914,
+      "learning_rate": 0.0001,
+      "loss": 0.3196,
+      "step": 540
+    },
+    {
+      "epoch": 3.6184210526315788,
+      "grad_norm": 14.867337226867676,
+      "learning_rate": 0.0001,
+      "loss": 0.3137,
+      "step": 550
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 12.948201179504395,
+      "learning_rate": 0.0001,
+      "loss": 0.2949,
+      "step": 560
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 13.009255409240723,
+      "learning_rate": 0.0001,
+      "loss": 0.2955,
+      "step": 570
+    },
+    {
+      "epoch": 3.8157894736842106,
+      "grad_norm": 13.705317497253418,
+      "learning_rate": 0.0001,
+      "loss": 0.2905,
+      "step": 580
+    },
+    {
+      "epoch": 3.8815789473684212,
+      "grad_norm": 17.22597885131836,
+      "learning_rate": 0.0001,
+      "loss": 0.2974,
+      "step": 590
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 14.919693946838379,
+      "learning_rate": 0.0001,
+      "loss": 0.2898,
+      "step": 600
+    },
+    {
+      "epoch": 4.0131578947368425,
+      "grad_norm": 18.780914306640625,
+      "learning_rate": 0.0001,
+      "loss": 0.2816,
+      "step": 610
+    },
+    {
+      "epoch": 4.078947368421052,
+      "grad_norm": 16.027746200561523,
+      "learning_rate": 0.0001,
+      "loss": 0.2761,
+      "step": 620
+    },
+    {
+      "epoch": 4.144736842105263,
+      "grad_norm": 13.083127975463867,
+      "learning_rate": 0.0001,
+      "loss": 0.2688,
+      "step": 630
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 13.55245590209961,
+      "learning_rate": 0.0001,
+      "loss": 0.2667,
+      "step": 640
+    },
+    {
+      "epoch": 4.276315789473684,
+      "grad_norm": 11.151076316833496,
+      "learning_rate": 0.0001,
+      "loss": 0.2677,
+      "step": 650
+    },
+    {
+      "epoch": 4.342105263157895,
+      "grad_norm": 12.883306503295898,
+      "learning_rate": 0.0001,
+      "loss": 0.2619,
+      "step": 660
+    },
+    {
+      "epoch": 4.407894736842105,
+      "grad_norm": 13.993839263916016,
+      "learning_rate": 0.0001,
+      "loss": 0.2626,
+      "step": 670
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 13.596793174743652,
+      "learning_rate": 0.0001,
+      "loss": 0.253,
+      "step": 680
+    },
+    {
+      "epoch": 4.5394736842105265,
+      "grad_norm": 12.375000953674316,
+      "learning_rate": 0.0001,
+      "loss": 0.2449,
+      "step": 690
+    },
+    {
+      "epoch": 4.605263157894737,
+      "grad_norm": 12.333367347717285,
+      "learning_rate": 0.0001,
+      "loss": 0.2449,
+      "step": 700
+    },
+    {
+      "epoch": 4.671052631578947,
+      "grad_norm": 14.13685131072998,
+      "learning_rate": 0.0001,
+      "loss": 0.2363,
+      "step": 710
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 12.23291015625,
+      "learning_rate": 0.0001,
+      "loss": 0.2374,
+      "step": 720
+    },
+    {
+      "epoch": 4.802631578947368,
+      "grad_norm": 11.502880096435547,
+      "learning_rate": 0.0001,
+      "loss": 0.2289,
+      "step": 730
+    },
+    {
+      "epoch": 4.868421052631579,
+      "grad_norm": 10.85533332824707,
+      "learning_rate": 0.0001,
+      "loss": 0.2266,
+      "step": 740
+    },
+    {
+      "epoch": 4.934210526315789,
+      "grad_norm": 12.013697624206543,
+      "learning_rate": 0.0001,
+      "loss": 0.2188,
+      "step": 750
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 13.170636177062988,
+      "learning_rate": 0.0001,
+      "loss": 0.2321,
+      "step": 760
+    },
+    {
+      "epoch": 5.065789473684211,
+      "grad_norm": 12.095258712768555,
+      "learning_rate": 0.0001,
+      "loss": 0.2229,
+      "step": 770
+    },
+    {
+      "epoch": 5.131578947368421,
+      "grad_norm": 10.180475234985352,
+      "learning_rate": 0.0001,
+      "loss": 0.2231,
+      "step": 780
+    },
+    {
+      "epoch": 5.197368421052632,
+      "grad_norm": 13.165190696716309,
+      "learning_rate": 0.0001,
+      "loss": 0.223,
+      "step": 790
+    },
+    {
+      "epoch": 5.2631578947368425,
+      "grad_norm": 10.063823699951172,
+      "learning_rate": 0.0001,
+      "loss": 0.2118,
+      "step": 800
+    },
+    {
+      "epoch": 5.328947368421053,
+      "grad_norm": 12.834861755371094,
+      "learning_rate": 0.0001,
+      "loss": 0.217,
+      "step": 810
+    },
+    {
+      "epoch": 5.394736842105263,
+      "grad_norm": 12.119063377380371,
+      "learning_rate": 0.0001,
+      "loss": 0.2053,
+      "step": 820
+    },
+    {
+      "epoch": 5.4605263157894735,
+      "grad_norm": 12.132641792297363,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 830
+    },
+    {
+      "epoch": 5.526315789473684,
+      "grad_norm": 11.347237586975098,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 840
+    },
+    {
+      "epoch": 5.592105263157895,
+      "grad_norm": 11.609687805175781,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 850
+    },
+    {
+      "epoch": 5.657894736842105,
+      "grad_norm": 10.659272193908691,
+      "learning_rate": 0.0001,
+      "loss": 0.203,
+      "step": 860
+    },
+    {
+      "epoch": 5.723684210526316,
+      "grad_norm": 12.028623580932617,
+      "learning_rate": 0.0001,
+      "loss": 0.21,
+      "step": 870
+    },
+    {
+      "epoch": 5.7894736842105265,
+      "grad_norm": 10.963968276977539,
+      "learning_rate": 0.0001,
+      "loss": 0.2019,
+      "step": 880
+    },
+    {
+      "epoch": 5.855263157894737,
+      "grad_norm": 10.573838233947754,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 890
+    },
+    {
+      "epoch": 5.921052631578947,
+      "grad_norm": 10.971415519714355,
+      "learning_rate": 0.0001,
+      "loss": 0.1975,
+      "step": 900
+    },
+    {
+      "epoch": 5.9868421052631575,
+      "grad_norm": 12.391592025756836,
+      "learning_rate": 0.0001,
+      "loss": 0.2002,
+      "step": 910
+    },
+    {
+      "epoch": 6.052631578947368,
+      "grad_norm": 11.677957534790039,
+      "learning_rate": 0.0001,
+      "loss": 0.1949,
+      "step": 920
+    },
+    {
+      "epoch": 6.118421052631579,
+      "grad_norm": 9.757022857666016,
+      "learning_rate": 0.0001,
+      "loss": 0.1887,
+      "step": 930
+    },
+    {
+      "epoch": 6.184210526315789,
+      "grad_norm": 10.96141529083252,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 940
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 11.207174301147461,
+      "learning_rate": 0.0001,
+      "loss": 0.1881,
+      "step": 950
+    },
+    {
+      "epoch": 6.315789473684211,
+      "grad_norm": 9.526222229003906,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 960
+    },
+    {
+      "epoch": 6.381578947368421,
+      "grad_norm": 10.266683578491211,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 970
+    },
+    {
+      "epoch": 6.447368421052632,
+      "grad_norm": 10.702817916870117,
+      "learning_rate": 0.0001,
+      "loss": 0.1877,
+      "step": 980
+    },
+    {
+      "epoch": 6.5131578947368425,
+      "grad_norm": 9.538935661315918,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 990
+    },
+    {
+      "epoch": 6.578947368421053,
+      "grad_norm": 9.56125545501709,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 1000
+    },
+    {
+      "epoch": 6.644736842105263,
+      "grad_norm": 9.874727249145508,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1010
+    },
+    {
+      "epoch": 6.7105263157894735,
+      "grad_norm": 9.713470458984375,
+      "learning_rate": 0.0001,
+      "loss": 0.1788,
+      "step": 1020
+    },
+    {
+      "epoch": 6.776315789473684,
+      "grad_norm": 9.037981986999512,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1030
+    },
+    {
+      "epoch": 6.842105263157895,
+      "grad_norm": 10.102965354919434,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 1040
+    },
+    {
+      "epoch": 6.907894736842105,
+      "grad_norm": 9.202942848205566,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1050
+    },
+    {
+      "epoch": 6.973684210526316,
+      "grad_norm": 10.504417419433594,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1060
+    },
+    {
+      "epoch": 7.0394736842105265,
+      "grad_norm": 8.70566463470459,
+      "learning_rate": 0.0001,
+      "loss": 0.1724,
+      "step": 1070
+    },
+    {
+      "epoch": 7.105263157894737,
+      "grad_norm": 10.730690956115723,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 1080
+    },
+    {
+      "epoch": 7.171052631578948,
+      "grad_norm": 9.544865608215332,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 1090
+    },
+    {
+      "epoch": 7.2368421052631575,
+      "grad_norm": 10.796236038208008,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1100
+    },
+    {
+      "epoch": 7.302631578947368,
+      "grad_norm": 9.174856185913086,
+      "learning_rate": 0.0001,
+      "loss": 0.1705,
+      "step": 1110
+    },
+    {
+      "epoch": 7.368421052631579,
+      "grad_norm": 10.11459732055664,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 1120
+    },
+    {
+      "epoch": 7.434210526315789,
+      "grad_norm": 10.245781898498535,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 1130
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 9.092036247253418,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 1140
+    },
+    {
+      "epoch": 7.565789473684211,
+      "grad_norm": 8.524528503417969,
+      "learning_rate": 0.0001,
+      "loss": 0.1528,
+      "step": 1150
+    },
+    {
+      "epoch": 7.631578947368421,
+      "grad_norm": 9.26682186126709,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 1160
+    },
+    {
+      "epoch": 7.697368421052632,
+      "grad_norm": 10.105231285095215,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 1170
+    },
+    {
+      "epoch": 7.7631578947368425,
+      "grad_norm": 9.100499153137207,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 1180
+    },
+    {
+      "epoch": 7.828947368421053,
+      "grad_norm": 7.819246768951416,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 1190
+    },
+    {
+      "epoch": 7.894736842105263,
+      "grad_norm": 8.291773796081543,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 1200
+    },
+    {
+      "epoch": 7.9605263157894735,
+      "grad_norm": 8.024109840393066,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 1210
+    },
+    {
+      "epoch": 8.026315789473685,
+      "grad_norm": 8.904583930969238,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 1220
+    },
+    {
+      "epoch": 8.092105263157896,
+      "grad_norm": 7.31992769241333,
+      "learning_rate": 0.0001,
+      "loss": 0.1453,
+      "step": 1230
+    },
+    {
+      "epoch": 8.157894736842104,
+      "grad_norm": 9.026426315307617,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 1240
+    },
+    {
+      "epoch": 8.223684210526315,
+      "grad_norm": 9.364782333374023,
+      "learning_rate": 0.0001,
+      "loss": 0.1504,
+      "step": 1250
+    },
+    {
+      "epoch": 8.289473684210526,
+      "grad_norm": 8.736329078674316,
+      "learning_rate": 0.0001,
+      "loss": 0.1445,
+      "step": 1260
+    },
+    {
+      "epoch": 8.355263157894736,
+      "grad_norm": 11.340313911437988,
+      "learning_rate": 0.0001,
+      "loss": 0.1469,
+      "step": 1270
+    },
+    {
+      "epoch": 8.421052631578947,
+      "grad_norm": 10.251652717590332,
+      "learning_rate": 0.0001,
+      "loss": 0.1559,
+      "step": 1280
+    },
+    {
+      "epoch": 8.486842105263158,
+      "grad_norm": 9.239073753356934,
+      "learning_rate": 0.0001,
+      "loss": 0.1456,
+      "step": 1290
+    },
+    {
+      "epoch": 8.552631578947368,
+      "grad_norm": 8.867827415466309,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 1300
+    },
+    {
+      "epoch": 8.618421052631579,
+      "grad_norm": 9.638751983642578,
+      "learning_rate": 0.0001,
+      "loss": 0.1521,
+      "step": 1310
+    },
+    {
+      "epoch": 8.68421052631579,
+      "grad_norm": 9.176225662231445,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 1320
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 8.754321098327637,
+      "learning_rate": 0.0001,
+      "loss": 0.1423,
+      "step": 1330
+    },
+    {
+      "epoch": 8.81578947368421,
+      "grad_norm": 9.146409034729004,
+      "learning_rate": 0.0001,
+      "loss": 0.1397,
+      "step": 1340
+    },
+    {
+      "epoch": 8.881578947368421,
+      "grad_norm": 9.756692886352539,
+      "learning_rate": 0.0001,
+      "loss": 0.1339,
+      "step": 1350
+    },
+    {
+      "epoch": 8.947368421052632,
+      "grad_norm": 8.44552230834961,
+      "learning_rate": 0.0001,
+      "loss": 0.1433,
+      "step": 1360
+    },
+    {
+      "epoch": 9.013157894736842,
+      "grad_norm": 9.708316802978516,
+      "learning_rate": 0.0001,
+      "loss": 0.1443,
+      "step": 1370
+    },
+    {
+      "epoch": 9.078947368421053,
+      "grad_norm": 10.184338569641113,
+      "learning_rate": 0.0001,
+      "loss": 0.1397,
+      "step": 1380
+    },
+    {
+      "epoch": 9.144736842105264,
+      "grad_norm": 8.404060363769531,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 1390
+    },
+    {
+      "epoch": 9.210526315789474,
+      "grad_norm": 6.799143314361572,
+      "learning_rate": 0.0001,
+      "loss": 0.1391,
+      "step": 1400
+    },
+    {
+      "epoch": 9.276315789473685,
+      "grad_norm": 6.993669033050537,
+      "learning_rate": 0.0001,
+      "loss": 0.1356,
+      "step": 1410
+    },
+    {
+      "epoch": 9.342105263157896,
+      "grad_norm": 7.0708537101745605,
+      "learning_rate": 0.0001,
+      "loss": 0.1348,
+      "step": 1420
+    },
+    {
+      "epoch": 9.407894736842104,
+      "grad_norm": 7.106939792633057,
+      "learning_rate": 0.0001,
+      "loss": 0.1395,
+      "step": 1430
+    },
+    {
+      "epoch": 9.473684210526315,
+      "grad_norm": 8.091696739196777,
+      "learning_rate": 0.0001,
+      "loss": 0.1347,
+      "step": 1440
+    },
+    {
+      "epoch": 9.539473684210526,
+      "grad_norm": 7.712802410125732,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 1450
+    },
+    {
+      "epoch": 9.605263157894736,
+      "grad_norm": 7.227226734161377,
+      "learning_rate": 0.0001,
+      "loss": 0.1302,
+      "step": 1460
+    },
+    {
+      "epoch": 9.671052631578947,
+      "grad_norm": 7.4794464111328125,
+      "learning_rate": 0.0001,
+      "loss": 0.1238,
+      "step": 1470
+    },
+    {
+      "epoch": 9.736842105263158,
+      "grad_norm": 8.05570125579834,
+      "learning_rate": 0.0001,
+      "loss": 0.134,
+      "step": 1480
+    },
+    {
+      "epoch": 9.802631578947368,
+      "grad_norm": 8.51727294921875,
+      "learning_rate": 0.0001,
+      "loss": 0.1262,
+      "step": 1490
+    },
+    {
+      "epoch": 9.868421052631579,
+      "grad_norm": 8.46773910522461,
+      "learning_rate": 0.0001,
+      "loss": 0.133,
+      "step": 1500
+    },
+    {
+      "epoch": 9.93421052631579,
+      "grad_norm": 8.314281463623047,
+      "learning_rate": 0.0001,
+      "loss": 0.1297,
+      "step": 1510
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 8.289139747619629,
+      "learning_rate": 0.0001,
+      "loss": 0.131,
+      "step": 1520
+    },
+    {
+      "epoch": 10.06578947368421,
+      "grad_norm": 8.90967082977295,
+      "learning_rate": 0.0001,
+      "loss": 0.1232,
+      "step": 1530
+    },
+    {
+      "epoch": 10.131578947368421,
+      "grad_norm": 8.999959945678711,
+      "learning_rate": 0.0001,
+      "loss": 0.1273,
+      "step": 1540
+    },
+    {
+      "epoch": 10.197368421052632,
+      "grad_norm": 8.188413619995117,
+      "learning_rate": 0.0001,
+      "loss": 0.1263,
+      "step": 1550
+    },
+    {
+      "epoch": 10.263157894736842,
+      "grad_norm": 8.14466381072998,
+      "learning_rate": 0.0001,
+      "loss": 0.1204,
+      "step": 1560
+    },
+    {
+      "epoch": 10.328947368421053,
+      "grad_norm": 8.005374908447266,
+      "learning_rate": 0.0001,
+      "loss": 0.1232,
+      "step": 1570
+    },
+    {
+      "epoch": 10.394736842105264,
+      "grad_norm": 8.354129791259766,
+      "learning_rate": 0.0001,
+      "loss": 0.1164,
+      "step": 1580
+    },
+    {
+      "epoch": 10.460526315789474,
+      "grad_norm": 7.809772491455078,
+      "learning_rate": 0.0001,
+      "loss": 0.1256,
+      "step": 1590
+    },
+    {
+      "epoch": 10.526315789473685,
+      "grad_norm": 7.232692241668701,
+      "learning_rate": 0.0001,
+      "loss": 0.1226,
+      "step": 1600
+    },
+    {
+      "epoch": 10.592105263157894,
+      "grad_norm": 7.019472122192383,
+      "learning_rate": 0.0001,
+      "loss": 0.1184,
+      "step": 1610
+    },
+    {
+      "epoch": 10.657894736842106,
+      "grad_norm": 6.870980262756348,
+      "learning_rate": 0.0001,
+      "loss": 0.1196,
+      "step": 1620
+    },
+    {
+      "epoch": 10.723684210526315,
+      "grad_norm": 7.686590671539307,
+      "learning_rate": 0.0001,
+      "loss": 0.123,
+      "step": 1630
+    },
+    {
+      "epoch": 10.789473684210526,
+      "grad_norm": 6.786712169647217,
+      "learning_rate": 0.0001,
+      "loss": 0.119,
+      "step": 1640
+    },
+    {
+      "epoch": 10.855263157894736,
+      "grad_norm": 8.563821792602539,
+      "learning_rate": 0.0001,
+      "loss": 0.1186,
+      "step": 1650
+    },
+    {
+      "epoch": 10.921052631578947,
+      "grad_norm": 8.261466026306152,
+      "learning_rate": 0.0001,
+      "loss": 0.1246,
+      "step": 1660
+    },
+    {
+      "epoch": 10.986842105263158,
+      "grad_norm": 6.972947120666504,
+      "learning_rate": 0.0001,
+      "loss": 0.123,
+      "step": 1670
+    },
+    {
+      "epoch": 11.052631578947368,
+      "grad_norm": 8.273945808410645,
+      "learning_rate": 0.0001,
+      "loss": 0.1213,
+      "step": 1680
+    },
+    {
+      "epoch": 11.118421052631579,
+      "grad_norm": 6.411956310272217,
+      "learning_rate": 0.0001,
+      "loss": 0.1123,
+      "step": 1690
+    },
+    {
+      "epoch": 11.18421052631579,
+      "grad_norm": 7.9852166175842285,
+      "learning_rate": 0.0001,
+      "loss": 0.1125,
+      "step": 1700
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 6.184416770935059,
+      "learning_rate": 0.0001,
+      "loss": 0.1113,
+      "step": 1710
+    },
+    {
+      "epoch": 11.31578947368421,
+      "grad_norm": 7.070769786834717,
+      "learning_rate": 0.0001,
+      "loss": 0.115,
+      "step": 1720
+    },
+    {
+      "epoch": 11.381578947368421,
+      "grad_norm": 6.914794921875,
+      "learning_rate": 0.0001,
+      "loss": 0.111,
+      "step": 1730
+    },
+    {
+      "epoch": 11.447368421052632,
+      "grad_norm": 7.74068546295166,
+      "learning_rate": 0.0001,
+      "loss": 0.113,
+      "step": 1740
+    },
+    {
+      "epoch": 11.513157894736842,
+      "grad_norm": 6.931029319763184,
+      "learning_rate": 0.0001,
+      "loss": 0.1152,
+      "step": 1750
+    },
+    {
+      "epoch": 11.578947368421053,
+      "grad_norm": 5.994091510772705,
+      "learning_rate": 0.0001,
+      "loss": 0.1119,
+      "step": 1760
+    },
+    {
+      "epoch": 11.644736842105264,
+      "grad_norm": 6.253146648406982,
+      "learning_rate": 0.0001,
+      "loss": 0.1093,
+      "step": 1770
+    },
+    {
+      "epoch": 11.710526315789474,
+      "grad_norm": 5.8790974617004395,
+      "learning_rate": 0.0001,
+      "loss": 0.1122,
+      "step": 1780
+    },
+    {
+      "epoch": 11.776315789473685,
+      "grad_norm": 6.187469959259033,
+      "learning_rate": 0.0001,
+      "loss": 0.1122,
+      "step": 1790
+    },
+    {
+      "epoch": 11.842105263157894,
+      "grad_norm": 5.537177562713623,
+      "learning_rate": 0.0001,
+      "loss": 0.1123,
+      "step": 1800
+    },
+    {
+      "epoch": 11.907894736842106,
+      "grad_norm": 7.146965503692627,
+      "learning_rate": 0.0001,
+      "loss": 0.1118,
+      "step": 1810
+    },
+    {
+      "epoch": 11.973684210526315,
+      "grad_norm": 6.597342491149902,
+      "learning_rate": 0.0001,
+      "loss": 0.1053,
+      "step": 1820
+    },
+    {
+      "epoch": 12.039473684210526,
+      "grad_norm": 6.711949825286865,
+      "learning_rate": 0.0001,
+      "loss": 0.1098,
+      "step": 1830
+    },
+    {
+      "epoch": 12.105263157894736,
+      "grad_norm": 5.853811264038086,
+      "learning_rate": 0.0001,
+      "loss": 0.1072,
+      "step": 1840
+    },
+    {
+      "epoch": 12.171052631578947,
+      "grad_norm": 5.994247913360596,
+      "learning_rate": 0.0001,
+      "loss": 0.1058,
+      "step": 1850
+    },
+    {
+      "epoch": 12.236842105263158,
+      "grad_norm": 4.701684951782227,
+      "learning_rate": 0.0001,
+      "loss": 0.1044,
+      "step": 1860
+    },
+    {
+      "epoch": 12.302631578947368,
+      "grad_norm": 6.289170742034912,
+      "learning_rate": 0.0001,
+      "loss": 0.1025,
+      "step": 1870
+    },
+    {
+      "epoch": 12.368421052631579,
+      "grad_norm": 4.761482238769531,
+      "learning_rate": 0.0001,
+      "loss": 0.1028,
+      "step": 1880
+    },
+    {
+      "epoch": 12.43421052631579,
+      "grad_norm": 7.5639448165893555,
+      "learning_rate": 0.0001,
+      "loss": 0.1028,
+      "step": 1890
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 5.687307357788086,
+      "learning_rate": 0.0001,
+      "loss": 0.1056,
+      "step": 1900
+    },
+    {
+      "epoch": 12.56578947368421,
+      "grad_norm": 6.268471717834473,
+      "learning_rate": 0.0001,
+      "loss": 0.0987,
+      "step": 1910
+    },
+    {
+      "epoch": 12.631578947368421,
+      "grad_norm": 7.3990254402160645,
+      "learning_rate": 0.0001,
+      "loss": 0.0981,
+      "step": 1920
+    },
+    {
+      "epoch": 12.697368421052632,
+      "grad_norm": 6.721970558166504,
+      "learning_rate": 0.0001,
+      "loss": 0.1024,
+      "step": 1930
+    },
+    {
+      "epoch": 12.763157894736842,
+      "grad_norm": 6.922434329986572,
+      "learning_rate": 0.0001,
+      "loss": 0.1034,
+      "step": 1940
+    },
+    {
+      "epoch": 12.828947368421053,
+      "grad_norm": 6.5644097328186035,
+      "learning_rate": 0.0001,
+      "loss": 0.1012,
+      "step": 1950
+    },
+    {
+      "epoch": 12.894736842105264,
+      "grad_norm": 5.8850908279418945,
+      "learning_rate": 0.0001,
+      "loss": 0.1011,
+      "step": 1960
+    },
+    {
+      "epoch": 12.960526315789474,
+      "grad_norm": 7.03394889831543,
+      "learning_rate": 0.0001,
+      "loss": 0.1031,
+      "step": 1970
+    },
+    {
+      "epoch": 13.026315789473685,
+      "grad_norm": 7.228178024291992,
+      "learning_rate": 0.0001,
+      "loss": 0.1021,
+      "step": 1980
+    },
+    {
+      "epoch": 13.092105263157896,
+      "grad_norm": 6.708334922790527,
+      "learning_rate": 0.0001,
+      "loss": 0.1003,
+      "step": 1990
+    },
+    {
+      "epoch": 13.157894736842104,
+      "grad_norm": 6.730384826660156,
+      "learning_rate": 0.0001,
+      "loss": 0.1034,
+      "step": 2000
+    },
+    {
+      "epoch": 13.223684210526315,
+      "grad_norm": 6.330638408660889,
+      "learning_rate": 0.0001,
+      "loss": 0.0981,
+      "step": 2010
+    },
+    {
+      "epoch": 13.289473684210526,
+      "grad_norm": 6.243671417236328,
+      "learning_rate": 0.0001,
+      "loss": 0.1005,
+      "step": 2020
+    },
+    {
+      "epoch": 13.355263157894736,
+      "grad_norm": 7.014003276824951,
+      "learning_rate": 0.0001,
+      "loss": 0.0937,
+      "step": 2030
+    },
+    {
+      "epoch": 13.421052631578947,
+      "grad_norm": 6.188398361206055,
+      "learning_rate": 0.0001,
+      "loss": 0.0967,
+      "step": 2040
+    },
+    {
+      "epoch": 13.486842105263158,
+      "grad_norm": 6.1966938972473145,
+      "learning_rate": 0.0001,
+      "loss": 0.1026,
+      "step": 2050
+    },
+    {
+      "epoch": 13.552631578947368,
+      "grad_norm": 5.604138374328613,
+      "learning_rate": 0.0001,
+      "loss": 0.1022,
+      "step": 2060
+    },
+    {
+      "epoch": 13.618421052631579,
+      "grad_norm": 6.079825401306152,
+      "learning_rate": 0.0001,
+      "loss": 0.096,
+      "step": 2070
+    },
+    {
+      "epoch": 13.68421052631579,
+      "grad_norm": 6.047208786010742,
+      "learning_rate": 0.0001,
+      "loss": 0.0943,
+      "step": 2080
+    },
+    {
+      "epoch": 13.75,
+      "grad_norm": 6.325393199920654,
+      "learning_rate": 0.0001,
+      "loss": 0.0975,
+      "step": 2090
+    },
+    {
+      "epoch": 13.81578947368421,
+      "grad_norm": 5.832712650299072,
+      "learning_rate": 0.0001,
+      "loss": 0.0944,
+      "step": 2100
+    },
+    {
+      "epoch": 13.881578947368421,
+      "grad_norm": 5.263245105743408,
+      "learning_rate": 0.0001,
+      "loss": 0.0933,
+      "step": 2110
+    },
+    {
+      "epoch": 13.947368421052632,
+      "grad_norm": 5.7324934005737305,
+      "learning_rate": 0.0001,
+      "loss": 0.0919,
+      "step": 2120
+    },
+    {
+      "epoch": 14.013157894736842,
+      "grad_norm": 4.923053741455078,
+      "learning_rate": 0.0001,
+      "loss": 0.0895,
+      "step": 2130
+    },
+    {
+      "epoch": 14.078947368421053,
+      "grad_norm": 5.948666095733643,
+      "learning_rate": 0.0001,
+      "loss": 0.0914,
+      "step": 2140
+    },
+    {
+      "epoch": 14.144736842105264,
+      "grad_norm": 5.863275051116943,
+      "learning_rate": 0.0001,
+      "loss": 0.0898,
+      "step": 2150
+    },
+    {
+      "epoch": 14.210526315789474,
+      "grad_norm": 5.876058101654053,
+      "learning_rate": 0.0001,
+      "loss": 0.0926,
+      "step": 2160
+    },
+    {
+      "epoch": 14.276315789473685,
+      "grad_norm": 5.606838703155518,
+      "learning_rate": 0.0001,
+      "loss": 0.0927,
+      "step": 2170
+    },
+    {
+      "epoch": 14.342105263157896,
+      "grad_norm": 5.575480937957764,
+      "learning_rate": 0.0001,
+      "loss": 0.0929,
+      "step": 2180
+    },
+    {
+      "epoch": 14.407894736842104,
+      "grad_norm": 5.6978440284729,
+      "learning_rate": 0.0001,
+      "loss": 0.0933,
+      "step": 2190
+    },
+    {
+      "epoch": 14.473684210526315,
+      "grad_norm": 6.567765712738037,
+      "learning_rate": 0.0001,
+      "loss": 0.09,
+      "step": 2200
+    },
+    {
+      "epoch": 14.539473684210526,
+      "grad_norm": 6.093446254730225,
+      "learning_rate": 0.0001,
+      "loss": 0.0933,
+      "step": 2210
+    },
+    {
+      "epoch": 14.605263157894736,
+      "grad_norm": 5.184361457824707,
+      "learning_rate": 0.0001,
+      "loss": 0.0942,
+      "step": 2220
+    },
+    {
+      "epoch": 14.671052631578947,
+      "grad_norm": 5.555819511413574,
+      "learning_rate": 0.0001,
+      "loss": 0.086,
+      "step": 2230
+    },
+    {
+      "epoch": 14.736842105263158,
+      "grad_norm": 5.339320182800293,
+      "learning_rate": 0.0001,
+      "loss": 0.0918,
+      "step": 2240
+    },
+    {
+      "epoch": 14.802631578947368,
+      "grad_norm": 5.596290588378906,
+      "learning_rate": 0.0001,
+      "loss": 0.0912,
+      "step": 2250
+    },
+    {
+      "epoch": 14.868421052631579,
+      "grad_norm": 4.995247840881348,
+      "learning_rate": 0.0001,
+      "loss": 0.0881,
+      "step": 2260
+    },
+    {
+      "epoch": 14.93421052631579,
+      "grad_norm": 5.6531548500061035,
+      "learning_rate": 0.0001,
+      "loss": 0.0871,
+      "step": 2270
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 5.662561893463135,
+      "learning_rate": 0.0001,
+      "loss": 0.0852,
+      "step": 2280
+    },
+    {
+      "epoch": 15.06578947368421,
+      "grad_norm": 5.818594932556152,
+      "learning_rate": 0.0001,
+      "loss": 0.087,
+      "step": 2290
+    },
+    {
+      "epoch": 15.131578947368421,
+      "grad_norm": 6.480447769165039,
+      "learning_rate": 0.0001,
+      "loss": 0.0856,
+      "step": 2300
+    },
+    {
+      "epoch": 15.197368421052632,
+      "grad_norm": 5.37237548828125,
+      "learning_rate": 0.0001,
+      "loss": 0.0855,
+      "step": 2310
+    },
+    {
+      "epoch": 15.263157894736842,
+      "grad_norm": 5.8733696937561035,
+      "learning_rate": 0.0001,
+      "loss": 0.0835,
+      "step": 2320
+    },
+    {
+      "epoch": 15.328947368421053,
+      "grad_norm": 6.209512710571289,
+      "learning_rate": 0.0001,
+      "loss": 0.0856,
+      "step": 2330
+    },
+    {
+      "epoch": 15.394736842105264,
+      "grad_norm": 6.127361297607422,
+      "learning_rate": 0.0001,
+      "loss": 0.0864,
+      "step": 2340
+    },
+    {
+      "epoch": 15.460526315789474,
+      "grad_norm": 5.621913433074951,
+      "learning_rate": 0.0001,
+      "loss": 0.0818,
+      "step": 2350
+    },
+    {
+      "epoch": 15.526315789473685,
+      "grad_norm": 5.90310525894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0811,
+      "step": 2360
+    },
+    {
+      "epoch": 15.592105263157894,
+      "grad_norm": 5.0243239402771,
+      "learning_rate": 0.0001,
+      "loss": 0.0819,
+      "step": 2370
+    },
+    {
+      "epoch": 15.657894736842106,
+      "grad_norm": 6.154341697692871,
+      "learning_rate": 0.0001,
+      "loss": 0.0839,
+      "step": 2380
+    },
+    {
+      "epoch": 15.723684210526315,
+      "grad_norm": 5.303261756896973,
+      "learning_rate": 0.0001,
+      "loss": 0.0798,
+      "step": 2390
+    },
+    {
+      "epoch": 15.789473684210526,
+      "grad_norm": 5.788863658905029,
+      "learning_rate": 0.0001,
+      "loss": 0.0804,
+      "step": 2400
+    },
+    {
+      "epoch": 15.855263157894736,
+      "grad_norm": 5.377126693725586,
+      "learning_rate": 0.0001,
+      "loss": 0.0812,
+      "step": 2410
+    },
+    {
+      "epoch": 15.921052631578947,
+      "grad_norm": 5.7089972496032715,
+      "learning_rate": 0.0001,
+      "loss": 0.0777,
+      "step": 2420
+    },
+    {
+      "epoch": 15.986842105263158,
+      "grad_norm": 5.824652671813965,
+      "learning_rate": 0.0001,
+      "loss": 0.0834,
+      "step": 2430
+    },
+    {
+      "epoch": 16.05263157894737,
+      "grad_norm": 4.787572383880615,
+      "learning_rate": 0.0001,
+      "loss": 0.078,
+      "step": 2440
+    },
+    {
+      "epoch": 16.11842105263158,
+      "grad_norm": 6.0126800537109375,
+      "learning_rate": 0.0001,
+      "loss": 0.0794,
+      "step": 2450
+    },
+    {
+      "epoch": 16.18421052631579,
+      "grad_norm": 6.0084919929504395,
+      "learning_rate": 0.0001,
+      "loss": 0.08,
+      "step": 2460
+    },
+    {
+      "epoch": 16.25,
+      "grad_norm": 5.7310380935668945,
+      "learning_rate": 0.0001,
+      "loss": 0.0795,
+      "step": 2470
+    },
+    {
+      "epoch": 16.31578947368421,
+      "grad_norm": 5.645249366760254,
+      "learning_rate": 0.0001,
+      "loss": 0.0772,
+      "step": 2480
+    },
+    {
+      "epoch": 16.38157894736842,
+      "grad_norm": 4.670828342437744,
+      "learning_rate": 0.0001,
+      "loss": 0.0838,
+      "step": 2490
+    },
+    {
+      "epoch": 16.44736842105263,
+      "grad_norm": 4.980633735656738,
+      "learning_rate": 0.0001,
+      "loss": 0.0772,
+      "step": 2500
+    },
+    {
+      "epoch": 16.513157894736842,
+      "grad_norm": 4.4033660888671875,
+      "learning_rate": 0.0001,
+      "loss": 0.0819,
+      "step": 2510
+    },
+    {
+      "epoch": 16.57894736842105,
+      "grad_norm": 5.503461837768555,
+      "learning_rate": 0.0001,
+      "loss": 0.0792,
+      "step": 2520
+    },
+    {
+      "epoch": 16.644736842105264,
+      "grad_norm": 4.978002071380615,
+      "learning_rate": 0.0001,
+      "loss": 0.0797,
+      "step": 2530
+    },
+    {
+      "epoch": 16.710526315789473,
+      "grad_norm": 4.877954006195068,
+      "learning_rate": 0.0001,
+      "loss": 0.0792,
+      "step": 2540
+    },
+    {
+      "epoch": 16.776315789473685,
+      "grad_norm": 4.718324184417725,
+      "learning_rate": 0.0001,
+      "loss": 0.0781,
+      "step": 2550
+    },
+    {
+      "epoch": 16.842105263157894,
+      "grad_norm": 4.839942455291748,
+      "learning_rate": 0.0001,
+      "loss": 0.0768,
+      "step": 2560
+    },
+    {
+      "epoch": 16.907894736842106,
+      "grad_norm": 4.971443176269531,
+      "learning_rate": 0.0001,
+      "loss": 0.081,
+      "step": 2570
+    },
+    {
+      "epoch": 16.973684210526315,
+      "grad_norm": 6.21569299697876,
+      "learning_rate": 0.0001,
+      "loss": 0.0835,
+      "step": 2580
+    },
+    {
+      "epoch": 17.039473684210527,
+      "grad_norm": 5.315542697906494,
+      "learning_rate": 0.0001,
+      "loss": 0.0771,
+      "step": 2590
+    },
+    {
+      "epoch": 17.105263157894736,
+      "grad_norm": 5.427717685699463,
+      "learning_rate": 0.0001,
+      "loss": 0.0807,
+      "step": 2600
+    },
+    {
+      "epoch": 17.17105263157895,
+      "grad_norm": 4.329200267791748,
+      "learning_rate": 0.0001,
+      "loss": 0.0737,
+      "step": 2610
+    },
+    {
+      "epoch": 17.236842105263158,
+      "grad_norm": 4.520540714263916,
+      "learning_rate": 0.0001,
+      "loss": 0.0826,
+      "step": 2620
+    },
+    {
+      "epoch": 17.30263157894737,
+      "grad_norm": 4.716585636138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0765,
+      "step": 2630
+    },
+    {
+      "epoch": 17.36842105263158,
+      "grad_norm": 5.242930889129639,
+      "learning_rate": 0.0001,
+      "loss": 0.0739,
+      "step": 2640
+    },
+    {
+      "epoch": 17.43421052631579,
+      "grad_norm": 4.814362525939941,
+      "learning_rate": 0.0001,
+      "loss": 0.0766,
+      "step": 2650
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 4.858095645904541,
+      "learning_rate": 0.0001,
+      "loss": 0.0761,
+      "step": 2660
+    },
+    {
+      "epoch": 17.56578947368421,
+      "grad_norm": 4.718443393707275,
+      "learning_rate": 0.0001,
+      "loss": 0.073,
+      "step": 2670
+    },
+    {
+      "epoch": 17.63157894736842,
+      "grad_norm": 4.559201240539551,
+      "learning_rate": 0.0001,
+      "loss": 0.0749,
+      "step": 2680
+    },
+    {
+      "epoch": 17.69736842105263,
+      "grad_norm": 3.9688594341278076,
+      "learning_rate": 0.0001,
+      "loss": 0.0782,
+      "step": 2690
+    },
+    {
+      "epoch": 17.763157894736842,
+      "grad_norm": 4.480958938598633,
+      "learning_rate": 0.0001,
+      "loss": 0.0776,
+      "step": 2700
+    },
+    {
+      "epoch": 17.82894736842105,
+      "grad_norm": 5.901112079620361,
+      "learning_rate": 0.0001,
+      "loss": 0.0756,
+      "step": 2710
+    },
+    {
+      "epoch": 17.894736842105264,
+      "grad_norm": 5.043224811553955,
+      "learning_rate": 0.0001,
+      "loss": 0.0784,
+      "step": 2720
+    },
+    {
+      "epoch": 17.960526315789473,
+      "grad_norm": 4.380178451538086,
+      "learning_rate": 0.0001,
+      "loss": 0.0725,
+      "step": 2730
+    },
+    {
+      "epoch": 18.026315789473685,
+      "grad_norm": 3.8729310035705566,
+      "learning_rate": 0.0001,
+      "loss": 0.0751,
+      "step": 2740
+    },
+    {
+      "epoch": 18.092105263157894,
+      "grad_norm": 4.4962639808654785,
+      "learning_rate": 0.0001,
+      "loss": 0.0727,
+      "step": 2750
+    },
+    {
+      "epoch": 18.157894736842106,
+      "grad_norm": 4.605258941650391,
+      "learning_rate": 0.0001,
+      "loss": 0.0745,
+      "step": 2760
+    },
+    {
+      "epoch": 18.223684210526315,
+      "grad_norm": 4.574273586273193,
+      "learning_rate": 0.0001,
+      "loss": 0.0758,
+      "step": 2770
+    },
+    {
+      "epoch": 18.289473684210527,
+      "grad_norm": 4.572812557220459,
+      "learning_rate": 0.0001,
+      "loss": 0.0755,
+      "step": 2780
+    },
+    {
+      "epoch": 18.355263157894736,
+      "grad_norm": 4.3042378425598145,
+      "learning_rate": 0.0001,
+      "loss": 0.0707,
+      "step": 2790
+    },
+    {
+      "epoch": 18.42105263157895,
+      "grad_norm": 4.969744682312012,
+      "learning_rate": 0.0001,
+      "loss": 0.0726,
+      "step": 2800
+    },
+    {
+      "epoch": 18.486842105263158,
+      "grad_norm": 5.771834850311279,
+      "learning_rate": 0.0001,
+      "loss": 0.0723,
+      "step": 2810
+    },
+    {
+      "epoch": 18.55263157894737,
+      "grad_norm": 4.393187046051025,
+      "learning_rate": 0.0001,
+      "loss": 0.0732,
+      "step": 2820
+    },
+    {
+      "epoch": 18.61842105263158,
+      "grad_norm": 4.077311038970947,
+      "learning_rate": 0.0001,
+      "loss": 0.0703,
+      "step": 2830
+    },
+    {
+      "epoch": 18.68421052631579,
+      "grad_norm": 3.5009090900421143,
+      "learning_rate": 0.0001,
+      "loss": 0.0705,
+      "step": 2840
+    },
+    {
+      "epoch": 18.75,
+      "grad_norm": 4.456014156341553,
+      "learning_rate": 0.0001,
+      "loss": 0.0685,
+      "step": 2850
+    },
+    {
+      "epoch": 18.81578947368421,
+      "grad_norm": 4.8032307624816895,
+      "learning_rate": 0.0001,
+      "loss": 0.0707,
+      "step": 2860
+    },
+    {
+      "epoch": 18.88157894736842,
+      "grad_norm": 4.6024603843688965,
+      "learning_rate": 0.0001,
+      "loss": 0.0754,
+      "step": 2870
+    },
+    {
+      "epoch": 18.94736842105263,
+      "grad_norm": 5.125010013580322,
+      "learning_rate": 0.0001,
+      "loss": 0.0707,
+      "step": 2880
+    },
+    {
+      "epoch": 19.013157894736842,
+      "grad_norm": 4.631536483764648,
+      "learning_rate": 0.0001,
+      "loss": 0.0743,
+      "step": 2890
+    },
+    {
+      "epoch": 19.07894736842105,
+      "grad_norm": 4.642434120178223,
+      "learning_rate": 0.0001,
+      "loss": 0.0679,
+      "step": 2900
+    },
+    {
+      "epoch": 19.144736842105264,
+      "grad_norm": 4.4804911613464355,
+      "learning_rate": 0.0001,
+      "loss": 0.0688,
+      "step": 2910
+    },
+    {
+      "epoch": 19.210526315789473,
+      "grad_norm": 3.7083773612976074,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "step": 2920
+    },
+    {
+      "epoch": 19.276315789473685,
+      "grad_norm": 4.256147384643555,
+      "learning_rate": 0.0001,
+      "loss": 0.0725,
+      "step": 2930
+    },
+    {
+      "epoch": 19.342105263157894,
+      "grad_norm": 4.125244140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0674,
+      "step": 2940
+    },
+    {
+      "epoch": 19.407894736842106,
+      "grad_norm": 4.567220211029053,
+      "learning_rate": 0.0001,
+      "loss": 0.0683,
+      "step": 2950
+    },
+    {
+      "epoch": 19.473684210526315,
+      "grad_norm": 4.298739433288574,
+      "learning_rate": 0.0001,
+      "loss": 0.0674,
+      "step": 2960
+    },
+    {
+      "epoch": 19.539473684210527,
+      "grad_norm": 4.303213596343994,
+      "learning_rate": 0.0001,
+      "loss": 0.0704,
+      "step": 2970
+    },
+    {
+      "epoch": 19.605263157894736,
+      "grad_norm": 4.842376232147217,
+      "learning_rate": 0.0001,
+      "loss": 0.0687,
+      "step": 2980
+    },
+    {
+      "epoch": 19.67105263157895,
+      "grad_norm": 4.336226940155029,
+      "learning_rate": 0.0001,
+      "loss": 0.0693,
+      "step": 2990
+    },
+    {
+      "epoch": 19.736842105263158,
+      "grad_norm": 4.425253868103027,
+      "learning_rate": 0.0001,
+      "loss": 0.0674,
+      "step": 3000
+    },
+    {
+      "epoch": 19.80263157894737,
+      "grad_norm": 4.074360370635986,
+      "learning_rate": 0.0001,
+      "loss": 0.0622,
+      "step": 3010
+    },
+    {
+      "epoch": 19.86842105263158,
+      "grad_norm": 4.191122531890869,
+      "learning_rate": 0.0001,
+      "loss": 0.0692,
+      "step": 3020
+    },
+    {
+      "epoch": 19.93421052631579,
+      "grad_norm": 4.672008514404297,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "step": 3030
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 4.377945423126221,
+      "learning_rate": 0.0001,
+      "loss": 0.0627,
+      "step": 3040
+    },
+    {
+      "epoch": 20.06578947368421,
+      "grad_norm": 4.234347820281982,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 3050
+    },
+    {
+      "epoch": 20.13157894736842,
+      "grad_norm": 4.540591716766357,
+      "learning_rate": 0.0001,
+      "loss": 0.0683,
+      "step": 3060
+    },
+    {
+      "epoch": 20.19736842105263,
+      "grad_norm": 4.472556114196777,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 3070
+    },
+    {
+      "epoch": 20.263157894736842,
+      "grad_norm": 4.038685321807861,
+      "learning_rate": 0.0001,
+      "loss": 0.0646,
+      "step": 3080
+    },
+    {
+      "epoch": 20.32894736842105,
+      "grad_norm": 4.310518741607666,
+      "learning_rate": 0.0001,
+      "loss": 0.066,
+      "step": 3090
+    },
+    {
+      "epoch": 20.394736842105264,
+      "grad_norm": 3.681676149368286,
+      "learning_rate": 0.0001,
+      "loss": 0.0628,
+      "step": 3100
+    },
+    {
+      "epoch": 20.460526315789473,
+      "grad_norm": 4.144742012023926,
+      "learning_rate": 0.0001,
+      "loss": 0.0632,
+      "step": 3110
+    },
+    {
+      "epoch": 20.526315789473685,
+      "grad_norm": 3.517277479171753,
+      "learning_rate": 0.0001,
+      "loss": 0.0645,
+      "step": 3120
+    },
+    {
+      "epoch": 20.592105263157894,
+      "grad_norm": 3.710193395614624,
+      "learning_rate": 0.0001,
+      "loss": 0.0645,
+      "step": 3130
+    },
+    {
+      "epoch": 20.657894736842106,
+      "grad_norm": 4.236420631408691,
+      "learning_rate": 0.0001,
+      "loss": 0.0644,
+      "step": 3140
+    },
+    {
+      "epoch": 20.723684210526315,
+      "grad_norm": 3.9308488368988037,
+      "learning_rate": 0.0001,
+      "loss": 0.0672,
+      "step": 3150
+    },
+    {
+      "epoch": 20.789473684210527,
+      "grad_norm": 4.319930553436279,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 3160
+    },
+    {
+      "epoch": 20.855263157894736,
+      "grad_norm": 4.539423942565918,
+      "learning_rate": 0.0001,
+      "loss": 0.0676,
+      "step": 3170
+    },
+    {
+      "epoch": 20.92105263157895,
+      "grad_norm": 4.0434956550598145,
+      "learning_rate": 0.0001,
+      "loss": 0.0683,
+      "step": 3180
+    },
+    {
+      "epoch": 20.986842105263158,
+      "grad_norm": 3.5958993434906006,
+      "learning_rate": 0.0001,
+      "loss": 0.0699,
+      "step": 3190
+    },
+    {
+      "epoch": 21.05263157894737,
+      "grad_norm": 3.827382802963257,
+      "learning_rate": 0.0001,
+      "loss": 0.0679,
+      "step": 3200
+    },
+    {
+      "epoch": 21.11842105263158,
+      "grad_norm": 4.15602445602417,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "step": 3210
+    },
+    {
+      "epoch": 21.18421052631579,
+      "grad_norm": 4.118391036987305,
+      "learning_rate": 0.0001,
+      "loss": 0.0676,
+      "step": 3220
+    },
+    {
+      "epoch": 21.25,
+      "grad_norm": 4.27808141708374,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 3230
+    },
+    {
+      "epoch": 21.31578947368421,
+      "grad_norm": 4.284823894500732,
+      "learning_rate": 0.0001,
+      "loss": 0.0621,
+      "step": 3240
+    },
+    {
+      "epoch": 21.38157894736842,
+      "grad_norm": 3.516188621520996,
+      "learning_rate": 0.0001,
+      "loss": 0.0621,
+      "step": 3250
+    },
+    {
+      "epoch": 21.44736842105263,
+      "grad_norm": 3.9423298835754395,
+      "learning_rate": 0.0001,
+      "loss": 0.0633,
+      "step": 3260
+    },
+    {
+      "epoch": 21.513157894736842,
+      "grad_norm": 4.444387435913086,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 3270
+    },
+    {
+      "epoch": 21.57894736842105,
+      "grad_norm": 4.271636486053467,
+      "learning_rate": 0.0001,
+      "loss": 0.0638,
+      "step": 3280
+    },
+    {
+      "epoch": 21.644736842105264,
+      "grad_norm": 4.359874725341797,
+      "learning_rate": 0.0001,
+      "loss": 0.0678,
+      "step": 3290
+    },
+    {
+      "epoch": 21.710526315789473,
+      "grad_norm": 4.128505229949951,
+      "learning_rate": 0.0001,
+      "loss": 0.0632,
+      "step": 3300
+    },
+    {
+      "epoch": 21.776315789473685,
+      "grad_norm": 3.6902530193328857,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 3310
+    },
+    {
+      "epoch": 21.842105263157894,
+      "grad_norm": 4.0358805656433105,
+      "learning_rate": 0.0001,
+      "loss": 0.059,
+      "step": 3320
+    },
+    {
+      "epoch": 21.907894736842106,
+      "grad_norm": 3.965226173400879,
+      "learning_rate": 0.0001,
+      "loss": 0.062,
+      "step": 3330
+    },
+    {
+      "epoch": 21.973684210526315,
+      "grad_norm": 3.338334083557129,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 3340
+    },
+    {
+      "epoch": 22.039473684210527,
+      "grad_norm": 4.442704200744629,
+      "learning_rate": 0.0001,
+      "loss": 0.0591,
+      "step": 3350
+    },
+    {
+      "epoch": 22.105263157894736,
+      "grad_norm": 4.312819480895996,
+      "learning_rate": 0.0001,
+      "loss": 0.057,
+      "step": 3360
+    },
+    {
+      "epoch": 22.17105263157895,
+      "grad_norm": 3.8974449634552,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 3370
+    },
+    {
+      "epoch": 22.236842105263158,
+      "grad_norm": 3.41093111038208,
+      "learning_rate": 0.0001,
+      "loss": 0.0557,
+      "step": 3380
+    },
+    {
+      "epoch": 22.30263157894737,
+      "grad_norm": 3.5688576698303223,
+      "learning_rate": 0.0001,
+      "loss": 0.0561,
+      "step": 3390
+    },
+    {
+      "epoch": 22.36842105263158,
+      "grad_norm": 3.9463613033294678,
+      "learning_rate": 0.0001,
+      "loss": 0.0595,
+      "step": 3400
+    },
+    {
+      "epoch": 22.43421052631579,
+      "grad_norm": 4.227533340454102,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 3410
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 3.651512861251831,
+      "learning_rate": 0.0001,
+      "loss": 0.0568,
+      "step": 3420
+    },
+    {
+      "epoch": 22.56578947368421,
+      "grad_norm": 3.8400089740753174,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 3430
+    },
+    {
+      "epoch": 22.63157894736842,
+      "grad_norm": 3.567005157470703,
+      "learning_rate": 0.0001,
+      "loss": 0.0592,
+      "step": 3440
+    },
+    {
+      "epoch": 22.69736842105263,
+      "grad_norm": 3.756096363067627,
+      "learning_rate": 0.0001,
+      "loss": 0.0565,
+      "step": 3450
+    },
+    {
+      "epoch": 22.763157894736842,
+      "grad_norm": 3.8833200931549072,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 3460
+    },
+    {
+      "epoch": 22.82894736842105,
+      "grad_norm": 3.5916409492492676,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 3470
+    },
+    {
+      "epoch": 22.894736842105264,
+      "grad_norm": 3.6396472454071045,
+      "learning_rate": 0.0001,
+      "loss": 0.0621,
+      "step": 3480
+    },
+    {
+      "epoch": 22.960526315789473,
+      "grad_norm": 4.026757717132568,
+      "learning_rate": 0.0001,
+      "loss": 0.0609,
+      "step": 3490
+    },
+    {
+      "epoch": 23.026315789473685,
+      "grad_norm": 3.412045955657959,
+      "learning_rate": 0.0001,
+      "loss": 0.0635,
+      "step": 3500
+    },
+    {
+      "epoch": 23.092105263157894,
+      "grad_norm": 3.8278250694274902,
+      "learning_rate": 0.0001,
+      "loss": 0.0607,
+      "step": 3510
+    },
+    {
+      "epoch": 23.157894736842106,
+      "grad_norm": 4.461212635040283,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 3520
+    },
+    {
+      "epoch": 23.223684210526315,
+      "grad_norm": 3.1517691612243652,
+      "learning_rate": 0.0001,
+      "loss": 0.0564,
+      "step": 3530
+    },
+    {
+      "epoch": 23.289473684210527,
+      "grad_norm": 3.9116435050964355,
+      "learning_rate": 0.0001,
+      "loss": 0.0614,
+      "step": 3540
+    },
+    {
+      "epoch": 23.355263157894736,
+      "grad_norm": 3.633558511734009,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 3550
+    },
+    {
+      "epoch": 23.42105263157895,
+      "grad_norm": 3.1927671432495117,
+      "learning_rate": 0.0001,
+      "loss": 0.0592,
+      "step": 3560
+    },
+    {
+      "epoch": 23.486842105263158,
+      "grad_norm": 3.9745633602142334,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 3570
+    },
+    {
+      "epoch": 23.55263157894737,
+      "grad_norm": 4.017717361450195,
+      "learning_rate": 0.0001,
+      "loss": 0.0599,
+      "step": 3580
+    },
+    {
+      "epoch": 23.61842105263158,
+      "grad_norm": 3.3736987113952637,
+      "learning_rate": 0.0001,
+      "loss": 0.0561,
+      "step": 3590
+    },
+    {
+      "epoch": 23.68421052631579,
+      "grad_norm": 3.3666791915893555,
+      "learning_rate": 0.0001,
+      "loss": 0.0589,
+      "step": 3600
+    },
+    {
+      "epoch": 23.75,
+      "grad_norm": 3.2723591327667236,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 3610
+    },
+    {
+      "epoch": 23.81578947368421,
+      "grad_norm": 3.429962396621704,
+      "learning_rate": 0.0001,
+      "loss": 0.0572,
+      "step": 3620
+    },
+    {
+      "epoch": 23.88157894736842,
+      "grad_norm": 4.048157691955566,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "step": 3630
+    },
+    {
+      "epoch": 23.94736842105263,
+      "grad_norm": 4.095017910003662,
+      "learning_rate": 0.0001,
+      "loss": 0.0532,
+      "step": 3640
+    },
+    {
+      "epoch": 24.013157894736842,
+      "grad_norm": 3.7212188243865967,
+      "learning_rate": 0.0001,
+      "loss": 0.0527,
+      "step": 3650
+    },
+    {
+      "epoch": 24.07894736842105,
+      "grad_norm": 3.424227237701416,
+      "learning_rate": 0.0001,
+      "loss": 0.0525,
+      "step": 3660
+    },
+    {
+      "epoch": 24.144736842105264,
+      "grad_norm": 3.869912624359131,
+      "learning_rate": 0.0001,
+      "loss": 0.0529,
+      "step": 3670
+    },
+    {
+      "epoch": 24.210526315789473,
+      "grad_norm": 3.593370199203491,
+      "learning_rate": 0.0001,
+      "loss": 0.0547,
+      "step": 3680
+    },
+    {
+      "epoch": 24.276315789473685,
+      "grad_norm": 3.430244207382202,
+      "learning_rate": 0.0001,
+      "loss": 0.0546,
+      "step": 3690
+    },
+    {
+      "epoch": 24.342105263157894,
+      "grad_norm": 3.3279993534088135,
+      "learning_rate": 0.0001,
+      "loss": 0.0561,
+      "step": 3700
+    },
+    {
+      "epoch": 24.407894736842106,
+      "grad_norm": 3.5217323303222656,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 3710
+    },
+    {
+      "epoch": 24.473684210526315,
+      "grad_norm": 3.2047455310821533,
+      "learning_rate": 0.0001,
+      "loss": 0.0528,
+      "step": 3720
+    },
+    {
+      "epoch": 24.539473684210527,
+      "grad_norm": 3.5667405128479004,
+      "learning_rate": 0.0001,
+      "loss": 0.0565,
+      "step": 3730
+    },
+    {
+      "epoch": 24.605263157894736,
+      "grad_norm": 3.1612064838409424,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 3740
+    },
+    {
+      "epoch": 24.67105263157895,
+      "grad_norm": 3.379948616027832,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 3750
+    },
+    {
+      "epoch": 24.736842105263158,
+      "grad_norm": 3.004784107208252,
+      "learning_rate": 0.0001,
+      "loss": 0.0534,
+      "step": 3760
+    },
+    {
+      "epoch": 24.80263157894737,
+      "grad_norm": 3.648918867111206,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 3770
+    },
+    {
+      "epoch": 24.86842105263158,
+      "grad_norm": 3.0311264991760254,
+      "learning_rate": 0.0001,
+      "loss": 0.06,
+      "step": 3780
+    },
+    {
+      "epoch": 24.93421052631579,
+      "grad_norm": 3.015455722808838,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "step": 3790
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 3.084538698196411,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 3800
+    },
+    {
+      "epoch": 25.06578947368421,
+      "grad_norm": 3.5729260444641113,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "step": 3810
+    },
+    {
+      "epoch": 25.13157894736842,
+      "grad_norm": 3.5942578315734863,
+      "learning_rate": 0.0001,
+      "loss": 0.0569,
+      "step": 3820
+    },
+    {
+      "epoch": 25.19736842105263,
+      "grad_norm": 3.8373870849609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0578,
+      "step": 3830
+    },
+    {
+      "epoch": 25.263157894736842,
+      "grad_norm": 3.1174817085266113,
+      "learning_rate": 0.0001,
+      "loss": 0.0539,
+      "step": 3840
+    },
+    {
+      "epoch": 25.32894736842105,
+      "grad_norm": 2.9916906356811523,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 3850
+    },
+    {
+      "epoch": 25.394736842105264,
+      "grad_norm": 3.7236101627349854,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 3860
+    },
+    {
+      "epoch": 25.460526315789473,
+      "grad_norm": 3.196599006652832,
+      "learning_rate": 0.0001,
+      "loss": 0.0613,
+      "step": 3870
+    },
+    {
+      "epoch": 25.526315789473685,
+      "grad_norm": 3.3245861530303955,
+      "learning_rate": 0.0001,
+      "loss": 0.0595,
+      "step": 3880
+    },
+    {
+      "epoch": 25.592105263157894,
+      "grad_norm": 3.1229844093322754,
+      "learning_rate": 0.0001,
+      "loss": 0.0581,
+      "step": 3890
+    },
+    {
+      "epoch": 25.657894736842106,
+      "grad_norm": 3.090834617614746,
+      "learning_rate": 0.0001,
+      "loss": 0.057,
+      "step": 3900
+    },
+    {
+      "epoch": 25.723684210526315,
+      "grad_norm": 3.7173891067504883,
+      "learning_rate": 0.0001,
+      "loss": 0.0553,
+      "step": 3910
+    },
+    {
+      "epoch": 25.789473684210527,
+      "grad_norm": 3.4529900550842285,
+      "learning_rate": 0.0001,
+      "loss": 0.06,
+      "step": 3920
+    },
+    {
+      "epoch": 25.855263157894736,
+      "grad_norm": 3.41372013092041,
+      "learning_rate": 0.0001,
+      "loss": 0.0534,
+      "step": 3930
+    },
+    {
+      "epoch": 25.92105263157895,
+      "grad_norm": 3.5578503608703613,
+      "learning_rate": 0.0001,
+      "loss": 0.0579,
+      "step": 3940
+    },
+    {
+      "epoch": 25.986842105263158,
+      "grad_norm": 2.5233664512634277,
+      "learning_rate": 0.0001,
+      "loss": 0.0571,
+      "step": 3950
+    },
+    {
+      "epoch": 26.05263157894737,
+      "grad_norm": 3.310196876525879,
+      "learning_rate": 0.0001,
+      "loss": 0.0505,
+      "step": 3960
+    },
+    {
+      "epoch": 26.11842105263158,
+      "grad_norm": 3.0970399379730225,
+      "learning_rate": 0.0001,
+      "loss": 0.0532,
+      "step": 3970
+    },
+    {
+      "epoch": 26.18421052631579,
+      "grad_norm": 2.9945688247680664,
+      "learning_rate": 0.0001,
+      "loss": 0.05,
+      "step": 3980
+    },
+    {
+      "epoch": 26.25,
+      "grad_norm": 2.966948986053467,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "step": 3990
+    },
+    {
+      "epoch": 26.31578947368421,
+      "grad_norm": 3.1473228931427,
+      "learning_rate": 0.0001,
+      "loss": 0.0548,
+      "step": 4000
+    },
+    {
+      "epoch": 26.38157894736842,
+      "grad_norm": 3.781965494155884,
+      "learning_rate": 0.0001,
+      "loss": 0.0513,
+      "step": 4010
+    },
+    {
+      "epoch": 26.44736842105263,
+      "grad_norm": 3.1109609603881836,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 4020
+    },
+    {
+      "epoch": 26.513157894736842,
+      "grad_norm": 3.153053045272827,
+      "learning_rate": 0.0001,
+      "loss": 0.0601,
+      "step": 4030
+    },
+    {
+      "epoch": 26.57894736842105,
+      "grad_norm": 2.897550582885742,
+      "learning_rate": 0.0001,
+      "loss": 0.0563,
+      "step": 4040
+    },
+    {
+      "epoch": 26.644736842105264,
+      "grad_norm": 3.540456771850586,
+      "learning_rate": 0.0001,
+      "loss": 0.0529,
+      "step": 4050
+    },
+    {
+      "epoch": 26.710526315789473,
+      "grad_norm": 3.526008129119873,
+      "learning_rate": 0.0001,
+      "loss": 0.0529,
+      "step": 4060
+    },
+    {
+      "epoch": 26.776315789473685,
+      "grad_norm": 2.84975004196167,
+      "learning_rate": 0.0001,
+      "loss": 0.0546,
+      "step": 4070
+    },
+    {
+      "epoch": 26.842105263157894,
+      "grad_norm": 2.8743505477905273,
+      "learning_rate": 0.0001,
+      "loss": 0.0516,
+      "step": 4080
+    },
+    {
+      "epoch": 26.907894736842106,
+      "grad_norm": 2.8977339267730713,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 4090
+    },
+    {
+      "epoch": 26.973684210526315,
+      "grad_norm": 2.9389920234680176,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 4100
+    },
+    {
+      "epoch": 27.039473684210527,
+      "grad_norm": 3.0356078147888184,
+      "learning_rate": 0.0001,
+      "loss": 0.0549,
+      "step": 4110
+    },
+    {
+      "epoch": 27.105263157894736,
+      "grad_norm": 2.933394193649292,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 4120
+    },
+    {
+      "epoch": 27.17105263157895,
+      "grad_norm": 3.342971086502075,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 4130
+    },
+    {
+      "epoch": 27.236842105263158,
+      "grad_norm": 3.144812822341919,
+      "learning_rate": 0.0001,
+      "loss": 0.0526,
+      "step": 4140
+    },
+    {
+      "epoch": 27.30263157894737,
+      "grad_norm": 2.595964193344116,
+      "learning_rate": 0.0001,
+      "loss": 0.0513,
+      "step": 4150
+    },
+    {
+      "epoch": 27.36842105263158,
+      "grad_norm": 3.5423874855041504,
+      "learning_rate": 0.0001,
+      "loss": 0.0504,
+      "step": 4160
+    },
+    {
+      "epoch": 27.43421052631579,
+      "grad_norm": 3.1405887603759766,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 4170
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 3.289807081222534,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 4180
+    },
+    {
+      "epoch": 27.56578947368421,
+      "grad_norm": 3.1032474040985107,
+      "learning_rate": 0.0001,
+      "loss": 0.0528,
+      "step": 4190
+    },
+    {
+      "epoch": 27.63157894736842,
+      "grad_norm": 2.8093903064727783,
+      "learning_rate": 0.0001,
+      "loss": 0.0496,
+      "step": 4200
+    },
+    {
+      "epoch": 27.69736842105263,
+      "grad_norm": 2.896862268447876,
+      "learning_rate": 0.0001,
+      "loss": 0.0486,
+      "step": 4210
+    },
+    {
+      "epoch": 27.763157894736842,
+      "grad_norm": 3.2978949546813965,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 4220
+    },
+    {
+      "epoch": 27.82894736842105,
+      "grad_norm": 3.006734609603882,
+      "learning_rate": 0.0001,
+      "loss": 0.0477,
+      "step": 4230
+    },
+    {
+      "epoch": 27.894736842105264,
+      "grad_norm": 3.1355373859405518,
+      "learning_rate": 0.0001,
+      "loss": 0.0491,
+      "step": 4240
+    },
+    {
+      "epoch": 27.960526315789473,
+      "grad_norm": 2.898782253265381,
+      "learning_rate": 0.0001,
+      "loss": 0.0476,
+      "step": 4250
+    },
+    {
+      "epoch": 28.026315789473685,
+      "grad_norm": 2.8229269981384277,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 4260
+    },
+    {
+      "epoch": 28.092105263157894,
+      "grad_norm": 2.915052890777588,
+      "learning_rate": 0.0001,
+      "loss": 0.0515,
+      "step": 4270
+    },
+    {
+      "epoch": 28.157894736842106,
+      "grad_norm": 3.278546094894409,
+      "learning_rate": 0.0001,
+      "loss": 0.0483,
+      "step": 4280
+    },
+    {
+      "epoch": 28.223684210526315,
+      "grad_norm": 3.0418665409088135,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "step": 4290
+    },
+    {
+      "epoch": 28.289473684210527,
+      "grad_norm": 3.2005770206451416,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "step": 4300
+    },
+    {
+      "epoch": 28.355263157894736,
+      "grad_norm": 3.2774760723114014,
+      "learning_rate": 0.0001,
+      "loss": 0.0503,
+      "step": 4310
+    },
+    {
+      "epoch": 28.42105263157895,
+      "grad_norm": 3.155532121658325,
+      "learning_rate": 0.0001,
+      "loss": 0.0469,
+      "step": 4320
+    },
+    {
+      "epoch": 28.486842105263158,
+      "grad_norm": 3.1832993030548096,
+      "learning_rate": 0.0001,
+      "loss": 0.0483,
+      "step": 4330
+    },
+    {
+      "epoch": 28.55263157894737,
+      "grad_norm": 3.1676037311553955,
+      "learning_rate": 0.0001,
+      "loss": 0.0497,
+      "step": 4340
+    },
+    {
+      "epoch": 28.61842105263158,
+      "grad_norm": 3.4129903316497803,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 4350
+    },
+    {
+      "epoch": 28.68421052631579,
+      "grad_norm": 3.1655561923980713,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 4360
+    },
+    {
+      "epoch": 28.75,
+      "grad_norm": 2.708021640777588,
+      "learning_rate": 0.0001,
+      "loss": 0.0466,
+      "step": 4370
+    },
+    {
+      "epoch": 28.81578947368421,
+      "grad_norm": 2.68147873878479,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 4380
+    },
+    {
+      "epoch": 28.88157894736842,
+      "grad_norm": 3.428018808364868,
+      "learning_rate": 0.0001,
+      "loss": 0.0489,
+      "step": 4390
+    },
+    {
+      "epoch": 28.94736842105263,
+      "grad_norm": 2.94804048538208,
+      "learning_rate": 0.0001,
+      "loss": 0.0513,
+      "step": 4400
+    },
+    {
+      "epoch": 29.013157894736842,
+      "grad_norm": 2.679884910583496,
+      "learning_rate": 0.0001,
+      "loss": 0.0524,
+      "step": 4410
+    },
+    {
+      "epoch": 29.07894736842105,
+      "grad_norm": 2.392934560775757,
+      "learning_rate": 0.0001,
+      "loss": 0.0483,
+      "step": 4420
+    },
+    {
+      "epoch": 29.144736842105264,
+      "grad_norm": 2.7461166381835938,
+      "learning_rate": 0.0001,
+      "loss": 0.0474,
+      "step": 4430
+    },
+    {
+      "epoch": 29.210526315789473,
+      "grad_norm": 2.788827896118164,
+      "learning_rate": 0.0001,
+      "loss": 0.0489,
+      "step": 4440
+    },
+    {
+      "epoch": 29.276315789473685,
+      "grad_norm": 3.0062499046325684,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 4450
+    },
+    {
+      "epoch": 29.342105263157894,
+      "grad_norm": 3.0573859214782715,
+      "learning_rate": 0.0001,
+      "loss": 0.05,
+      "step": 4460
+    },
+    {
+      "epoch": 29.407894736842106,
+      "grad_norm": 2.665456533432007,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "step": 4470
+    },
+    {
+      "epoch": 29.473684210526315,
+      "grad_norm": 2.522658586502075,
+      "learning_rate": 0.0001,
+      "loss": 0.0467,
+      "step": 4480
+    },
+    {
+      "epoch": 29.539473684210527,
+      "grad_norm": 2.600645065307617,
+      "learning_rate": 0.0001,
+      "loss": 0.0489,
+      "step": 4490
+    },
+    {
+      "epoch": 29.605263157894736,
+      "grad_norm": 2.6609113216400146,
+      "learning_rate": 0.0001,
+      "loss": 0.0476,
+      "step": 4500
+    },
+    {
+      "epoch": 29.67105263157895,
+      "grad_norm": 2.691256284713745,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 4510
+    },
+    {
+      "epoch": 29.736842105263158,
+      "grad_norm": 2.8132810592651367,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 4520
+    },
+    {
+      "epoch": 29.80263157894737,
+      "grad_norm": 2.914437770843506,
+      "learning_rate": 0.0001,
+      "loss": 0.0486,
+      "step": 4530
+    },
+    {
+      "epoch": 29.86842105263158,
+      "grad_norm": 2.497580051422119,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 4540
+    },
+    {
+      "epoch": 29.93421052631579,
+      "grad_norm": 2.6248748302459717,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 4550
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 2.8012468814849854,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 4560
+    },
+    {
+      "epoch": 30.06578947368421,
+      "grad_norm": 2.505524158477783,
+      "learning_rate": 0.0001,
+      "loss": 0.047,
+      "step": 4570
+    },
+    {
+      "epoch": 30.13157894736842,
+      "grad_norm": 2.6117448806762695,
+      "learning_rate": 0.0001,
+      "loss": 0.0503,
+      "step": 4580
+    },
+    {
+      "epoch": 30.19736842105263,
+      "grad_norm": 2.7365365028381348,
+      "learning_rate": 0.0001,
+      "loss": 0.0424,
+      "step": 4590
+    },
+    {
+      "epoch": 30.263157894736842,
+      "grad_norm": 2.8292315006256104,
+      "learning_rate": 0.0001,
+      "loss": 0.0467,
+      "step": 4600
+    },
+    {
+      "epoch": 30.32894736842105,
+      "grad_norm": 2.5775272846221924,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 4610
+    },
+    {
+      "epoch": 30.394736842105264,
+      "grad_norm": 2.720210313796997,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 4620
+    },
+    {
+      "epoch": 30.460526315789473,
+      "grad_norm": 2.5070748329162598,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 4630
+    },
+    {
+      "epoch": 30.526315789473685,
+      "grad_norm": 2.8825769424438477,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 4640
+    },
+    {
+      "epoch": 30.592105263157894,
+      "grad_norm": 2.553520441055298,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 4650
+    },
+    {
+      "epoch": 30.657894736842106,
+      "grad_norm": 2.818341016769409,
+      "learning_rate": 0.0001,
+      "loss": 0.0488,
+      "step": 4660
+    },
+    {
+      "epoch": 30.723684210526315,
+      "grad_norm": 3.132817268371582,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 4670
+    },
+    {
+      "epoch": 30.789473684210527,
+      "grad_norm": 2.931819200515747,
+      "learning_rate": 0.0001,
+      "loss": 0.0488,
+      "step": 4680
+    },
+    {
+      "epoch": 30.855263157894736,
+      "grad_norm": 2.6675987243652344,
+      "learning_rate": 0.0001,
+      "loss": 0.0449,
+      "step": 4690
+    },
+    {
+      "epoch": 30.92105263157895,
+      "grad_norm": 3.1334962844848633,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 4700
+    },
+    {
+      "epoch": 30.986842105263158,
+      "grad_norm": 2.9134418964385986,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 4710
+    },
+    {
+      "epoch": 31.05263157894737,
+      "grad_norm": 2.820773124694824,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "step": 4720
+    },
+    {
+      "epoch": 31.11842105263158,
+      "grad_norm": 2.4287455081939697,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 4730
+    },
+    {
+      "epoch": 31.18421052631579,
+      "grad_norm": 2.476181745529175,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 4740
+    },
+    {
+      "epoch": 31.25,
+      "grad_norm": 2.6724979877471924,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 4750
+    },
+    {
+      "epoch": 31.31578947368421,
+      "grad_norm": 2.832930326461792,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 4760
+    },
+    {
+      "epoch": 31.38157894736842,
+      "grad_norm": 2.3849406242370605,
+      "learning_rate": 0.0001,
+      "loss": 0.0481,
+      "step": 4770
+    },
+    {
+      "epoch": 31.44736842105263,
+      "grad_norm": 2.4562463760375977,
+      "learning_rate": 0.0001,
+      "loss": 0.0444,
+      "step": 4780
+    },
+    {
+      "epoch": 31.513157894736842,
+      "grad_norm": 2.278359889984131,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 4790
+    },
+    {
+      "epoch": 31.57894736842105,
+      "grad_norm": 2.701538562774658,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 4800
+    },
+    {
+      "epoch": 31.644736842105264,
+      "grad_norm": 3.181910276412964,
+      "learning_rate": 0.0001,
+      "loss": 0.0451,
+      "step": 4810
+    },
+    {
+      "epoch": 31.710526315789473,
+      "grad_norm": 2.4333291053771973,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 4820
+    },
+    {
+      "epoch": 31.776315789473685,
+      "grad_norm": 2.6533899307250977,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 4830
+    },
+    {
+      "epoch": 31.842105263157894,
+      "grad_norm": 2.8121790885925293,
+      "learning_rate": 0.0001,
+      "loss": 0.0476,
+      "step": 4840
+    },
+    {
+      "epoch": 31.907894736842106,
+      "grad_norm": 2.6064651012420654,
+      "learning_rate": 0.0001,
+      "loss": 0.0479,
+      "step": 4850
+    },
+    {
+      "epoch": 31.973684210526315,
+      "grad_norm": 2.4942471981048584,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 4860
+    },
+    {
+      "epoch": 32.03947368421053,
+      "grad_norm": 2.7892401218414307,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 4870
+    },
+    {
+      "epoch": 32.10526315789474,
+      "grad_norm": 2.6274139881134033,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 4880
+    },
+    {
+      "epoch": 32.171052631578945,
+      "grad_norm": 2.729278802871704,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "step": 4890
+    },
+    {
+      "epoch": 32.23684210526316,
+      "grad_norm": 2.4070258140563965,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 4900
+    },
+    {
+      "epoch": 32.30263157894737,
+      "grad_norm": 2.6891207695007324,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 4910
+    },
+    {
+      "epoch": 32.36842105263158,
+      "grad_norm": 2.2059195041656494,
+      "learning_rate": 0.0001,
+      "loss": 0.0466,
+      "step": 4920
+    },
+    {
+      "epoch": 32.43421052631579,
+      "grad_norm": 2.6894748210906982,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 4930
+    },
+    {
+      "epoch": 32.5,
+      "grad_norm": 2.526822805404663,
+      "learning_rate": 0.0001,
+      "loss": 0.0422,
+      "step": 4940
+    },
+    {
+      "epoch": 32.56578947368421,
+      "grad_norm": 2.509107828140259,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 4950
+    },
+    {
+      "epoch": 32.63157894736842,
+      "grad_norm": 2.662522792816162,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 4960
+    },
+    {
+      "epoch": 32.69736842105263,
+      "grad_norm": 2.4000043869018555,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 4970
+    },
+    {
+      "epoch": 32.76315789473684,
+      "grad_norm": 2.2575981616973877,
+      "learning_rate": 0.0001,
+      "loss": 0.0427,
+      "step": 4980
+    },
+    {
+      "epoch": 32.828947368421055,
+      "grad_norm": 2.4728028774261475,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "step": 4990
+    },
+    {
+      "epoch": 32.89473684210526,
+      "grad_norm": 2.4107658863067627,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 5000
+    },
+    {
+      "epoch": 32.96052631578947,
+      "grad_norm": 2.1474099159240723,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 5010
+    },
+    {
+      "epoch": 33.026315789473685,
+      "grad_norm": 2.5930263996124268,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 5020
+    },
+    {
+      "epoch": 33.0921052631579,
+      "grad_norm": 2.652590036392212,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "step": 5030
+    },
+    {
+      "epoch": 33.1578947368421,
+      "grad_norm": 2.6589486598968506,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 5040
+    },
+    {
+      "epoch": 33.223684210526315,
+      "grad_norm": 2.6637887954711914,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 5050
+    },
+    {
+      "epoch": 33.28947368421053,
+      "grad_norm": 2.2335708141326904,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 5060
+    },
+    {
+      "epoch": 33.35526315789474,
+      "grad_norm": 2.3757381439208984,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 5070
+    },
+    {
+      "epoch": 33.421052631578945,
+      "grad_norm": 2.569889545440674,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 5080
+    },
+    {
+      "epoch": 33.48684210526316,
+      "grad_norm": 2.7300987243652344,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 5090
+    },
+    {
+      "epoch": 33.55263157894737,
+      "grad_norm": 2.591104745864868,
+      "learning_rate": 0.0001,
+      "loss": 0.0446,
+      "step": 5100
+    },
+    {
+      "epoch": 33.61842105263158,
+      "grad_norm": 2.4056169986724854,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 5110
+    },
+    {
+      "epoch": 33.68421052631579,
+      "grad_norm": 2.4610161781311035,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 5120
+    },
+    {
+      "epoch": 33.75,
+      "grad_norm": 2.3633317947387695,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 5130
+    },
+    {
+      "epoch": 33.81578947368421,
+      "grad_norm": 2.227242946624756,
+      "learning_rate": 0.0001,
+      "loss": 0.0451,
+      "step": 5140
+    },
+    {
+      "epoch": 33.88157894736842,
+      "grad_norm": 2.552421808242798,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 5150
+    },
+    {
+      "epoch": 33.94736842105263,
+      "grad_norm": 2.2416305541992188,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 5160
+    },
+    {
+      "epoch": 34.01315789473684,
+      "grad_norm": 2.6720187664031982,
+      "learning_rate": 0.0001,
+      "loss": 0.0449,
+      "step": 5170
+    },
+    {
+      "epoch": 34.078947368421055,
+      "grad_norm": 2.6561310291290283,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 5180
+    },
+    {
+      "epoch": 34.14473684210526,
+      "grad_norm": 2.019381523132324,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 5190
+    },
+    {
+      "epoch": 34.21052631578947,
+      "grad_norm": 2.435896873474121,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 5200
+    },
+    {
+      "epoch": 34.276315789473685,
+      "grad_norm": 2.5385663509368896,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 5210
+    },
+    {
+      "epoch": 34.3421052631579,
+      "grad_norm": 2.449340343475342,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 5220
+    },
+    {
+      "epoch": 34.4078947368421,
+      "grad_norm": 2.6000161170959473,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 5230
+    },
+    {
+      "epoch": 34.473684210526315,
+      "grad_norm": 2.6699042320251465,
+      "learning_rate": 0.0001,
+      "loss": 0.0446,
+      "step": 5240
+    },
+    {
+      "epoch": 34.53947368421053,
+      "grad_norm": 2.4187822341918945,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 5250
+    },
+    {
+      "epoch": 34.60526315789474,
+      "grad_norm": 2.450216770172119,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 5260
+    },
+    {
+      "epoch": 34.671052631578945,
+      "grad_norm": 2.4974207878112793,
+      "learning_rate": 0.0001,
+      "loss": 0.042,
+      "step": 5270
+    },
+    {
+      "epoch": 34.73684210526316,
+      "grad_norm": 2.525740623474121,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 5280
+    },
+    {
+      "epoch": 34.80263157894737,
+      "grad_norm": 2.640798568725586,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 5290
+    },
+    {
+      "epoch": 34.86842105263158,
+      "grad_norm": 2.456847667694092,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 5300
+    },
+    {
+      "epoch": 34.93421052631579,
+      "grad_norm": 2.523120164871216,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 5310
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 1.9672609567642212,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 5320
+    },
+    {
+      "epoch": 35.06578947368421,
+      "grad_norm": 2.4729208946228027,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 5330
+    },
+    {
+      "epoch": 35.13157894736842,
+      "grad_norm": 2.3115689754486084,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 5340
+    },
+    {
+      "epoch": 35.19736842105263,
+      "grad_norm": 2.2606678009033203,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 5350
+    },
+    {
+      "epoch": 35.26315789473684,
+      "grad_norm": 2.3886468410491943,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 5360
+    },
+    {
+      "epoch": 35.328947368421055,
+      "grad_norm": 2.1863503456115723,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 5370
+    },
+    {
+      "epoch": 35.39473684210526,
+      "grad_norm": 2.3488121032714844,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 5380
+    },
+    {
+      "epoch": 35.46052631578947,
+      "grad_norm": 2.360938310623169,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 5390
+    },
+    {
+      "epoch": 35.526315789473685,
+      "grad_norm": 2.5234484672546387,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 5400
+    },
+    {
+      "epoch": 35.5921052631579,
+      "grad_norm": 2.612522602081299,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 5410
+    },
+    {
+      "epoch": 35.6578947368421,
+      "grad_norm": 2.419776678085327,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 5420
+    },
+    {
+      "epoch": 35.723684210526315,
+      "grad_norm": 2.3333466053009033,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 5430
+    },
+    {
+      "epoch": 35.78947368421053,
+      "grad_norm": 2.3613359928131104,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 5440
+    },
+    {
+      "epoch": 35.85526315789474,
+      "grad_norm": 1.962768316268921,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 5450
+    },
+    {
+      "epoch": 35.921052631578945,
+      "grad_norm": 2.3597044944763184,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 5460
+    },
+    {
+      "epoch": 35.98684210526316,
+      "grad_norm": 2.2614336013793945,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 5470
+    },
+    {
+      "epoch": 36.05263157894737,
+      "grad_norm": 2.350679397583008,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 5480
+    },
+    {
+      "epoch": 36.11842105263158,
+      "grad_norm": 2.374591588973999,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 5490
+    },
+    {
+      "epoch": 36.18421052631579,
+      "grad_norm": 2.3889739513397217,
+      "learning_rate": 0.0001,
+      "loss": 0.0422,
+      "step": 5500
+    },
+    {
+      "epoch": 36.25,
+      "grad_norm": 2.4520390033721924,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 5510
+    },
+    {
+      "epoch": 36.31578947368421,
+      "grad_norm": 2.349119186401367,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 5520
+    },
+    {
+      "epoch": 36.38157894736842,
+      "grad_norm": 2.202120780944824,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 5530
+    },
+    {
+      "epoch": 36.44736842105263,
+      "grad_norm": 2.2714016437530518,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 5540
+    },
+    {
+      "epoch": 36.51315789473684,
+      "grad_norm": 2.2632577419281006,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 5550
+    },
+    {
+      "epoch": 36.578947368421055,
+      "grad_norm": 2.608222007751465,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 5560
+    },
+    {
+      "epoch": 36.64473684210526,
+      "grad_norm": 2.298335552215576,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 5570
+    },
+    {
+      "epoch": 36.71052631578947,
+      "grad_norm": 2.498187303543091,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 5580
+    },
+    {
+      "epoch": 36.776315789473685,
+      "grad_norm": 2.62980580329895,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 5590
+    },
+    {
+      "epoch": 36.8421052631579,
+      "grad_norm": 2.6925511360168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 5600
+    },
+    {
+      "epoch": 36.9078947368421,
+      "grad_norm": 2.493591070175171,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 5610
+    },
+    {
+      "epoch": 36.973684210526315,
+      "grad_norm": 2.0936026573181152,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "step": 5620
+    },
+    {
+      "epoch": 37.03947368421053,
+      "grad_norm": 2.3660683631896973,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 5630
+    },
+    {
+      "epoch": 37.10526315789474,
+      "grad_norm": 2.5689046382904053,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 5640
+    },
+    {
+      "epoch": 37.171052631578945,
+      "grad_norm": 2.339864730834961,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 5650
+    },
+    {
+      "epoch": 37.23684210526316,
+      "grad_norm": 1.9542019367218018,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 5660
+    },
+    {
+      "epoch": 37.30263157894737,
+      "grad_norm": 2.236436128616333,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 5670
+    },
+    {
+      "epoch": 37.36842105263158,
+      "grad_norm": 2.5332870483398438,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 5680
+    },
+    {
+      "epoch": 37.43421052631579,
+      "grad_norm": 2.4465630054473877,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 5690
+    },
+    {
+      "epoch": 37.5,
+      "grad_norm": 2.6095876693725586,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 5700
+    },
+    {
+      "epoch": 37.56578947368421,
+      "grad_norm": 2.1583876609802246,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 5710
+    },
+    {
+      "epoch": 37.63157894736842,
+      "grad_norm": 2.2866389751434326,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 5720
+    },
+    {
+      "epoch": 37.69736842105263,
+      "grad_norm": 2.5979535579681396,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 5730
+    },
+    {
+      "epoch": 37.76315789473684,
+      "grad_norm": 2.4028756618499756,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 5740
+    },
+    {
+      "epoch": 37.828947368421055,
+      "grad_norm": 2.268890619277954,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 5750
+    },
+    {
+      "epoch": 37.89473684210526,
+      "grad_norm": 2.3834378719329834,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 5760
+    },
+    {
+      "epoch": 37.96052631578947,
+      "grad_norm": 2.4863436222076416,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "step": 5770
+    },
+    {
+      "epoch": 38.026315789473685,
+      "grad_norm": 2.2150402069091797,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 5780
+    },
+    {
+      "epoch": 38.0921052631579,
+      "grad_norm": 2.104177474975586,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 5790
+    },
+    {
+      "epoch": 38.1578947368421,
+      "grad_norm": 2.027252674102783,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 5800
+    },
+    {
+      "epoch": 38.223684210526315,
+      "grad_norm": 2.5156586170196533,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 5810
+    },
+    {
+      "epoch": 38.28947368421053,
+      "grad_norm": 2.186910629272461,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 5820
+    },
+    {
+      "epoch": 38.35526315789474,
+      "grad_norm": 2.3481905460357666,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 5830
+    },
+    {
+      "epoch": 38.421052631578945,
+      "grad_norm": 2.13398814201355,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 5840
+    },
+    {
+      "epoch": 38.48684210526316,
+      "grad_norm": 2.2620978355407715,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 5850
+    },
+    {
+      "epoch": 38.55263157894737,
+      "grad_norm": 2.134782075881958,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 5860
+    },
+    {
+      "epoch": 38.61842105263158,
+      "grad_norm": 1.88546884059906,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 5870
+    },
+    {
+      "epoch": 38.68421052631579,
+      "grad_norm": 2.0278677940368652,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "step": 5880
+    },
+    {
+      "epoch": 38.75,
+      "grad_norm": 2.1986711025238037,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 5890
+    },
+    {
+      "epoch": 38.81578947368421,
+      "grad_norm": 2.5093231201171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 5900
+    },
+    {
+      "epoch": 38.88157894736842,
+      "grad_norm": 2.205291509628296,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 5910
+    },
+    {
+      "epoch": 38.94736842105263,
+      "grad_norm": 2.2630720138549805,
+      "learning_rate": 0.0001,
+      "loss": 0.042,
+      "step": 5920
+    },
+    {
+      "epoch": 39.01315789473684,
+      "grad_norm": 2.1553659439086914,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 5930
+    },
+    {
+      "epoch": 39.078947368421055,
+      "grad_norm": 2.4210903644561768,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 5940
+    },
+    {
+      "epoch": 39.14473684210526,
+      "grad_norm": 2.5638651847839355,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 5950
+    },
+    {
+      "epoch": 39.21052631578947,
+      "grad_norm": 2.3888773918151855,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 5960
+    },
+    {
+      "epoch": 39.276315789473685,
+      "grad_norm": 2.2515876293182373,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 5970
+    },
+    {
+      "epoch": 39.3421052631579,
+      "grad_norm": 2.554680347442627,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 5980
+    },
+    {
+      "epoch": 39.4078947368421,
+      "grad_norm": 2.454371690750122,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 5990
+    },
+    {
+      "epoch": 39.473684210526315,
+      "grad_norm": 2.179824113845825,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 6000
+    },
+    {
+      "epoch": 39.53947368421053,
+      "grad_norm": 2.5795116424560547,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 6010
+    },
+    {
+      "epoch": 39.60526315789474,
+      "grad_norm": 2.0184319019317627,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 6020
+    },
+    {
+      "epoch": 39.671052631578945,
+      "grad_norm": 2.221646547317505,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 6030
+    },
+    {
+      "epoch": 39.73684210526316,
+      "grad_norm": 2.4176130294799805,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 6040
+    },
+    {
+      "epoch": 39.80263157894737,
+      "grad_norm": 2.258746385574341,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 6050
+    },
+    {
+      "epoch": 39.86842105263158,
+      "grad_norm": 2.2716710567474365,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 6060
+    },
+    {
+      "epoch": 39.93421052631579,
+      "grad_norm": 2.2062830924987793,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 6070
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 2.2521860599517822,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 6080
+    },
+    {
+      "epoch": 40.06578947368421,
+      "grad_norm": 2.0055341720581055,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 6090
+    },
+    {
+      "epoch": 40.13157894736842,
+      "grad_norm": 2.273428201675415,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 6100
+    },
+    {
+      "epoch": 40.19736842105263,
+      "grad_norm": 2.1523613929748535,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 6110
+    },
+    {
+      "epoch": 40.26315789473684,
+      "grad_norm": 2.2679247856140137,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 6120
+    },
+    {
+      "epoch": 40.328947368421055,
+      "grad_norm": 2.01351261138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 6130
+    },
+    {
+      "epoch": 40.39473684210526,
+      "grad_norm": 2.1786999702453613,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 6140
+    },
+    {
+      "epoch": 40.46052631578947,
+      "grad_norm": 2.424417734146118,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 6150
+    },
+    {
+      "epoch": 40.526315789473685,
+      "grad_norm": 1.980931043624878,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 6160
+    },
+    {
+      "epoch": 40.5921052631579,
+      "grad_norm": 2.2606894969940186,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 6170
+    },
+    {
+      "epoch": 40.6578947368421,
+      "grad_norm": 2.0380303859710693,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 6180
+    },
+    {
+      "epoch": 40.723684210526315,
+      "grad_norm": 2.143332004547119,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 6190
+    },
+    {
+      "epoch": 40.78947368421053,
+      "grad_norm": 2.299978017807007,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 6200
+    },
+    {
+      "epoch": 40.85526315789474,
+      "grad_norm": 2.454505205154419,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 6210
+    },
+    {
+      "epoch": 40.921052631578945,
+      "grad_norm": 2.5504910945892334,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 6220
+    },
+    {
+      "epoch": 40.98684210526316,
+      "grad_norm": 2.2120444774627686,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 6230
+    },
+    {
+      "epoch": 41.05263157894737,
+      "grad_norm": 2.1109986305236816,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 6240
+    },
+    {
+      "epoch": 41.11842105263158,
+      "grad_norm": 2.4341702461242676,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 6250
+    },
+    {
+      "epoch": 41.18421052631579,
+      "grad_norm": 2.1559791564941406,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 6260
+    },
+    {
+      "epoch": 41.25,
+      "grad_norm": 2.233827590942383,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 6270
+    },
+    {
+      "epoch": 41.31578947368421,
+      "grad_norm": 2.0615148544311523,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 6280
+    },
+    {
+      "epoch": 41.38157894736842,
+      "grad_norm": 2.319382667541504,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 6290
+    },
+    {
+      "epoch": 41.44736842105263,
+      "grad_norm": 2.304764986038208,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 6300
+    },
+    {
+      "epoch": 41.51315789473684,
+      "grad_norm": 2.4223990440368652,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 6310
+    },
+    {
+      "epoch": 41.578947368421055,
+      "grad_norm": 2.1475377082824707,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 6320
+    },
+    {
+      "epoch": 41.64473684210526,
+      "grad_norm": 2.036505699157715,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 6330
+    },
+    {
+      "epoch": 41.71052631578947,
+      "grad_norm": 1.9919437170028687,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 6340
+    },
+    {
+      "epoch": 41.776315789473685,
+      "grad_norm": 2.1360974311828613,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 6350
+    },
+    {
+      "epoch": 41.8421052631579,
+      "grad_norm": 2.1707985401153564,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 6360
+    },
+    {
+      "epoch": 41.9078947368421,
+      "grad_norm": 2.1546218395233154,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 6370
+    },
+    {
+      "epoch": 41.973684210526315,
+      "grad_norm": 1.8794069290161133,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 6380
+    },
+    {
+      "epoch": 42.03947368421053,
+      "grad_norm": 2.156665086746216,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 6390
+    },
+    {
+      "epoch": 42.10526315789474,
+      "grad_norm": 2.243816614151001,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 6400
+    },
+    {
+      "epoch": 42.171052631578945,
+      "grad_norm": 2.2223100662231445,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 6410
+    },
+    {
+      "epoch": 42.23684210526316,
+      "grad_norm": 2.0650203227996826,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 6420
+    },
+    {
+      "epoch": 42.30263157894737,
+      "grad_norm": 2.1891047954559326,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 6430
+    },
+    {
+      "epoch": 42.36842105263158,
+      "grad_norm": 2.168792963027954,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 6440
+    },
+    {
+      "epoch": 42.43421052631579,
+      "grad_norm": 2.247410774230957,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 6450
+    },
+    {
+      "epoch": 42.5,
+      "grad_norm": 1.878936529159546,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 6460
+    },
+    {
+      "epoch": 42.56578947368421,
+      "grad_norm": 2.108396530151367,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 6470
+    },
+    {
+      "epoch": 42.63157894736842,
+      "grad_norm": 1.6226104497909546,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 6480
+    },
+    {
+      "epoch": 42.69736842105263,
+      "grad_norm": 2.6258723735809326,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 6490
+    },
+    {
+      "epoch": 42.76315789473684,
+      "grad_norm": 2.1555979251861572,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 6500
+    },
+    {
+      "epoch": 42.828947368421055,
+      "grad_norm": 2.5804638862609863,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 6510
+    },
+    {
+      "epoch": 42.89473684210526,
+      "grad_norm": 2.313135862350464,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 6520
+    },
+    {
+      "epoch": 42.96052631578947,
+      "grad_norm": 2.659311056137085,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 6530
+    },
+    {
+      "epoch": 43.026315789473685,
+      "grad_norm": 2.1776773929595947,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 6540
+    },
+    {
+      "epoch": 43.0921052631579,
+      "grad_norm": 2.1495778560638428,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 6550
+    },
+    {
+      "epoch": 43.1578947368421,
+      "grad_norm": 2.05283522605896,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 6560
+    },
+    {
+      "epoch": 43.223684210526315,
+      "grad_norm": 2.5983822345733643,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 6570
+    },
+    {
+      "epoch": 43.28947368421053,
+      "grad_norm": 2.5054686069488525,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 6580
+    },
+    {
+      "epoch": 43.35526315789474,
+      "grad_norm": 1.933837890625,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 6590
+    },
+    {
+      "epoch": 43.421052631578945,
+      "grad_norm": 2.3833420276641846,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 6600
+    },
+    {
+      "epoch": 43.48684210526316,
+      "grad_norm": 2.4395833015441895,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 6610
+    },
+    {
+      "epoch": 43.55263157894737,
+      "grad_norm": 2.5590438842773438,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 6620
+    },
+    {
+      "epoch": 43.61842105263158,
+      "grad_norm": 2.5415804386138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 6630
+    },
+    {
+      "epoch": 43.68421052631579,
+      "grad_norm": 1.9072297811508179,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 6640
+    },
+    {
+      "epoch": 43.75,
+      "grad_norm": 1.8998295068740845,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 6650
+    },
+    {
+      "epoch": 43.81578947368421,
+      "grad_norm": 1.905118465423584,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 6660
+    },
+    {
+      "epoch": 43.88157894736842,
+      "grad_norm": 2.132871627807617,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 6670
+    },
+    {
+      "epoch": 43.94736842105263,
+      "grad_norm": 2.0413613319396973,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 6680
+    },
+    {
+      "epoch": 44.01315789473684,
+      "grad_norm": 2.2114999294281006,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 6690
+    },
+    {
+      "epoch": 44.078947368421055,
+      "grad_norm": 2.3088066577911377,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 6700
+    },
+    {
+      "epoch": 44.14473684210526,
+      "grad_norm": 1.784862756729126,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 6710
+    },
+    {
+      "epoch": 44.21052631578947,
+      "grad_norm": 2.099011182785034,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 6720
+    },
+    {
+      "epoch": 44.276315789473685,
+      "grad_norm": 1.912520170211792,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 6730
+    },
+    {
+      "epoch": 44.3421052631579,
+      "grad_norm": 1.9815537929534912,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 6740
+    },
+    {
+      "epoch": 44.4078947368421,
+      "grad_norm": 2.396345853805542,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 6750
+    },
+    {
+      "epoch": 44.473684210526315,
+      "grad_norm": 2.2811830043792725,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 6760
+    },
+    {
+      "epoch": 44.53947368421053,
+      "grad_norm": 2.1888914108276367,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 6770
+    },
+    {
+      "epoch": 44.60526315789474,
+      "grad_norm": 2.2141542434692383,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 6780
+    },
+    {
+      "epoch": 44.671052631578945,
+      "grad_norm": 1.9573568105697632,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 6790
+    },
+    {
+      "epoch": 44.73684210526316,
+      "grad_norm": 2.2900731563568115,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 6800
+    },
+    {
+      "epoch": 44.80263157894737,
+      "grad_norm": 2.2001209259033203,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 6810
+    },
+    {
+      "epoch": 44.86842105263158,
+      "grad_norm": 2.0125033855438232,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 6820
+    },
+    {
+      "epoch": 44.93421052631579,
+      "grad_norm": 1.908859372138977,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 6830
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 2.0976343154907227,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 6840
+    },
+    {
+      "epoch": 45.06578947368421,
+      "grad_norm": 2.10421085357666,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 6850
+    },
+    {
+      "epoch": 45.13157894736842,
+      "grad_norm": 1.7706947326660156,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 6860
+    },
+    {
+      "epoch": 45.19736842105263,
+      "grad_norm": 2.1556997299194336,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 6870
+    },
+    {
+      "epoch": 45.26315789473684,
+      "grad_norm": 2.2130556106567383,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 6880
+    },
+    {
+      "epoch": 45.328947368421055,
+      "grad_norm": 2.1554789543151855,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 6890
+    },
+    {
+      "epoch": 45.39473684210526,
+      "grad_norm": 1.984050989151001,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 6900
+    },
+    {
+      "epoch": 45.46052631578947,
+      "grad_norm": 1.7056621313095093,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 6910
+    },
+    {
+      "epoch": 45.526315789473685,
+      "grad_norm": 1.9881770610809326,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 6920
+    },
+    {
+      "epoch": 45.5921052631579,
+      "grad_norm": 1.829092264175415,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 6930
+    },
+    {
+      "epoch": 45.6578947368421,
+      "grad_norm": 1.8903696537017822,
+      "learning_rate": 0.0001,
+      "loss": 0.0299,
+      "step": 6940
+    },
+    {
+      "epoch": 45.723684210526315,
+      "grad_norm": 2.089012861251831,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 6950
+    },
+    {
+      "epoch": 45.78947368421053,
+      "grad_norm": 2.2835586071014404,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 6960
+    },
+    {
+      "epoch": 45.85526315789474,
+      "grad_norm": 2.3608603477478027,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 6970
+    },
+    {
+      "epoch": 45.921052631578945,
+      "grad_norm": 2.000751495361328,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 6980
+    },
+    {
+      "epoch": 45.98684210526316,
+      "grad_norm": 2.0586018562316895,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 6990
+    },
+    {
+      "epoch": 46.05263157894737,
+      "grad_norm": 2.077038526535034,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 7000
+    },
+    {
+      "epoch": 46.11842105263158,
+      "grad_norm": 2.457906484603882,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 7010
+    },
+    {
+      "epoch": 46.18421052631579,
+      "grad_norm": 2.286679267883301,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 7020
+    },
+    {
+      "epoch": 46.25,
+      "grad_norm": 2.5469555854797363,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 7030
+    },
+    {
+      "epoch": 46.31578947368421,
+      "grad_norm": 2.1791422367095947,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 7040
+    },
+    {
+      "epoch": 46.38157894736842,
+      "grad_norm": 3.0373752117156982,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 7050
+    },
+    {
+      "epoch": 46.44736842105263,
+      "grad_norm": 2.291721820831299,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 7060
+    },
+    {
+      "epoch": 46.51315789473684,
+      "grad_norm": 2.208719253540039,
+      "learning_rate": 0.0001,
+      "loss": 0.0308,
+      "step": 7070
+    },
+    {
+      "epoch": 46.578947368421055,
+      "grad_norm": 2.194161891937256,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 7080
+    },
+    {
+      "epoch": 46.64473684210526,
+      "grad_norm": 2.2741446495056152,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 7090
+    },
+    {
+      "epoch": 46.71052631578947,
+      "grad_norm": 2.3912644386291504,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 7100
+    },
+    {
+      "epoch": 46.776315789473685,
+      "grad_norm": 2.1767451763153076,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 7110
+    },
+    {
+      "epoch": 46.8421052631579,
+      "grad_norm": 2.3903443813323975,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 7120
+    },
+    {
+      "epoch": 46.9078947368421,
+      "grad_norm": 2.259274482727051,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 7130
+    },
+    {
+      "epoch": 46.973684210526315,
+      "grad_norm": 2.3009188175201416,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 7140
+    },
+    {
+      "epoch": 47.03947368421053,
+      "grad_norm": 2.3467867374420166,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 7150
+    },
+    {
+      "epoch": 47.10526315789474,
+      "grad_norm": 2.426922082901001,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 7160
+    },
+    {
+      "epoch": 47.171052631578945,
+      "grad_norm": 2.1517276763916016,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 7170
+    },
+    {
+      "epoch": 47.23684210526316,
+      "grad_norm": 2.0987422466278076,
+      "learning_rate": 0.0001,
+      "loss": 0.0281,
+      "step": 7180
+    },
+    {
+      "epoch": 47.30263157894737,
+      "grad_norm": 1.8270368576049805,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 7190
+    },
+    {
+      "epoch": 47.36842105263158,
+      "grad_norm": 2.406790018081665,
+      "learning_rate": 0.0001,
+      "loss": 0.0296,
+      "step": 7200
+    },
+    {
+      "epoch": 47.43421052631579,
+      "grad_norm": 1.9560794830322266,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 7210
+    },
+    {
+      "epoch": 47.5,
+      "grad_norm": 1.9922142028808594,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 7220
+    },
+    {
+      "epoch": 47.56578947368421,
+      "grad_norm": 1.9082542657852173,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 7230
+    },
+    {
+      "epoch": 47.63157894736842,
+      "grad_norm": 2.251033306121826,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 7240
+    },
+    {
+      "epoch": 47.69736842105263,
+      "grad_norm": 1.8770537376403809,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 7250
+    },
+    {
+      "epoch": 47.76315789473684,
+      "grad_norm": 2.259347438812256,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 7260
+    },
+    {
+      "epoch": 47.828947368421055,
+      "grad_norm": 2.3631973266601562,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 7270
+    },
+    {
+      "epoch": 47.89473684210526,
+      "grad_norm": 2.6666479110717773,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 7280
+    },
+    {
+      "epoch": 47.96052631578947,
+      "grad_norm": 2.482814073562622,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 7290
+    },
+    {
+      "epoch": 48.026315789473685,
+      "grad_norm": 2.0329630374908447,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 7300
+    },
+    {
+      "epoch": 48.0921052631579,
+      "grad_norm": 2.2136952877044678,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 7310
+    },
+    {
+      "epoch": 48.1578947368421,
+      "grad_norm": 2.1611454486846924,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 7320
+    },
+    {
+      "epoch": 48.223684210526315,
+      "grad_norm": 2.0240912437438965,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 7330
+    },
+    {
+      "epoch": 48.28947368421053,
+      "grad_norm": 2.6601762771606445,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 7340
+    },
+    {
+      "epoch": 48.35526315789474,
+      "grad_norm": 2.2921714782714844,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 7350
+    },
+    {
+      "epoch": 48.421052631578945,
+      "grad_norm": 2.367628812789917,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 7360
+    },
+    {
+      "epoch": 48.48684210526316,
+      "grad_norm": 2.152510166168213,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 7370
+    },
+    {
+      "epoch": 48.55263157894737,
+      "grad_norm": 2.1306285858154297,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 7380
+    },
+    {
+      "epoch": 48.61842105263158,
+      "grad_norm": 2.3841469287872314,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 7390
+    },
+    {
+      "epoch": 48.68421052631579,
+      "grad_norm": 2.2278404235839844,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 7400
+    },
+    {
+      "epoch": 48.75,
+      "grad_norm": 2.0181496143341064,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 7410
+    },
+    {
+      "epoch": 48.81578947368421,
+      "grad_norm": 1.9607185125350952,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 7420
+    },
+    {
+      "epoch": 48.88157894736842,
+      "grad_norm": 2.1952409744262695,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 7430
+    },
+    {
+      "epoch": 48.94736842105263,
+      "grad_norm": 1.9687610864639282,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 7440
+    },
+    {
+      "epoch": 49.01315789473684,
+      "grad_norm": 2.2068471908569336,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 7450
+    },
+    {
+      "epoch": 49.078947368421055,
+      "grad_norm": 2.1997668743133545,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 7460
+    },
+    {
+      "epoch": 49.14473684210526,
+      "grad_norm": 2.183655261993408,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 7470
+    },
+    {
+      "epoch": 49.21052631578947,
+      "grad_norm": 1.942799687385559,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 7480
+    },
+    {
+      "epoch": 49.276315789473685,
+      "grad_norm": 1.9380285739898682,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 7490
+    },
+    {
+      "epoch": 49.3421052631579,
+      "grad_norm": 2.13240122795105,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 7500
+    },
+    {
+      "epoch": 49.4078947368421,
+      "grad_norm": 2.2831900119781494,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 7510
+    },
+    {
+      "epoch": 49.473684210526315,
+      "grad_norm": 2.127020835876465,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 7520
+    },
+    {
+      "epoch": 49.53947368421053,
+      "grad_norm": 2.251577138900757,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 7530
+    },
+    {
+      "epoch": 49.60526315789474,
+      "grad_norm": 2.340786933898926,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 7540
+    },
+    {
+      "epoch": 49.671052631578945,
+      "grad_norm": 2.267075300216675,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 7550
+    },
+    {
+      "epoch": 49.73684210526316,
+      "grad_norm": 2.334462881088257,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 7560
+    },
+    {
+      "epoch": 49.80263157894737,
+      "grad_norm": 2.038785696029663,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 7570
+    },
+    {
+      "epoch": 49.86842105263158,
+      "grad_norm": 2.192054510116577,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 7580
+    },
+    {
+      "epoch": 49.93421052631579,
+      "grad_norm": 2.1006948947906494,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 7590
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 1.9702825546264648,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 7600
+    },
+    {
+      "epoch": 50.06578947368421,
+      "grad_norm": 1.8587983846664429,
+      "learning_rate": 0.0001,
+      "loss": 0.0297,
+      "step": 7610
+    },
+    {
+      "epoch": 50.13157894736842,
+      "grad_norm": 2.164700984954834,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 7620
+    },
+    {
+      "epoch": 50.19736842105263,
+      "grad_norm": 2.0936055183410645,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 7630
+    },
+    {
+      "epoch": 50.26315789473684,
+      "grad_norm": 2.155484914779663,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 7640
+    },
+    {
+      "epoch": 50.328947368421055,
+      "grad_norm": 2.1522960662841797,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 7650
+    },
+    {
+      "epoch": 50.39473684210526,
+      "grad_norm": 1.9459351301193237,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 7660
+    },
+    {
+      "epoch": 50.46052631578947,
+      "grad_norm": 2.1670145988464355,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 7670
+    },
+    {
+      "epoch": 50.526315789473685,
+      "grad_norm": 2.069352626800537,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 7680
+    },
+    {
+      "epoch": 50.5921052631579,
+      "grad_norm": 1.8716386556625366,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 7690
+    },
+    {
+      "epoch": 50.6578947368421,
+      "grad_norm": 2.056234836578369,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "step": 7700
+    },
+    {
+      "epoch": 50.723684210526315,
+      "grad_norm": 2.560983896255493,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 7710
+    },
+    {
+      "epoch": 50.78947368421053,
+      "grad_norm": 2.280230760574341,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 7720
+    },
+    {
+      "epoch": 50.85526315789474,
+      "grad_norm": 2.542572498321533,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 7730
+    },
+    {
+      "epoch": 50.921052631578945,
+      "grad_norm": 2.2160181999206543,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 7740
+    },
+    {
+      "epoch": 50.98684210526316,
+      "grad_norm": 1.9975266456604004,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 7750
+    },
+    {
+      "epoch": 51.05263157894737,
+      "grad_norm": 2.1476669311523438,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 7760
+    },
+    {
+      "epoch": 51.11842105263158,
+      "grad_norm": 1.9486243724822998,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 7770
+    },
+    {
+      "epoch": 51.18421052631579,
+      "grad_norm": 1.993624210357666,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 7780
+    },
+    {
+      "epoch": 51.25,
+      "grad_norm": 2.0168542861938477,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 7790
+    },
+    {
+      "epoch": 51.31578947368421,
+      "grad_norm": 2.1617634296417236,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 7800
+    },
+    {
+      "epoch": 51.38157894736842,
+      "grad_norm": 1.82743239402771,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 7810
+    },
+    {
+      "epoch": 51.44736842105263,
+      "grad_norm": 1.7217556238174438,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 7820
+    },
+    {
+      "epoch": 51.51315789473684,
+      "grad_norm": 1.9004185199737549,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 7830
+    },
+    {
+      "epoch": 51.578947368421055,
+      "grad_norm": 2.210575580596924,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 7840
+    },
+    {
+      "epoch": 51.64473684210526,
+      "grad_norm": 1.8974437713623047,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 7850
+    },
+    {
+      "epoch": 51.71052631578947,
+      "grad_norm": 1.8807051181793213,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 7860
+    },
+    {
+      "epoch": 51.776315789473685,
+      "grad_norm": 1.9475070238113403,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 7870
+    },
+    {
+      "epoch": 51.8421052631579,
+      "grad_norm": 1.9319655895233154,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 7880
+    },
+    {
+      "epoch": 51.9078947368421,
+      "grad_norm": 1.9801864624023438,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 7890
+    },
+    {
+      "epoch": 51.973684210526315,
+      "grad_norm": 1.8490853309631348,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 7900
+    },
+    {
+      "epoch": 52.03947368421053,
+      "grad_norm": 1.844796061515808,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 7910
+    },
+    {
+      "epoch": 52.10526315789474,
+      "grad_norm": 2.170541763305664,
+      "learning_rate": 0.0001,
+      "loss": 0.0304,
+      "step": 7920
+    },
+    {
+      "epoch": 52.171052631578945,
+      "grad_norm": 1.8265526294708252,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 7930
+    },
+    {
+      "epoch": 52.23684210526316,
+      "grad_norm": 1.8391716480255127,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 7940
+    },
+    {
+      "epoch": 52.30263157894737,
+      "grad_norm": 2.0495717525482178,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 7950
+    },
+    {
+      "epoch": 52.36842105263158,
+      "grad_norm": 2.0973620414733887,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 7960
+    },
+    {
+      "epoch": 52.43421052631579,
+      "grad_norm": 2.0247726440429688,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 7970
+    },
+    {
+      "epoch": 52.5,
+      "grad_norm": 2.3339807987213135,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 7980
+    },
+    {
+      "epoch": 52.56578947368421,
+      "grad_norm": 2.4479763507843018,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 7990
+    },
+    {
+      "epoch": 52.63157894736842,
+      "grad_norm": 2.1954760551452637,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 8000
+    },
+    {
+      "epoch": 52.69736842105263,
+      "grad_norm": 2.01871919631958,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 8010
+    },
+    {
+      "epoch": 52.76315789473684,
+      "grad_norm": 1.8598896265029907,
+      "learning_rate": 0.0001,
+      "loss": 0.0304,
+      "step": 8020
+    },
+    {
+      "epoch": 52.828947368421055,
+      "grad_norm": 2.171518087387085,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 8030
+    },
+    {
+      "epoch": 52.89473684210526,
+      "grad_norm": 2.029583692550659,
+      "learning_rate": 0.0001,
+      "loss": 0.0308,
+      "step": 8040
+    },
+    {
+      "epoch": 52.96052631578947,
+      "grad_norm": 1.7323065996170044,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 8050
+    },
+    {
+      "epoch": 53.026315789473685,
+      "grad_norm": 2.0981595516204834,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 8060
+    },
+    {
+      "epoch": 53.0921052631579,
+      "grad_norm": 1.671781063079834,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 8070
+    },
+    {
+      "epoch": 53.1578947368421,
+      "grad_norm": 1.8050312995910645,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 8080
+    },
+    {
+      "epoch": 53.223684210526315,
+      "grad_norm": 2.084514617919922,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 8090
+    },
+    {
+      "epoch": 53.28947368421053,
+      "grad_norm": 1.9709466695785522,
+      "learning_rate": 0.0001,
+      "loss": 0.0304,
+      "step": 8100
+    },
+    {
+      "epoch": 53.35526315789474,
+      "grad_norm": 1.856992244720459,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "step": 8110
+    },
+    {
+      "epoch": 53.421052631578945,
+      "grad_norm": 1.892453670501709,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 8120
+    },
+    {
+      "epoch": 53.48684210526316,
+      "grad_norm": 1.9388136863708496,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "step": 8130
+    },
+    {
+      "epoch": 53.55263157894737,
+      "grad_norm": 1.8416920900344849,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 8140
+    },
+    {
+      "epoch": 53.61842105263158,
+      "grad_norm": 2.1083364486694336,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 8150
+    },
+    {
+      "epoch": 53.68421052631579,
+      "grad_norm": 2.071626663208008,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 8160
+    },
+    {
+      "epoch": 53.75,
+      "grad_norm": 1.8106697797775269,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 8170
+    },
+    {
+      "epoch": 53.81578947368421,
+      "grad_norm": 1.9116227626800537,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 8180
+    },
+    {
+      "epoch": 53.88157894736842,
+      "grad_norm": 1.6516140699386597,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "step": 8190
+    },
+    {
+      "epoch": 53.94736842105263,
+      "grad_norm": 2.0792529582977295,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 8200
+    },
+    {
+      "epoch": 54.01315789473684,
+      "grad_norm": 2.022188425064087,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 8210
+    },
+    {
+      "epoch": 54.078947368421055,
+      "grad_norm": 1.862682819366455,
+      "learning_rate": 0.0001,
+      "loss": 0.029,
+      "step": 8220
+    },
+    {
+      "epoch": 54.14473684210526,
+      "grad_norm": 1.7206437587738037,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 8230
+    },
+    {
+      "epoch": 54.21052631578947,
+      "grad_norm": 1.6664676666259766,
+      "learning_rate": 0.0001,
+      "loss": 0.0306,
+      "step": 8240
+    },
+    {
+      "epoch": 54.276315789473685,
+      "grad_norm": 2.4094207286834717,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 8250
+    },
+    {
+      "epoch": 54.3421052631579,
+      "grad_norm": 1.9840387105941772,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 8260
+    },
+    {
+      "epoch": 54.4078947368421,
+      "grad_norm": 2.449998140335083,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 8270
+    },
+    {
+      "epoch": 54.473684210526315,
+      "grad_norm": 2.195594072341919,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 8280
+    },
+    {
+      "epoch": 54.53947368421053,
+      "grad_norm": 2.2626521587371826,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 8290
+    },
+    {
+      "epoch": 54.60526315789474,
+      "grad_norm": 1.693280816078186,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 8300
+    },
+    {
+      "epoch": 54.671052631578945,
+      "grad_norm": 1.8923887014389038,
+      "learning_rate": 0.0001,
+      "loss": 0.0296,
+      "step": 8310
+    },
+    {
+      "epoch": 54.73684210526316,
+      "grad_norm": 1.639238953590393,
+      "learning_rate": 0.0001,
+      "loss": 0.0289,
+      "step": 8320
+    },
+    {
+      "epoch": 54.80263157894737,
+      "grad_norm": 1.9876776933670044,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 8330
+    },
+    {
+      "epoch": 54.86842105263158,
+      "grad_norm": 1.686047077178955,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 8340
+    },
+    {
+      "epoch": 54.93421052631579,
+      "grad_norm": 2.186433792114258,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 8350
+    },
+    {
+      "epoch": 55.0,
+      "grad_norm": 1.9404265880584717,
+      "learning_rate": 0.0001,
+      "loss": 0.0302,
+      "step": 8360
+    },
+    {
+      "epoch": 55.06578947368421,
+      "grad_norm": 2.0368666648864746,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 8370
+    },
+    {
+      "epoch": 55.13157894736842,
+      "grad_norm": 1.385018229484558,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 8380
+    },
+    {
+      "epoch": 55.19736842105263,
+      "grad_norm": 2.025214433670044,
+      "learning_rate": 0.0001,
+      "loss": 0.0291,
+      "step": 8390
+    },
+    {
+      "epoch": 55.26315789473684,
+      "grad_norm": 1.7362509965896606,
+      "learning_rate": 0.0001,
+      "loss": 0.0302,
+      "step": 8400
+    },
+    {
+      "epoch": 55.328947368421055,
+      "grad_norm": 1.9747296571731567,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 8410
+    },
+    {
+      "epoch": 55.39473684210526,
+      "grad_norm": 1.8088208436965942,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 8420
+    },
+    {
+      "epoch": 55.46052631578947,
+      "grad_norm": 1.8630869388580322,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 8430
+    },
+    {
+      "epoch": 55.526315789473685,
+      "grad_norm": 1.8169368505477905,
+      "learning_rate": 0.0001,
+      "loss": 0.0294,
+      "step": 8440
+    },
+    {
+      "epoch": 55.5921052631579,
+      "grad_norm": 1.717041254043579,
+      "learning_rate": 0.0001,
+      "loss": 0.0304,
+      "step": 8450
+    },
+    {
+      "epoch": 55.6578947368421,
+      "grad_norm": 2.054370164871216,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 8460
+    },
+    {
+      "epoch": 55.723684210526315,
+      "grad_norm": 1.8718016147613525,
+      "learning_rate": 0.0001,
+      "loss": 0.0294,
+      "step": 8470
+    },
+    {
+      "epoch": 55.78947368421053,
+      "grad_norm": 2.4793713092803955,
+      "learning_rate": 0.0001,
+      "loss": 0.0302,
+      "step": 8480
+    },
+    {
+      "epoch": 55.85526315789474,
+      "grad_norm": 2.1499078273773193,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 8490
+    },
+    {
+      "epoch": 55.921052631578945,
+      "grad_norm": 2.1316916942596436,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 8500
+    },
+    {
+      "epoch": 55.98684210526316,
+      "grad_norm": 2.108400583267212,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 8510
+    },
+    {
+      "epoch": 56.05263157894737,
+      "grad_norm": 1.7546727657318115,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 8520
+    },
+    {
+      "epoch": 56.11842105263158,
+      "grad_norm": 1.8155990839004517,
+      "learning_rate": 0.0001,
+      "loss": 0.0285,
+      "step": 8530
+    },
+    {
+      "epoch": 56.18421052631579,
+      "grad_norm": 1.8450534343719482,
+      "learning_rate": 0.0001,
+      "loss": 0.0299,
+      "step": 8540
+    },
+    {
+      "epoch": 56.25,
+      "grad_norm": 1.9651875495910645,
+      "learning_rate": 0.0001,
+      "loss": 0.0308,
+      "step": 8550
+    },
+    {
+      "epoch": 56.31578947368421,
+      "grad_norm": 1.9930384159088135,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 8560
+    },
+    {
+      "epoch": 56.38157894736842,
+      "grad_norm": 2.035289764404297,
+      "learning_rate": 0.0001,
+      "loss": 0.0296,
+      "step": 8570
+    },
+    {
+      "epoch": 56.44736842105263,
+      "grad_norm": 2.206793785095215,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 8580
+    },
+    {
+      "epoch": 56.51315789473684,
+      "grad_norm": 2.2738990783691406,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 8590
+    },
+    {
+      "epoch": 56.578947368421055,
+      "grad_norm": 1.6985399723052979,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 8600
+    },
+    {
+      "epoch": 56.64473684210526,
+      "grad_norm": 1.7506465911865234,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 8610
+    },
+    {
+      "epoch": 56.71052631578947,
+      "grad_norm": 1.8177592754364014,
+      "learning_rate": 0.0001,
+      "loss": 0.0299,
+      "step": 8620
+    },
+    {
+      "epoch": 56.776315789473685,
+      "grad_norm": 2.0824978351593018,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 8630
+    },
+    {
+      "epoch": 56.8421052631579,
+      "grad_norm": 1.8708689212799072,
+      "learning_rate": 0.0001,
+      "loss": 0.0294,
+      "step": 8640
+    },
+    {
+      "epoch": 56.9078947368421,
+      "grad_norm": 1.536291480064392,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 8650
+    },
+    {
+      "epoch": 56.973684210526315,
+      "grad_norm": 1.5639556646347046,
+      "learning_rate": 0.0001,
+      "loss": 0.0292,
+      "step": 8660
+    },
+    {
+      "epoch": 57.03947368421053,
+      "grad_norm": 1.836817979812622,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 8670
+    },
+    {
+      "epoch": 57.10526315789474,
+      "grad_norm": 1.6977314949035645,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 8680
+    },
+    {
+      "epoch": 57.171052631578945,
+      "grad_norm": 1.9491312503814697,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 8690
+    },
+    {
+      "epoch": 57.23684210526316,
+      "grad_norm": 1.9810190200805664,
+      "learning_rate": 0.0001,
+      "loss": 0.0299,
+      "step": 8700
+    },
+    {
+      "epoch": 57.30263157894737,
+      "grad_norm": 3.380073308944702,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 8710
+    },
+    {
+      "epoch": 57.36842105263158,
+      "grad_norm": 2.4485232830047607,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 8720
+    },
+    {
+      "epoch": 57.43421052631579,
+      "grad_norm": 2.817413330078125,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 8730
+    },
+    {
+      "epoch": 57.5,
+      "grad_norm": 2.479456901550293,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 8740
+    },
+    {
+      "epoch": 57.56578947368421,
+      "grad_norm": 2.279740571975708,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 8750
+    },
+    {
+      "epoch": 57.63157894736842,
+      "grad_norm": 2.4175727367401123,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 8760
+    },
+    {
+      "epoch": 57.69736842105263,
+      "grad_norm": 1.8929423093795776,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 8770
+    },
+    {
+      "epoch": 57.76315789473684,
+      "grad_norm": 1.9583559036254883,
+      "learning_rate": 0.0001,
+      "loss": 0.0271,
+      "step": 8780
+    },
+    {
+      "epoch": 57.828947368421055,
+      "grad_norm": 2.2703166007995605,
+      "learning_rate": 0.0001,
+      "loss": 0.0274,
+      "step": 8790
+    },
+    {
+      "epoch": 57.89473684210526,
+      "grad_norm": 2.0460057258605957,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "step": 8800
+    },
+    {
+      "epoch": 57.96052631578947,
+      "grad_norm": 2.1975557804107666,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 8810
+    },
+    {
+      "epoch": 58.026315789473685,
+      "grad_norm": 1.971722960472107,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 8820
+    },
+    {
+      "epoch": 58.0921052631579,
+      "grad_norm": 2.0381031036376953,
+      "learning_rate": 0.0001,
+      "loss": 0.0289,
+      "step": 8830
+    },
+    {
+      "epoch": 58.1578947368421,
+      "grad_norm": 1.650619626045227,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 8840
+    },
+    {
+      "epoch": 58.223684210526315,
+      "grad_norm": 2.122473955154419,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 8850
+    },
+    {
+      "epoch": 58.28947368421053,
+      "grad_norm": 1.7196934223175049,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 8860
+    },
+    {
+      "epoch": 58.35526315789474,
+      "grad_norm": 1.6575008630752563,
+      "learning_rate": 0.0001,
+      "loss": 0.0277,
+      "step": 8870
+    },
+    {
+      "epoch": 58.421052631578945,
+      "grad_norm": 1.7115286588668823,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 8880
+    },
+    {
+      "epoch": 58.48684210526316,
+      "grad_norm": 1.908919095993042,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 8890
+    },
+    {
+      "epoch": 58.55263157894737,
+      "grad_norm": 1.7055505514144897,
+      "learning_rate": 0.0001,
+      "loss": 0.0299,
+      "step": 8900
+    },
+    {
+      "epoch": 58.61842105263158,
+      "grad_norm": 1.72140371799469,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 8910
+    },
+    {
+      "epoch": 58.68421052631579,
+      "grad_norm": 1.752676248550415,
+      "learning_rate": 0.0001,
+      "loss": 0.0288,
+      "step": 8920
+    },
+    {
+      "epoch": 58.75,
+      "grad_norm": 1.886494517326355,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 8930
+    },
+    {
+      "epoch": 58.81578947368421,
+      "grad_norm": 1.7200045585632324,
+      "learning_rate": 0.0001,
+      "loss": 0.0308,
+      "step": 8940
+    },
+    {
+      "epoch": 58.88157894736842,
+      "grad_norm": 1.7977442741394043,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 8950
+    },
+    {
+      "epoch": 58.94736842105263,
+      "grad_norm": 1.9197369813919067,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 8960
+    },
+    {
+      "epoch": 59.01315789473684,
+      "grad_norm": 1.717230200767517,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 8970
+    },
+    {
+      "epoch": 59.078947368421055,
+      "grad_norm": 1.8992480039596558,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 8980
+    },
+    {
+      "epoch": 59.14473684210526,
+      "grad_norm": 1.9328705072402954,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 8990
+    },
+    {
+      "epoch": 59.21052631578947,
+      "grad_norm": 1.7343662977218628,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 9000
+    },
+    {
+      "epoch": 59.276315789473685,
+      "grad_norm": 1.6927893161773682,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 9010
+    },
+    {
+      "epoch": 59.3421052631579,
+      "grad_norm": 1.746718406677246,
+      "learning_rate": 0.0001,
+      "loss": 0.0308,
+      "step": 9020
+    },
+    {
+      "epoch": 59.4078947368421,
+      "grad_norm": 2.095132827758789,
+      "learning_rate": 0.0001,
+      "loss": 0.0296,
+      "step": 9030
+    },
+    {
+      "epoch": 59.473684210526315,
+      "grad_norm": 1.837533950805664,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 9040
+    },
+    {
+      "epoch": 59.53947368421053,
+      "grad_norm": 1.8166559934616089,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 9050
+    },
+    {
+      "epoch": 59.60526315789474,
+      "grad_norm": 1.6477857828140259,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 9060
+    },
+    {
+      "epoch": 59.671052631578945,
+      "grad_norm": 1.8890045881271362,
+      "learning_rate": 0.0001,
+      "loss": 0.0294,
+      "step": 9070
+    },
+    {
+      "epoch": 59.73684210526316,
+      "grad_norm": 2.0278661251068115,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 9080
+    },
+    {
+      "epoch": 59.80263157894737,
+      "grad_norm": 2.0316243171691895,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 9090
+    },
+    {
+      "epoch": 59.86842105263158,
+      "grad_norm": 1.8405317068099976,
+      "learning_rate": 0.0001,
+      "loss": 0.0282,
+      "step": 9100
+    },
+    {
+      "epoch": 59.93421052631579,
+      "grad_norm": 2.397641897201538,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 9110
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 1.7151198387145996,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 9120
+    },
+    {
+      "epoch": 60.06578947368421,
+      "grad_norm": 2.201002359390259,
+      "learning_rate": 0.0001,
+      "loss": 0.0294,
+      "step": 9130
+    },
+    {
+      "epoch": 60.13157894736842,
+      "grad_norm": 1.911597490310669,
+      "learning_rate": 0.0001,
+      "loss": 0.0289,
+      "step": 9140
+    },
+    {
+      "epoch": 60.19736842105263,
+      "grad_norm": 2.3099772930145264,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 9150
+    },
+    {
+      "epoch": 60.26315789473684,
+      "grad_norm": 2.0865139961242676,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 9160
+    },
+    {
+      "epoch": 60.328947368421055,
+      "grad_norm": 2.1491951942443848,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 9170
+    },
+    {
+      "epoch": 60.39473684210526,
+      "grad_norm": 1.7886714935302734,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 9180
+    },
+    {
+      "epoch": 60.46052631578947,
+      "grad_norm": 2.1274571418762207,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 9190
+    },
+    {
+      "epoch": 60.526315789473685,
+      "grad_norm": 2.1661720275878906,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 9200
+    },
+    {
+      "epoch": 60.5921052631579,
+      "grad_norm": 1.6278198957443237,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 9210
+    },
+    {
+      "epoch": 60.6578947368421,
+      "grad_norm": 1.8151003122329712,
+      "learning_rate": 0.0001,
+      "loss": 0.0259,
+      "step": 9220
+    },
+    {
+      "epoch": 60.723684210526315,
+      "grad_norm": 1.9003932476043701,
+      "learning_rate": 0.0001,
+      "loss": 0.0291,
+      "step": 9230
+    },
+    {
+      "epoch": 60.78947368421053,
+      "grad_norm": 1.6437238454818726,
+      "learning_rate": 0.0001,
+      "loss": 0.0274,
+      "step": 9240
+    },
+    {
+      "epoch": 60.85526315789474,
+      "grad_norm": 1.418359637260437,
+      "learning_rate": 0.0001,
+      "loss": 0.0285,
+      "step": 9250
+    },
+    {
+      "epoch": 60.921052631578945,
+      "grad_norm": 1.5304903984069824,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 9260
+    },
+    {
+      "epoch": 60.98684210526316,
+      "grad_norm": 1.7247731685638428,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 9270
+    },
+    {
+      "epoch": 61.05263157894737,
+      "grad_norm": 1.9069534540176392,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 9280
+    },
+    {
+      "epoch": 61.11842105263158,
+      "grad_norm": 1.6469271183013916,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 9290
+    },
+    {
+      "epoch": 61.18421052631579,
+      "grad_norm": 1.5234088897705078,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 9300
+    },
+    {
+      "epoch": 61.25,
+      "grad_norm": 1.9155454635620117,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 9310
+    },
+    {
+      "epoch": 61.31578947368421,
+      "grad_norm": 1.3420042991638184,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 9320
+    },
+    {
+      "epoch": 61.38157894736842,
+      "grad_norm": 1.73533296585083,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 9330
+    },
+    {
+      "epoch": 61.44736842105263,
+      "grad_norm": 2.0660176277160645,
+      "learning_rate": 0.0001,
+      "loss": 0.0302,
+      "step": 9340
+    },
+    {
+      "epoch": 61.51315789473684,
+      "grad_norm": 1.8069857358932495,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 9350
+    },
+    {
+      "epoch": 61.578947368421055,
+      "grad_norm": 2.0904011726379395,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "step": 9360
+    },
+    {
+      "epoch": 61.64473684210526,
+      "grad_norm": 1.8158031702041626,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 9370
+    },
+    {
+      "epoch": 61.71052631578947,
+      "grad_norm": 1.410555362701416,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 9380
+    },
+    {
+      "epoch": 61.776315789473685,
+      "grad_norm": 1.4610170125961304,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 9390
+    },
+    {
+      "epoch": 61.8421052631579,
+      "grad_norm": 1.5231578350067139,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 9400
+    },
+    {
+      "epoch": 61.9078947368421,
+      "grad_norm": 1.829412817955017,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 9410
+    },
+    {
+      "epoch": 61.973684210526315,
+      "grad_norm": 1.9624525308609009,
+      "learning_rate": 0.0001,
+      "loss": 0.0302,
+      "step": 9420
+    },
+    {
+      "epoch": 62.03947368421053,
+      "grad_norm": 1.5953247547149658,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 9430
+    },
+    {
+      "epoch": 62.10526315789474,
+      "grad_norm": 1.8386650085449219,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 9440
+    },
+    {
+      "epoch": 62.171052631578945,
+      "grad_norm": 2.2038257122039795,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 9450
+    },
+    {
+      "epoch": 62.23684210526316,
+      "grad_norm": 2.08797550201416,
+      "learning_rate": 0.0001,
+      "loss": 0.0288,
+      "step": 9460
+    },
+    {
+      "epoch": 62.30263157894737,
+      "grad_norm": 1.9139946699142456,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 9470
+    },
+    {
+      "epoch": 62.36842105263158,
+      "grad_norm": 2.331393241882324,
+      "learning_rate": 0.0001,
+      "loss": 0.0292,
+      "step": 9480
+    },
+    {
+      "epoch": 62.43421052631579,
+      "grad_norm": 1.5585893392562866,
+      "learning_rate": 0.0001,
+      "loss": 0.0285,
+      "step": 9490
+    },
+    {
+      "epoch": 62.5,
+      "grad_norm": 1.67988920211792,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 9500
+    },
+    {
+      "epoch": 62.56578947368421,
+      "grad_norm": 1.9753636121749878,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 9510
+    },
+    {
+      "epoch": 62.63157894736842,
+      "grad_norm": 1.5838857889175415,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 9520
+    },
+    {
+      "epoch": 62.69736842105263,
+      "grad_norm": 1.9455825090408325,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 9530
+    },
+    {
+      "epoch": 62.76315789473684,
+      "grad_norm": 2.0940582752227783,
+      "learning_rate": 0.0001,
+      "loss": 0.029,
+      "step": 9540
+    },
+    {
+      "epoch": 62.828947368421055,
+      "grad_norm": 2.082871198654175,
+      "learning_rate": 0.0001,
+      "loss": 0.0271,
+      "step": 9550
+    },
+    {
+      "epoch": 62.89473684210526,
+      "grad_norm": 1.8131487369537354,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 9560
+    },
+    {
+      "epoch": 62.96052631578947,
+      "grad_norm": 1.8822680711746216,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 9570
+    },
+    {
+      "epoch": 63.026315789473685,
+      "grad_norm": 1.9734561443328857,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 9580
+    },
+    {
+      "epoch": 63.0921052631579,
+      "grad_norm": 1.6446675062179565,
+      "learning_rate": 0.0001,
+      "loss": 0.0285,
+      "step": 9590
+    },
+    {
+      "epoch": 63.1578947368421,
+      "grad_norm": 1.587969422340393,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 9600
+    },
+    {
+      "epoch": 63.223684210526315,
+      "grad_norm": 2.2087881565093994,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 9610
+    },
+    {
+      "epoch": 63.28947368421053,
+      "grad_norm": 2.1909730434417725,
+      "learning_rate": 0.0001,
+      "loss": 0.0296,
+      "step": 9620
+    },
+    {
+      "epoch": 63.35526315789474,
+      "grad_norm": 1.4683645963668823,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "step": 9630
+    },
+    {
+      "epoch": 63.421052631578945,
+      "grad_norm": 1.7564547061920166,
+      "learning_rate": 0.0001,
+      "loss": 0.0273,
+      "step": 9640
+    },
+    {
+      "epoch": 63.48684210526316,
+      "grad_norm": 1.922224998474121,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 9650
+    },
+    {
+      "epoch": 63.55263157894737,
+      "grad_norm": 1.8191381692886353,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 9660
+    },
+    {
+      "epoch": 63.61842105263158,
+      "grad_norm": 1.686124324798584,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 9670
+    },
+    {
+      "epoch": 63.68421052631579,
+      "grad_norm": 1.717603325843811,
+      "learning_rate": 0.0001,
+      "loss": 0.0285,
+      "step": 9680
+    },
+    {
+      "epoch": 63.75,
+      "grad_norm": 1.855259656906128,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "step": 9690
+    },
+    {
+      "epoch": 63.81578947368421,
+      "grad_norm": 1.6836774349212646,
+      "learning_rate": 0.0001,
+      "loss": 0.0275,
+      "step": 9700
+    },
+    {
+      "epoch": 63.88157894736842,
+      "grad_norm": 1.776174545288086,
+      "learning_rate": 0.0001,
+      "loss": 0.0288,
+      "step": 9710
+    },
+    {
+      "epoch": 63.94736842105263,
+      "grad_norm": 1.868103265762329,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 9720
+    },
+    {
+      "epoch": 64.01315789473684,
+      "grad_norm": 1.6347148418426514,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 9730
+    },
+    {
+      "epoch": 64.07894736842105,
+      "grad_norm": 1.8645349740982056,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 9740
+    },
+    {
+      "epoch": 64.14473684210526,
+      "grad_norm": 2.106388807296753,
+      "learning_rate": 0.0001,
+      "loss": 0.0289,
+      "step": 9750
+    },
+    {
+      "epoch": 64.21052631578948,
+      "grad_norm": 1.4443974494934082,
+      "learning_rate": 0.0001,
+      "loss": 0.0289,
+      "step": 9760
+    },
+    {
+      "epoch": 64.27631578947368,
+      "grad_norm": 1.8115508556365967,
+      "learning_rate": 0.0001,
+      "loss": 0.0298,
+      "step": 9770
+    },
+    {
+      "epoch": 64.34210526315789,
+      "grad_norm": 1.6376737356185913,
+      "learning_rate": 0.0001,
+      "loss": 0.0275,
+      "step": 9780
+    },
+    {
+      "epoch": 64.40789473684211,
+      "grad_norm": 1.7943278551101685,
+      "learning_rate": 0.0001,
+      "loss": 0.0297,
+      "step": 9790
+    },
+    {
+      "epoch": 64.47368421052632,
+      "grad_norm": 1.6659159660339355,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "step": 9800
+    },
+    {
+      "epoch": 64.53947368421052,
+      "grad_norm": 1.792837381362915,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 9810
+    },
+    {
+      "epoch": 64.60526315789474,
+      "grad_norm": 1.9182575941085815,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 9820
+    },
+    {
+      "epoch": 64.67105263157895,
+      "grad_norm": 1.6181617975234985,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 9830
+    },
+    {
+      "epoch": 64.73684210526316,
+      "grad_norm": 1.5407286882400513,
+      "learning_rate": 0.0001,
+      "loss": 0.0298,
+      "step": 9840
+    },
+    {
+      "epoch": 64.80263157894737,
+      "grad_norm": 1.621099591255188,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 9850
+    },
+    {
+      "epoch": 64.86842105263158,
+      "grad_norm": 1.7339770793914795,
+      "learning_rate": 0.0001,
+      "loss": 0.0306,
+      "step": 9860
+    },
+    {
+      "epoch": 64.9342105263158,
+      "grad_norm": 1.9840919971466064,
+      "learning_rate": 0.0001,
+      "loss": 0.0291,
+      "step": 9870
+    },
+    {
+      "epoch": 65.0,
+      "grad_norm": 1.8245599269866943,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 9880
+    },
+    {
+      "epoch": 65.0657894736842,
+      "grad_norm": 1.6021732091903687,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 9890
+    },
+    {
+      "epoch": 65.13157894736842,
+      "grad_norm": 2.1346256732940674,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 9900
+    },
+    {
+      "epoch": 65.19736842105263,
+      "grad_norm": 1.537832498550415,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 9910
+    },
+    {
+      "epoch": 65.26315789473684,
+      "grad_norm": 1.70017671585083,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 9920
+    },
+    {
+      "epoch": 65.32894736842105,
+      "grad_norm": 2.058269500732422,
+      "learning_rate": 0.0001,
+      "loss": 0.0289,
+      "step": 9930
+    },
+    {
+      "epoch": 65.39473684210526,
+      "grad_norm": 1.592274785041809,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 9940
+    },
+    {
+      "epoch": 65.46052631578948,
+      "grad_norm": 1.6225742101669312,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 9950
+    },
+    {
+      "epoch": 65.52631578947368,
+      "grad_norm": 1.3477301597595215,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 9960
+    },
+    {
+      "epoch": 65.59210526315789,
+      "grad_norm": 1.6625304222106934,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 9970
+    },
+    {
+      "epoch": 65.65789473684211,
+      "grad_norm": 1.5278533697128296,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 9980
+    },
+    {
+      "epoch": 65.72368421052632,
+      "grad_norm": 1.9545159339904785,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 9990
+    },
+    {
+      "epoch": 65.78947368421052,
+      "grad_norm": 2.0630109310150146,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 10000
+    },
+    {
+      "epoch": 65.85526315789474,
+      "grad_norm": 2.0734527111053467,
+      "learning_rate": 0.0001,
+      "loss": 0.0304,
+      "step": 10010
+    },
+    {
+      "epoch": 65.92105263157895,
+      "grad_norm": 1.8130292892456055,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 10020
+    },
+    {
+      "epoch": 65.98684210526316,
+      "grad_norm": 1.9439656734466553,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 10030
+    },
+    {
+      "epoch": 66.05263157894737,
+      "grad_norm": 1.825518012046814,
+      "learning_rate": 0.0001,
+      "loss": 0.0306,
+      "step": 10040
+    },
+    {
+      "epoch": 66.11842105263158,
+      "grad_norm": 1.7734483480453491,
+      "learning_rate": 0.0001,
+      "loss": 0.0288,
+      "step": 10050
+    },
+    {
+      "epoch": 66.1842105263158,
+      "grad_norm": 1.8506808280944824,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 10060
+    },
+    {
+      "epoch": 66.25,
+      "grad_norm": 1.4367343187332153,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 10070
+    },
+    {
+      "epoch": 66.3157894736842,
+      "grad_norm": 2.0297582149505615,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 10080
+    },
+    {
+      "epoch": 66.38157894736842,
+      "grad_norm": 2.1654398441314697,
+      "learning_rate": 0.0001,
+      "loss": 0.0282,
+      "step": 10090
+    },
+    {
+      "epoch": 66.44736842105263,
+      "grad_norm": 1.8896561861038208,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "step": 10100
+    },
+    {
+      "epoch": 66.51315789473684,
+      "grad_norm": 2.0529367923736572,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 10110
+    },
+    {
+      "epoch": 66.57894736842105,
+      "grad_norm": 2.187924385070801,
+      "learning_rate": 0.0001,
+      "loss": 0.029,
+      "step": 10120
+    },
+    {
+      "epoch": 66.64473684210526,
+      "grad_norm": 1.9992048740386963,
+      "learning_rate": 0.0001,
+      "loss": 0.0277,
+      "step": 10130
+    },
+    {
+      "epoch": 66.71052631578948,
+      "grad_norm": 1.930832862854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0274,
+      "step": 10140
+    },
+    {
+      "epoch": 66.77631578947368,
+      "grad_norm": 2.112302780151367,
+      "learning_rate": 0.0001,
+      "loss": 0.0292,
+      "step": 10150
+    },
+    {
+      "epoch": 66.84210526315789,
+      "grad_norm": 2.1016273498535156,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 10160
+    },
+    {
+      "epoch": 66.90789473684211,
+      "grad_norm": 1.839139699935913,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 10170
+    },
+    {
+      "epoch": 66.97368421052632,
+      "grad_norm": 2.132181406021118,
+      "learning_rate": 0.0001,
+      "loss": 0.0264,
+      "step": 10180
+    },
+    {
+      "epoch": 67.03947368421052,
+      "grad_norm": 1.6180822849273682,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 10190
+    },
+    {
+      "epoch": 67.10526315789474,
+      "grad_norm": 2.0724172592163086,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "step": 10200
+    },
+    {
+      "epoch": 67.17105263157895,
+      "grad_norm": 1.807228446006775,
+      "learning_rate": 0.0001,
+      "loss": 0.0264,
+      "step": 10210
+    },
+    {
+      "epoch": 67.23684210526316,
+      "grad_norm": 1.818137764930725,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 10220
+    },
+    {
+      "epoch": 67.30263157894737,
+      "grad_norm": 1.9353524446487427,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 10230
+    },
+    {
+      "epoch": 67.36842105263158,
+      "grad_norm": 1.9978927373886108,
+      "learning_rate": 0.0001,
+      "loss": 0.0259,
+      "step": 10240
+    },
+    {
+      "epoch": 67.4342105263158,
+      "grad_norm": 1.6602277755737305,
+      "learning_rate": 0.0001,
+      "loss": 0.0287,
+      "step": 10250
+    },
+    {
+      "epoch": 67.5,
+      "grad_norm": 1.6295214891433716,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 10260
+    },
+    {
+      "epoch": 67.5657894736842,
+      "grad_norm": 1.5134141445159912,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 10270
+    },
+    {
+      "epoch": 67.63157894736842,
+      "grad_norm": 1.7062309980392456,
+      "learning_rate": 0.0001,
+      "loss": 0.0288,
+      "step": 10280
+    },
+    {
+      "epoch": 67.69736842105263,
+      "grad_norm": 2.065563678741455,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 10290
+    },
+    {
+      "epoch": 67.76315789473684,
+      "grad_norm": 1.7206897735595703,
+      "learning_rate": 0.0001,
+      "loss": 0.0304,
+      "step": 10300
+    },
+    {
+      "epoch": 67.82894736842105,
+      "grad_norm": 1.927570104598999,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 10310
+    },
+    {
+      "epoch": 67.89473684210526,
+      "grad_norm": 1.7813613414764404,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "step": 10320
+    },
+    {
+      "epoch": 67.96052631578948,
+      "grad_norm": 1.7587003707885742,
+      "learning_rate": 0.0001,
+      "loss": 0.0264,
+      "step": 10330
+    },
+    {
+      "epoch": 68.02631578947368,
+      "grad_norm": 1.5077179670333862,
+      "learning_rate": 0.0001,
+      "loss": 0.0282,
+      "step": 10340
+    },
+    {
+      "epoch": 68.09210526315789,
+      "grad_norm": 1.6387265920639038,
+      "learning_rate": 0.0001,
+      "loss": 0.0308,
+      "step": 10350
+    },
+    {
+      "epoch": 68.15789473684211,
+      "grad_norm": 1.3568955659866333,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "step": 10360
+    },
+    {
+      "epoch": 68.22368421052632,
+      "grad_norm": 1.4115484952926636,
+      "learning_rate": 0.0001,
+      "loss": 0.0267,
+      "step": 10370
+    },
+    {
+      "epoch": 68.28947368421052,
+      "grad_norm": 1.726250171661377,
+      "learning_rate": 0.0001,
+      "loss": 0.0264,
+      "step": 10380
+    },
+    {
+      "epoch": 68.35526315789474,
+      "grad_norm": 1.825256586074829,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 10390
+    },
+    {
+      "epoch": 68.42105263157895,
+      "grad_norm": 1.7078465223312378,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 10400
+    },
+    {
+      "epoch": 68.48684210526316,
+      "grad_norm": 1.6383626461029053,
+      "learning_rate": 0.0001,
+      "loss": 0.0277,
+      "step": 10410
+    },
+    {
+      "epoch": 68.55263157894737,
+      "grad_norm": 1.635048270225525,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 10420
+    },
+    {
+      "epoch": 68.61842105263158,
+      "grad_norm": 1.7052724361419678,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 10430
+    },
+    {
+      "epoch": 68.6842105263158,
+      "grad_norm": 1.6822491884231567,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "step": 10440
+    },
+    {
+      "epoch": 68.75,
+      "grad_norm": 1.7057929039001465,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 10450
+    },
+    {
+      "epoch": 68.8157894736842,
+      "grad_norm": 1.5500861406326294,
+      "learning_rate": 0.0001,
+      "loss": 0.0282,
+      "step": 10460
+    },
+    {
+      "epoch": 68.88157894736842,
+      "grad_norm": 1.7019840478897095,
+      "learning_rate": 0.0001,
+      "loss": 0.0296,
+      "step": 10470
+    },
+    {
+      "epoch": 68.94736842105263,
+      "grad_norm": 1.6525379419326782,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 10480
+    },
+    {
+      "epoch": 69.01315789473684,
+      "grad_norm": 2.0071771144866943,
+      "learning_rate": 0.0001,
+      "loss": 0.029,
+      "step": 10490
+    },
+    {
+      "epoch": 69.07894736842105,
+      "grad_norm": 2.170626640319824,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 10500
+    },
+    {
+      "epoch": 69.14473684210526,
+      "grad_norm": 1.7918614149093628,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 10510
+    },
+    {
+      "epoch": 69.21052631578948,
+      "grad_norm": 1.633731722831726,
+      "learning_rate": 0.0001,
+      "loss": 0.0277,
+      "step": 10520
+    },
+    {
+      "epoch": 69.27631578947368,
+      "grad_norm": 1.8682153224945068,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 10530
+    },
+    {
+      "epoch": 69.34210526315789,
+      "grad_norm": 1.8960853815078735,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "step": 10540
+    },
+    {
+      "epoch": 69.40789473684211,
+      "grad_norm": 1.7412015199661255,
+      "learning_rate": 0.0001,
+      "loss": 0.0267,
+      "step": 10550
+    },
+    {
+      "epoch": 69.47368421052632,
+      "grad_norm": 1.7792584896087646,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "step": 10560
+    },
+    {
+      "epoch": 69.53947368421052,
+      "grad_norm": 1.578399896621704,
+      "learning_rate": 0.0001,
+      "loss": 0.0274,
+      "step": 10570
+    },
+    {
+      "epoch": 69.60526315789474,
+      "grad_norm": 1.7477600574493408,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 10580
+    },
+    {
+      "epoch": 69.67105263157895,
+      "grad_norm": 1.8707166910171509,
+      "learning_rate": 0.0001,
+      "loss": 0.0264,
+      "step": 10590
+    },
+    {
+      "epoch": 69.73684210526316,
+      "grad_norm": 1.953047275543213,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 10600
+    },
+    {
+      "epoch": 69.80263157894737,
+      "grad_norm": 1.7453070878982544,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 10610
+    },
+    {
+      "epoch": 69.86842105263158,
+      "grad_norm": 1.3676609992980957,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 10620
+    },
+    {
+      "epoch": 69.9342105263158,
+      "grad_norm": 1.8007850646972656,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 10630
+    },
+    {
+      "epoch": 70.0,
+      "grad_norm": 1.5974959135055542,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 10640
+    },
+    {
+      "epoch": 70.0657894736842,
+      "grad_norm": 1.7148112058639526,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 10650
+    },
+    {
+      "epoch": 70.13157894736842,
+      "grad_norm": 1.8458138704299927,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 10660
+    },
+    {
+      "epoch": 70.19736842105263,
+      "grad_norm": 1.8393112421035767,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 10670
+    },
+    {
+      "epoch": 70.26315789473684,
+      "grad_norm": 2.173353433609009,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 10680
+    },
+    {
+      "epoch": 70.32894736842105,
+      "grad_norm": 1.9548739194869995,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 10690
+    },
+    {
+      "epoch": 70.39473684210526,
+      "grad_norm": 2.02528715133667,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 10700
+    },
+    {
+      "epoch": 70.46052631578948,
+      "grad_norm": 1.9165359735488892,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 10710
+    },
+    {
+      "epoch": 70.52631578947368,
+      "grad_norm": 1.8980177640914917,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 10720
+    },
+    {
+      "epoch": 70.59210526315789,
+      "grad_norm": 1.5036553144454956,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 10730
+    },
+    {
+      "epoch": 70.65789473684211,
+      "grad_norm": 1.9608731269836426,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "step": 10740
+    },
+    {
+      "epoch": 70.72368421052632,
+      "grad_norm": 1.665083885192871,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 10750
+    },
+    {
+      "epoch": 70.78947368421052,
+      "grad_norm": 1.6562914848327637,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 10760
+    },
+    {
+      "epoch": 70.85526315789474,
+      "grad_norm": 1.5882655382156372,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 10770
+    },
+    {
+      "epoch": 70.92105263157895,
+      "grad_norm": 1.7006151676177979,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 10780
+    },
+    {
+      "epoch": 70.98684210526316,
+      "grad_norm": 1.5319530963897705,
+      "learning_rate": 0.0001,
+      "loss": 0.0281,
+      "step": 10790
+    },
+    {
+      "epoch": 71.05263157894737,
+      "grad_norm": 1.9043638706207275,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 10800
+    },
+    {
+      "epoch": 71.11842105263158,
+      "grad_norm": 1.5125269889831543,
+      "learning_rate": 0.0001,
+      "loss": 0.0267,
+      "step": 10810
+    },
+    {
+      "epoch": 71.1842105263158,
+      "grad_norm": 1.8229715824127197,
+      "learning_rate": 0.0001,
+      "loss": 0.0299,
+      "step": 10820
+    },
+    {
+      "epoch": 71.25,
+      "grad_norm": 1.796139121055603,
+      "learning_rate": 0.0001,
+      "loss": 0.0264,
+      "step": 10830
+    },
+    {
+      "epoch": 71.3157894736842,
+      "grad_norm": 1.8449422121047974,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 10840
+    },
+    {
+      "epoch": 71.38157894736842,
+      "grad_norm": 1.7261402606964111,
+      "learning_rate": 0.0001,
+      "loss": 0.0287,
+      "step": 10850
+    },
+    {
+      "epoch": 71.44736842105263,
+      "grad_norm": 1.5883243083953857,
+      "learning_rate": 0.0001,
+      "loss": 0.0298,
+      "step": 10860
+    },
+    {
+      "epoch": 71.51315789473684,
+      "grad_norm": 1.3273059129714966,
+      "learning_rate": 0.0001,
+      "loss": 0.0277,
+      "step": 10870
+    },
+    {
+      "epoch": 71.57894736842105,
+      "grad_norm": 1.3199368715286255,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 10880
+    },
+    {
+      "epoch": 71.64473684210526,
+      "grad_norm": 1.3597865104675293,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 10890
+    },
+    {
+      "epoch": 71.71052631578948,
+      "grad_norm": 1.6409027576446533,
+      "learning_rate": 0.0001,
+      "loss": 0.0274,
+      "step": 10900
+    },
+    {
+      "epoch": 71.77631578947368,
+      "grad_norm": 1.7969907522201538,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 10910
+    },
+    {
+      "epoch": 71.84210526315789,
+      "grad_norm": 1.7779749631881714,
+      "learning_rate": 0.0001,
+      "loss": 0.0285,
+      "step": 10920
+    },
+    {
+      "epoch": 71.90789473684211,
+      "grad_norm": 1.8789730072021484,
+      "learning_rate": 0.0001,
+      "loss": 0.0281,
+      "step": 10930
+    },
+    {
+      "epoch": 71.97368421052632,
+      "grad_norm": 1.832434058189392,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 10940
+    },
+    {
+      "epoch": 72.03947368421052,
+      "grad_norm": 1.6766541004180908,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 10950
+    },
+    {
+      "epoch": 72.10526315789474,
+      "grad_norm": 1.7368793487548828,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 10960
+    },
+    {
+      "epoch": 72.17105263157895,
+      "grad_norm": 1.8270950317382812,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 10970
+    },
+    {
+      "epoch": 72.23684210526316,
+      "grad_norm": 1.5050655603408813,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 10980
+    },
+    {
+      "epoch": 72.30263157894737,
+      "grad_norm": 1.7033870220184326,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 10990
+    },
+    {
+      "epoch": 72.36842105263158,
+      "grad_norm": 1.559373378753662,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 11000
+    },
+    {
+      "epoch": 72.4342105263158,
+      "grad_norm": 1.8907803297042847,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "step": 11010
+    },
+    {
+      "epoch": 72.5,
+      "grad_norm": 1.5523408651351929,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 11020
+    },
+    {
+      "epoch": 72.5657894736842,
+      "grad_norm": 1.4263194799423218,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 11030
+    },
+    {
+      "epoch": 72.63157894736842,
+      "grad_norm": 1.5710784196853638,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 11040
+    },
+    {
+      "epoch": 72.69736842105263,
+      "grad_norm": 1.9038443565368652,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 11050
+    },
+    {
+      "epoch": 72.76315789473684,
+      "grad_norm": 1.9687570333480835,
+      "learning_rate": 0.0001,
+      "loss": 0.0271,
+      "step": 11060
+    },
+    {
+      "epoch": 72.82894736842105,
+      "grad_norm": 1.8713347911834717,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 11070
+    },
+    {
+      "epoch": 72.89473684210526,
+      "grad_norm": 1.827601671218872,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 11080
+    },
+    {
+      "epoch": 72.96052631578948,
+      "grad_norm": 1.7612308263778687,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 11090
+    },
+    {
+      "epoch": 73.02631578947368,
+      "grad_norm": 1.6651338338851929,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 11100
+    },
+    {
+      "epoch": 73.09210526315789,
+      "grad_norm": 1.7223801612854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 11110
+    },
+    {
+      "epoch": 73.15789473684211,
+      "grad_norm": 2.2473108768463135,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 11120
+    },
+    {
+      "epoch": 73.22368421052632,
+      "grad_norm": 1.6378623247146606,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 11130
+    },
+    {
+      "epoch": 73.28947368421052,
+      "grad_norm": 1.7509886026382446,
+      "learning_rate": 0.0001,
+      "loss": 0.0274,
+      "step": 11140
+    },
+    {
+      "epoch": 73.35526315789474,
+      "grad_norm": 1.7899543046951294,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 11150
+    },
+    {
+      "epoch": 73.42105263157895,
+      "grad_norm": 1.7126433849334717,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 11160
+    },
+    {
+      "epoch": 73.48684210526316,
+      "grad_norm": 1.624640941619873,
+      "learning_rate": 0.0001,
+      "loss": 0.0258,
+      "step": 11170
+    },
+    {
+      "epoch": 73.55263157894737,
+      "grad_norm": 1.6922892332077026,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 11180
+    },
+    {
+      "epoch": 73.61842105263158,
+      "grad_norm": 1.6018348932266235,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 11190
+    },
+    {
+      "epoch": 73.6842105263158,
+      "grad_norm": 1.7257733345031738,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 11200
+    },
+    {
+      "epoch": 73.75,
+      "grad_norm": 1.5125445127487183,
+      "learning_rate": 0.0001,
+      "loss": 0.0274,
+      "step": 11210
+    },
+    {
+      "epoch": 73.8157894736842,
+      "grad_norm": 2.309644937515259,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 11220
+    },
+    {
+      "epoch": 73.88157894736842,
+      "grad_norm": 2.2980353832244873,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 11230
+    },
+    {
+      "epoch": 73.94736842105263,
+      "grad_norm": 2.155663251876831,
+      "learning_rate": 0.0001,
+      "loss": 0.0271,
+      "step": 11240
+    },
+    {
+      "epoch": 74.01315789473684,
+      "grad_norm": 1.5754151344299316,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 11250
+    },
+    {
+      "epoch": 74.07894736842105,
+      "grad_norm": 1.6232088804244995,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 11260
+    },
+    {
+      "epoch": 74.14473684210526,
+      "grad_norm": 2.0059304237365723,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 11270
+    },
+    {
+      "epoch": 74.21052631578948,
+      "grad_norm": 2.166524648666382,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 11280
+    },
+    {
+      "epoch": 74.27631578947368,
+      "grad_norm": 1.8639678955078125,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "step": 11290
+    },
+    {
+      "epoch": 74.34210526315789,
+      "grad_norm": 2.540382146835327,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 11300
+    },
+    {
+      "epoch": 74.40789473684211,
+      "grad_norm": 1.8720356225967407,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 11310
+    },
+    {
+      "epoch": 74.47368421052632,
+      "grad_norm": 1.6244946718215942,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 11320
+    },
+    {
+      "epoch": 74.53947368421052,
+      "grad_norm": 1.4546319246292114,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 11330
+    },
+    {
+      "epoch": 74.60526315789474,
+      "grad_norm": 1.374340534210205,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 11340
+    },
+    {
+      "epoch": 74.67105263157895,
+      "grad_norm": 1.5473732948303223,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 11350
+    },
+    {
+      "epoch": 74.73684210526316,
+      "grad_norm": 1.646817922592163,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 11360
+    },
+    {
+      "epoch": 74.80263157894737,
+      "grad_norm": 1.6465305089950562,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 11370
+    },
+    {
+      "epoch": 74.86842105263158,
+      "grad_norm": 1.6561059951782227,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 11380
+    },
+    {
+      "epoch": 74.9342105263158,
+      "grad_norm": 1.7500685453414917,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "step": 11390
+    },
+    {
+      "epoch": 75.0,
+      "grad_norm": 1.3135260343551636,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 11400
+    },
+    {
+      "epoch": 75.0657894736842,
+      "grad_norm": 1.6338132619857788,
+      "learning_rate": 0.0001,
+      "loss": 0.0259,
+      "step": 11410
+    },
+    {
+      "epoch": 75.13157894736842,
+      "grad_norm": 1.5177695751190186,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 11420
+    },
+    {
+      "epoch": 75.19736842105263,
+      "grad_norm": 1.5215574502944946,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 11430
+    },
+    {
+      "epoch": 75.26315789473684,
+      "grad_norm": 1.4262157678604126,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 11440
+    },
+    {
+      "epoch": 75.32894736842105,
+      "grad_norm": 1.7465966939926147,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 11450
+    },
+    {
+      "epoch": 75.39473684210526,
+      "grad_norm": 1.7781189680099487,
+      "learning_rate": 0.0001,
+      "loss": 0.0258,
+      "step": 11460
+    },
+    {
+      "epoch": 75.46052631578948,
+      "grad_norm": 1.861718773841858,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 11470
+    },
+    {
+      "epoch": 75.52631578947368,
+      "grad_norm": 1.34523606300354,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "step": 11480
+    },
+    {
+      "epoch": 75.59210526315789,
+      "grad_norm": 1.6540141105651855,
+      "learning_rate": 0.0001,
+      "loss": 0.0277,
+      "step": 11490
+    },
+    {
+      "epoch": 75.65789473684211,
+      "grad_norm": 1.8050261735916138,
+      "learning_rate": 0.0001,
+      "loss": 0.0258,
+      "step": 11500
+    },
+    {
+      "epoch": 75.72368421052632,
+      "grad_norm": 1.8646867275238037,
+      "learning_rate": 0.0001,
+      "loss": 0.0285,
+      "step": 11510
+    },
+    {
+      "epoch": 75.78947368421052,
+      "grad_norm": 1.4578933715820312,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 11520
+    },
+    {
+      "epoch": 75.85526315789474,
+      "grad_norm": 1.5991803407669067,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 11530
+    },
+    {
+      "epoch": 75.92105263157895,
+      "grad_norm": 1.9933110475540161,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 11540
+    },
+    {
+      "epoch": 75.98684210526316,
+      "grad_norm": 1.5704941749572754,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 11550
+    },
+    {
+      "epoch": 76.05263157894737,
+      "grad_norm": 1.813393235206604,
+      "learning_rate": 0.0001,
+      "loss": 0.0259,
+      "step": 11560
+    },
+    {
+      "epoch": 76.11842105263158,
+      "grad_norm": 2.1795897483825684,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 11570
+    },
+    {
+      "epoch": 76.1842105263158,
+      "grad_norm": 1.719844102859497,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 11580
+    },
+    {
+      "epoch": 76.25,
+      "grad_norm": 1.964848518371582,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 11590
+    },
+    {
+      "epoch": 76.3157894736842,
+      "grad_norm": 2.0262434482574463,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "step": 11600
+    },
+    {
+      "epoch": 76.38157894736842,
+      "grad_norm": 2.0098633766174316,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 11610
+    },
+    {
+      "epoch": 76.44736842105263,
+      "grad_norm": 1.8454413414001465,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 11620
+    },
+    {
+      "epoch": 76.51315789473684,
+      "grad_norm": 1.8662750720977783,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 11630
+    },
+    {
+      "epoch": 76.57894736842105,
+      "grad_norm": 1.9370334148406982,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 11640
+    },
+    {
+      "epoch": 76.64473684210526,
+      "grad_norm": 1.6803852319717407,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 11650
+    },
+    {
+      "epoch": 76.71052631578948,
+      "grad_norm": 1.9592006206512451,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 11660
+    },
+    {
+      "epoch": 76.77631578947368,
+      "grad_norm": 1.9452179670333862,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 11670
+    },
+    {
+      "epoch": 76.84210526315789,
+      "grad_norm": 1.9002512693405151,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 11680
+    },
+    {
+      "epoch": 76.90789473684211,
+      "grad_norm": 1.9148516654968262,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 11690
+    },
+    {
+      "epoch": 76.97368421052632,
+      "grad_norm": 1.7390122413635254,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 11700
+    },
+    {
+      "epoch": 77.03947368421052,
+      "grad_norm": 1.5007272958755493,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 11710
+    },
+    {
+      "epoch": 77.10526315789474,
+      "grad_norm": 1.6768064498901367,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 11720
+    },
+    {
+      "epoch": 77.17105263157895,
+      "grad_norm": 1.8090136051177979,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 11730
+    },
+    {
+      "epoch": 77.23684210526316,
+      "grad_norm": 1.8841345310211182,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 11740
+    },
+    {
+      "epoch": 77.30263157894737,
+      "grad_norm": 1.282132863998413,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 11750
+    },
+    {
+      "epoch": 77.36842105263158,
+      "grad_norm": 1.8869783878326416,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 11760
+    },
+    {
+      "epoch": 77.4342105263158,
+      "grad_norm": 1.700897216796875,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 11770
+    },
+    {
+      "epoch": 77.5,
+      "grad_norm": 1.763547420501709,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 11780
+    },
+    {
+      "epoch": 77.5657894736842,
+      "grad_norm": 1.661258339881897,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 11790
+    },
+    {
+      "epoch": 77.63157894736842,
+      "grad_norm": 1.2386846542358398,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 11800
+    },
+    {
+      "epoch": 77.69736842105263,
+      "grad_norm": 1.4917902946472168,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 11810
+    },
+    {
+      "epoch": 77.76315789473684,
+      "grad_norm": 1.4909294843673706,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 11820
+    },
+    {
+      "epoch": 77.82894736842105,
+      "grad_norm": 1.6174029111862183,
+      "learning_rate": 0.0001,
+      "loss": 0.0282,
+      "step": 11830
+    },
+    {
+      "epoch": 77.89473684210526,
+      "grad_norm": 1.75030517578125,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 11840
+    },
+    {
+      "epoch": 77.96052631578948,
+      "grad_norm": 1.3512479066848755,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 11850
+    },
+    {
+      "epoch": 78.02631578947368,
+      "grad_norm": 1.5821315050125122,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 11860
+    },
+    {
+      "epoch": 78.09210526315789,
+      "grad_norm": 1.8378440141677856,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 11870
+    },
+    {
+      "epoch": 78.15789473684211,
+      "grad_norm": 1.536332130432129,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 11880
+    },
+    {
+      "epoch": 78.22368421052632,
+      "grad_norm": 1.5582396984100342,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 11890
+    },
+    {
+      "epoch": 78.28947368421052,
+      "grad_norm": 1.637697458267212,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 11900
+    },
+    {
+      "epoch": 78.35526315789474,
+      "grad_norm": 1.7529138326644897,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 11910
+    },
+    {
+      "epoch": 78.42105263157895,
+      "grad_norm": 1.9082077741622925,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 11920
+    },
+    {
+      "epoch": 78.48684210526316,
+      "grad_norm": 1.5302765369415283,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 11930
+    },
+    {
+      "epoch": 78.55263157894737,
+      "grad_norm": 1.9385207891464233,
+      "learning_rate": 0.0001,
+      "loss": 0.0298,
+      "step": 11940
+    },
+    {
+      "epoch": 78.61842105263158,
+      "grad_norm": 1.6783729791641235,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 11950
+    },
+    {
+      "epoch": 78.6842105263158,
+      "grad_norm": 1.7037731409072876,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 11960
+    },
+    {
+      "epoch": 78.75,
+      "grad_norm": 1.3808066844940186,
+      "learning_rate": 0.0001,
+      "loss": 0.0273,
+      "step": 11970
+    },
+    {
+      "epoch": 78.8157894736842,
+      "grad_norm": 1.4401555061340332,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 11980
+    },
+    {
+      "epoch": 78.88157894736842,
+      "grad_norm": 1.8740235567092896,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 11990
+    },
+    {
+      "epoch": 78.94736842105263,
+      "grad_norm": 1.5552549362182617,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 12000
+    },
+    {
+      "epoch": 79.01315789473684,
+      "grad_norm": 1.5589462518692017,
+      "learning_rate": 0.0001,
+      "loss": 0.0258,
+      "step": 12010
+    },
+    {
+      "epoch": 79.07894736842105,
+      "grad_norm": 1.641662359237671,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 12020
+    },
+    {
+      "epoch": 79.14473684210526,
+      "grad_norm": 1.6106501817703247,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 12030
+    },
+    {
+      "epoch": 79.21052631578948,
+      "grad_norm": 1.9273698329925537,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 12040
+    },
+    {
+      "epoch": 79.27631578947368,
+      "grad_norm": 1.7661243677139282,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 12050
+    },
+    {
+      "epoch": 79.34210526315789,
+      "grad_norm": 1.7702202796936035,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 12060
+    },
+    {
+      "epoch": 79.40789473684211,
+      "grad_norm": 1.2317006587982178,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 12070
+    },
+    {
+      "epoch": 79.47368421052632,
+      "grad_norm": 1.6304638385772705,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 12080
+    },
+    {
+      "epoch": 79.53947368421052,
+      "grad_norm": 1.764953851699829,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 12090
+    },
+    {
+      "epoch": 79.60526315789474,
+      "grad_norm": 1.6987632513046265,
+      "learning_rate": 0.0001,
+      "loss": 0.0281,
+      "step": 12100
+    },
+    {
+      "epoch": 79.67105263157895,
+      "grad_norm": 1.7687889337539673,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 12110
+    },
+    {
+      "epoch": 79.73684210526316,
+      "grad_norm": 1.714768648147583,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 12120
+    },
+    {
+      "epoch": 79.80263157894737,
+      "grad_norm": 1.5168180465698242,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 12130
+    },
+    {
+      "epoch": 79.86842105263158,
+      "grad_norm": 1.6321282386779785,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 12140
+    },
+    {
+      "epoch": 79.9342105263158,
+      "grad_norm": 1.8023918867111206,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 12150
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 1.475740909576416,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 12160
+    },
+    {
+      "epoch": 80.0657894736842,
+      "grad_norm": 1.8622595071792603,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 12170
+    },
+    {
+      "epoch": 80.13157894736842,
+      "grad_norm": 1.6295769214630127,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 12180
+    },
+    {
+      "epoch": 80.19736842105263,
+      "grad_norm": 1.6116666793823242,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 12190
+    },
+    {
+      "epoch": 80.26315789473684,
+      "grad_norm": 1.599142074584961,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 12200
+    },
+    {
+      "epoch": 80.32894736842105,
+      "grad_norm": 1.5806418657302856,
+      "learning_rate": 0.0001,
+      "loss": 0.0259,
+      "step": 12210
+    },
+    {
+      "epoch": 80.39473684210526,
+      "grad_norm": 1.4835152626037598,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "step": 12220
+    },
+    {
+      "epoch": 80.46052631578948,
+      "grad_norm": 1.2303491830825806,
+      "learning_rate": 0.0001,
+      "loss": 0.0274,
+      "step": 12230
+    },
+    {
+      "epoch": 80.52631578947368,
+      "grad_norm": 1.4486466646194458,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 12240
+    },
+    {
+      "epoch": 80.59210526315789,
+      "grad_norm": 1.5165975093841553,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 12250
+    },
+    {
+      "epoch": 80.65789473684211,
+      "grad_norm": 1.4321902990341187,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 12260
+    },
+    {
+      "epoch": 80.72368421052632,
+      "grad_norm": 1.3290172815322876,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 12270
+    },
+    {
+      "epoch": 80.78947368421052,
+      "grad_norm": 1.4528899192810059,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 12280
+    },
+    {
+      "epoch": 80.85526315789474,
+      "grad_norm": 1.4954471588134766,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 12290
+    },
+    {
+      "epoch": 80.92105263157895,
+      "grad_norm": 1.686185359954834,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 12300
+    },
+    {
+      "epoch": 80.98684210526316,
+      "grad_norm": 1.8050355911254883,
+      "learning_rate": 0.0001,
+      "loss": 0.0273,
+      "step": 12310
+    },
+    {
+      "epoch": 81.05263157894737,
+      "grad_norm": 1.667278528213501,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 12320
+    },
+    {
+      "epoch": 81.11842105263158,
+      "grad_norm": 1.8313591480255127,
+      "learning_rate": 0.0001,
+      "loss": 0.0275,
+      "step": 12330
+    },
+    {
+      "epoch": 81.1842105263158,
+      "grad_norm": 1.5865910053253174,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 12340
+    },
+    {
+      "epoch": 81.25,
+      "grad_norm": 1.4617630243301392,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 12350
+    },
+    {
+      "epoch": 81.3157894736842,
+      "grad_norm": 1.6927311420440674,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 12360
+    },
+    {
+      "epoch": 81.38157894736842,
+      "grad_norm": 1.7742199897766113,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 12370
+    },
+    {
+      "epoch": 81.44736842105263,
+      "grad_norm": 1.5240896940231323,
+      "learning_rate": 0.0001,
+      "loss": 0.0273,
+      "step": 12380
+    },
+    {
+      "epoch": 81.51315789473684,
+      "grad_norm": 1.980691909790039,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 12390
+    },
+    {
+      "epoch": 81.57894736842105,
+      "grad_norm": 1.9031286239624023,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 12400
+    },
+    {
+      "epoch": 81.64473684210526,
+      "grad_norm": 1.8978111743927002,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 12410
+    },
+    {
+      "epoch": 81.71052631578948,
+      "grad_norm": 2.1107935905456543,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 12420
+    },
+    {
+      "epoch": 81.77631578947368,
+      "grad_norm": 1.7692242860794067,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 12430
+    },
+    {
+      "epoch": 81.84210526315789,
+      "grad_norm": 1.9986506700515747,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 12440
+    },
+    {
+      "epoch": 81.90789473684211,
+      "grad_norm": 2.076988935470581,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 12450
+    },
+    {
+      "epoch": 81.97368421052632,
+      "grad_norm": 1.7011905908584595,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 12460
+    },
+    {
+      "epoch": 82.03947368421052,
+      "grad_norm": 1.8305109739303589,
+      "learning_rate": 0.0001,
+      "loss": 0.0264,
+      "step": 12470
+    },
+    {
+      "epoch": 82.10526315789474,
+      "grad_norm": 1.6795586347579956,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 12480
+    },
+    {
+      "epoch": 82.17105263157895,
+      "grad_norm": 1.4467089176177979,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 12490
+    },
+    {
+      "epoch": 82.23684210526316,
+      "grad_norm": 1.618145227432251,
+      "learning_rate": 0.0001,
+      "loss": 0.0258,
+      "step": 12500
+    },
+    {
+      "epoch": 82.30263157894737,
+      "grad_norm": 1.5068567991256714,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 12510
+    },
+    {
+      "epoch": 82.36842105263158,
+      "grad_norm": 1.7931199073791504,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 12520
+    },
+    {
+      "epoch": 82.4342105263158,
+      "grad_norm": 1.7545214891433716,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 12530
+    },
+    {
+      "epoch": 82.5,
+      "grad_norm": 1.677007794380188,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 12540
+    },
+    {
+      "epoch": 82.5657894736842,
+      "grad_norm": 1.8432847261428833,
+      "learning_rate": 0.0001,
+      "loss": 0.0275,
+      "step": 12550
+    },
+    {
+      "epoch": 82.63157894736842,
+      "grad_norm": 1.6245819330215454,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 12560
+    },
+    {
+      "epoch": 82.69736842105263,
+      "grad_norm": 1.837830901145935,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 12570
+    },
+    {
+      "epoch": 82.76315789473684,
+      "grad_norm": 1.7958941459655762,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 12580
+    },
+    {
+      "epoch": 82.82894736842105,
+      "grad_norm": 1.5123578310012817,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 12590
+    },
+    {
+      "epoch": 82.89473684210526,
+      "grad_norm": 1.5450695753097534,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 12600
+    },
+    {
+      "epoch": 82.96052631578948,
+      "grad_norm": 1.3524283170700073,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 12610
+    },
+    {
+      "epoch": 83.02631578947368,
+      "grad_norm": 1.588718295097351,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 12620
+    },
+    {
+      "epoch": 83.09210526315789,
+      "grad_norm": 1.4721720218658447,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 12630
+    },
+    {
+      "epoch": 83.15789473684211,
+      "grad_norm": 1.411702036857605,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 12640
+    },
+    {
+      "epoch": 83.22368421052632,
+      "grad_norm": 1.5033427476882935,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 12650
+    },
+    {
+      "epoch": 83.28947368421052,
+      "grad_norm": 1.7150914669036865,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 12660
+    },
+    {
+      "epoch": 83.35526315789474,
+      "grad_norm": 1.4532182216644287,
+      "learning_rate": 0.0001,
+      "loss": 0.0275,
+      "step": 12670
+    },
+    {
+      "epoch": 83.42105263157895,
+      "grad_norm": 1.678092360496521,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 12680
+    },
+    {
+      "epoch": 83.48684210526316,
+      "grad_norm": 1.542473554611206,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "step": 12690
+    },
+    {
+      "epoch": 83.55263157894737,
+      "grad_norm": 1.6771373748779297,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "step": 12700
+    },
+    {
+      "epoch": 83.61842105263158,
+      "grad_norm": 1.5920690298080444,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 12710
+    },
+    {
+      "epoch": 83.6842105263158,
+      "grad_norm": 1.29483962059021,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 12720
+    },
+    {
+      "epoch": 83.75,
+      "grad_norm": 1.575247883796692,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 12730
+    },
+    {
+      "epoch": 83.8157894736842,
+      "grad_norm": 1.4278053045272827,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 12740
+    },
+    {
+      "epoch": 83.88157894736842,
+      "grad_norm": 1.0531659126281738,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 12750
+    },
+    {
+      "epoch": 83.94736842105263,
+      "grad_norm": 1.2762305736541748,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 12760
+    },
+    {
+      "epoch": 84.01315789473684,
+      "grad_norm": 1.573049545288086,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 12770
+    },
+    {
+      "epoch": 84.07894736842105,
+      "grad_norm": 1.2428479194641113,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 12780
+    },
+    {
+      "epoch": 84.14473684210526,
+      "grad_norm": 1.6974382400512695,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 12790
+    },
+    {
+      "epoch": 84.21052631578948,
+      "grad_norm": 1.1859573125839233,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 12800
+    },
+    {
+      "epoch": 84.27631578947368,
+      "grad_norm": 1.5613579750061035,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 12810
+    },
+    {
+      "epoch": 84.34210526315789,
+      "grad_norm": 1.359305739402771,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 12820
+    },
+    {
+      "epoch": 84.40789473684211,
+      "grad_norm": 1.6808149814605713,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 12830
+    },
+    {
+      "epoch": 84.47368421052632,
+      "grad_norm": 1.8556708097457886,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 12840
+    },
+    {
+      "epoch": 84.53947368421052,
+      "grad_norm": 1.8008233308792114,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 12850
+    },
+    {
+      "epoch": 84.60526315789474,
+      "grad_norm": 1.5086830854415894,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 12860
+    },
+    {
+      "epoch": 84.67105263157895,
+      "grad_norm": 1.6331901550292969,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 12870
+    },
+    {
+      "epoch": 84.73684210526316,
+      "grad_norm": 1.1920050382614136,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 12880
+    },
+    {
+      "epoch": 84.80263157894737,
+      "grad_norm": 1.910048484802246,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 12890
+    },
+    {
+      "epoch": 84.86842105263158,
+      "grad_norm": 1.4449541568756104,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 12900
+    },
+    {
+      "epoch": 84.9342105263158,
+      "grad_norm": 1.3787635564804077,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 12910
+    },
+    {
+      "epoch": 85.0,
+      "grad_norm": 1.3375296592712402,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 12920
+    },
+    {
+      "epoch": 85.0657894736842,
+      "grad_norm": 1.44197416305542,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 12930
+    },
+    {
+      "epoch": 85.13157894736842,
+      "grad_norm": 1.2884734869003296,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 12940
+    },
+    {
+      "epoch": 85.19736842105263,
+      "grad_norm": 1.1492596864700317,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 12950
+    },
+    {
+      "epoch": 85.26315789473684,
+      "grad_norm": 1.6913357973098755,
+      "learning_rate": 0.0001,
+      "loss": 0.0278,
+      "step": 12960
+    },
+    {
+      "epoch": 85.32894736842105,
+      "grad_norm": 1.591509461402893,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 12970
+    },
+    {
+      "epoch": 85.39473684210526,
+      "grad_norm": 1.9279398918151855,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 12980
+    },
+    {
+      "epoch": 85.46052631578948,
+      "grad_norm": 1.8467780351638794,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 12990
+    },
+    {
+      "epoch": 85.52631578947368,
+      "grad_norm": 1.6500190496444702,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 13000
+    },
+    {
+      "epoch": 85.59210526315789,
+      "grad_norm": 2.0222458839416504,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 13010
+    },
+    {
+      "epoch": 85.65789473684211,
+      "grad_norm": 1.445722222328186,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 13020
+    },
+    {
+      "epoch": 85.72368421052632,
+      "grad_norm": 1.3406612873077393,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 13030
+    },
+    {
+      "epoch": 85.78947368421052,
+      "grad_norm": 1.2125831842422485,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 13040
+    },
+    {
+      "epoch": 85.85526315789474,
+      "grad_norm": 1.3329436779022217,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 13050
+    },
+    {
+      "epoch": 85.92105263157895,
+      "grad_norm": 1.7026501893997192,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 13060
+    },
+    {
+      "epoch": 85.98684210526316,
+      "grad_norm": 1.6232529878616333,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 13070
+    },
+    {
+      "epoch": 86.05263157894737,
+      "grad_norm": 1.4894771575927734,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 13080
+    },
+    {
+      "epoch": 86.11842105263158,
+      "grad_norm": 1.3056237697601318,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 13090
+    },
+    {
+      "epoch": 86.1842105263158,
+      "grad_norm": 1.318498134613037,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 13100
+    },
+    {
+      "epoch": 86.25,
+      "grad_norm": 1.4497151374816895,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 13110
+    },
+    {
+      "epoch": 86.3157894736842,
+      "grad_norm": 1.6557412147521973,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 13120
+    },
+    {
+      "epoch": 86.38157894736842,
+      "grad_norm": 1.569457769393921,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 13130
+    },
+    {
+      "epoch": 86.44736842105263,
+      "grad_norm": 1.7060179710388184,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 13140
+    },
+    {
+      "epoch": 86.51315789473684,
+      "grad_norm": 1.4572563171386719,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 13150
+    },
+    {
+      "epoch": 86.57894736842105,
+      "grad_norm": 1.4536716938018799,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 13160
+    },
+    {
+      "epoch": 86.64473684210526,
+      "grad_norm": 1.6783726215362549,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 13170
+    },
+    {
+      "epoch": 86.71052631578948,
+      "grad_norm": 1.4602693319320679,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 13180
+    },
+    {
+      "epoch": 86.77631578947368,
+      "grad_norm": 1.6970704793930054,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 13190
+    },
+    {
+      "epoch": 86.84210526315789,
+      "grad_norm": 1.3816075325012207,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 13200
+    },
+    {
+      "epoch": 86.90789473684211,
+      "grad_norm": 1.6789470911026,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 13210
+    },
+    {
+      "epoch": 86.97368421052632,
+      "grad_norm": 1.5384844541549683,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 13220
+    },
+    {
+      "epoch": 87.03947368421052,
+      "grad_norm": 1.5757942199707031,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 13230
+    },
+    {
+      "epoch": 87.10526315789474,
+      "grad_norm": 1.7311173677444458,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 13240
+    },
+    {
+      "epoch": 87.17105263157895,
+      "grad_norm": 1.5452516078948975,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 13250
+    },
+    {
+      "epoch": 87.23684210526316,
+      "grad_norm": 1.495526909828186,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 13260
+    },
+    {
+      "epoch": 87.30263157894737,
+      "grad_norm": 1.5925183296203613,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 13270
+    },
+    {
+      "epoch": 87.36842105263158,
+      "grad_norm": 1.58163321018219,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 13280
+    },
+    {
+      "epoch": 87.4342105263158,
+      "grad_norm": 1.8590866327285767,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 13290
+    },
+    {
+      "epoch": 87.5,
+      "grad_norm": 1.5774991512298584,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 13300
+    },
+    {
+      "epoch": 87.5657894736842,
+      "grad_norm": 1.7061364650726318,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 13310
+    },
+    {
+      "epoch": 87.63157894736842,
+      "grad_norm": 1.4900331497192383,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 13320
+    },
+    {
+      "epoch": 87.69736842105263,
+      "grad_norm": 1.8623180389404297,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 13330
+    },
+    {
+      "epoch": 87.76315789473684,
+      "grad_norm": 1.6893125772476196,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 13340
+    },
+    {
+      "epoch": 87.82894736842105,
+      "grad_norm": 1.8551734685897827,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 13350
+    },
+    {
+      "epoch": 87.89473684210526,
+      "grad_norm": 1.7966662645339966,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 13360
+    },
+    {
+      "epoch": 87.96052631578948,
+      "grad_norm": 1.7751268148422241,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 13370
+    },
+    {
+      "epoch": 88.02631578947368,
+      "grad_norm": 1.7870049476623535,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 13380
+    },
+    {
+      "epoch": 88.09210526315789,
+      "grad_norm": 1.6607742309570312,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 13390
+    },
+    {
+      "epoch": 88.15789473684211,
+      "grad_norm": 1.734802007675171,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 13400
+    },
+    {
+      "epoch": 88.22368421052632,
+      "grad_norm": 2.1333210468292236,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 13410
+    },
+    {
+      "epoch": 88.28947368421052,
+      "grad_norm": 1.473213791847229,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 13420
+    },
+    {
+      "epoch": 88.35526315789474,
+      "grad_norm": 1.6745879650115967,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 13430
+    },
+    {
+      "epoch": 88.42105263157895,
+      "grad_norm": 1.3163137435913086,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 13440
+    },
+    {
+      "epoch": 88.48684210526316,
+      "grad_norm": 1.573905110359192,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 13450
+    },
+    {
+      "epoch": 88.55263157894737,
+      "grad_norm": 1.672709345817566,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 13460
+    },
+    {
+      "epoch": 88.61842105263158,
+      "grad_norm": 1.473910927772522,
+      "learning_rate": 0.0001,
+      "loss": 0.0262,
+      "step": 13470
+    },
+    {
+      "epoch": 88.6842105263158,
+      "grad_norm": 1.4131242036819458,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 13480
+    },
+    {
+      "epoch": 88.75,
+      "grad_norm": 1.6035963296890259,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 13490
+    },
+    {
+      "epoch": 88.8157894736842,
+      "grad_norm": 1.489512324333191,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 13500
+    },
+    {
+      "epoch": 88.88157894736842,
+      "grad_norm": 1.45292329788208,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 13510
+    },
+    {
+      "epoch": 88.94736842105263,
+      "grad_norm": 1.6451809406280518,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 13520
+    },
+    {
+      "epoch": 89.01315789473684,
+      "grad_norm": 1.826170563697815,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 13530
+    },
+    {
+      "epoch": 89.07894736842105,
+      "grad_norm": 1.5260390043258667,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 13540
+    },
+    {
+      "epoch": 89.14473684210526,
+      "grad_norm": 1.5344808101654053,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 13550
+    },
+    {
+      "epoch": 89.21052631578948,
+      "grad_norm": 1.3127132654190063,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 13560
+    },
+    {
+      "epoch": 89.27631578947368,
+      "grad_norm": 1.71023428440094,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 13570
+    },
+    {
+      "epoch": 89.34210526315789,
+      "grad_norm": 1.7097973823547363,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 13580
+    },
+    {
+      "epoch": 89.40789473684211,
+      "grad_norm": 1.8148266077041626,
+      "learning_rate": 0.0001,
+      "loss": 0.0258,
+      "step": 13590
+    },
+    {
+      "epoch": 89.47368421052632,
+      "grad_norm": 1.3944507837295532,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 13600
+    },
+    {
+      "epoch": 89.53947368421052,
+      "grad_norm": 1.421279788017273,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 13610
+    },
+    {
+      "epoch": 89.60526315789474,
+      "grad_norm": 1.6334819793701172,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 13620
+    },
+    {
+      "epoch": 89.67105263157895,
+      "grad_norm": 1.9979926347732544,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 13630
+    },
+    {
+      "epoch": 89.73684210526316,
+      "grad_norm": 1.5678966045379639,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 13640
+    },
+    {
+      "epoch": 89.80263157894737,
+      "grad_norm": 1.6031328439712524,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 13650
+    },
+    {
+      "epoch": 89.86842105263158,
+      "grad_norm": 1.8089882135391235,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 13660
+    },
+    {
+      "epoch": 89.9342105263158,
+      "grad_norm": 1.2982338666915894,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 13670
+    },
+    {
+      "epoch": 90.0,
+      "grad_norm": 1.3500252962112427,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 13680
+    },
+    {
+      "epoch": 90.0657894736842,
+      "grad_norm": 1.7127699851989746,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 13690
+    },
+    {
+      "epoch": 90.13157894736842,
+      "grad_norm": 1.1409753561019897,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 13700
+    },
+    {
+      "epoch": 90.19736842105263,
+      "grad_norm": 1.4835864305496216,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "step": 13710
+    },
+    {
+      "epoch": 90.26315789473684,
+      "grad_norm": 1.676373839378357,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 13720
+    },
+    {
+      "epoch": 90.32894736842105,
+      "grad_norm": 1.247381329536438,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 13730
+    },
+    {
+      "epoch": 90.39473684210526,
+      "grad_norm": 1.1744779348373413,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 13740
+    },
+    {
+      "epoch": 90.46052631578948,
+      "grad_norm": 1.8650901317596436,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 13750
+    },
+    {
+      "epoch": 90.52631578947368,
+      "grad_norm": 1.7591873407363892,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 13760
+    },
+    {
+      "epoch": 90.59210526315789,
+      "grad_norm": 1.6825592517852783,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 13770
+    },
+    {
+      "epoch": 90.65789473684211,
+      "grad_norm": 1.285715937614441,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 13780
+    },
+    {
+      "epoch": 90.72368421052632,
+      "grad_norm": 1.576466679573059,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 13790
+    },
+    {
+      "epoch": 90.78947368421052,
+      "grad_norm": 1.6532832384109497,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 13800
+    },
+    {
+      "epoch": 90.85526315789474,
+      "grad_norm": 1.4761638641357422,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 13810
+    },
+    {
+      "epoch": 90.92105263157895,
+      "grad_norm": 1.490132451057434,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 13820
+    },
+    {
+      "epoch": 90.98684210526316,
+      "grad_norm": 1.9364569187164307,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 13830
+    },
+    {
+      "epoch": 91.05263157894737,
+      "grad_norm": 1.6803781986236572,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 13840
+    },
+    {
+      "epoch": 91.11842105263158,
+      "grad_norm": 1.5090696811676025,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 13850
+    },
+    {
+      "epoch": 91.1842105263158,
+      "grad_norm": 1.7960859537124634,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 13860
+    },
+    {
+      "epoch": 91.25,
+      "grad_norm": 1.7104315757751465,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 13870
+    },
+    {
+      "epoch": 91.3157894736842,
+      "grad_norm": 1.793373703956604,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 13880
+    },
+    {
+      "epoch": 91.38157894736842,
+      "grad_norm": 2.0539634227752686,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 13890
+    },
+    {
+      "epoch": 91.44736842105263,
+      "grad_norm": 2.1168859004974365,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 13900
+    },
+    {
+      "epoch": 91.51315789473684,
+      "grad_norm": 1.8195390701293945,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 13910
+    },
+    {
+      "epoch": 91.57894736842105,
+      "grad_norm": 1.714077353477478,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 13920
+    },
+    {
+      "epoch": 91.64473684210526,
+      "grad_norm": 1.7049275636672974,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 13930
+    },
+    {
+      "epoch": 91.71052631578948,
+      "grad_norm": 2.158938407897949,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 13940
+    },
+    {
+      "epoch": 91.77631578947368,
+      "grad_norm": 2.1923866271972656,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 13950
+    },
+    {
+      "epoch": 91.84210526315789,
+      "grad_norm": 1.7512110471725464,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 13960
+    },
+    {
+      "epoch": 91.90789473684211,
+      "grad_norm": 1.8669230937957764,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 13970
+    },
+    {
+      "epoch": 91.97368421052632,
+      "grad_norm": 1.6305128335952759,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 13980
+    },
+    {
+      "epoch": 92.03947368421052,
+      "grad_norm": 1.5910475254058838,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 13990
+    },
+    {
+      "epoch": 92.10526315789474,
+      "grad_norm": 1.5273504257202148,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 14000
+    },
+    {
+      "epoch": 92.17105263157895,
+      "grad_norm": 1.7885582447052002,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 14010
+    },
+    {
+      "epoch": 92.23684210526316,
+      "grad_norm": 2.1910252571105957,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 14020
+    },
+    {
+      "epoch": 92.30263157894737,
+      "grad_norm": 1.755102515220642,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 14030
+    },
+    {
+      "epoch": 92.36842105263158,
+      "grad_norm": 1.6435348987579346,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 14040
+    },
+    {
+      "epoch": 92.4342105263158,
+      "grad_norm": 2.323174238204956,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 14050
+    },
+    {
+      "epoch": 92.5,
+      "grad_norm": 3.1456828117370605,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 14060
+    },
+    {
+      "epoch": 92.5657894736842,
+      "grad_norm": 2.456490993499756,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 14070
+    },
+    {
+      "epoch": 92.63157894736842,
+      "grad_norm": 1.9856231212615967,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 14080
+    },
+    {
+      "epoch": 92.69736842105263,
+      "grad_norm": 1.8175798654556274,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 14090
+    },
+    {
+      "epoch": 92.76315789473684,
+      "grad_norm": 1.905171513557434,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 14100
+    },
+    {
+      "epoch": 92.82894736842105,
+      "grad_norm": 1.7334188222885132,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 14110
+    },
+    {
+      "epoch": 92.89473684210526,
+      "grad_norm": 1.6266708374023438,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 14120
+    },
+    {
+      "epoch": 92.96052631578948,
+      "grad_norm": 1.4972394704818726,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 14130
+    },
+    {
+      "epoch": 93.02631578947368,
+      "grad_norm": 1.655664086341858,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 14140
+    },
+    {
+      "epoch": 93.09210526315789,
+      "grad_norm": 1.6464513540267944,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 14150
+    },
+    {
+      "epoch": 93.15789473684211,
+      "grad_norm": 1.4033807516098022,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 14160
+    },
+    {
+      "epoch": 93.22368421052632,
+      "grad_norm": 1.5057168006896973,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 14170
+    },
+    {
+      "epoch": 93.28947368421052,
+      "grad_norm": 1.735034465789795,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 14180
+    },
+    {
+      "epoch": 93.35526315789474,
+      "grad_norm": 1.6805418729782104,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 14190
+    },
+    {
+      "epoch": 93.42105263157895,
+      "grad_norm": 1.3809490203857422,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 14200
+    },
+    {
+      "epoch": 93.48684210526316,
+      "grad_norm": 1.5541129112243652,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 14210
+    },
+    {
+      "epoch": 93.55263157894737,
+      "grad_norm": 1.6721612215042114,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 14220
+    },
+    {
+      "epoch": 93.61842105263158,
+      "grad_norm": 1.4749832153320312,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 14230
+    },
+    {
+      "epoch": 93.6842105263158,
+      "grad_norm": 1.229355812072754,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 14240
+    },
+    {
+      "epoch": 93.75,
+      "grad_norm": 1.3654001951217651,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 14250
+    },
+    {
+      "epoch": 93.8157894736842,
+      "grad_norm": 1.3455557823181152,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 14260
+    },
+    {
+      "epoch": 93.88157894736842,
+      "grad_norm": 1.2095595598220825,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 14270
+    },
+    {
+      "epoch": 93.94736842105263,
+      "grad_norm": 1.2949072122573853,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 14280
+    },
+    {
+      "epoch": 94.01315789473684,
+      "grad_norm": 1.7537165880203247,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 14290
+    },
+    {
+      "epoch": 94.07894736842105,
+      "grad_norm": 1.4385466575622559,
+      "learning_rate": 0.0001,
+      "loss": 0.0282,
+      "step": 14300
+    },
+    {
+      "epoch": 94.14473684210526,
+      "grad_norm": 1.1989967823028564,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 14310
+    },
+    {
+      "epoch": 94.21052631578948,
+      "grad_norm": 1.478391408920288,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 14320
+    },
+    {
+      "epoch": 94.27631578947368,
+      "grad_norm": 1.5552846193313599,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 14330
+    },
+    {
+      "epoch": 94.34210526315789,
+      "grad_norm": 1.5184147357940674,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 14340
+    },
+    {
+      "epoch": 94.40789473684211,
+      "grad_norm": 1.4583368301391602,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 14350
+    },
+    {
+      "epoch": 94.47368421052632,
+      "grad_norm": 1.5319230556488037,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 14360
+    },
+    {
+      "epoch": 94.53947368421052,
+      "grad_norm": 1.505807876586914,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 14370
+    },
+    {
+      "epoch": 94.60526315789474,
+      "grad_norm": 1.2531390190124512,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 14380
+    },
+    {
+      "epoch": 94.67105263157895,
+      "grad_norm": 1.9696604013442993,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 14390
+    },
+    {
+      "epoch": 94.73684210526316,
+      "grad_norm": 1.7615699768066406,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 14400
+    },
+    {
+      "epoch": 94.80263157894737,
+      "grad_norm": 1.9439195394515991,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "step": 14410
+    },
+    {
+      "epoch": 94.86842105263158,
+      "grad_norm": 2.943202018737793,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 14420
+    },
+    {
+      "epoch": 94.9342105263158,
+      "grad_norm": 1.8300869464874268,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 14430
+    },
+    {
+      "epoch": 95.0,
+      "grad_norm": 2.4264817237854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 14440
+    },
+    {
+      "epoch": 95.0657894736842,
+      "grad_norm": 2.130911350250244,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 14450
+    },
+    {
+      "epoch": 95.13157894736842,
+      "grad_norm": 2.132805109024048,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 14460
+    },
+    {
+      "epoch": 95.19736842105263,
+      "grad_norm": 2.0448882579803467,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 14470
+    },
+    {
+      "epoch": 95.26315789473684,
+      "grad_norm": 1.6117432117462158,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 14480
+    },
+    {
+      "epoch": 95.32894736842105,
+      "grad_norm": 1.5426193475723267,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 14490
+    },
+    {
+      "epoch": 95.39473684210526,
+      "grad_norm": 1.6238819360733032,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 14500
+    },
+    {
+      "epoch": 95.46052631578948,
+      "grad_norm": 1.608115553855896,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 14510
+    },
+    {
+      "epoch": 95.52631578947368,
+      "grad_norm": 1.6221609115600586,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 14520
+    },
+    {
+      "epoch": 95.59210526315789,
+      "grad_norm": 1.772195816040039,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 14530
+    },
+    {
+      "epoch": 95.65789473684211,
+      "grad_norm": 1.5958181619644165,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 14540
+    },
+    {
+      "epoch": 95.72368421052632,
+      "grad_norm": 1.7433152198791504,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 14550
+    },
+    {
+      "epoch": 95.78947368421052,
+      "grad_norm": 1.7404011487960815,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 14560
+    },
+    {
+      "epoch": 95.85526315789474,
+      "grad_norm": 1.8106964826583862,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 14570
+    },
+    {
+      "epoch": 95.92105263157895,
+      "grad_norm": 1.8747131824493408,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 14580
+    },
+    {
+      "epoch": 95.98684210526316,
+      "grad_norm": 1.9325027465820312,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 14590
+    },
+    {
+      "epoch": 96.05263157894737,
+      "grad_norm": 1.6936956644058228,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 14600
+    },
+    {
+      "epoch": 96.11842105263158,
+      "grad_norm": 1.6250704526901245,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 14610
+    },
+    {
+      "epoch": 96.1842105263158,
+      "grad_norm": 2.4609410762786865,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 14620
+    },
+    {
+      "epoch": 96.25,
+      "grad_norm": 2.147636890411377,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 14630
+    },
+    {
+      "epoch": 96.3157894736842,
+      "grad_norm": 1.605124831199646,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 14640
+    },
+    {
+      "epoch": 96.38157894736842,
+      "grad_norm": 1.8906571865081787,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 14650
+    },
+    {
+      "epoch": 96.44736842105263,
+      "grad_norm": 1.9277275800704956,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 14660
+    },
+    {
+      "epoch": 96.51315789473684,
+      "grad_norm": 1.7315579652786255,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 14670
+    },
+    {
+      "epoch": 96.57894736842105,
+      "grad_norm": 1.8327442407608032,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 14680
+    },
+    {
+      "epoch": 96.64473684210526,
+      "grad_norm": 1.5039429664611816,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 14690
+    },
+    {
+      "epoch": 96.71052631578948,
+      "grad_norm": 1.958951473236084,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 14700
+    },
+    {
+      "epoch": 96.77631578947368,
+      "grad_norm": 1.7764480113983154,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 14710
+    },
+    {
+      "epoch": 96.84210526315789,
+      "grad_norm": 1.4950709342956543,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 14720
+    },
+    {
+      "epoch": 96.90789473684211,
+      "grad_norm": 2.5329701900482178,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 14730
+    },
+    {
+      "epoch": 96.97368421052632,
+      "grad_norm": 1.7658509016036987,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 14740
+    },
+    {
+      "epoch": 97.03947368421052,
+      "grad_norm": 1.8286138772964478,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 14750
+    },
+    {
+      "epoch": 97.10526315789474,
+      "grad_norm": 1.5775742530822754,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 14760
+    },
+    {
+      "epoch": 97.17105263157895,
+      "grad_norm": 1.4077131748199463,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 14770
+    },
+    {
+      "epoch": 97.23684210526316,
+      "grad_norm": 1.6287063360214233,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 14780
+    },
+    {
+      "epoch": 97.30263157894737,
+      "grad_norm": 1.9090032577514648,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 14790
+    },
+    {
+      "epoch": 97.36842105263158,
+      "grad_norm": 1.7820619344711304,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 14800
+    },
+    {
+      "epoch": 97.4342105263158,
+      "grad_norm": 1.4791796207427979,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 14810
+    },
+    {
+      "epoch": 97.5,
+      "grad_norm": 1.6727066040039062,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 14820
+    },
+    {
+      "epoch": 97.5657894736842,
+      "grad_norm": 1.720504641532898,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 14830
+    },
+    {
+      "epoch": 97.63157894736842,
+      "grad_norm": 2.1600022315979004,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 14840
+    },
+    {
+      "epoch": 97.69736842105263,
+      "grad_norm": 2.139911413192749,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 14850
+    },
+    {
+      "epoch": 97.76315789473684,
+      "grad_norm": 1.762117624282837,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 14860
+    },
+    {
+      "epoch": 97.82894736842105,
+      "grad_norm": 1.6421223878860474,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 14870
+    },
+    {
+      "epoch": 97.89473684210526,
+      "grad_norm": 1.2390879392623901,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 14880
+    },
+    {
+      "epoch": 97.96052631578948,
+      "grad_norm": 1.3071866035461426,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 14890
+    },
+    {
+      "epoch": 98.02631578947368,
+      "grad_norm": 1.6224156618118286,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 14900
+    },
+    {
+      "epoch": 98.09210526315789,
+      "grad_norm": 1.703869342803955,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 14910
+    },
+    {
+      "epoch": 98.15789473684211,
+      "grad_norm": 1.475963830947876,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 14920
+    },
+    {
+      "epoch": 98.22368421052632,
+      "grad_norm": 1.451790690422058,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 14930
+    },
+    {
+      "epoch": 98.28947368421052,
+      "grad_norm": 1.3597410917282104,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 14940
+    },
+    {
+      "epoch": 98.35526315789474,
+      "grad_norm": 1.047415018081665,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 14950
+    },
+    {
+      "epoch": 98.42105263157895,
+      "grad_norm": 1.6275583505630493,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 14960
+    },
+    {
+      "epoch": 98.48684210526316,
+      "grad_norm": 1.3515644073486328,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 14970
+    },
+    {
+      "epoch": 98.55263157894737,
+      "grad_norm": 1.5167436599731445,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 14980
+    },
+    {
+      "epoch": 98.61842105263158,
+      "grad_norm": 1.5004242658615112,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 14990
+    },
+    {
+      "epoch": 98.6842105263158,
+      "grad_norm": 1.2739217281341553,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 15000
+    },
+    {
+      "epoch": 98.75,
+      "grad_norm": 1.6664555072784424,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 15010
+    },
+    {
+      "epoch": 98.8157894736842,
+      "grad_norm": 1.635536789894104,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 15020
+    },
+    {
+      "epoch": 98.88157894736842,
+      "grad_norm": 1.3948959112167358,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 15030
+    },
+    {
+      "epoch": 98.94736842105263,
+      "grad_norm": 1.6810851097106934,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 15040
+    },
+    {
+      "epoch": 99.01315789473684,
+      "grad_norm": 1.4952811002731323,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 15050
+    },
+    {
+      "epoch": 99.07894736842105,
+      "grad_norm": 1.319809913635254,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 15060
+    },
+    {
+      "epoch": 99.14473684210526,
+      "grad_norm": 1.7210019826889038,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 15070
+    },
+    {
+      "epoch": 99.21052631578948,
+      "grad_norm": 1.4392873048782349,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 15080
+    },
+    {
+      "epoch": 99.27631578947368,
+      "grad_norm": 1.3925666809082031,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 15090
+    },
+    {
+      "epoch": 99.34210526315789,
+      "grad_norm": 1.5854908227920532,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 15100
+    },
+    {
+      "epoch": 99.40789473684211,
+      "grad_norm": 1.3266748189926147,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 15110
+    },
+    {
+      "epoch": 99.47368421052632,
+      "grad_norm": 1.3466399908065796,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 15120
+    },
+    {
+      "epoch": 99.53947368421052,
+      "grad_norm": 1.1351549625396729,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 15130
+    },
+    {
+      "epoch": 99.60526315789474,
+      "grad_norm": 1.2604902982711792,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 15140
+    },
+    {
+      "epoch": 99.67105263157895,
+      "grad_norm": 1.4526969194412231,
+      "learning_rate": 0.0001,
+      "loss": 0.0282,
+      "step": 15150
+    },
+    {
+      "epoch": 99.73684210526316,
+      "grad_norm": 1.7058517932891846,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 15160
+    },
+    {
+      "epoch": 99.80263157894737,
+      "grad_norm": 1.7141302824020386,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 15170
+    },
+    {
+      "epoch": 99.86842105263158,
+      "grad_norm": 1.6191295385360718,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 15180
+    },
+    {
+      "epoch": 99.9342105263158,
+      "grad_norm": 1.258483648300171,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 15190
+    },
+    {
+      "epoch": 100.0,
+      "grad_norm": 1.4220373630523682,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 15200
+    },
+    {
+      "epoch": 100.0657894736842,
+      "grad_norm": 1.295554518699646,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 15210
+    },
+    {
+      "epoch": 100.13157894736842,
+      "grad_norm": 1.5223777294158936,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 15220
+    },
+    {
+      "epoch": 100.19736842105263,
+      "grad_norm": 0.9295150637626648,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 15230
+    },
+    {
+      "epoch": 100.26315789473684,
+      "grad_norm": 1.522321105003357,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 15240
+    },
+    {
+      "epoch": 100.32894736842105,
+      "grad_norm": 1.380570650100708,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 15250
+    },
+    {
+      "epoch": 100.39473684210526,
+      "grad_norm": 1.745607614517212,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 15260
+    },
+    {
+      "epoch": 100.46052631578948,
+      "grad_norm": 1.6587855815887451,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 15270
+    },
+    {
+      "epoch": 100.52631578947368,
+      "grad_norm": 1.592822551727295,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 15280
+    },
+    {
+      "epoch": 100.59210526315789,
+      "grad_norm": 1.649880290031433,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 15290
+    },
+    {
+      "epoch": 100.65789473684211,
+      "grad_norm": 1.350723385810852,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 15300
+    },
+    {
+      "epoch": 100.72368421052632,
+      "grad_norm": 1.5590640306472778,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 15310
+    },
+    {
+      "epoch": 100.78947368421052,
+      "grad_norm": 1.4502315521240234,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 15320
+    },
+    {
+      "epoch": 100.85526315789474,
+      "grad_norm": 1.206421136856079,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 15330
+    },
+    {
+      "epoch": 100.92105263157895,
+      "grad_norm": 1.6681444644927979,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 15340
+    },
+    {
+      "epoch": 100.98684210526316,
+      "grad_norm": 1.7128868103027344,
+      "learning_rate": 0.0001,
+      "loss": 0.0277,
+      "step": 15350
+    },
+    {
+      "epoch": 101.05263157894737,
+      "grad_norm": 1.7957628965377808,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 15360
+    },
+    {
+      "epoch": 101.11842105263158,
+      "grad_norm": 1.8973162174224854,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 15370
+    },
+    {
+      "epoch": 101.1842105263158,
+      "grad_norm": 1.4725027084350586,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 15380
+    },
+    {
+      "epoch": 101.25,
+      "grad_norm": 1.2609061002731323,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 15390
+    },
+    {
+      "epoch": 101.3157894736842,
+      "grad_norm": 1.3644044399261475,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 15400
+    },
+    {
+      "epoch": 101.38157894736842,
+      "grad_norm": 1.4843460321426392,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 15410
+    },
+    {
+      "epoch": 101.44736842105263,
+      "grad_norm": 1.5529197454452515,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 15420
+    },
+    {
+      "epoch": 101.51315789473684,
+      "grad_norm": 1.5046693086624146,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 15430
+    },
+    {
+      "epoch": 101.57894736842105,
+      "grad_norm": 1.6078213453292847,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 15440
+    },
+    {
+      "epoch": 101.64473684210526,
+      "grad_norm": 1.9121206998825073,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 15450
+    },
+    {
+      "epoch": 101.71052631578948,
+      "grad_norm": 1.69439697265625,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 15460
+    },
+    {
+      "epoch": 101.77631578947368,
+      "grad_norm": 1.3743191957473755,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 15470
+    },
+    {
+      "epoch": 101.84210526315789,
+      "grad_norm": 1.4057716131210327,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 15480
+    },
+    {
+      "epoch": 101.90789473684211,
+      "grad_norm": 1.2741807699203491,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 15490
+    },
+    {
+      "epoch": 101.97368421052632,
+      "grad_norm": 1.209885597229004,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 15500
+    },
+    {
+      "epoch": 102.03947368421052,
+      "grad_norm": 1.3488867282867432,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 15510
+    },
+    {
+      "epoch": 102.10526315789474,
+      "grad_norm": 1.4938287734985352,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 15520
+    },
+    {
+      "epoch": 102.17105263157895,
+      "grad_norm": 1.3326101303100586,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 15530
+    },
+    {
+      "epoch": 102.23684210526316,
+      "grad_norm": 1.462449550628662,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 15540
+    },
+    {
+      "epoch": 102.30263157894737,
+      "grad_norm": 1.2210637331008911,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 15550
+    },
+    {
+      "epoch": 102.36842105263158,
+      "grad_norm": 1.4737353324890137,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 15560
+    },
+    {
+      "epoch": 102.4342105263158,
+      "grad_norm": 1.4356292486190796,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 15570
+    },
+    {
+      "epoch": 102.5,
+      "grad_norm": 1.472893238067627,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 15580
+    },
+    {
+      "epoch": 102.5657894736842,
+      "grad_norm": 1.5573779344558716,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 15590
+    },
+    {
+      "epoch": 102.63157894736842,
+      "grad_norm": 1.5440502166748047,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 15600
+    },
+    {
+      "epoch": 102.69736842105263,
+      "grad_norm": 1.5142933130264282,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 15610
+    },
+    {
+      "epoch": 102.76315789473684,
+      "grad_norm": 1.6982897520065308,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 15620
+    },
+    {
+      "epoch": 102.82894736842105,
+      "grad_norm": 1.8263667821884155,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 15630
+    },
+    {
+      "epoch": 102.89473684210526,
+      "grad_norm": 1.2521591186523438,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 15640
+    },
+    {
+      "epoch": 102.96052631578948,
+      "grad_norm": 1.5161454677581787,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 15650
+    },
+    {
+      "epoch": 103.02631578947368,
+      "grad_norm": 1.4727702140808105,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 15660
+    },
+    {
+      "epoch": 103.09210526315789,
+      "grad_norm": 1.3148939609527588,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 15670
+    },
+    {
+      "epoch": 103.15789473684211,
+      "grad_norm": 1.3606066703796387,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 15680
+    },
+    {
+      "epoch": 103.22368421052632,
+      "grad_norm": 1.4023685455322266,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 15690
+    },
+    {
+      "epoch": 103.28947368421052,
+      "grad_norm": 1.424401879310608,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 15700
+    },
+    {
+      "epoch": 103.35526315789474,
+      "grad_norm": 1.1225404739379883,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 15710
+    },
+    {
+      "epoch": 103.42105263157895,
+      "grad_norm": 1.4364515542984009,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 15720
+    },
+    {
+      "epoch": 103.48684210526316,
+      "grad_norm": 1.8251726627349854,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 15730
+    },
+    {
+      "epoch": 103.55263157894737,
+      "grad_norm": 1.6310752630233765,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 15740
+    },
+    {
+      "epoch": 103.61842105263158,
+      "grad_norm": 1.3311368227005005,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 15750
+    },
+    {
+      "epoch": 103.6842105263158,
+      "grad_norm": 1.7061165571212769,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 15760
+    },
+    {
+      "epoch": 103.75,
+      "grad_norm": 1.737336277961731,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 15770
+    },
+    {
+      "epoch": 103.8157894736842,
+      "grad_norm": 1.429793119430542,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 15780
+    },
+    {
+      "epoch": 103.88157894736842,
+      "grad_norm": 1.2859523296356201,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 15790
+    },
+    {
+      "epoch": 103.94736842105263,
+      "grad_norm": 1.5139811038970947,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 15800
+    },
+    {
+      "epoch": 104.01315789473684,
+      "grad_norm": 1.5388349294662476,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 15810
+    },
+    {
+      "epoch": 104.07894736842105,
+      "grad_norm": 1.2440307140350342,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 15820
+    },
+    {
+      "epoch": 104.14473684210526,
+      "grad_norm": 1.590995192527771,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 15830
+    },
+    {
+      "epoch": 104.21052631578948,
+      "grad_norm": 1.5027881860733032,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 15840
+    },
+    {
+      "epoch": 104.27631578947368,
+      "grad_norm": 1.2226101160049438,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 15850
+    },
+    {
+      "epoch": 104.34210526315789,
+      "grad_norm": 1.3308379650115967,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 15860
+    },
+    {
+      "epoch": 104.40789473684211,
+      "grad_norm": 1.2079100608825684,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 15870
+    },
+    {
+      "epoch": 104.47368421052632,
+      "grad_norm": 1.1981853246688843,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 15880
+    },
+    {
+      "epoch": 104.53947368421052,
+      "grad_norm": 1.3664216995239258,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 15890
+    },
+    {
+      "epoch": 104.60526315789474,
+      "grad_norm": 1.8628100156784058,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 15900
+    },
+    {
+      "epoch": 104.67105263157895,
+      "grad_norm": 1.2317039966583252,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "step": 15910
+    },
+    {
+      "epoch": 104.73684210526316,
+      "grad_norm": 1.513042688369751,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 15920
+    },
+    {
+      "epoch": 104.80263157894737,
+      "grad_norm": 1.6402066946029663,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 15930
+    },
+    {
+      "epoch": 104.86842105263158,
+      "grad_norm": 1.7423036098480225,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 15940
+    },
+    {
+      "epoch": 104.9342105263158,
+      "grad_norm": 1.2021863460540771,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 15950
+    },
+    {
+      "epoch": 105.0,
+      "grad_norm": 1.5771297216415405,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 15960
+    },
+    {
+      "epoch": 105.0657894736842,
+      "grad_norm": 1.2136021852493286,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 15970
+    },
+    {
+      "epoch": 105.13157894736842,
+      "grad_norm": 1.7530934810638428,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 15980
+    },
+    {
+      "epoch": 105.19736842105263,
+      "grad_norm": 1.727021336555481,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 15990
+    },
+    {
+      "epoch": 105.26315789473684,
+      "grad_norm": 1.689032793045044,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 16000
+    },
+    {
+      "epoch": 105.32894736842105,
+      "grad_norm": 2.0343551635742188,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 16010
+    },
+    {
+      "epoch": 105.39473684210526,
+      "grad_norm": 1.865867257118225,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 16020
+    },
+    {
+      "epoch": 105.46052631578948,
+      "grad_norm": 1.5226266384124756,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 16030
+    },
+    {
+      "epoch": 105.52631578947368,
+      "grad_norm": 1.8167226314544678,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 16040
+    },
+    {
+      "epoch": 105.59210526315789,
+      "grad_norm": 1.2684314250946045,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 16050
+    },
+    {
+      "epoch": 105.65789473684211,
+      "grad_norm": 1.7727292776107788,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 16060
+    },
+    {
+      "epoch": 105.72368421052632,
+      "grad_norm": 1.3268436193466187,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 16070
+    },
+    {
+      "epoch": 105.78947368421052,
+      "grad_norm": 1.5590864419937134,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 16080
+    },
+    {
+      "epoch": 105.85526315789474,
+      "grad_norm": 1.4346693754196167,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 16090
+    },
+    {
+      "epoch": 105.92105263157895,
+      "grad_norm": 1.5737860202789307,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 16100
+    },
+    {
+      "epoch": 105.98684210526316,
+      "grad_norm": 1.5371108055114746,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 16110
+    },
+    {
+      "epoch": 106.05263157894737,
+      "grad_norm": 1.2796305418014526,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 16120
+    },
+    {
+      "epoch": 106.11842105263158,
+      "grad_norm": 1.506816029548645,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 16130
+    },
+    {
+      "epoch": 106.1842105263158,
+      "grad_norm": 1.4355484247207642,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 16140
+    },
+    {
+      "epoch": 106.25,
+      "grad_norm": 1.6774417161941528,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 16150
+    },
+    {
+      "epoch": 106.3157894736842,
+      "grad_norm": 1.5477979183197021,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 16160
+    },
+    {
+      "epoch": 106.38157894736842,
+      "grad_norm": 1.6077028512954712,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 16170
+    },
+    {
+      "epoch": 106.44736842105263,
+      "grad_norm": 1.9327597618103027,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 16180
+    },
+    {
+      "epoch": 106.51315789473684,
+      "grad_norm": 1.2858346700668335,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 16190
+    },
+    {
+      "epoch": 106.57894736842105,
+      "grad_norm": 1.3751716613769531,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 16200
+    },
+    {
+      "epoch": 106.64473684210526,
+      "grad_norm": 1.340836763381958,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 16210
+    },
+    {
+      "epoch": 106.71052631578948,
+      "grad_norm": 1.5455154180526733,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 16220
+    },
+    {
+      "epoch": 106.77631578947368,
+      "grad_norm": 1.066322922706604,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 16230
+    },
+    {
+      "epoch": 106.84210526315789,
+      "grad_norm": 1.5077837705612183,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 16240
+    },
+    {
+      "epoch": 106.90789473684211,
+      "grad_norm": 1.211094856262207,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 16250
+    },
+    {
+      "epoch": 106.97368421052632,
+      "grad_norm": 2.008754014968872,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 16260
+    },
+    {
+      "epoch": 107.03947368421052,
+      "grad_norm": 1.3147034645080566,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 16270
+    },
+    {
+      "epoch": 107.10526315789474,
+      "grad_norm": 1.1280360221862793,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 16280
+    },
+    {
+      "epoch": 107.17105263157895,
+      "grad_norm": 1.3991897106170654,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 16290
+    },
+    {
+      "epoch": 107.23684210526316,
+      "grad_norm": 1.2809622287750244,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 16300
+    },
+    {
+      "epoch": 107.30263157894737,
+      "grad_norm": 1.4164490699768066,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 16310
+    },
+    {
+      "epoch": 107.36842105263158,
+      "grad_norm": 1.374470829963684,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 16320
+    },
+    {
+      "epoch": 107.4342105263158,
+      "grad_norm": 1.3295577764511108,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 16330
+    },
+    {
+      "epoch": 107.5,
+      "grad_norm": 1.4546465873718262,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 16340
+    },
+    {
+      "epoch": 107.5657894736842,
+      "grad_norm": 1.6725012063980103,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 16350
+    },
+    {
+      "epoch": 107.63157894736842,
+      "grad_norm": 1.4987870454788208,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 16360
+    },
+    {
+      "epoch": 107.69736842105263,
+      "grad_norm": 1.4768329858779907,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 16370
+    },
+    {
+      "epoch": 107.76315789473684,
+      "grad_norm": 1.509992003440857,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 16380
+    },
+    {
+      "epoch": 107.82894736842105,
+      "grad_norm": 1.5696868896484375,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 16390
+    },
+    {
+      "epoch": 107.89473684210526,
+      "grad_norm": 1.9608714580535889,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 16400
+    },
+    {
+      "epoch": 107.96052631578948,
+      "grad_norm": 1.597530722618103,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 16410
+    },
+    {
+      "epoch": 108.02631578947368,
+      "grad_norm": 1.544816493988037,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 16420
+    },
+    {
+      "epoch": 108.09210526315789,
+      "grad_norm": 1.8198386430740356,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 16430
+    },
+    {
+      "epoch": 108.15789473684211,
+      "grad_norm": 1.1581134796142578,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 16440
+    },
+    {
+      "epoch": 108.22368421052632,
+      "grad_norm": 1.273421049118042,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 16450
+    },
+    {
+      "epoch": 108.28947368421052,
+      "grad_norm": 1.438262939453125,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 16460
+    },
+    {
+      "epoch": 108.35526315789474,
+      "grad_norm": 1.5786799192428589,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 16470
+    },
+    {
+      "epoch": 108.42105263157895,
+      "grad_norm": 1.8563624620437622,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 16480
+    },
+    {
+      "epoch": 108.48684210526316,
+      "grad_norm": 1.6424920558929443,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 16490
+    },
+    {
+      "epoch": 108.55263157894737,
+      "grad_norm": 1.5354336500167847,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 16500
+    },
+    {
+      "epoch": 108.61842105263158,
+      "grad_norm": 1.7186359167099,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 16510
+    },
+    {
+      "epoch": 108.6842105263158,
+      "grad_norm": 1.3204076290130615,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 16520
+    },
+    {
+      "epoch": 108.75,
+      "grad_norm": 1.1582022905349731,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 16530
+    },
+    {
+      "epoch": 108.8157894736842,
+      "grad_norm": 1.2648552656173706,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 16540
+    },
+    {
+      "epoch": 108.88157894736842,
+      "grad_norm": 1.2796598672866821,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 16550
+    },
+    {
+      "epoch": 108.94736842105263,
+      "grad_norm": 1.9214965105056763,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 16560
+    },
+    {
+      "epoch": 109.01315789473684,
+      "grad_norm": 1.449998140335083,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 16570
+    },
+    {
+      "epoch": 109.07894736842105,
+      "grad_norm": 1.1618794202804565,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 16580
+    },
+    {
+      "epoch": 109.14473684210526,
+      "grad_norm": 1.4211252927780151,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 16590
+    },
+    {
+      "epoch": 109.21052631578948,
+      "grad_norm": 1.3328765630722046,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 16600
+    },
+    {
+      "epoch": 109.27631578947368,
+      "grad_norm": 1.715414047241211,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 16610
+    },
+    {
+      "epoch": 109.34210526315789,
+      "grad_norm": 1.2124247550964355,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 16620
+    },
+    {
+      "epoch": 109.40789473684211,
+      "grad_norm": 1.3251482248306274,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 16630
+    },
+    {
+      "epoch": 109.47368421052632,
+      "grad_norm": 1.4357656240463257,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 16640
+    },
+    {
+      "epoch": 109.53947368421052,
+      "grad_norm": 1.2938634157180786,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 16650
+    },
+    {
+      "epoch": 109.60526315789474,
+      "grad_norm": 1.2677600383758545,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 16660
+    },
+    {
+      "epoch": 109.67105263157895,
+      "grad_norm": 1.1516252756118774,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 16670
+    },
+    {
+      "epoch": 109.73684210526316,
+      "grad_norm": 1.4747333526611328,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 16680
+    },
+    {
+      "epoch": 109.80263157894737,
+      "grad_norm": 1.3867018222808838,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 16690
+    },
+    {
+      "epoch": 109.86842105263158,
+      "grad_norm": 1.8688561916351318,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 16700
+    },
+    {
+      "epoch": 109.9342105263158,
+      "grad_norm": 1.686368465423584,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 16710
+    },
+    {
+      "epoch": 110.0,
+      "grad_norm": 1.311707615852356,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 16720
+    },
+    {
+      "epoch": 110.0657894736842,
+      "grad_norm": 1.1597585678100586,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 16730
+    },
+    {
+      "epoch": 110.13157894736842,
+      "grad_norm": 1.41483736038208,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 16740
+    },
+    {
+      "epoch": 110.19736842105263,
+      "grad_norm": 1.4122625589370728,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 16750
+    },
+    {
+      "epoch": 110.26315789473684,
+      "grad_norm": 1.336828589439392,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 16760
+    },
+    {
+      "epoch": 110.32894736842105,
+      "grad_norm": 1.5870128870010376,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 16770
+    },
+    {
+      "epoch": 110.39473684210526,
+      "grad_norm": 1.2915987968444824,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 16780
+    },
+    {
+      "epoch": 110.46052631578948,
+      "grad_norm": 1.4546549320220947,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 16790
+    },
+    {
+      "epoch": 110.52631578947368,
+      "grad_norm": 1.3749148845672607,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 16800
+    },
+    {
+      "epoch": 110.59210526315789,
+      "grad_norm": 1.8686497211456299,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 16810
+    },
+    {
+      "epoch": 110.65789473684211,
+      "grad_norm": 1.5140857696533203,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 16820
+    },
+    {
+      "epoch": 110.72368421052632,
+      "grad_norm": 1.222466230392456,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 16830
+    },
+    {
+      "epoch": 110.78947368421052,
+      "grad_norm": 1.4809081554412842,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 16840
+    },
+    {
+      "epoch": 110.85526315789474,
+      "grad_norm": 1.4631025791168213,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 16850
+    },
+    {
+      "epoch": 110.92105263157895,
+      "grad_norm": 1.5033537149429321,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 16860
+    },
+    {
+      "epoch": 110.98684210526316,
+      "grad_norm": 1.5242358446121216,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 16870
+    },
+    {
+      "epoch": 111.05263157894737,
+      "grad_norm": 1.4843236207962036,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 16880
+    },
+    {
+      "epoch": 111.11842105263158,
+      "grad_norm": 1.1458847522735596,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 16890
+    },
+    {
+      "epoch": 111.1842105263158,
+      "grad_norm": 1.1731573343276978,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 16900
+    },
+    {
+      "epoch": 111.25,
+      "grad_norm": 1.4374542236328125,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 16910
+    },
+    {
+      "epoch": 111.3157894736842,
+      "grad_norm": 1.523205041885376,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 16920
+    },
+    {
+      "epoch": 111.38157894736842,
+      "grad_norm": 1.5758098363876343,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 16930
+    },
+    {
+      "epoch": 111.44736842105263,
+      "grad_norm": 1.42531156539917,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 16940
+    },
+    {
+      "epoch": 111.51315789473684,
+      "grad_norm": 1.332920789718628,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 16950
+    },
+    {
+      "epoch": 111.57894736842105,
+      "grad_norm": 1.5037161111831665,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 16960
+    },
+    {
+      "epoch": 111.64473684210526,
+      "grad_norm": 1.2761225700378418,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 16970
+    },
+    {
+      "epoch": 111.71052631578948,
+      "grad_norm": 1.315170168876648,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 16980
+    },
+    {
+      "epoch": 111.77631578947368,
+      "grad_norm": 1.6541320085525513,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 16990
+    },
+    {
+      "epoch": 111.84210526315789,
+      "grad_norm": 1.4604840278625488,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 17000
+    },
+    {
+      "epoch": 111.90789473684211,
+      "grad_norm": 1.7079929113388062,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 17010
+    },
+    {
+      "epoch": 111.97368421052632,
+      "grad_norm": 1.5881198644638062,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 17020
+    },
+    {
+      "epoch": 112.03947368421052,
+      "grad_norm": 1.1928815841674805,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 17030
+    },
+    {
+      "epoch": 112.10526315789474,
+      "grad_norm": 1.283767819404602,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 17040
+    },
+    {
+      "epoch": 112.17105263157895,
+      "grad_norm": 1.0753800868988037,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 17050
+    },
+    {
+      "epoch": 112.23684210526316,
+      "grad_norm": 1.3679052591323853,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 17060
+    },
+    {
+      "epoch": 112.30263157894737,
+      "grad_norm": 1.5422084331512451,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 17070
+    },
+    {
+      "epoch": 112.36842105263158,
+      "grad_norm": 1.0112746953964233,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 17080
+    },
+    {
+      "epoch": 112.4342105263158,
+      "grad_norm": 1.7527899742126465,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 17090
+    },
+    {
+      "epoch": 112.5,
+      "grad_norm": 2.344182252883911,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 17100
+    },
+    {
+      "epoch": 112.5657894736842,
+      "grad_norm": 2.5137453079223633,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 17110
+    },
+    {
+      "epoch": 112.63157894736842,
+      "grad_norm": 2.357712507247925,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 17120
+    },
+    {
+      "epoch": 112.69736842105263,
+      "grad_norm": 2.287658929824829,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 17130
+    },
+    {
+      "epoch": 112.76315789473684,
+      "grad_norm": 1.9697659015655518,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 17140
+    },
+    {
+      "epoch": 112.82894736842105,
+      "grad_norm": 1.8423213958740234,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 17150
+    },
+    {
+      "epoch": 112.89473684210526,
+      "grad_norm": 1.7179113626480103,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 17160
+    },
+    {
+      "epoch": 112.96052631578948,
+      "grad_norm": 1.2885282039642334,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 17170
+    },
+    {
+      "epoch": 113.02631578947368,
+      "grad_norm": 2.470750331878662,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 17180
+    },
+    {
+      "epoch": 113.09210526315789,
+      "grad_norm": 1.9972500801086426,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 17190
+    },
+    {
+      "epoch": 113.15789473684211,
+      "grad_norm": 1.7662476301193237,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 17200
+    },
+    {
+      "epoch": 113.22368421052632,
+      "grad_norm": 2.526029586791992,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 17210
+    },
+    {
+      "epoch": 113.28947368421052,
+      "grad_norm": 1.8449680805206299,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 17220
+    },
+    {
+      "epoch": 113.35526315789474,
+      "grad_norm": 1.9079487323760986,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 17230
+    },
+    {
+      "epoch": 113.42105263157895,
+      "grad_norm": 1.5968166589736938,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 17240
+    },
+    {
+      "epoch": 113.48684210526316,
+      "grad_norm": 1.7774184942245483,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 17250
+    },
+    {
+      "epoch": 113.55263157894737,
+      "grad_norm": 1.8195269107818604,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 17260
+    },
+    {
+      "epoch": 113.61842105263158,
+      "grad_norm": 2.0075738430023193,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 17270
+    },
+    {
+      "epoch": 113.6842105263158,
+      "grad_norm": 1.6541519165039062,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 17280
+    },
+    {
+      "epoch": 113.75,
+      "grad_norm": 1.6574792861938477,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 17290
+    },
+    {
+      "epoch": 113.8157894736842,
+      "grad_norm": 1.459970235824585,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 17300
+    },
+    {
+      "epoch": 113.88157894736842,
+      "grad_norm": 1.072644829750061,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 17310
+    },
+    {
+      "epoch": 113.94736842105263,
+      "grad_norm": 1.4994428157806396,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 17320
+    },
+    {
+      "epoch": 114.01315789473684,
+      "grad_norm": 1.1813583374023438,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 17330
+    },
+    {
+      "epoch": 114.07894736842105,
+      "grad_norm": 1.115692377090454,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 17340
+    },
+    {
+      "epoch": 114.14473684210526,
+      "grad_norm": 1.6557120084762573,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 17350
+    },
+    {
+      "epoch": 114.21052631578948,
+      "grad_norm": 1.287843942642212,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 17360
+    },
+    {
+      "epoch": 114.27631578947368,
+      "grad_norm": 1.5140693187713623,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 17370
+    },
+    {
+      "epoch": 114.34210526315789,
+      "grad_norm": 1.3632303476333618,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 17380
+    },
+    {
+      "epoch": 114.40789473684211,
+      "grad_norm": 1.5650956630706787,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 17390
+    },
+    {
+      "epoch": 114.47368421052632,
+      "grad_norm": 1.413570761680603,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 17400
+    },
+    {
+      "epoch": 114.53947368421052,
+      "grad_norm": 1.2383382320404053,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 17410
+    },
+    {
+      "epoch": 114.60526315789474,
+      "grad_norm": 1.6793911457061768,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 17420
+    },
+    {
+      "epoch": 114.67105263157895,
+      "grad_norm": 1.5027910470962524,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 17430
+    },
+    {
+      "epoch": 114.73684210526316,
+      "grad_norm": 1.4984204769134521,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 17440
+    },
+    {
+      "epoch": 114.80263157894737,
+      "grad_norm": 1.712581992149353,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 17450
+    },
+    {
+      "epoch": 114.86842105263158,
+      "grad_norm": 1.507750153541565,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 17460
+    },
+    {
+      "epoch": 114.9342105263158,
+      "grad_norm": 1.515345573425293,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 17470
+    },
+    {
+      "epoch": 115.0,
+      "grad_norm": 1.804507851600647,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 17480
+    },
+    {
+      "epoch": 115.0657894736842,
+      "grad_norm": 1.6192446947097778,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 17490
+    },
+    {
+      "epoch": 115.13157894736842,
+      "grad_norm": 1.3027291297912598,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 17500
+    },
+    {
+      "epoch": 115.19736842105263,
+      "grad_norm": 1.5303033590316772,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 17510
+    },
+    {
+      "epoch": 115.26315789473684,
+      "grad_norm": 1.245597004890442,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 17520
+    },
+    {
+      "epoch": 115.32894736842105,
+      "grad_norm": 1.048466444015503,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 17530
+    },
+    {
+      "epoch": 115.39473684210526,
+      "grad_norm": 1.5736318826675415,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 17540
+    },
+    {
+      "epoch": 115.46052631578948,
+      "grad_norm": 1.8937121629714966,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 17550
+    },
+    {
+      "epoch": 115.52631578947368,
+      "grad_norm": 1.350262999534607,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 17560
+    },
+    {
+      "epoch": 115.59210526315789,
+      "grad_norm": 1.4281275272369385,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 17570
+    },
+    {
+      "epoch": 115.65789473684211,
+      "grad_norm": 1.6036561727523804,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 17580
+    },
+    {
+      "epoch": 115.72368421052632,
+      "grad_norm": 1.1876840591430664,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 17590
+    },
+    {
+      "epoch": 115.78947368421052,
+      "grad_norm": 1.1769388914108276,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 17600
+    },
+    {
+      "epoch": 115.85526315789474,
+      "grad_norm": 1.621465802192688,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 17610
+    },
+    {
+      "epoch": 115.92105263157895,
+      "grad_norm": 3.2968623638153076,
+      "learning_rate": 0.0001,
+      "loss": 0.0258,
+      "step": 17620
+    },
+    {
+      "epoch": 115.98684210526316,
+      "grad_norm": 1.8271244764328003,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 17630
+    },
+    {
+      "epoch": 116.05263157894737,
+      "grad_norm": 1.8423553705215454,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 17640
+    },
+    {
+      "epoch": 116.11842105263158,
+      "grad_norm": 1.65046226978302,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 17650
+    },
+    {
+      "epoch": 116.1842105263158,
+      "grad_norm": 1.7557287216186523,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 17660
+    },
+    {
+      "epoch": 116.25,
+      "grad_norm": 2.067875385284424,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 17670
+    },
+    {
+      "epoch": 116.3157894736842,
+      "grad_norm": 1.5921411514282227,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 17680
+    },
+    {
+      "epoch": 116.38157894736842,
+      "grad_norm": 1.153791069984436,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 17690
+    },
+    {
+      "epoch": 116.44736842105263,
+      "grad_norm": 1.2742387056350708,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 17700
+    },
+    {
+      "epoch": 116.51315789473684,
+      "grad_norm": 1.384451150894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 17710
+    },
+    {
+      "epoch": 116.57894736842105,
+      "grad_norm": 1.3240984678268433,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 17720
+    },
+    {
+      "epoch": 116.64473684210526,
+      "grad_norm": 1.3543736934661865,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 17730
+    },
+    {
+      "epoch": 116.71052631578948,
+      "grad_norm": 1.4543614387512207,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 17740
+    },
+    {
+      "epoch": 116.77631578947368,
+      "grad_norm": 1.2755897045135498,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 17750
+    },
+    {
+      "epoch": 116.84210526315789,
+      "grad_norm": 1.4515208005905151,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 17760
+    },
+    {
+      "epoch": 116.90789473684211,
+      "grad_norm": 1.4449197053909302,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 17770
+    },
+    {
+      "epoch": 116.97368421052632,
+      "grad_norm": 1.1386222839355469,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 17780
+    },
+    {
+      "epoch": 117.03947368421052,
+      "grad_norm": 1.416370153427124,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 17790
+    },
+    {
+      "epoch": 117.10526315789474,
+      "grad_norm": 1.1340827941894531,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 17800
+    },
+    {
+      "epoch": 117.17105263157895,
+      "grad_norm": 1.6249243021011353,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 17810
+    },
+    {
+      "epoch": 117.23684210526316,
+      "grad_norm": 1.2724114656448364,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 17820
+    },
+    {
+      "epoch": 117.30263157894737,
+      "grad_norm": 1.7152994871139526,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 17830
+    },
+    {
+      "epoch": 117.36842105263158,
+      "grad_norm": 1.4186722040176392,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 17840
+    },
+    {
+      "epoch": 117.4342105263158,
+      "grad_norm": 1.5989038944244385,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 17850
+    },
+    {
+      "epoch": 117.5,
+      "grad_norm": 1.5071998834609985,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 17860
+    },
+    {
+      "epoch": 117.5657894736842,
+      "grad_norm": 1.528257131576538,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 17870
+    },
+    {
+      "epoch": 117.63157894736842,
+      "grad_norm": 1.3654208183288574,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 17880
+    },
+    {
+      "epoch": 117.69736842105263,
+      "grad_norm": 2.163045883178711,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 17890
+    },
+    {
+      "epoch": 117.76315789473684,
+      "grad_norm": 1.3992512226104736,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 17900
+    },
+    {
+      "epoch": 117.82894736842105,
+      "grad_norm": 1.3927537202835083,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 17910
+    },
+    {
+      "epoch": 117.89473684210526,
+      "grad_norm": 1.443617343902588,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 17920
+    },
+    {
+      "epoch": 117.96052631578948,
+      "grad_norm": 1.7115437984466553,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 17930
+    },
+    {
+      "epoch": 118.02631578947368,
+      "grad_norm": 1.743646502494812,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 17940
+    },
+    {
+      "epoch": 118.09210526315789,
+      "grad_norm": 1.8455554246902466,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 17950
+    },
+    {
+      "epoch": 118.15789473684211,
+      "grad_norm": 1.5267865657806396,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 17960
+    },
+    {
+      "epoch": 118.22368421052632,
+      "grad_norm": 1.3369451761245728,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 17970
+    },
+    {
+      "epoch": 118.28947368421052,
+      "grad_norm": 1.7073959112167358,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 17980
+    },
+    {
+      "epoch": 118.35526315789474,
+      "grad_norm": 1.3288969993591309,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 17990
+    },
+    {
+      "epoch": 118.42105263157895,
+      "grad_norm": 1.353158950805664,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 18000
+    },
+    {
+      "epoch": 118.48684210526316,
+      "grad_norm": 1.260098934173584,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 18010
+    },
+    {
+      "epoch": 118.55263157894737,
+      "grad_norm": 1.0010889768600464,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 18020
+    },
+    {
+      "epoch": 118.61842105263158,
+      "grad_norm": 1.4269499778747559,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 18030
+    },
+    {
+      "epoch": 118.6842105263158,
+      "grad_norm": 1.6440515518188477,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 18040
+    },
+    {
+      "epoch": 118.75,
+      "grad_norm": 1.2828301191329956,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 18050
+    },
+    {
+      "epoch": 118.8157894736842,
+      "grad_norm": 1.632926106452942,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 18060
+    },
+    {
+      "epoch": 118.88157894736842,
+      "grad_norm": 1.6261667013168335,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 18070
+    },
+    {
+      "epoch": 118.94736842105263,
+      "grad_norm": 1.401406168937683,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 18080
+    },
+    {
+      "epoch": 119.01315789473684,
+      "grad_norm": 1.7001771926879883,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 18090
+    },
+    {
+      "epoch": 119.07894736842105,
+      "grad_norm": 1.5548099279403687,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 18100
+    },
+    {
+      "epoch": 119.14473684210526,
+      "grad_norm": 1.1037205457687378,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 18110
+    },
+    {
+      "epoch": 119.21052631578948,
+      "grad_norm": 1.5918753147125244,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 18120
+    },
+    {
+      "epoch": 119.27631578947368,
+      "grad_norm": 1.6971482038497925,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 18130
+    },
+    {
+      "epoch": 119.34210526315789,
+      "grad_norm": 2.4634158611297607,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 18140
+    },
+    {
+      "epoch": 119.40789473684211,
+      "grad_norm": 1.9838364124298096,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 18150
+    },
+    {
+      "epoch": 119.47368421052632,
+      "grad_norm": 1.578308343887329,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 18160
+    },
+    {
+      "epoch": 119.53947368421052,
+      "grad_norm": 1.6289037466049194,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 18170
+    },
+    {
+      "epoch": 119.60526315789474,
+      "grad_norm": 1.2893215417861938,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 18180
+    },
+    {
+      "epoch": 119.67105263157895,
+      "grad_norm": 1.372152328491211,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 18190
+    },
+    {
+      "epoch": 119.73684210526316,
+      "grad_norm": 1.3960975408554077,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 18200
+    },
+    {
+      "epoch": 119.80263157894737,
+      "grad_norm": 2.3912997245788574,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 18210
+    },
+    {
+      "epoch": 119.86842105263158,
+      "grad_norm": 2.194430112838745,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 18220
+    },
+    {
+      "epoch": 119.9342105263158,
+      "grad_norm": 1.995544195175171,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 18230
+    },
+    {
+      "epoch": 120.0,
+      "grad_norm": 2.018141031265259,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 18240
+    },
+    {
+      "epoch": 120.0657894736842,
+      "grad_norm": 2.009296178817749,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 18250
+    },
+    {
+      "epoch": 120.13157894736842,
+      "grad_norm": 1.7988462448120117,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 18260
+    },
+    {
+      "epoch": 120.19736842105263,
+      "grad_norm": 1.3400589227676392,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 18270
+    },
+    {
+      "epoch": 120.26315789473684,
+      "grad_norm": 1.8497594594955444,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 18280
+    },
+    {
+      "epoch": 120.32894736842105,
+      "grad_norm": 1.826569676399231,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 18290
+    },
+    {
+      "epoch": 120.39473684210526,
+      "grad_norm": 1.4031181335449219,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 18300
+    },
+    {
+      "epoch": 120.46052631578948,
+      "grad_norm": 1.3137474060058594,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 18310
+    },
+    {
+      "epoch": 120.52631578947368,
+      "grad_norm": 1.4643906354904175,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 18320
+    },
+    {
+      "epoch": 120.59210526315789,
+      "grad_norm": 1.4734902381896973,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 18330
+    },
+    {
+      "epoch": 120.65789473684211,
+      "grad_norm": 1.2963664531707764,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 18340
+    },
+    {
+      "epoch": 120.72368421052632,
+      "grad_norm": 1.8689401149749756,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 18350
+    },
+    {
+      "epoch": 120.78947368421052,
+      "grad_norm": 1.442025899887085,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 18360
+    },
+    {
+      "epoch": 120.85526315789474,
+      "grad_norm": 1.0878459215164185,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 18370
+    },
+    {
+      "epoch": 120.92105263157895,
+      "grad_norm": 1.1373950242996216,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 18380
+    },
+    {
+      "epoch": 120.98684210526316,
+      "grad_norm": 1.359013319015503,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 18390
+    },
+    {
+      "epoch": 121.05263157894737,
+      "grad_norm": 1.154805302619934,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 18400
+    },
+    {
+      "epoch": 121.11842105263158,
+      "grad_norm": 1.2236026525497437,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 18410
+    },
+    {
+      "epoch": 121.1842105263158,
+      "grad_norm": 1.6635630130767822,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 18420
+    },
+    {
+      "epoch": 121.25,
+      "grad_norm": 1.2726155519485474,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 18430
+    },
+    {
+      "epoch": 121.3157894736842,
+      "grad_norm": 1.6328818798065186,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 18440
+    },
+    {
+      "epoch": 121.38157894736842,
+      "grad_norm": 1.1731946468353271,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 18450
+    },
+    {
+      "epoch": 121.44736842105263,
+      "grad_norm": 1.302308440208435,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 18460
+    },
+    {
+      "epoch": 121.51315789473684,
+      "grad_norm": 1.287872076034546,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 18470
+    },
+    {
+      "epoch": 121.57894736842105,
+      "grad_norm": 1.315284013748169,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 18480
+    },
+    {
+      "epoch": 121.64473684210526,
+      "grad_norm": 1.6699914932250977,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 18490
+    },
+    {
+      "epoch": 121.71052631578948,
+      "grad_norm": 1.2884074449539185,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 18500
+    },
+    {
+      "epoch": 121.77631578947368,
+      "grad_norm": 1.6029454469680786,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 18510
+    },
+    {
+      "epoch": 121.84210526315789,
+      "grad_norm": 1.4443784952163696,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 18520
+    },
+    {
+      "epoch": 121.90789473684211,
+      "grad_norm": 1.3697898387908936,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 18530
+    },
+    {
+      "epoch": 121.97368421052632,
+      "grad_norm": 1.4503991603851318,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 18540
+    },
+    {
+      "epoch": 122.03947368421052,
+      "grad_norm": 2.0889556407928467,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 18550
+    },
+    {
+      "epoch": 122.10526315789474,
+      "grad_norm": 2.1557774543762207,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 18560
+    },
+    {
+      "epoch": 122.17105263157895,
+      "grad_norm": 1.9375776052474976,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 18570
+    },
+    {
+      "epoch": 122.23684210526316,
+      "grad_norm": 1.9048449993133545,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 18580
+    },
+    {
+      "epoch": 122.30263157894737,
+      "grad_norm": 1.9082547426223755,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 18590
+    },
+    {
+      "epoch": 122.36842105263158,
+      "grad_norm": 1.3321175575256348,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 18600
+    },
+    {
+      "epoch": 122.4342105263158,
+      "grad_norm": 1.7989875078201294,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 18610
+    },
+    {
+      "epoch": 122.5,
+      "grad_norm": 1.6709932088851929,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 18620
+    },
+    {
+      "epoch": 122.5657894736842,
+      "grad_norm": 2.0253422260284424,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 18630
+    },
+    {
+      "epoch": 122.63157894736842,
+      "grad_norm": 1.6843767166137695,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 18640
+    },
+    {
+      "epoch": 122.69736842105263,
+      "grad_norm": 1.6629420518875122,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 18650
+    },
+    {
+      "epoch": 122.76315789473684,
+      "grad_norm": 1.5332101583480835,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 18660
+    },
+    {
+      "epoch": 122.82894736842105,
+      "grad_norm": 1.3796052932739258,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 18670
+    },
+    {
+      "epoch": 122.89473684210526,
+      "grad_norm": 1.4094196557998657,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 18680
+    },
+    {
+      "epoch": 122.96052631578948,
+      "grad_norm": 1.7053511142730713,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 18690
+    },
+    {
+      "epoch": 123.02631578947368,
+      "grad_norm": 1.2467565536499023,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 18700
+    },
+    {
+      "epoch": 123.09210526315789,
+      "grad_norm": 1.2807624340057373,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 18710
+    },
+    {
+      "epoch": 123.15789473684211,
+      "grad_norm": 2.067713499069214,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 18720
+    },
+    {
+      "epoch": 123.22368421052632,
+      "grad_norm": 1.4443409442901611,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 18730
+    },
+    {
+      "epoch": 123.28947368421052,
+      "grad_norm": 1.0428988933563232,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 18740
+    },
+    {
+      "epoch": 123.35526315789474,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 18750
+    },
+    {
+      "epoch": 123.42105263157895,
+      "grad_norm": 1.2120792865753174,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 18760
+    },
+    {
+      "epoch": 123.48684210526316,
+      "grad_norm": 1.4466508626937866,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 18770
+    },
+    {
+      "epoch": 123.55263157894737,
+      "grad_norm": 1.5829192399978638,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 18780
+    },
+    {
+      "epoch": 123.61842105263158,
+      "grad_norm": 1.7138676643371582,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 18790
+    },
+    {
+      "epoch": 123.6842105263158,
+      "grad_norm": 1.3385616540908813,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 18800
+    },
+    {
+      "epoch": 123.75,
+      "grad_norm": 1.5545923709869385,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 18810
+    },
+    {
+      "epoch": 123.8157894736842,
+      "grad_norm": 1.66548752784729,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 18820
+    },
+    {
+      "epoch": 123.88157894736842,
+      "grad_norm": 1.5020532608032227,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 18830
+    },
+    {
+      "epoch": 123.94736842105263,
+      "grad_norm": 1.4988902807235718,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 18840
+    },
+    {
+      "epoch": 124.01315789473684,
+      "grad_norm": 1.415207862854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 18850
+    },
+    {
+      "epoch": 124.07894736842105,
+      "grad_norm": 1.9516119956970215,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 18860
+    },
+    {
+      "epoch": 124.14473684210526,
+      "grad_norm": 1.5451509952545166,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 18870
+    },
+    {
+      "epoch": 124.21052631578948,
+      "grad_norm": 2.0787911415100098,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 18880
+    },
+    {
+      "epoch": 124.27631578947368,
+      "grad_norm": 1.574526071548462,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 18890
+    },
+    {
+      "epoch": 124.34210526315789,
+      "grad_norm": 1.3802706003189087,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 18900
+    },
+    {
+      "epoch": 124.40789473684211,
+      "grad_norm": 1.6244707107543945,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 18910
+    },
+    {
+      "epoch": 124.47368421052632,
+      "grad_norm": 2.145286798477173,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 18920
+    },
+    {
+      "epoch": 124.53947368421052,
+      "grad_norm": 1.4865652322769165,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 18930
+    },
+    {
+      "epoch": 124.60526315789474,
+      "grad_norm": 1.6736758947372437,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 18940
+    },
+    {
+      "epoch": 124.67105263157895,
+      "grad_norm": 1.779492735862732,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 18950
+    },
+    {
+      "epoch": 124.73684210526316,
+      "grad_norm": 1.4367070198059082,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 18960
+    },
+    {
+      "epoch": 124.80263157894737,
+      "grad_norm": 2.134930372238159,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 18970
+    },
+    {
+      "epoch": 124.86842105263158,
+      "grad_norm": 1.5714412927627563,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 18980
+    },
+    {
+      "epoch": 124.9342105263158,
+      "grad_norm": 1.3601255416870117,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 18990
+    },
+    {
+      "epoch": 125.0,
+      "grad_norm": 1.4811025857925415,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 19000
+    },
+    {
+      "epoch": 125.0657894736842,
+      "grad_norm": 1.156876564025879,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 19010
+    },
+    {
+      "epoch": 125.13157894736842,
+      "grad_norm": 1.284725546836853,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 19020
+    },
+    {
+      "epoch": 125.19736842105263,
+      "grad_norm": 1.6561628580093384,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 19030
+    },
+    {
+      "epoch": 125.26315789473684,
+      "grad_norm": 1.505050778388977,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 19040
+    },
+    {
+      "epoch": 125.32894736842105,
+      "grad_norm": 1.6309056282043457,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 19050
+    },
+    {
+      "epoch": 125.39473684210526,
+      "grad_norm": 1.736699104309082,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 19060
+    },
+    {
+      "epoch": 125.46052631578948,
+      "grad_norm": 1.9428766965866089,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 19070
+    },
+    {
+      "epoch": 125.52631578947368,
+      "grad_norm": 1.7686620950698853,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 19080
+    },
+    {
+      "epoch": 125.59210526315789,
+      "grad_norm": 1.7622973918914795,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 19090
+    },
+    {
+      "epoch": 125.65789473684211,
+      "grad_norm": 1.5690042972564697,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 19100
+    },
+    {
+      "epoch": 125.72368421052632,
+      "grad_norm": 1.8227565288543701,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 19110
+    },
+    {
+      "epoch": 125.78947368421052,
+      "grad_norm": 1.6870558261871338,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 19120
+    },
+    {
+      "epoch": 125.85526315789474,
+      "grad_norm": 1.479076862335205,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 19130
+    },
+    {
+      "epoch": 125.92105263157895,
+      "grad_norm": 1.2277092933654785,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 19140
+    },
+    {
+      "epoch": 125.98684210526316,
+      "grad_norm": 1.489528775215149,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 19150
+    },
+    {
+      "epoch": 126.05263157894737,
+      "grad_norm": 1.0452980995178223,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 19160
+    },
+    {
+      "epoch": 126.11842105263158,
+      "grad_norm": 1.3360462188720703,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 19170
+    },
+    {
+      "epoch": 126.1842105263158,
+      "grad_norm": 1.3254673480987549,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 19180
+    },
+    {
+      "epoch": 126.25,
+      "grad_norm": 1.2667726278305054,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 19190
+    },
+    {
+      "epoch": 126.3157894736842,
+      "grad_norm": 1.4161853790283203,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 19200
+    },
+    {
+      "epoch": 126.38157894736842,
+      "grad_norm": 1.2914543151855469,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 19210
+    },
+    {
+      "epoch": 126.44736842105263,
+      "grad_norm": 1.231687307357788,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 19220
+    },
+    {
+      "epoch": 126.51315789473684,
+      "grad_norm": 1.5462013483047485,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 19230
+    },
+    {
+      "epoch": 126.57894736842105,
+      "grad_norm": 1.4174609184265137,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 19240
+    },
+    {
+      "epoch": 126.64473684210526,
+      "grad_norm": 1.4570950269699097,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 19250
+    },
+    {
+      "epoch": 126.71052631578948,
+      "grad_norm": 1.3737624883651733,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 19260
+    },
+    {
+      "epoch": 126.77631578947368,
+      "grad_norm": 1.2217376232147217,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 19270
+    },
+    {
+      "epoch": 126.84210526315789,
+      "grad_norm": 1.2423151731491089,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 19280
+    },
+    {
+      "epoch": 126.90789473684211,
+      "grad_norm": 0.9811283349990845,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 19290
+    },
+    {
+      "epoch": 126.97368421052632,
+      "grad_norm": 1.0705347061157227,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 19300
+    },
+    {
+      "epoch": 127.03947368421052,
+      "grad_norm": 1.479683518409729,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 19310
+    },
+    {
+      "epoch": 127.10526315789474,
+      "grad_norm": 1.0638930797576904,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 19320
+    },
+    {
+      "epoch": 127.17105263157895,
+      "grad_norm": 1.0353878736495972,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 19330
+    },
+    {
+      "epoch": 127.23684210526316,
+      "grad_norm": 1.5880337953567505,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 19340
+    },
+    {
+      "epoch": 127.30263157894737,
+      "grad_norm": 1.4664833545684814,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 19350
+    },
+    {
+      "epoch": 127.36842105263158,
+      "grad_norm": 1.302016019821167,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 19360
+    },
+    {
+      "epoch": 127.4342105263158,
+      "grad_norm": 1.1787173748016357,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 19370
+    },
+    {
+      "epoch": 127.5,
+      "grad_norm": 1.5672588348388672,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 19380
+    },
+    {
+      "epoch": 127.5657894736842,
+      "grad_norm": 1.434388518333435,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 19390
+    },
+    {
+      "epoch": 127.63157894736842,
+      "grad_norm": 1.571679949760437,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 19400
+    },
+    {
+      "epoch": 127.69736842105263,
+      "grad_norm": 1.2628552913665771,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 19410
+    },
+    {
+      "epoch": 127.76315789473684,
+      "grad_norm": 1.4900308847427368,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 19420
+    },
+    {
+      "epoch": 127.82894736842105,
+      "grad_norm": 1.2159069776535034,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 19430
+    },
+    {
+      "epoch": 127.89473684210526,
+      "grad_norm": 1.3993160724639893,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 19440
+    },
+    {
+      "epoch": 127.96052631578948,
+      "grad_norm": 1.1536054611206055,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 19450
+    },
+    {
+      "epoch": 128.02631578947367,
+      "grad_norm": 1.333641767501831,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 19460
+    },
+    {
+      "epoch": 128.0921052631579,
+      "grad_norm": 1.1968632936477661,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 19470
+    },
+    {
+      "epoch": 128.1578947368421,
+      "grad_norm": 1.3755112886428833,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 19480
+    },
+    {
+      "epoch": 128.22368421052633,
+      "grad_norm": 1.2440341711044312,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 19490
+    },
+    {
+      "epoch": 128.28947368421052,
+      "grad_norm": 1.4638385772705078,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 19500
+    },
+    {
+      "epoch": 128.35526315789474,
+      "grad_norm": 1.5661355257034302,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 19510
+    },
+    {
+      "epoch": 128.42105263157896,
+      "grad_norm": 1.3939542770385742,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 19520
+    },
+    {
+      "epoch": 128.48684210526315,
+      "grad_norm": 1.1770873069763184,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 19530
+    },
+    {
+      "epoch": 128.55263157894737,
+      "grad_norm": 1.2070765495300293,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 19540
+    },
+    {
+      "epoch": 128.6184210526316,
+      "grad_norm": 1.6263254880905151,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 19550
+    },
+    {
+      "epoch": 128.68421052631578,
+      "grad_norm": 1.2368308305740356,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 19560
+    },
+    {
+      "epoch": 128.75,
+      "grad_norm": 1.4440819025039673,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 19570
+    },
+    {
+      "epoch": 128.81578947368422,
+      "grad_norm": 1.1472845077514648,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 19580
+    },
+    {
+      "epoch": 128.8815789473684,
+      "grad_norm": 1.410331130027771,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 19590
+    },
+    {
+      "epoch": 128.94736842105263,
+      "grad_norm": 1.265761137008667,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 19600
+    },
+    {
+      "epoch": 129.01315789473685,
+      "grad_norm": 1.2967408895492554,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 19610
+    },
+    {
+      "epoch": 129.07894736842104,
+      "grad_norm": 1.5309879779815674,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 19620
+    },
+    {
+      "epoch": 129.14473684210526,
+      "grad_norm": 1.5370832681655884,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 19630
+    },
+    {
+      "epoch": 129.21052631578948,
+      "grad_norm": 1.3466033935546875,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 19640
+    },
+    {
+      "epoch": 129.27631578947367,
+      "grad_norm": 1.4539897441864014,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 19650
+    },
+    {
+      "epoch": 129.3421052631579,
+      "grad_norm": 1.633927583694458,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 19660
+    },
+    {
+      "epoch": 129.4078947368421,
+      "grad_norm": 1.246416449546814,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 19670
+    },
+    {
+      "epoch": 129.47368421052633,
+      "grad_norm": 1.3420946598052979,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 19680
+    },
+    {
+      "epoch": 129.53947368421052,
+      "grad_norm": 1.3653072118759155,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 19690
+    },
+    {
+      "epoch": 129.60526315789474,
+      "grad_norm": 1.0461795330047607,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 19700
+    },
+    {
+      "epoch": 129.67105263157896,
+      "grad_norm": 1.4482468366622925,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 19710
+    },
+    {
+      "epoch": 129.73684210526315,
+      "grad_norm": 1.0564301013946533,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 19720
+    },
+    {
+      "epoch": 129.80263157894737,
+      "grad_norm": 1.5804928541183472,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 19730
+    },
+    {
+      "epoch": 129.8684210526316,
+      "grad_norm": 1.6412838697433472,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 19740
+    },
+    {
+      "epoch": 129.93421052631578,
+      "grad_norm": 1.5836437940597534,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 19750
+    },
+    {
+      "epoch": 130.0,
+      "grad_norm": 1.197196125984192,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 19760
+    },
+    {
+      "epoch": 130.06578947368422,
+      "grad_norm": 0.96909499168396,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 19770
+    },
+    {
+      "epoch": 130.1315789473684,
+      "grad_norm": 1.4088032245635986,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 19780
+    },
+    {
+      "epoch": 130.19736842105263,
+      "grad_norm": 1.3101266622543335,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 19790
+    },
+    {
+      "epoch": 130.26315789473685,
+      "grad_norm": 1.375200867652893,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 19800
+    },
+    {
+      "epoch": 130.32894736842104,
+      "grad_norm": 1.243313193321228,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 19810
+    },
+    {
+      "epoch": 130.39473684210526,
+      "grad_norm": 1.121595025062561,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 19820
+    },
+    {
+      "epoch": 130.46052631578948,
+      "grad_norm": 2.108215093612671,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 19830
+    },
+    {
+      "epoch": 130.52631578947367,
+      "grad_norm": 2.1370813846588135,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 19840
+    },
+    {
+      "epoch": 130.5921052631579,
+      "grad_norm": 1.728285789489746,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 19850
+    },
+    {
+      "epoch": 130.6578947368421,
+      "grad_norm": 2.018031358718872,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 19860
+    },
+    {
+      "epoch": 130.72368421052633,
+      "grad_norm": 1.6380020380020142,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 19870
+    },
+    {
+      "epoch": 130.78947368421052,
+      "grad_norm": 1.7799023389816284,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 19880
+    },
+    {
+      "epoch": 130.85526315789474,
+      "grad_norm": 1.419378638267517,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 19890
+    },
+    {
+      "epoch": 130.92105263157896,
+      "grad_norm": 1.769496202468872,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 19900
+    },
+    {
+      "epoch": 130.98684210526315,
+      "grad_norm": 1.9092391729354858,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 19910
+    },
+    {
+      "epoch": 131.05263157894737,
+      "grad_norm": 1.5821865797042847,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 19920
+    },
+    {
+      "epoch": 131.1184210526316,
+      "grad_norm": 1.5526002645492554,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 19930
+    },
+    {
+      "epoch": 131.18421052631578,
+      "grad_norm": 1.4830267429351807,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 19940
+    },
+    {
+      "epoch": 131.25,
+      "grad_norm": 1.3607596158981323,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 19950
+    },
+    {
+      "epoch": 131.31578947368422,
+      "grad_norm": 1.4321061372756958,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 19960
+    },
+    {
+      "epoch": 131.3815789473684,
+      "grad_norm": 1.343908667564392,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 19970
+    },
+    {
+      "epoch": 131.44736842105263,
+      "grad_norm": 1.047028660774231,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 19980
+    },
+    {
+      "epoch": 131.51315789473685,
+      "grad_norm": 1.2941601276397705,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 19990
+    },
+    {
+      "epoch": 131.57894736842104,
+      "grad_norm": 1.480362892150879,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 20000
+    },
+    {
+      "epoch": 131.64473684210526,
+      "grad_norm": 1.8121306896209717,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 20010
+    },
+    {
+      "epoch": 131.71052631578948,
+      "grad_norm": 1.2839956283569336,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 20020
+    },
+    {
+      "epoch": 131.77631578947367,
+      "grad_norm": 1.1678411960601807,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 20030
+    },
+    {
+      "epoch": 131.8421052631579,
+      "grad_norm": 1.2261030673980713,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 20040
+    },
+    {
+      "epoch": 131.9078947368421,
+      "grad_norm": 1.1341450214385986,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 20050
+    },
+    {
+      "epoch": 131.97368421052633,
+      "grad_norm": 1.0404419898986816,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 20060
+    },
+    {
+      "epoch": 132.03947368421052,
+      "grad_norm": 1.3259118795394897,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 20070
+    },
+    {
+      "epoch": 132.10526315789474,
+      "grad_norm": 1.6572375297546387,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 20080
+    },
+    {
+      "epoch": 132.17105263157896,
+      "grad_norm": 1.5282135009765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 20090
+    },
+    {
+      "epoch": 132.23684210526315,
+      "grad_norm": 1.3130202293395996,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 20100
+    },
+    {
+      "epoch": 132.30263157894737,
+      "grad_norm": 1.7606295347213745,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 20110
+    },
+    {
+      "epoch": 132.3684210526316,
+      "grad_norm": 1.5770694017410278,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 20120
+    },
+    {
+      "epoch": 132.43421052631578,
+      "grad_norm": 1.3118009567260742,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 20130
+    },
+    {
+      "epoch": 132.5,
+      "grad_norm": 1.220281958580017,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 20140
+    },
+    {
+      "epoch": 132.56578947368422,
+      "grad_norm": 1.5257964134216309,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 20150
+    },
+    {
+      "epoch": 132.6315789473684,
+      "grad_norm": 1.567521333694458,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 20160
+    },
+    {
+      "epoch": 132.69736842105263,
+      "grad_norm": 1.1980758905410767,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 20170
+    },
+    {
+      "epoch": 132.76315789473685,
+      "grad_norm": 1.453986644744873,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 20180
+    },
+    {
+      "epoch": 132.82894736842104,
+      "grad_norm": 1.4468498229980469,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 20190
+    },
+    {
+      "epoch": 132.89473684210526,
+      "grad_norm": 1.3943188190460205,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 20200
+    },
+    {
+      "epoch": 132.96052631578948,
+      "grad_norm": 1.7826237678527832,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 20210
+    },
+    {
+      "epoch": 133.02631578947367,
+      "grad_norm": 1.3166335821151733,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 20220
+    },
+    {
+      "epoch": 133.0921052631579,
+      "grad_norm": 1.3654383420944214,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 20230
+    },
+    {
+      "epoch": 133.1578947368421,
+      "grad_norm": 1.322871208190918,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 20240
+    },
+    {
+      "epoch": 133.22368421052633,
+      "grad_norm": 1.2936899662017822,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 20250
+    },
+    {
+      "epoch": 133.28947368421052,
+      "grad_norm": 0.9831916093826294,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 20260
+    },
+    {
+      "epoch": 133.35526315789474,
+      "grad_norm": 1.345542311668396,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 20270
+    },
+    {
+      "epoch": 133.42105263157896,
+      "grad_norm": 1.4100401401519775,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 20280
+    },
+    {
+      "epoch": 133.48684210526315,
+      "grad_norm": 1.1066662073135376,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 20290
+    },
+    {
+      "epoch": 133.55263157894737,
+      "grad_norm": 1.3817609548568726,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 20300
+    },
+    {
+      "epoch": 133.6184210526316,
+      "grad_norm": 1.107347846031189,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 20310
+    },
+    {
+      "epoch": 133.68421052631578,
+      "grad_norm": 0.8152048587799072,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 20320
+    },
+    {
+      "epoch": 133.75,
+      "grad_norm": 1.1018624305725098,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 20330
+    },
+    {
+      "epoch": 133.81578947368422,
+      "grad_norm": 1.5733144283294678,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 20340
+    },
+    {
+      "epoch": 133.8815789473684,
+      "grad_norm": 1.2820268869400024,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 20350
+    },
+    {
+      "epoch": 133.94736842105263,
+      "grad_norm": 1.4256007671356201,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 20360
+    },
+    {
+      "epoch": 134.01315789473685,
+      "grad_norm": 1.2284027338027954,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 20370
+    },
+    {
+      "epoch": 134.07894736842104,
+      "grad_norm": 1.1770983934402466,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 20380
+    },
+    {
+      "epoch": 134.14473684210526,
+      "grad_norm": 1.1191388368606567,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 20390
+    },
+    {
+      "epoch": 134.21052631578948,
+      "grad_norm": 1.1628490686416626,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 20400
+    },
+    {
+      "epoch": 134.27631578947367,
+      "grad_norm": 1.2943140268325806,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 20410
+    },
+    {
+      "epoch": 134.3421052631579,
+      "grad_norm": 1.0791293382644653,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 20420
+    },
+    {
+      "epoch": 134.4078947368421,
+      "grad_norm": 1.308792233467102,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 20430
+    },
+    {
+      "epoch": 134.47368421052633,
+      "grad_norm": 1.2937871217727661,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 20440
+    },
+    {
+      "epoch": 134.53947368421052,
+      "grad_norm": 1.2588351964950562,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 20450
+    },
+    {
+      "epoch": 134.60526315789474,
+      "grad_norm": 1.3284668922424316,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 20460
+    },
+    {
+      "epoch": 134.67105263157896,
+      "grad_norm": 1.3896305561065674,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 20470
+    },
+    {
+      "epoch": 134.73684210526315,
+      "grad_norm": 1.4380180835723877,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 20480
+    },
+    {
+      "epoch": 134.80263157894737,
+      "grad_norm": 1.1442885398864746,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 20490
+    },
+    {
+      "epoch": 134.8684210526316,
+      "grad_norm": 1.3140195608139038,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 20500
+    },
+    {
+      "epoch": 134.93421052631578,
+      "grad_norm": 1.2796006202697754,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 20510
+    },
+    {
+      "epoch": 135.0,
+      "grad_norm": 1.2553552389144897,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 20520
+    },
+    {
+      "epoch": 135.06578947368422,
+      "grad_norm": 1.573137640953064,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 20530
+    },
+    {
+      "epoch": 135.1315789473684,
+      "grad_norm": 1.4506351947784424,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 20540
+    },
+    {
+      "epoch": 135.19736842105263,
+      "grad_norm": 1.219154953956604,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 20550
+    },
+    {
+      "epoch": 135.26315789473685,
+      "grad_norm": 1.235091209411621,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 20560
+    },
+    {
+      "epoch": 135.32894736842104,
+      "grad_norm": 1.7018022537231445,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 20570
+    },
+    {
+      "epoch": 135.39473684210526,
+      "grad_norm": 1.2762569189071655,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 20580
+    },
+    {
+      "epoch": 135.46052631578948,
+      "grad_norm": 1.5831774473190308,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 20590
+    },
+    {
+      "epoch": 135.52631578947367,
+      "grad_norm": 1.3347238302230835,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 20600
+    },
+    {
+      "epoch": 135.5921052631579,
+      "grad_norm": 1.5127307176589966,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 20610
+    },
+    {
+      "epoch": 135.6578947368421,
+      "grad_norm": 1.3495934009552002,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 20620
+    },
+    {
+      "epoch": 135.72368421052633,
+      "grad_norm": 1.6535933017730713,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 20630
+    },
+    {
+      "epoch": 135.78947368421052,
+      "grad_norm": 1.341562032699585,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 20640
+    },
+    {
+      "epoch": 135.85526315789474,
+      "grad_norm": 1.2143185138702393,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 20650
+    },
+    {
+      "epoch": 135.92105263157896,
+      "grad_norm": 1.1850231885910034,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 20660
+    },
+    {
+      "epoch": 135.98684210526315,
+      "grad_norm": 1.3171666860580444,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 20670
+    },
+    {
+      "epoch": 136.05263157894737,
+      "grad_norm": 1.7448344230651855,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 20680
+    },
+    {
+      "epoch": 136.1184210526316,
+      "grad_norm": 1.0638881921768188,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 20690
+    },
+    {
+      "epoch": 136.18421052631578,
+      "grad_norm": 0.8317880034446716,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 20700
+    },
+    {
+      "epoch": 136.25,
+      "grad_norm": 1.029839277267456,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 20710
+    },
+    {
+      "epoch": 136.31578947368422,
+      "grad_norm": 0.9803492426872253,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 20720
+    },
+    {
+      "epoch": 136.3815789473684,
+      "grad_norm": 1.7811894416809082,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 20730
+    },
+    {
+      "epoch": 136.44736842105263,
+      "grad_norm": 1.0440852642059326,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 20740
+    },
+    {
+      "epoch": 136.51315789473685,
+      "grad_norm": 1.3539479970932007,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 20750
+    },
+    {
+      "epoch": 136.57894736842104,
+      "grad_norm": 1.2563800811767578,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 20760
+    },
+    {
+      "epoch": 136.64473684210526,
+      "grad_norm": 0.9620851874351501,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 20770
+    },
+    {
+      "epoch": 136.71052631578948,
+      "grad_norm": 1.340428352355957,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 20780
+    },
+    {
+      "epoch": 136.77631578947367,
+      "grad_norm": 1.2113215923309326,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 20790
+    },
+    {
+      "epoch": 136.8421052631579,
+      "grad_norm": 1.3657537698745728,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 20800
+    },
+    {
+      "epoch": 136.9078947368421,
+      "grad_norm": 1.3240277767181396,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 20810
+    },
+    {
+      "epoch": 136.97368421052633,
+      "grad_norm": 1.2962788343429565,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 20820
+    },
+    {
+      "epoch": 137.03947368421052,
+      "grad_norm": 0.9829322695732117,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 20830
+    },
+    {
+      "epoch": 137.10526315789474,
+      "grad_norm": 1.1034486293792725,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 20840
+    },
+    {
+      "epoch": 137.17105263157896,
+      "grad_norm": 1.254267692565918,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 20850
+    },
+    {
+      "epoch": 137.23684210526315,
+      "grad_norm": 1.059151530265808,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 20860
+    },
+    {
+      "epoch": 137.30263157894737,
+      "grad_norm": 1.5754255056381226,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 20870
+    },
+    {
+      "epoch": 137.3684210526316,
+      "grad_norm": 1.3503801822662354,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 20880
+    },
+    {
+      "epoch": 137.43421052631578,
+      "grad_norm": 1.3268914222717285,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 20890
+    },
+    {
+      "epoch": 137.5,
+      "grad_norm": 1.3608646392822266,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 20900
+    },
+    {
+      "epoch": 137.56578947368422,
+      "grad_norm": 1.1551671028137207,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 20910
+    },
+    {
+      "epoch": 137.6315789473684,
+      "grad_norm": 0.8847004175186157,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 20920
+    },
+    {
+      "epoch": 137.69736842105263,
+      "grad_norm": 1.4831230640411377,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 20930
+    },
+    {
+      "epoch": 137.76315789473685,
+      "grad_norm": 1.1870131492614746,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 20940
+    },
+    {
+      "epoch": 137.82894736842104,
+      "grad_norm": 1.3143157958984375,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 20950
+    },
+    {
+      "epoch": 137.89473684210526,
+      "grad_norm": 1.1769039630889893,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 20960
+    },
+    {
+      "epoch": 137.96052631578948,
+      "grad_norm": 1.852351427078247,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 20970
+    },
+    {
+      "epoch": 138.02631578947367,
+      "grad_norm": 1.6631145477294922,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 20980
+    },
+    {
+      "epoch": 138.0921052631579,
+      "grad_norm": 1.4467190504074097,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 20990
+    },
+    {
+      "epoch": 138.1578947368421,
+      "grad_norm": 1.4761419296264648,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 21000
+    },
+    {
+      "epoch": 138.22368421052633,
+      "grad_norm": 1.160988211631775,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 21010
+    },
+    {
+      "epoch": 138.28947368421052,
+      "grad_norm": 1.591023325920105,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 21020
+    },
+    {
+      "epoch": 138.35526315789474,
+      "grad_norm": 1.4847773313522339,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 21030
+    },
+    {
+      "epoch": 138.42105263157896,
+      "grad_norm": 1.3351836204528809,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 21040
+    },
+    {
+      "epoch": 138.48684210526315,
+      "grad_norm": 1.6248340606689453,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 21050
+    },
+    {
+      "epoch": 138.55263157894737,
+      "grad_norm": 1.3487075567245483,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 21060
+    },
+    {
+      "epoch": 138.6184210526316,
+      "grad_norm": 1.291077733039856,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 21070
+    },
+    {
+      "epoch": 138.68421052631578,
+      "grad_norm": 1.6398788690567017,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 21080
+    },
+    {
+      "epoch": 138.75,
+      "grad_norm": 1.8707720041275024,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 21090
+    },
+    {
+      "epoch": 138.81578947368422,
+      "grad_norm": 1.5530805587768555,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 21100
+    },
+    {
+      "epoch": 138.8815789473684,
+      "grad_norm": 1.3435490131378174,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 21110
+    },
+    {
+      "epoch": 138.94736842105263,
+      "grad_norm": 1.51068115234375,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 21120
+    },
+    {
+      "epoch": 139.01315789473685,
+      "grad_norm": 1.293702244758606,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 21130
+    },
+    {
+      "epoch": 139.07894736842104,
+      "grad_norm": 1.422736406326294,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 21140
+    },
+    {
+      "epoch": 139.14473684210526,
+      "grad_norm": 1.303194522857666,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 21150
+    },
+    {
+      "epoch": 139.21052631578948,
+      "grad_norm": 1.724423885345459,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 21160
+    },
+    {
+      "epoch": 139.27631578947367,
+      "grad_norm": 1.5041424036026,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 21170
+    },
+    {
+      "epoch": 139.3421052631579,
+      "grad_norm": 1.4980031251907349,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 21180
+    },
+    {
+      "epoch": 139.4078947368421,
+      "grad_norm": 1.6551101207733154,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 21190
+    },
+    {
+      "epoch": 139.47368421052633,
+      "grad_norm": 1.4190785884857178,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 21200
+    },
+    {
+      "epoch": 139.53947368421052,
+      "grad_norm": 1.5306510925292969,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 21210
+    },
+    {
+      "epoch": 139.60526315789474,
+      "grad_norm": 1.395179033279419,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 21220
+    },
+    {
+      "epoch": 139.67105263157896,
+      "grad_norm": 1.9223673343658447,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 21230
+    },
+    {
+      "epoch": 139.73684210526315,
+      "grad_norm": 1.42075777053833,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 21240
+    },
+    {
+      "epoch": 139.80263157894737,
+      "grad_norm": 1.3706669807434082,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 21250
+    },
+    {
+      "epoch": 139.8684210526316,
+      "grad_norm": 1.5455049276351929,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 21260
+    },
+    {
+      "epoch": 139.93421052631578,
+      "grad_norm": 1.6650147438049316,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 21270
+    },
+    {
+      "epoch": 140.0,
+      "grad_norm": 1.3898757696151733,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 21280
+    },
+    {
+      "epoch": 140.06578947368422,
+      "grad_norm": 1.516356110572815,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 21290
+    },
+    {
+      "epoch": 140.1315789473684,
+      "grad_norm": 1.162001609802246,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 21300
+    },
+    {
+      "epoch": 140.19736842105263,
+      "grad_norm": 1.5784013271331787,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 21310
+    },
+    {
+      "epoch": 140.26315789473685,
+      "grad_norm": 1.3628367185592651,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 21320
+    },
+    {
+      "epoch": 140.32894736842104,
+      "grad_norm": 1.2204854488372803,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 21330
+    },
+    {
+      "epoch": 140.39473684210526,
+      "grad_norm": 1.3196606636047363,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 21340
+    },
+    {
+      "epoch": 140.46052631578948,
+      "grad_norm": 1.3891950845718384,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 21350
+    },
+    {
+      "epoch": 140.52631578947367,
+      "grad_norm": 0.994315505027771,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 21360
+    },
+    {
+      "epoch": 140.5921052631579,
+      "grad_norm": 1.3768020868301392,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 21370
+    },
+    {
+      "epoch": 140.6578947368421,
+      "grad_norm": 0.8438733816146851,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 21380
+    },
+    {
+      "epoch": 140.72368421052633,
+      "grad_norm": 1.5164715051651,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 21390
+    },
+    {
+      "epoch": 140.78947368421052,
+      "grad_norm": 1.3153681755065918,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 21400
+    },
+    {
+      "epoch": 140.85526315789474,
+      "grad_norm": 1.4804083108901978,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 21410
+    },
+    {
+      "epoch": 140.92105263157896,
+      "grad_norm": 1.6807725429534912,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 21420
+    },
+    {
+      "epoch": 140.98684210526315,
+      "grad_norm": 1.5503164529800415,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 21430
+    },
+    {
+      "epoch": 141.05263157894737,
+      "grad_norm": 1.3315825462341309,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 21440
+    },
+    {
+      "epoch": 141.1184210526316,
+      "grad_norm": 1.2883803844451904,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 21450
+    },
+    {
+      "epoch": 141.18421052631578,
+      "grad_norm": 1.409468173980713,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 21460
+    },
+    {
+      "epoch": 141.25,
+      "grad_norm": 1.7659132480621338,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 21470
+    },
+    {
+      "epoch": 141.31578947368422,
+      "grad_norm": 1.2036762237548828,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 21480
+    },
+    {
+      "epoch": 141.3815789473684,
+      "grad_norm": 1.3691455125808716,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 21490
+    },
+    {
+      "epoch": 141.44736842105263,
+      "grad_norm": 1.2026904821395874,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 21500
+    },
+    {
+      "epoch": 141.51315789473685,
+      "grad_norm": 1.3784390687942505,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 21510
+    },
+    {
+      "epoch": 141.57894736842104,
+      "grad_norm": 1.1604245901107788,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 21520
+    },
+    {
+      "epoch": 141.64473684210526,
+      "grad_norm": 1.6574337482452393,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 21530
+    },
+    {
+      "epoch": 141.71052631578948,
+      "grad_norm": 1.6064780950546265,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 21540
+    },
+    {
+      "epoch": 141.77631578947367,
+      "grad_norm": 1.4753048419952393,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 21550
+    },
+    {
+      "epoch": 141.8421052631579,
+      "grad_norm": 1.6422748565673828,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 21560
+    },
+    {
+      "epoch": 141.9078947368421,
+      "grad_norm": 1.4057989120483398,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 21570
+    },
+    {
+      "epoch": 141.97368421052633,
+      "grad_norm": 1.7741460800170898,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 21580
+    },
+    {
+      "epoch": 142.03947368421052,
+      "grad_norm": 1.278627872467041,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 21590
+    },
+    {
+      "epoch": 142.10526315789474,
+      "grad_norm": 1.6472762823104858,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 21600
+    },
+    {
+      "epoch": 142.17105263157896,
+      "grad_norm": 1.5067228078842163,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 21610
+    },
+    {
+      "epoch": 142.23684210526315,
+      "grad_norm": 1.04597008228302,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 21620
+    },
+    {
+      "epoch": 142.30263157894737,
+      "grad_norm": 1.3342257738113403,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 21630
+    },
+    {
+      "epoch": 142.3684210526316,
+      "grad_norm": 1.3006070852279663,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 21640
+    },
+    {
+      "epoch": 142.43421052631578,
+      "grad_norm": 1.0908511877059937,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 21650
+    },
+    {
+      "epoch": 142.5,
+      "grad_norm": 1.156730055809021,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 21660
+    },
+    {
+      "epoch": 142.56578947368422,
+      "grad_norm": 1.5827045440673828,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 21670
+    },
+    {
+      "epoch": 142.6315789473684,
+      "grad_norm": 1.2614614963531494,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 21680
+    },
+    {
+      "epoch": 142.69736842105263,
+      "grad_norm": 1.386438012123108,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 21690
+    },
+    {
+      "epoch": 142.76315789473685,
+      "grad_norm": 1.0151652097702026,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 21700
+    },
+    {
+      "epoch": 142.82894736842104,
+      "grad_norm": 1.2075785398483276,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 21710
+    },
+    {
+      "epoch": 142.89473684210526,
+      "grad_norm": 1.2344067096710205,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 21720
+    },
+    {
+      "epoch": 142.96052631578948,
+      "grad_norm": 1.3653433322906494,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 21730
+    },
+    {
+      "epoch": 143.02631578947367,
+      "grad_norm": 1.1140528917312622,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 21740
+    },
+    {
+      "epoch": 143.0921052631579,
+      "grad_norm": 1.543173909187317,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 21750
+    },
+    {
+      "epoch": 143.1578947368421,
+      "grad_norm": 1.6480592489242554,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 21760
+    },
+    {
+      "epoch": 143.22368421052633,
+      "grad_norm": 1.5367249250411987,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 21770
+    },
+    {
+      "epoch": 143.28947368421052,
+      "grad_norm": 1.3302656412124634,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 21780
+    },
+    {
+      "epoch": 143.35526315789474,
+      "grad_norm": 1.3698561191558838,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 21790
+    },
+    {
+      "epoch": 143.42105263157896,
+      "grad_norm": 1.6688646078109741,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 21800
+    },
+    {
+      "epoch": 143.48684210526315,
+      "grad_norm": 0.9735197424888611,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 21810
+    },
+    {
+      "epoch": 143.55263157894737,
+      "grad_norm": 1.394163727760315,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 21820
+    },
+    {
+      "epoch": 143.6184210526316,
+      "grad_norm": 1.4685237407684326,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 21830
+    },
+    {
+      "epoch": 143.68421052631578,
+      "grad_norm": 1.2259248495101929,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 21840
+    },
+    {
+      "epoch": 143.75,
+      "grad_norm": 1.0960712432861328,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 21850
+    },
+    {
+      "epoch": 143.81578947368422,
+      "grad_norm": 1.4755326509475708,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 21860
+    },
+    {
+      "epoch": 143.8815789473684,
+      "grad_norm": 1.1997923851013184,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 21870
+    },
+    {
+      "epoch": 143.94736842105263,
+      "grad_norm": 1.0807325839996338,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 21880
+    },
+    {
+      "epoch": 144.01315789473685,
+      "grad_norm": 1.120715618133545,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 21890
+    },
+    {
+      "epoch": 144.07894736842104,
+      "grad_norm": 1.401428461074829,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 21900
+    },
+    {
+      "epoch": 144.14473684210526,
+      "grad_norm": 1.0058794021606445,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 21910
+    },
+    {
+      "epoch": 144.21052631578948,
+      "grad_norm": 1.4190456867218018,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 21920
+    },
+    {
+      "epoch": 144.27631578947367,
+      "grad_norm": 1.0206199884414673,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 21930
+    },
+    {
+      "epoch": 144.3421052631579,
+      "grad_norm": 1.3440083265304565,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 21940
+    },
+    {
+      "epoch": 144.4078947368421,
+      "grad_norm": 1.0611833333969116,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 21950
+    },
+    {
+      "epoch": 144.47368421052633,
+      "grad_norm": 1.4812649488449097,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 21960
+    },
+    {
+      "epoch": 144.53947368421052,
+      "grad_norm": 1.3473953008651733,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 21970
+    },
+    {
+      "epoch": 144.60526315789474,
+      "grad_norm": 1.229573130607605,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 21980
+    },
+    {
+      "epoch": 144.67105263157896,
+      "grad_norm": 1.0524625778198242,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 21990
+    },
+    {
+      "epoch": 144.73684210526315,
+      "grad_norm": 1.317591905593872,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 22000
+    },
+    {
+      "epoch": 144.80263157894737,
+      "grad_norm": 1.3425631523132324,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 22010
+    },
+    {
+      "epoch": 144.8684210526316,
+      "grad_norm": 1.0790326595306396,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 22020
+    },
+    {
+      "epoch": 144.93421052631578,
+      "grad_norm": 1.0952657461166382,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 22030
+    },
+    {
+      "epoch": 145.0,
+      "grad_norm": 1.3941048383712769,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 22040
+    },
+    {
+      "epoch": 145.06578947368422,
+      "grad_norm": 1.4520479440689087,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 22050
+    },
+    {
+      "epoch": 145.1315789473684,
+      "grad_norm": 1.3955209255218506,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 22060
+    },
+    {
+      "epoch": 145.19736842105263,
+      "grad_norm": 1.3725961446762085,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 22070
+    },
+    {
+      "epoch": 145.26315789473685,
+      "grad_norm": 1.536729335784912,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 22080
+    },
+    {
+      "epoch": 145.32894736842104,
+      "grad_norm": 1.2806124687194824,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 22090
+    },
+    {
+      "epoch": 145.39473684210526,
+      "grad_norm": 1.2393194437026978,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 22100
+    },
+    {
+      "epoch": 145.46052631578948,
+      "grad_norm": 1.6960039138793945,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 22110
+    },
+    {
+      "epoch": 145.52631578947367,
+      "grad_norm": 1.5707310438156128,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 22120
+    },
+    {
+      "epoch": 145.5921052631579,
+      "grad_norm": 1.2192633152008057,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 22130
+    },
+    {
+      "epoch": 145.6578947368421,
+      "grad_norm": 1.4482849836349487,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 22140
+    },
+    {
+      "epoch": 145.72368421052633,
+      "grad_norm": 1.1989715099334717,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 22150
+    },
+    {
+      "epoch": 145.78947368421052,
+      "grad_norm": 1.3403609991073608,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 22160
+    },
+    {
+      "epoch": 145.85526315789474,
+      "grad_norm": 1.599914789199829,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 22170
+    },
+    {
+      "epoch": 145.92105263157896,
+      "grad_norm": 1.602010726928711,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 22180
+    },
+    {
+      "epoch": 145.98684210526315,
+      "grad_norm": 1.2780475616455078,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 22190
+    },
+    {
+      "epoch": 146.05263157894737,
+      "grad_norm": 1.600659728050232,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 22200
+    },
+    {
+      "epoch": 146.1184210526316,
+      "grad_norm": 1.0254467725753784,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 22210
+    },
+    {
+      "epoch": 146.18421052631578,
+      "grad_norm": 1.2700848579406738,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 22220
+    },
+    {
+      "epoch": 146.25,
+      "grad_norm": 1.2667077779769897,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 22230
+    },
+    {
+      "epoch": 146.31578947368422,
+      "grad_norm": 1.057997703552246,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 22240
+    },
+    {
+      "epoch": 146.3815789473684,
+      "grad_norm": 1.3093483448028564,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 22250
+    },
+    {
+      "epoch": 146.44736842105263,
+      "grad_norm": 1.0066701173782349,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 22260
+    },
+    {
+      "epoch": 146.51315789473685,
+      "grad_norm": 1.212918758392334,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 22270
+    },
+    {
+      "epoch": 146.57894736842104,
+      "grad_norm": 1.3692690134048462,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 22280
+    },
+    {
+      "epoch": 146.64473684210526,
+      "grad_norm": 1.1842776536941528,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 22290
+    },
+    {
+      "epoch": 146.71052631578948,
+      "grad_norm": 1.0390976667404175,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 22300
+    },
+    {
+      "epoch": 146.77631578947367,
+      "grad_norm": 0.8841727375984192,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 22310
+    },
+    {
+      "epoch": 146.8421052631579,
+      "grad_norm": 1.2545442581176758,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 22320
+    },
+    {
+      "epoch": 146.9078947368421,
+      "grad_norm": 1.3116580247879028,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 22330
+    },
+    {
+      "epoch": 146.97368421052633,
+      "grad_norm": 0.9272478222846985,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 22340
+    },
+    {
+      "epoch": 147.03947368421052,
+      "grad_norm": 1.2526774406433105,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 22350
+    },
+    {
+      "epoch": 147.10526315789474,
+      "grad_norm": 1.276404857635498,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 22360
+    },
+    {
+      "epoch": 147.17105263157896,
+      "grad_norm": 1.035636305809021,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 22370
+    },
+    {
+      "epoch": 147.23684210526315,
+      "grad_norm": 1.1112178564071655,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 22380
+    },
+    {
+      "epoch": 147.30263157894737,
+      "grad_norm": 1.3240015506744385,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 22390
+    },
+    {
+      "epoch": 147.3684210526316,
+      "grad_norm": 1.081715703010559,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 22400
+    },
+    {
+      "epoch": 147.43421052631578,
+      "grad_norm": 1.243489384651184,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 22410
+    },
+    {
+      "epoch": 147.5,
+      "grad_norm": 1.5017218589782715,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 22420
+    },
+    {
+      "epoch": 147.56578947368422,
+      "grad_norm": 0.9764308929443359,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 22430
+    },
+    {
+      "epoch": 147.6315789473684,
+      "grad_norm": 1.4389960765838623,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 22440
+    },
+    {
+      "epoch": 147.69736842105263,
+      "grad_norm": 1.4686914682388306,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 22450
+    },
+    {
+      "epoch": 147.76315789473685,
+      "grad_norm": 1.3333423137664795,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 22460
+    },
+    {
+      "epoch": 147.82894736842104,
+      "grad_norm": 1.3306684494018555,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 22470
+    },
+    {
+      "epoch": 147.89473684210526,
+      "grad_norm": 1.4141942262649536,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 22480
+    },
+    {
+      "epoch": 147.96052631578948,
+      "grad_norm": 1.542427659034729,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 22490
+    },
+    {
+      "epoch": 148.02631578947367,
+      "grad_norm": 1.7996158599853516,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 22500
+    },
+    {
+      "epoch": 148.0921052631579,
+      "grad_norm": 0.9638149738311768,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 22510
+    },
+    {
+      "epoch": 148.1578947368421,
+      "grad_norm": 1.074876070022583,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 22520
+    },
+    {
+      "epoch": 148.22368421052633,
+      "grad_norm": 1.0280288457870483,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 22530
+    },
+    {
+      "epoch": 148.28947368421052,
+      "grad_norm": 1.0511701107025146,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 22540
+    },
+    {
+      "epoch": 148.35526315789474,
+      "grad_norm": 1.678219199180603,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 22550
+    },
+    {
+      "epoch": 148.42105263157896,
+      "grad_norm": 1.0990034341812134,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 22560
+    },
+    {
+      "epoch": 148.48684210526315,
+      "grad_norm": 1.601233720779419,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 22570
+    },
+    {
+      "epoch": 148.55263157894737,
+      "grad_norm": 1.4822688102722168,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 22580
+    },
+    {
+      "epoch": 148.6184210526316,
+      "grad_norm": 1.492310881614685,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 22590
+    },
+    {
+      "epoch": 148.68421052631578,
+      "grad_norm": 1.3018604516983032,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 22600
+    },
+    {
+      "epoch": 148.75,
+      "grad_norm": 1.4880129098892212,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 22610
+    },
+    {
+      "epoch": 148.81578947368422,
+      "grad_norm": 1.6518675088882446,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 22620
+    },
+    {
+      "epoch": 148.8815789473684,
+      "grad_norm": 1.6061004400253296,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 22630
+    },
+    {
+      "epoch": 148.94736842105263,
+      "grad_norm": 1.5879257917404175,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 22640
+    },
+    {
+      "epoch": 149.01315789473685,
+      "grad_norm": 1.7779239416122437,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 22650
+    },
+    {
+      "epoch": 149.07894736842104,
+      "grad_norm": 1.654737949371338,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 22660
+    },
+    {
+      "epoch": 149.14473684210526,
+      "grad_norm": 1.3545973300933838,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 22670
+    },
+    {
+      "epoch": 149.21052631578948,
+      "grad_norm": 1.3350344896316528,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 22680
+    },
+    {
+      "epoch": 149.27631578947367,
+      "grad_norm": 1.253635287284851,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 22690
+    },
+    {
+      "epoch": 149.3421052631579,
+      "grad_norm": 1.4453550577163696,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 22700
+    },
+    {
+      "epoch": 149.4078947368421,
+      "grad_norm": 1.421644926071167,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 22710
+    },
+    {
+      "epoch": 149.47368421052633,
+      "grad_norm": 1.5055420398712158,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 22720
+    },
+    {
+      "epoch": 149.53947368421052,
+      "grad_norm": 1.4135552644729614,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 22730
+    },
+    {
+      "epoch": 149.60526315789474,
+      "grad_norm": 1.5352897644042969,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 22740
+    },
+    {
+      "epoch": 149.67105263157896,
+      "grad_norm": 1.561647891998291,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 22750
+    },
+    {
+      "epoch": 149.73684210526315,
+      "grad_norm": 1.3064342737197876,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 22760
+    },
+    {
+      "epoch": 149.80263157894737,
+      "grad_norm": 1.1616345643997192,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 22770
+    },
+    {
+      "epoch": 149.8684210526316,
+      "grad_norm": 1.3709357976913452,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 22780
+    },
+    {
+      "epoch": 149.93421052631578,
+      "grad_norm": 1.3530012369155884,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 22790
+    },
+    {
+      "epoch": 150.0,
+      "grad_norm": 1.3412154912948608,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 22800
+    },
+    {
+      "epoch": 150.06578947368422,
+      "grad_norm": 1.2247885465621948,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 22810
+    },
+    {
+      "epoch": 150.1315789473684,
+      "grad_norm": 1.5520859956741333,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 22820
+    },
+    {
+      "epoch": 150.19736842105263,
+      "grad_norm": 1.8670061826705933,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 22830
+    },
+    {
+      "epoch": 150.26315789473685,
+      "grad_norm": 2.187220573425293,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 22840
+    },
+    {
+      "epoch": 150.32894736842104,
+      "grad_norm": 1.9027713537216187,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 22850
+    },
+    {
+      "epoch": 150.39473684210526,
+      "grad_norm": 1.7409223318099976,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 22860
+    },
+    {
+      "epoch": 150.46052631578948,
+      "grad_norm": 1.7857537269592285,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 22870
+    },
+    {
+      "epoch": 150.52631578947367,
+      "grad_norm": 1.6792347431182861,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 22880
+    },
+    {
+      "epoch": 150.5921052631579,
+      "grad_norm": 1.2226988077163696,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 22890
+    },
+    {
+      "epoch": 150.6578947368421,
+      "grad_norm": 2.2091164588928223,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 22900
+    },
+    {
+      "epoch": 150.72368421052633,
+      "grad_norm": 2.2626655101776123,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 22910
+    },
+    {
+      "epoch": 150.78947368421052,
+      "grad_norm": 2.42525315284729,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 22920
+    },
+    {
+      "epoch": 150.85526315789474,
+      "grad_norm": 1.7006795406341553,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 22930
+    },
+    {
+      "epoch": 150.92105263157896,
+      "grad_norm": 1.7463268041610718,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 22940
+    },
+    {
+      "epoch": 150.98684210526315,
+      "grad_norm": 1.5665723085403442,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 22950
+    },
+    {
+      "epoch": 151.05263157894737,
+      "grad_norm": 1.7291148900985718,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 22960
+    },
+    {
+      "epoch": 151.1184210526316,
+      "grad_norm": 1.3778690099716187,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 22970
+    },
+    {
+      "epoch": 151.18421052631578,
+      "grad_norm": 1.612257480621338,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 22980
+    },
+    {
+      "epoch": 151.25,
+      "grad_norm": 1.3075007200241089,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 22990
+    },
+    {
+      "epoch": 151.31578947368422,
+      "grad_norm": 1.2902040481567383,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 23000
+    },
+    {
+      "epoch": 151.3815789473684,
+      "grad_norm": 1.3315701484680176,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 23010
+    },
+    {
+      "epoch": 151.44736842105263,
+      "grad_norm": 1.0458896160125732,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 23020
+    },
+    {
+      "epoch": 151.51315789473685,
+      "grad_norm": 1.1520403623580933,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 23030
+    },
+    {
+      "epoch": 151.57894736842104,
+      "grad_norm": 1.5853742361068726,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 23040
+    },
+    {
+      "epoch": 151.64473684210526,
+      "grad_norm": 1.2538849115371704,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 23050
+    },
+    {
+      "epoch": 151.71052631578948,
+      "grad_norm": 1.6762452125549316,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 23060
+    },
+    {
+      "epoch": 151.77631578947367,
+      "grad_norm": 1.7446788549423218,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 23070
+    },
+    {
+      "epoch": 151.8421052631579,
+      "grad_norm": 0.9342503547668457,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 23080
+    },
+    {
+      "epoch": 151.9078947368421,
+      "grad_norm": 1.1515876054763794,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 23090
+    },
+    {
+      "epoch": 151.97368421052633,
+      "grad_norm": 1.6669334173202515,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 23100
+    },
+    {
+      "epoch": 152.03947368421052,
+      "grad_norm": 1.5257030725479126,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 23110
+    },
+    {
+      "epoch": 152.10526315789474,
+      "grad_norm": 1.6215834617614746,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 23120
+    },
+    {
+      "epoch": 152.17105263157896,
+      "grad_norm": 1.7707445621490479,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 23130
+    },
+    {
+      "epoch": 152.23684210526315,
+      "grad_norm": 1.1109079122543335,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 23140
+    },
+    {
+      "epoch": 152.30263157894737,
+      "grad_norm": 1.362051010131836,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 23150
+    },
+    {
+      "epoch": 152.3684210526316,
+      "grad_norm": 1.205551028251648,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 23160
+    },
+    {
+      "epoch": 152.43421052631578,
+      "grad_norm": 1.1164907217025757,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 23170
+    },
+    {
+      "epoch": 152.5,
+      "grad_norm": 1.4803227186203003,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 23180
+    },
+    {
+      "epoch": 152.56578947368422,
+      "grad_norm": 1.282638430595398,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 23190
+    },
+    {
+      "epoch": 152.6315789473684,
+      "grad_norm": 1.3923507928848267,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 23200
+    },
+    {
+      "epoch": 152.69736842105263,
+      "grad_norm": 1.3010979890823364,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 23210
+    },
+    {
+      "epoch": 152.76315789473685,
+      "grad_norm": 1.3184207677841187,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 23220
+    },
+    {
+      "epoch": 152.82894736842104,
+      "grad_norm": 1.2217152118682861,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 23230
+    },
+    {
+      "epoch": 152.89473684210526,
+      "grad_norm": 3.6102325916290283,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 23240
+    },
+    {
+      "epoch": 152.96052631578948,
+      "grad_norm": 1.7692070007324219,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 23250
+    },
+    {
+      "epoch": 153.02631578947367,
+      "grad_norm": 1.6135303974151611,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 23260
+    },
+    {
+      "epoch": 153.0921052631579,
+      "grad_norm": 1.7957147359848022,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 23270
+    },
+    {
+      "epoch": 153.1578947368421,
+      "grad_norm": 1.8844883441925049,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 23280
+    },
+    {
+      "epoch": 153.22368421052633,
+      "grad_norm": 1.8668259382247925,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 23290
+    },
+    {
+      "epoch": 153.28947368421052,
+      "grad_norm": 1.8679157495498657,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 23300
+    },
+    {
+      "epoch": 153.35526315789474,
+      "grad_norm": 1.436536192893982,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 23310
+    },
+    {
+      "epoch": 153.42105263157896,
+      "grad_norm": 1.4835171699523926,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 23320
+    },
+    {
+      "epoch": 153.48684210526315,
+      "grad_norm": 1.3552966117858887,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 23330
+    },
+    {
+      "epoch": 153.55263157894737,
+      "grad_norm": 1.5863488912582397,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 23340
+    },
+    {
+      "epoch": 153.6184210526316,
+      "grad_norm": 1.2556802034378052,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 23350
+    },
+    {
+      "epoch": 153.68421052631578,
+      "grad_norm": 1.7405736446380615,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 23360
+    },
+    {
+      "epoch": 153.75,
+      "grad_norm": 1.4233875274658203,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 23370
+    },
+    {
+      "epoch": 153.81578947368422,
+      "grad_norm": 1.3720194101333618,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 23380
+    },
+    {
+      "epoch": 153.8815789473684,
+      "grad_norm": 1.481738567352295,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 23390
+    },
+    {
+      "epoch": 153.94736842105263,
+      "grad_norm": 1.8668274879455566,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 23400
+    },
+    {
+      "epoch": 154.01315789473685,
+      "grad_norm": 1.2795329093933105,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 23410
+    },
+    {
+      "epoch": 154.07894736842104,
+      "grad_norm": 1.5393232107162476,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 23420
+    },
+    {
+      "epoch": 154.14473684210526,
+      "grad_norm": 1.5710614919662476,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 23430
+    },
+    {
+      "epoch": 154.21052631578948,
+      "grad_norm": 1.5256733894348145,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 23440
+    },
+    {
+      "epoch": 154.27631578947367,
+      "grad_norm": 1.4345033168792725,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 23450
+    },
+    {
+      "epoch": 154.3421052631579,
+      "grad_norm": 1.4099990129470825,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 23460
+    },
+    {
+      "epoch": 154.4078947368421,
+      "grad_norm": 1.6725177764892578,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 23470
+    },
+    {
+      "epoch": 154.47368421052633,
+      "grad_norm": 1.6998838186264038,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 23480
+    },
+    {
+      "epoch": 154.53947368421052,
+      "grad_norm": 1.105302095413208,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 23490
+    },
+    {
+      "epoch": 154.60526315789474,
+      "grad_norm": 1.0682851076126099,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 23500
+    },
+    {
+      "epoch": 154.67105263157896,
+      "grad_norm": 1.4762492179870605,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 23510
+    },
+    {
+      "epoch": 154.73684210526315,
+      "grad_norm": 1.3171006441116333,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 23520
+    },
+    {
+      "epoch": 154.80263157894737,
+      "grad_norm": 1.1788697242736816,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 23530
+    },
+    {
+      "epoch": 154.8684210526316,
+      "grad_norm": 1.425657868385315,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 23540
+    },
+    {
+      "epoch": 154.93421052631578,
+      "grad_norm": 1.547235369682312,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 23550
+    },
+    {
+      "epoch": 155.0,
+      "grad_norm": 1.6571357250213623,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 23560
+    },
+    {
+      "epoch": 155.06578947368422,
+      "grad_norm": 3.123242139816284,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 23570
+    },
+    {
+      "epoch": 155.1315789473684,
+      "grad_norm": 1.6413636207580566,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 23580
+    },
+    {
+      "epoch": 155.19736842105263,
+      "grad_norm": 1.3690893650054932,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 23590
+    },
+    {
+      "epoch": 155.26315789473685,
+      "grad_norm": 1.3904107809066772,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 23600
+    },
+    {
+      "epoch": 155.32894736842104,
+      "grad_norm": 1.816664457321167,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 23610
+    },
+    {
+      "epoch": 155.39473684210526,
+      "grad_norm": 1.4407306909561157,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 23620
+    },
+    {
+      "epoch": 155.46052631578948,
+      "grad_norm": 1.6005239486694336,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 23630
+    },
+    {
+      "epoch": 155.52631578947367,
+      "grad_norm": 1.4299267530441284,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 23640
+    },
+    {
+      "epoch": 155.5921052631579,
+      "grad_norm": 1.0841288566589355,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 23650
+    },
+    {
+      "epoch": 155.6578947368421,
+      "grad_norm": 1.3383357524871826,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 23660
+    },
+    {
+      "epoch": 155.72368421052633,
+      "grad_norm": 1.0255568027496338,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 23670
+    },
+    {
+      "epoch": 155.78947368421052,
+      "grad_norm": 1.336609959602356,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 23680
+    },
+    {
+      "epoch": 155.85526315789474,
+      "grad_norm": 1.3863122463226318,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 23690
+    },
+    {
+      "epoch": 155.92105263157896,
+      "grad_norm": 1.3259609937667847,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 23700
+    },
+    {
+      "epoch": 155.98684210526315,
+      "grad_norm": 1.0066629648208618,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 23710
+    },
+    {
+      "epoch": 156.05263157894737,
+      "grad_norm": 1.1141328811645508,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 23720
+    },
+    {
+      "epoch": 156.1184210526316,
+      "grad_norm": 1.0179667472839355,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 23730
+    },
+    {
+      "epoch": 156.18421052631578,
+      "grad_norm": 1.498150110244751,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 23740
+    },
+    {
+      "epoch": 156.25,
+      "grad_norm": 1.844730257987976,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 23750
+    },
+    {
+      "epoch": 156.31578947368422,
+      "grad_norm": 1.3886311054229736,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 23760
+    },
+    {
+      "epoch": 156.3815789473684,
+      "grad_norm": 1.3924232721328735,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 23770
+    },
+    {
+      "epoch": 156.44736842105263,
+      "grad_norm": 1.7710256576538086,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 23780
+    },
+    {
+      "epoch": 156.51315789473685,
+      "grad_norm": 1.4801042079925537,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 23790
+    },
+    {
+      "epoch": 156.57894736842104,
+      "grad_norm": 1.1061052083969116,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 23800
+    },
+    {
+      "epoch": 156.64473684210526,
+      "grad_norm": 1.3435893058776855,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 23810
+    },
+    {
+      "epoch": 156.71052631578948,
+      "grad_norm": 1.2930083274841309,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 23820
+    },
+    {
+      "epoch": 156.77631578947367,
+      "grad_norm": 1.1935070753097534,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 23830
+    },
+    {
+      "epoch": 156.8421052631579,
+      "grad_norm": 1.4816370010375977,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 23840
+    },
+    {
+      "epoch": 156.9078947368421,
+      "grad_norm": 1.39240300655365,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 23850
+    },
+    {
+      "epoch": 156.97368421052633,
+      "grad_norm": 1.3462928533554077,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 23860
+    },
+    {
+      "epoch": 157.03947368421052,
+      "grad_norm": 1.2192853689193726,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 23870
+    },
+    {
+      "epoch": 157.10526315789474,
+      "grad_norm": 1.2664424180984497,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 23880
+    },
+    {
+      "epoch": 157.17105263157896,
+      "grad_norm": 1.3305250406265259,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 23890
+    },
+    {
+      "epoch": 157.23684210526315,
+      "grad_norm": 1.512176513671875,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 23900
+    },
+    {
+      "epoch": 157.30263157894737,
+      "grad_norm": 1.5272941589355469,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 23910
+    },
+    {
+      "epoch": 157.3684210526316,
+      "grad_norm": 1.281473994255066,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 23920
+    },
+    {
+      "epoch": 157.43421052631578,
+      "grad_norm": 1.4534245729446411,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 23930
+    },
+    {
+      "epoch": 157.5,
+      "grad_norm": 1.065725564956665,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 23940
+    },
+    {
+      "epoch": 157.56578947368422,
+      "grad_norm": 1.2938498258590698,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 23950
+    },
+    {
+      "epoch": 157.6315789473684,
+      "grad_norm": 1.5139905214309692,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 23960
+    },
+    {
+      "epoch": 157.69736842105263,
+      "grad_norm": 1.1256234645843506,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 23970
+    },
+    {
+      "epoch": 157.76315789473685,
+      "grad_norm": 1.3171970844268799,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 23980
+    },
+    {
+      "epoch": 157.82894736842104,
+      "grad_norm": 1.7550626993179321,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 23990
+    },
+    {
+      "epoch": 157.89473684210526,
+      "grad_norm": 1.2105461359024048,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 24000
+    },
+    {
+      "epoch": 157.96052631578948,
+      "grad_norm": 1.7495108842849731,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 24010
+    },
+    {
+      "epoch": 158.02631578947367,
+      "grad_norm": 1.5057815313339233,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 24020
+    },
+    {
+      "epoch": 158.0921052631579,
+      "grad_norm": 1.364128828048706,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 24030
+    },
+    {
+      "epoch": 158.1578947368421,
+      "grad_norm": 1.6452478170394897,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 24040
+    },
+    {
+      "epoch": 158.22368421052633,
+      "grad_norm": 1.2079576253890991,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 24050
+    },
+    {
+      "epoch": 158.28947368421052,
+      "grad_norm": 1.1615091562271118,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 24060
+    },
+    {
+      "epoch": 158.35526315789474,
+      "grad_norm": 1.7258861064910889,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 24070
+    },
+    {
+      "epoch": 158.42105263157896,
+      "grad_norm": 1.5624957084655762,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 24080
+    },
+    {
+      "epoch": 158.48684210526315,
+      "grad_norm": 1.3713676929473877,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 24090
+    },
+    {
+      "epoch": 158.55263157894737,
+      "grad_norm": 0.9600792527198792,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 24100
+    },
+    {
+      "epoch": 158.6184210526316,
+      "grad_norm": 1.1764682531356812,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 24110
+    },
+    {
+      "epoch": 158.68421052631578,
+      "grad_norm": 0.8807045221328735,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 24120
+    },
+    {
+      "epoch": 158.75,
+      "grad_norm": 1.495734691619873,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 24130
+    },
+    {
+      "epoch": 158.81578947368422,
+      "grad_norm": 1.2996577024459839,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 24140
+    },
+    {
+      "epoch": 158.8815789473684,
+      "grad_norm": 1.3335479497909546,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 24150
+    },
+    {
+      "epoch": 158.94736842105263,
+      "grad_norm": 1.3486828804016113,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 24160
+    },
+    {
+      "epoch": 159.01315789473685,
+      "grad_norm": 1.1436874866485596,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 24170
+    },
+    {
+      "epoch": 159.07894736842104,
+      "grad_norm": 1.0566798448562622,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 24180
+    },
+    {
+      "epoch": 159.14473684210526,
+      "grad_norm": 1.3232210874557495,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 24190
+    },
+    {
+      "epoch": 159.21052631578948,
+      "grad_norm": 1.1854642629623413,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 24200
+    },
+    {
+      "epoch": 159.27631578947367,
+      "grad_norm": 1.3037360906600952,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 24210
+    },
+    {
+      "epoch": 159.3421052631579,
+      "grad_norm": 1.1624709367752075,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 24220
+    },
+    {
+      "epoch": 159.4078947368421,
+      "grad_norm": 1.279187798500061,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 24230
+    },
+    {
+      "epoch": 159.47368421052633,
+      "grad_norm": 1.1820547580718994,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 24240
+    },
+    {
+      "epoch": 159.53947368421052,
+      "grad_norm": 1.3856960535049438,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 24250
+    },
+    {
+      "epoch": 159.60526315789474,
+      "grad_norm": 1.538688063621521,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 24260
+    },
+    {
+      "epoch": 159.67105263157896,
+      "grad_norm": 1.3758467435836792,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 24270
+    },
+    {
+      "epoch": 159.73684210526315,
+      "grad_norm": 1.637129306793213,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 24280
+    },
+    {
+      "epoch": 159.80263157894737,
+      "grad_norm": 1.3896411657333374,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 24290
+    },
+    {
+      "epoch": 159.8684210526316,
+      "grad_norm": 1.1721456050872803,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 24300
+    },
+    {
+      "epoch": 159.93421052631578,
+      "grad_norm": 1.9932554960250854,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 24310
+    },
+    {
+      "epoch": 160.0,
+      "grad_norm": 1.3891863822937012,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 24320
+    },
+    {
+      "epoch": 160.06578947368422,
+      "grad_norm": 1.4516315460205078,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 24330
+    },
+    {
+      "epoch": 160.1315789473684,
+      "grad_norm": 1.426121473312378,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 24340
+    },
+    {
+      "epoch": 160.19736842105263,
+      "grad_norm": 1.5220839977264404,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 24350
+    },
+    {
+      "epoch": 160.26315789473685,
+      "grad_norm": 1.4482239484786987,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 24360
+    },
+    {
+      "epoch": 160.32894736842104,
+      "grad_norm": 1.5488955974578857,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 24370
+    },
+    {
+      "epoch": 160.39473684210526,
+      "grad_norm": 1.3807427883148193,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 24380
+    },
+    {
+      "epoch": 160.46052631578948,
+      "grad_norm": 1.0160894393920898,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 24390
+    },
+    {
+      "epoch": 160.52631578947367,
+      "grad_norm": 1.2377235889434814,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 24400
+    },
+    {
+      "epoch": 160.5921052631579,
+      "grad_norm": 0.987001895904541,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 24410
+    },
+    {
+      "epoch": 160.6578947368421,
+      "grad_norm": 1.317531704902649,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 24420
+    },
+    {
+      "epoch": 160.72368421052633,
+      "grad_norm": 1.521753191947937,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 24430
+    },
+    {
+      "epoch": 160.78947368421052,
+      "grad_norm": 1.166321039199829,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 24440
+    },
+    {
+      "epoch": 160.85526315789474,
+      "grad_norm": 1.1440471410751343,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 24450
+    },
+    {
+      "epoch": 160.92105263157896,
+      "grad_norm": 1.3936820030212402,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 24460
+    },
+    {
+      "epoch": 160.98684210526315,
+      "grad_norm": 1.1521437168121338,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 24470
+    },
+    {
+      "epoch": 161.05263157894737,
+      "grad_norm": 1.0062541961669922,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 24480
+    },
+    {
+      "epoch": 161.1184210526316,
+      "grad_norm": 1.126672625541687,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 24490
+    },
+    {
+      "epoch": 161.18421052631578,
+      "grad_norm": 0.8864695429801941,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 24500
+    },
+    {
+      "epoch": 161.25,
+      "grad_norm": 1.3956273794174194,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 24510
+    },
+    {
+      "epoch": 161.31578947368422,
+      "grad_norm": 1.4042059183120728,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 24520
+    },
+    {
+      "epoch": 161.3815789473684,
+      "grad_norm": 1.1768255233764648,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 24530
+    },
+    {
+      "epoch": 161.44736842105263,
+      "grad_norm": 1.468287467956543,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 24540
+    },
+    {
+      "epoch": 161.51315789473685,
+      "grad_norm": 1.4978389739990234,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 24550
+    },
+    {
+      "epoch": 161.57894736842104,
+      "grad_norm": 0.9304846525192261,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 24560
+    },
+    {
+      "epoch": 161.64473684210526,
+      "grad_norm": 1.3616132736206055,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 24570
+    },
+    {
+      "epoch": 161.71052631578948,
+      "grad_norm": 1.0822436809539795,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 24580
+    },
+    {
+      "epoch": 161.77631578947367,
+      "grad_norm": 1.1985223293304443,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 24590
+    },
+    {
+      "epoch": 161.8421052631579,
+      "grad_norm": 1.1661827564239502,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 24600
+    },
+    {
+      "epoch": 161.9078947368421,
+      "grad_norm": 1.2504955530166626,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 24610
+    },
+    {
+      "epoch": 161.97368421052633,
+      "grad_norm": 1.2397123575210571,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 24620
+    },
+    {
+      "epoch": 162.03947368421052,
+      "grad_norm": 1.0715943574905396,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 24630
+    },
+    {
+      "epoch": 162.10526315789474,
+      "grad_norm": 1.1244536638259888,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 24640
+    },
+    {
+      "epoch": 162.17105263157896,
+      "grad_norm": 1.3353065252304077,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 24650
+    },
+    {
+      "epoch": 162.23684210526315,
+      "grad_norm": 1.408204436302185,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 24660
+    },
+    {
+      "epoch": 162.30263157894737,
+      "grad_norm": 1.3387621641159058,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 24670
+    },
+    {
+      "epoch": 162.3684210526316,
+      "grad_norm": 1.7283587455749512,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 24680
+    },
+    {
+      "epoch": 162.43421052631578,
+      "grad_norm": 1.5374019145965576,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 24690
+    },
+    {
+      "epoch": 162.5,
+      "grad_norm": 1.2795945405960083,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 24700
+    },
+    {
+      "epoch": 162.56578947368422,
+      "grad_norm": 1.537787675857544,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 24710
+    },
+    {
+      "epoch": 162.6315789473684,
+      "grad_norm": 1.3099159002304077,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 24720
+    },
+    {
+      "epoch": 162.69736842105263,
+      "grad_norm": 1.4521063566207886,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 24730
+    },
+    {
+      "epoch": 162.76315789473685,
+      "grad_norm": 1.0471484661102295,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 24740
+    },
+    {
+      "epoch": 162.82894736842104,
+      "grad_norm": 1.6101007461547852,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 24750
+    },
+    {
+      "epoch": 162.89473684210526,
+      "grad_norm": 1.3530237674713135,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 24760
+    },
+    {
+      "epoch": 162.96052631578948,
+      "grad_norm": 1.4511295557022095,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 24770
+    },
+    {
+      "epoch": 163.02631578947367,
+      "grad_norm": 1.6022499799728394,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 24780
+    },
+    {
+      "epoch": 163.0921052631579,
+      "grad_norm": 1.4935600757598877,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 24790
+    },
+    {
+      "epoch": 163.1578947368421,
+      "grad_norm": 1.327233910560608,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 24800
+    },
+    {
+      "epoch": 163.22368421052633,
+      "grad_norm": 1.6861796379089355,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 24810
+    },
+    {
+      "epoch": 163.28947368421052,
+      "grad_norm": 1.4635359048843384,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 24820
+    },
+    {
+      "epoch": 163.35526315789474,
+      "grad_norm": 1.0945830345153809,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 24830
+    },
+    {
+      "epoch": 163.42105263157896,
+      "grad_norm": 1.359348177909851,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 24840
+    },
+    {
+      "epoch": 163.48684210526315,
+      "grad_norm": 1.269120216369629,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 24850
+    },
+    {
+      "epoch": 163.55263157894737,
+      "grad_norm": 1.2218973636627197,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 24860
+    },
+    {
+      "epoch": 163.6184210526316,
+      "grad_norm": 1.1602321863174438,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 24870
+    },
+    {
+      "epoch": 163.68421052631578,
+      "grad_norm": 1.5064598321914673,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 24880
+    },
+    {
+      "epoch": 163.75,
+      "grad_norm": 1.5285717248916626,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 24890
+    },
+    {
+      "epoch": 163.81578947368422,
+      "grad_norm": 1.0176746845245361,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 24900
+    },
+    {
+      "epoch": 163.8815789473684,
+      "grad_norm": 1.0195715427398682,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 24910
+    },
+    {
+      "epoch": 163.94736842105263,
+      "grad_norm": 1.253736972808838,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 24920
+    },
+    {
+      "epoch": 164.01315789473685,
+      "grad_norm": 1.370121717453003,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 24930
+    },
+    {
+      "epoch": 164.07894736842104,
+      "grad_norm": 1.2714835405349731,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 24940
+    },
+    {
+      "epoch": 164.14473684210526,
+      "grad_norm": 1.4336186647415161,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 24950
+    },
+    {
+      "epoch": 164.21052631578948,
+      "grad_norm": 1.3978872299194336,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 24960
+    },
+    {
+      "epoch": 164.27631578947367,
+      "grad_norm": 1.19012451171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 24970
+    },
+    {
+      "epoch": 164.3421052631579,
+      "grad_norm": 1.0551053285598755,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 24980
+    },
+    {
+      "epoch": 164.4078947368421,
+      "grad_norm": 1.8229308128356934,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 24990
+    },
+    {
+      "epoch": 164.47368421052633,
+      "grad_norm": 0.7936663627624512,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 25000
+    },
+    {
+      "epoch": 164.53947368421052,
+      "grad_norm": 1.1978473663330078,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 25010
+    },
+    {
+      "epoch": 164.60526315789474,
+      "grad_norm": 1.305327296257019,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 25020
+    },
+    {
+      "epoch": 164.67105263157896,
+      "grad_norm": 1.2756356000900269,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 25030
+    },
+    {
+      "epoch": 164.73684210526315,
+      "grad_norm": 0.8239328861236572,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 25040
+    },
+    {
+      "epoch": 164.80263157894737,
+      "grad_norm": 1.0603528022766113,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 25050
+    },
+    {
+      "epoch": 164.8684210526316,
+      "grad_norm": 1.0184378623962402,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 25060
+    },
+    {
+      "epoch": 164.93421052631578,
+      "grad_norm": 1.0379877090454102,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 25070
+    },
+    {
+      "epoch": 165.0,
+      "grad_norm": 1.1607532501220703,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 25080
+    },
+    {
+      "epoch": 165.06578947368422,
+      "grad_norm": 1.4055289030075073,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 25090
+    },
+    {
+      "epoch": 165.1315789473684,
+      "grad_norm": 1.3009381294250488,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 25100
+    },
+    {
+      "epoch": 165.19736842105263,
+      "grad_norm": 1.4090827703475952,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 25110
+    },
+    {
+      "epoch": 165.26315789473685,
+      "grad_norm": 1.2835659980773926,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 25120
+    },
+    {
+      "epoch": 165.32894736842104,
+      "grad_norm": 1.6038185358047485,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 25130
+    },
+    {
+      "epoch": 165.39473684210526,
+      "grad_norm": 1.2217997312545776,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 25140
+    },
+    {
+      "epoch": 165.46052631578948,
+      "grad_norm": 1.1289339065551758,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 25150
+    },
+    {
+      "epoch": 165.52631578947367,
+      "grad_norm": 1.2177705764770508,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 25160
+    },
+    {
+      "epoch": 165.5921052631579,
+      "grad_norm": 1.6059798002243042,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 25170
+    },
+    {
+      "epoch": 165.6578947368421,
+      "grad_norm": 1.921615481376648,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 25180
+    },
+    {
+      "epoch": 165.72368421052633,
+      "grad_norm": 1.20645010471344,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 25190
+    },
+    {
+      "epoch": 165.78947368421052,
+      "grad_norm": 1.2785398960113525,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 25200
+    },
+    {
+      "epoch": 165.85526315789474,
+      "grad_norm": 1.6781924962997437,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 25210
+    },
+    {
+      "epoch": 165.92105263157896,
+      "grad_norm": 0.9301381707191467,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 25220
+    },
+    {
+      "epoch": 165.98684210526315,
+      "grad_norm": 1.2392312288284302,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 25230
+    },
+    {
+      "epoch": 166.05263157894737,
+      "grad_norm": 1.5881590843200684,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 25240
+    },
+    {
+      "epoch": 166.1184210526316,
+      "grad_norm": 1.0536082983016968,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 25250
+    },
+    {
+      "epoch": 166.18421052631578,
+      "grad_norm": 1.475879192352295,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 25260
+    },
+    {
+      "epoch": 166.25,
+      "grad_norm": 1.0223356485366821,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 25270
+    },
+    {
+      "epoch": 166.31578947368422,
+      "grad_norm": 1.4303029775619507,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 25280
+    },
+    {
+      "epoch": 166.3815789473684,
+      "grad_norm": 1.3278456926345825,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 25290
+    },
+    {
+      "epoch": 166.44736842105263,
+      "grad_norm": 1.6226589679718018,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 25300
+    },
+    {
+      "epoch": 166.51315789473685,
+      "grad_norm": 1.4535454511642456,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 25310
+    },
+    {
+      "epoch": 166.57894736842104,
+      "grad_norm": 1.3479703664779663,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 25320
+    },
+    {
+      "epoch": 166.64473684210526,
+      "grad_norm": 1.2640882730484009,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 25330
+    },
+    {
+      "epoch": 166.71052631578948,
+      "grad_norm": 0.9760680198669434,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 25340
+    },
+    {
+      "epoch": 166.77631578947367,
+      "grad_norm": 1.1063249111175537,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 25350
+    },
+    {
+      "epoch": 166.8421052631579,
+      "grad_norm": 1.4282013177871704,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 25360
+    },
+    {
+      "epoch": 166.9078947368421,
+      "grad_norm": 1.7794874906539917,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 25370
+    },
+    {
+      "epoch": 166.97368421052633,
+      "grad_norm": 1.1673434972763062,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 25380
+    },
+    {
+      "epoch": 167.03947368421052,
+      "grad_norm": 1.1143444776535034,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 25390
+    },
+    {
+      "epoch": 167.10526315789474,
+      "grad_norm": 1.4228686094284058,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 25400
+    },
+    {
+      "epoch": 167.17105263157896,
+      "grad_norm": 1.154889464378357,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 25410
+    },
+    {
+      "epoch": 167.23684210526315,
+      "grad_norm": 1.4954147338867188,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 25420
+    },
+    {
+      "epoch": 167.30263157894737,
+      "grad_norm": 1.0341747999191284,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 25430
+    },
+    {
+      "epoch": 167.3684210526316,
+      "grad_norm": 1.333762526512146,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 25440
+    },
+    {
+      "epoch": 167.43421052631578,
+      "grad_norm": 1.4361519813537598,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 25450
+    },
+    {
+      "epoch": 167.5,
+      "grad_norm": 1.2971889972686768,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 25460
+    },
+    {
+      "epoch": 167.56578947368422,
+      "grad_norm": 1.3549476861953735,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 25470
+    },
+    {
+      "epoch": 167.6315789473684,
+      "grad_norm": 1.283470630645752,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 25480
+    },
+    {
+      "epoch": 167.69736842105263,
+      "grad_norm": 1.2950304746627808,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 25490
+    },
+    {
+      "epoch": 167.76315789473685,
+      "grad_norm": 1.133094072341919,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 25500
+    },
+    {
+      "epoch": 167.82894736842104,
+      "grad_norm": 1.097690463066101,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 25510
+    },
+    {
+      "epoch": 167.89473684210526,
+      "grad_norm": 1.6458208560943604,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 25520
+    },
+    {
+      "epoch": 167.96052631578948,
+      "grad_norm": 1.4381608963012695,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 25530
+    },
+    {
+      "epoch": 168.02631578947367,
+      "grad_norm": 1.225541591644287,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 25540
+    },
+    {
+      "epoch": 168.0921052631579,
+      "grad_norm": 1.082775592803955,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 25550
+    },
+    {
+      "epoch": 168.1578947368421,
+      "grad_norm": 1.2318446636199951,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 25560
+    },
+    {
+      "epoch": 168.22368421052633,
+      "grad_norm": 1.5468242168426514,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 25570
+    },
+    {
+      "epoch": 168.28947368421052,
+      "grad_norm": 1.3016541004180908,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 25580
+    },
+    {
+      "epoch": 168.35526315789474,
+      "grad_norm": 1.2177584171295166,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 25590
+    },
+    {
+      "epoch": 168.42105263157896,
+      "grad_norm": 1.384822130203247,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 25600
+    },
+    {
+      "epoch": 168.48684210526315,
+      "grad_norm": 1.5428193807601929,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 25610
+    },
+    {
+      "epoch": 168.55263157894737,
+      "grad_norm": 1.5225193500518799,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 25620
+    },
+    {
+      "epoch": 168.6184210526316,
+      "grad_norm": 1.4411053657531738,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 25630
+    },
+    {
+      "epoch": 168.68421052631578,
+      "grad_norm": 1.386864185333252,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 25640
+    },
+    {
+      "epoch": 168.75,
+      "grad_norm": 1.2823785543441772,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 25650
+    },
+    {
+      "epoch": 168.81578947368422,
+      "grad_norm": 1.1967064142227173,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 25660
+    },
+    {
+      "epoch": 168.8815789473684,
+      "grad_norm": 1.3014733791351318,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 25670
+    },
+    {
+      "epoch": 168.94736842105263,
+      "grad_norm": 1.2117063999176025,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 25680
+    },
+    {
+      "epoch": 169.01315789473685,
+      "grad_norm": 1.3568642139434814,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 25690
+    },
+    {
+      "epoch": 169.07894736842104,
+      "grad_norm": 0.8560911417007446,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 25700
+    },
+    {
+      "epoch": 169.14473684210526,
+      "grad_norm": 1.223082184791565,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 25710
+    },
+    {
+      "epoch": 169.21052631578948,
+      "grad_norm": 1.027945637702942,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 25720
+    },
+    {
+      "epoch": 169.27631578947367,
+      "grad_norm": 1.2788870334625244,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 25730
+    },
+    {
+      "epoch": 169.3421052631579,
+      "grad_norm": 1.5068094730377197,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 25740
+    },
+    {
+      "epoch": 169.4078947368421,
+      "grad_norm": 1.1870827674865723,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 25750
+    },
+    {
+      "epoch": 169.47368421052633,
+      "grad_norm": 1.2578105926513672,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 25760
+    },
+    {
+      "epoch": 169.53947368421052,
+      "grad_norm": 1.2269151210784912,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 25770
+    },
+    {
+      "epoch": 169.60526315789474,
+      "grad_norm": 1.482102870941162,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 25780
+    },
+    {
+      "epoch": 169.67105263157896,
+      "grad_norm": 1.2745014429092407,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 25790
+    },
+    {
+      "epoch": 169.73684210526315,
+      "grad_norm": 1.4906076192855835,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 25800
+    },
+    {
+      "epoch": 169.80263157894737,
+      "grad_norm": 0.9935338497161865,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 25810
+    },
+    {
+      "epoch": 169.8684210526316,
+      "grad_norm": 1.6479783058166504,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 25820
+    },
+    {
+      "epoch": 169.93421052631578,
+      "grad_norm": 1.0743275880813599,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 25830
+    },
+    {
+      "epoch": 170.0,
+      "grad_norm": 1.0243290662765503,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 25840
+    },
+    {
+      "epoch": 170.06578947368422,
+      "grad_norm": 1.5142858028411865,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 25850
+    },
+    {
+      "epoch": 170.1315789473684,
+      "grad_norm": 0.9342432618141174,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 25860
+    },
+    {
+      "epoch": 170.19736842105263,
+      "grad_norm": 1.4976930618286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 25870
+    },
+    {
+      "epoch": 170.26315789473685,
+      "grad_norm": 1.565292477607727,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 25880
+    },
+    {
+      "epoch": 170.32894736842104,
+      "grad_norm": 1.1127831935882568,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 25890
+    },
+    {
+      "epoch": 170.39473684210526,
+      "grad_norm": 1.198778748512268,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 25900
+    },
+    {
+      "epoch": 170.46052631578948,
+      "grad_norm": 1.2573102712631226,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 25910
+    },
+    {
+      "epoch": 170.52631578947367,
+      "grad_norm": 1.2907289266586304,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 25920
+    },
+    {
+      "epoch": 170.5921052631579,
+      "grad_norm": 1.4278053045272827,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 25930
+    },
+    {
+      "epoch": 170.6578947368421,
+      "grad_norm": 1.3394439220428467,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 25940
+    },
+    {
+      "epoch": 170.72368421052633,
+      "grad_norm": 1.1726481914520264,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 25950
+    },
+    {
+      "epoch": 170.78947368421052,
+      "grad_norm": 1.369882583618164,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 25960
+    },
+    {
+      "epoch": 170.85526315789474,
+      "grad_norm": 1.3367946147918701,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 25970
+    },
+    {
+      "epoch": 170.92105263157896,
+      "grad_norm": 1.3555912971496582,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 25980
+    },
+    {
+      "epoch": 170.98684210526315,
+      "grad_norm": 1.1395307779312134,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 25990
+    },
+    {
+      "epoch": 171.05263157894737,
+      "grad_norm": 1.1588562726974487,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 26000
+    },
+    {
+      "epoch": 171.1184210526316,
+      "grad_norm": 0.930858850479126,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 26010
+    },
+    {
+      "epoch": 171.18421052631578,
+      "grad_norm": 0.9874327778816223,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 26020
+    },
+    {
+      "epoch": 171.25,
+      "grad_norm": 1.0993543863296509,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 26030
+    },
+    {
+      "epoch": 171.31578947368422,
+      "grad_norm": 1.1052964925765991,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 26040
+    },
+    {
+      "epoch": 171.3815789473684,
+      "grad_norm": 1.25705885887146,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 26050
+    },
+    {
+      "epoch": 171.44736842105263,
+      "grad_norm": 0.7609035968780518,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 26060
+    },
+    {
+      "epoch": 171.51315789473685,
+      "grad_norm": 0.779438316822052,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 26070
+    },
+    {
+      "epoch": 171.57894736842104,
+      "grad_norm": 0.9345868229866028,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 26080
+    },
+    {
+      "epoch": 171.64473684210526,
+      "grad_norm": 1.416816234588623,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 26090
+    },
+    {
+      "epoch": 171.71052631578948,
+      "grad_norm": 1.5457994937896729,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 26100
+    },
+    {
+      "epoch": 171.77631578947367,
+      "grad_norm": 1.3715592622756958,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 26110
+    },
+    {
+      "epoch": 171.8421052631579,
+      "grad_norm": 1.3556997776031494,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 26120
+    },
+    {
+      "epoch": 171.9078947368421,
+      "grad_norm": 1.2203997373580933,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 26130
+    },
+    {
+      "epoch": 171.97368421052633,
+      "grad_norm": 1.4606289863586426,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 26140
+    },
+    {
+      "epoch": 172.03947368421052,
+      "grad_norm": 1.5704396963119507,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 26150
+    },
+    {
+      "epoch": 172.10526315789474,
+      "grad_norm": 1.2381999492645264,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 26160
+    },
+    {
+      "epoch": 172.17105263157896,
+      "grad_norm": 1.518921136856079,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 26170
+    },
+    {
+      "epoch": 172.23684210526315,
+      "grad_norm": 1.4631500244140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 26180
+    },
+    {
+      "epoch": 172.30263157894737,
+      "grad_norm": 1.0997428894042969,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 26190
+    },
+    {
+      "epoch": 172.3684210526316,
+      "grad_norm": 1.2960116863250732,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 26200
+    },
+    {
+      "epoch": 172.43421052631578,
+      "grad_norm": 1.2885228395462036,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 26210
+    },
+    {
+      "epoch": 172.5,
+      "grad_norm": 1.0690890550613403,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 26220
+    },
+    {
+      "epoch": 172.56578947368422,
+      "grad_norm": 1.4473509788513184,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 26230
+    },
+    {
+      "epoch": 172.6315789473684,
+      "grad_norm": 0.9004012942314148,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 26240
+    },
+    {
+      "epoch": 172.69736842105263,
+      "grad_norm": 1.4914005994796753,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 26250
+    },
+    {
+      "epoch": 172.76315789473685,
+      "grad_norm": 1.30367112159729,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 26260
+    },
+    {
+      "epoch": 172.82894736842104,
+      "grad_norm": 1.3740594387054443,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 26270
+    },
+    {
+      "epoch": 172.89473684210526,
+      "grad_norm": 1.5620779991149902,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 26280
+    },
+    {
+      "epoch": 172.96052631578948,
+      "grad_norm": 1.559910774230957,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 26290
+    },
+    {
+      "epoch": 173.02631578947367,
+      "grad_norm": 1.3926976919174194,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 26300
+    },
+    {
+      "epoch": 173.0921052631579,
+      "grad_norm": 1.1394861936569214,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 26310
+    },
+    {
+      "epoch": 173.1578947368421,
+      "grad_norm": 1.7277958393096924,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 26320
+    },
+    {
+      "epoch": 173.22368421052633,
+      "grad_norm": 1.507736325263977,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 26330
+    },
+    {
+      "epoch": 173.28947368421052,
+      "grad_norm": 1.7787176370620728,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 26340
+    },
+    {
+      "epoch": 173.35526315789474,
+      "grad_norm": 1.5719088315963745,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 26350
+    },
+    {
+      "epoch": 173.42105263157896,
+      "grad_norm": 1.6575725078582764,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 26360
+    },
+    {
+      "epoch": 173.48684210526315,
+      "grad_norm": 1.9089598655700684,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 26370
+    },
+    {
+      "epoch": 173.55263157894737,
+      "grad_norm": 1.8127702474594116,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 26380
+    },
+    {
+      "epoch": 173.6184210526316,
+      "grad_norm": 2.279390573501587,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 26390
+    },
+    {
+      "epoch": 173.68421052631578,
+      "grad_norm": 1.7834361791610718,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 26400
+    },
+    {
+      "epoch": 173.75,
+      "grad_norm": 2.3080971240997314,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 26410
+    },
+    {
+      "epoch": 173.81578947368422,
+      "grad_norm": 1.2840913534164429,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 26420
+    },
+    {
+      "epoch": 173.8815789473684,
+      "grad_norm": 1.8161916732788086,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 26430
+    },
+    {
+      "epoch": 173.94736842105263,
+      "grad_norm": 2.031912088394165,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 26440
+    },
+    {
+      "epoch": 174.01315789473685,
+      "grad_norm": 1.5057018995285034,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 26450
+    },
+    {
+      "epoch": 174.07894736842104,
+      "grad_norm": 1.6998172998428345,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 26460
+    },
+    {
+      "epoch": 174.14473684210526,
+      "grad_norm": 1.6755802631378174,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 26470
+    },
+    {
+      "epoch": 174.21052631578948,
+      "grad_norm": 1.45081627368927,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 26480
+    },
+    {
+      "epoch": 174.27631578947367,
+      "grad_norm": 1.4480472803115845,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 26490
+    },
+    {
+      "epoch": 174.3421052631579,
+      "grad_norm": 1.2978633642196655,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 26500
+    },
+    {
+      "epoch": 174.4078947368421,
+      "grad_norm": 1.4956743717193604,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 26510
+    },
+    {
+      "epoch": 174.47368421052633,
+      "grad_norm": 1.184404969215393,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 26520
+    },
+    {
+      "epoch": 174.53947368421052,
+      "grad_norm": 1.209301233291626,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 26530
+    },
+    {
+      "epoch": 174.60526315789474,
+      "grad_norm": 1.2950125932693481,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 26540
+    },
+    {
+      "epoch": 174.67105263157896,
+      "grad_norm": 1.8026961088180542,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 26550
+    },
+    {
+      "epoch": 174.73684210526315,
+      "grad_norm": 1.2904764413833618,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 26560
+    },
+    {
+      "epoch": 174.80263157894737,
+      "grad_norm": 1.365148663520813,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 26570
+    },
+    {
+      "epoch": 174.8684210526316,
+      "grad_norm": 1.2511751651763916,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 26580
+    },
+    {
+      "epoch": 174.93421052631578,
+      "grad_norm": 1.4057785272598267,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 26590
+    },
+    {
+      "epoch": 175.0,
+      "grad_norm": 1.572277545928955,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 26600
+    },
+    {
+      "epoch": 175.06578947368422,
+      "grad_norm": 1.2321019172668457,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 26610
+    },
+    {
+      "epoch": 175.1315789473684,
+      "grad_norm": 1.2357227802276611,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 26620
+    },
+    {
+      "epoch": 175.19736842105263,
+      "grad_norm": 1.3698261976242065,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 26630
+    },
+    {
+      "epoch": 175.26315789473685,
+      "grad_norm": 1.2478382587432861,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 26640
+    },
+    {
+      "epoch": 175.32894736842104,
+      "grad_norm": 1.074097990989685,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 26650
+    },
+    {
+      "epoch": 175.39473684210526,
+      "grad_norm": 0.874796986579895,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 26660
+    },
+    {
+      "epoch": 175.46052631578948,
+      "grad_norm": 1.2885454893112183,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 26670
+    },
+    {
+      "epoch": 175.52631578947367,
+      "grad_norm": 1.255919337272644,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 26680
+    },
+    {
+      "epoch": 175.5921052631579,
+      "grad_norm": 1.7973171472549438,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 26690
+    },
+    {
+      "epoch": 175.6578947368421,
+      "grad_norm": 1.4382481575012207,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 26700
+    },
+    {
+      "epoch": 175.72368421052633,
+      "grad_norm": 1.3824963569641113,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 26710
+    },
+    {
+      "epoch": 175.78947368421052,
+      "grad_norm": 1.2797788381576538,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 26720
+    },
+    {
+      "epoch": 175.85526315789474,
+      "grad_norm": 1.3257101774215698,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 26730
+    },
+    {
+      "epoch": 175.92105263157896,
+      "grad_norm": 1.3095109462738037,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 26740
+    },
+    {
+      "epoch": 175.98684210526315,
+      "grad_norm": 1.2997289896011353,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 26750
+    },
+    {
+      "epoch": 176.05263157894737,
+      "grad_norm": 1.637865662574768,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 26760
+    },
+    {
+      "epoch": 176.1184210526316,
+      "grad_norm": 1.384114384651184,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 26770
+    },
+    {
+      "epoch": 176.18421052631578,
+      "grad_norm": 1.4303343296051025,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 26780
+    },
+    {
+      "epoch": 176.25,
+      "grad_norm": 1.5146687030792236,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 26790
+    },
+    {
+      "epoch": 176.31578947368422,
+      "grad_norm": 1.5877227783203125,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 26800
+    },
+    {
+      "epoch": 176.3815789473684,
+      "grad_norm": 1.3442049026489258,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 26810
+    },
+    {
+      "epoch": 176.44736842105263,
+      "grad_norm": 1.266574740409851,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 26820
+    },
+    {
+      "epoch": 176.51315789473685,
+      "grad_norm": 1.4955312013626099,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 26830
+    },
+    {
+      "epoch": 176.57894736842104,
+      "grad_norm": 0.9446787238121033,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 26840
+    },
+    {
+      "epoch": 176.64473684210526,
+      "grad_norm": 0.9993739724159241,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 26850
+    },
+    {
+      "epoch": 176.71052631578948,
+      "grad_norm": 1.365684151649475,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 26860
+    },
+    {
+      "epoch": 176.77631578947367,
+      "grad_norm": 1.4473934173583984,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 26870
+    },
+    {
+      "epoch": 176.8421052631579,
+      "grad_norm": 1.6463544368743896,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 26880
+    },
+    {
+      "epoch": 176.9078947368421,
+      "grad_norm": 1.3166974782943726,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 26890
+    },
+    {
+      "epoch": 176.97368421052633,
+      "grad_norm": 1.214211106300354,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 26900
+    },
+    {
+      "epoch": 177.03947368421052,
+      "grad_norm": 1.529462218284607,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 26910
+    },
+    {
+      "epoch": 177.10526315789474,
+      "grad_norm": 1.0761915445327759,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 26920
+    },
+    {
+      "epoch": 177.17105263157896,
+      "grad_norm": 1.2023556232452393,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 26930
+    },
+    {
+      "epoch": 177.23684210526315,
+      "grad_norm": 1.5598783493041992,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 26940
+    },
+    {
+      "epoch": 177.30263157894737,
+      "grad_norm": 1.2606966495513916,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 26950
+    },
+    {
+      "epoch": 177.3684210526316,
+      "grad_norm": 1.6140894889831543,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 26960
+    },
+    {
+      "epoch": 177.43421052631578,
+      "grad_norm": 1.3021560907363892,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 26970
+    },
+    {
+      "epoch": 177.5,
+      "grad_norm": 1.0216045379638672,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 26980
+    },
+    {
+      "epoch": 177.56578947368422,
+      "grad_norm": 0.9635993242263794,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 26990
+    },
+    {
+      "epoch": 177.6315789473684,
+      "grad_norm": 1.2217941284179688,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 27000
+    },
+    {
+      "epoch": 177.69736842105263,
+      "grad_norm": 1.3411375284194946,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 27010
+    },
+    {
+      "epoch": 177.76315789473685,
+      "grad_norm": 0.7628015279769897,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 27020
+    },
+    {
+      "epoch": 177.82894736842104,
+      "grad_norm": 0.9084556698799133,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 27030
+    },
+    {
+      "epoch": 177.89473684210526,
+      "grad_norm": 1.1490297317504883,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 27040
+    },
+    {
+      "epoch": 177.96052631578948,
+      "grad_norm": 1.4998666048049927,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 27050
+    },
+    {
+      "epoch": 178.02631578947367,
+      "grad_norm": 1.2536399364471436,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 27060
+    },
+    {
+      "epoch": 178.0921052631579,
+      "grad_norm": 1.217274785041809,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 27070
+    },
+    {
+      "epoch": 178.1578947368421,
+      "grad_norm": 1.3491017818450928,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 27080
+    },
+    {
+      "epoch": 178.22368421052633,
+      "grad_norm": 1.25252366065979,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 27090
+    },
+    {
+      "epoch": 178.28947368421052,
+      "grad_norm": 1.0801442861557007,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 27100
+    },
+    {
+      "epoch": 178.35526315789474,
+      "grad_norm": 1.302809238433838,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 27110
+    },
+    {
+      "epoch": 178.42105263157896,
+      "grad_norm": 1.4321649074554443,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 27120
+    },
+    {
+      "epoch": 178.48684210526315,
+      "grad_norm": 1.3368439674377441,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 27130
+    },
+    {
+      "epoch": 178.55263157894737,
+      "grad_norm": 1.2113497257232666,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 27140
+    },
+    {
+      "epoch": 178.6184210526316,
+      "grad_norm": 1.0615336894989014,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 27150
+    },
+    {
+      "epoch": 178.68421052631578,
+      "grad_norm": 1.075102686882019,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 27160
+    },
+    {
+      "epoch": 178.75,
+      "grad_norm": 0.9051514863967896,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 27170
+    },
+    {
+      "epoch": 178.81578947368422,
+      "grad_norm": 1.019240140914917,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 27180
+    },
+    {
+      "epoch": 178.8815789473684,
+      "grad_norm": 1.2126598358154297,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 27190
+    },
+    {
+      "epoch": 178.94736842105263,
+      "grad_norm": 1.2346746921539307,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 27200
+    },
+    {
+      "epoch": 179.01315789473685,
+      "grad_norm": 1.2096296548843384,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 27210
+    },
+    {
+      "epoch": 179.07894736842104,
+      "grad_norm": 0.973487913608551,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 27220
+    },
+    {
+      "epoch": 179.14473684210526,
+      "grad_norm": 1.368411660194397,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 27230
+    },
+    {
+      "epoch": 179.21052631578948,
+      "grad_norm": 1.2822602987289429,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 27240
+    },
+    {
+      "epoch": 179.27631578947367,
+      "grad_norm": 1.2959389686584473,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 27250
+    },
+    {
+      "epoch": 179.3421052631579,
+      "grad_norm": 1.3127459287643433,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 27260
+    },
+    {
+      "epoch": 179.4078947368421,
+      "grad_norm": 1.122056007385254,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 27270
+    },
+    {
+      "epoch": 179.47368421052633,
+      "grad_norm": 1.3028178215026855,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 27280
+    },
+    {
+      "epoch": 179.53947368421052,
+      "grad_norm": 0.9754687547683716,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 27290
+    },
+    {
+      "epoch": 179.60526315789474,
+      "grad_norm": 1.238944411277771,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 27300
+    },
+    {
+      "epoch": 179.67105263157896,
+      "grad_norm": 1.3136788606643677,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 27310
+    },
+    {
+      "epoch": 179.73684210526315,
+      "grad_norm": 1.3043994903564453,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 27320
+    },
+    {
+      "epoch": 179.80263157894737,
+      "grad_norm": 1.0866022109985352,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 27330
+    },
+    {
+      "epoch": 179.8684210526316,
+      "grad_norm": 1.4017601013183594,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 27340
+    },
+    {
+      "epoch": 179.93421052631578,
+      "grad_norm": 1.2640719413757324,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 27350
+    },
+    {
+      "epoch": 180.0,
+      "grad_norm": 1.5062077045440674,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 27360
+    },
+    {
+      "epoch": 180.06578947368422,
+      "grad_norm": 1.3499562740325928,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 27370
+    },
+    {
+      "epoch": 180.1315789473684,
+      "grad_norm": 1.028475284576416,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 27380
+    },
+    {
+      "epoch": 180.19736842105263,
+      "grad_norm": 1.1006380319595337,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 27390
+    },
+    {
+      "epoch": 180.26315789473685,
+      "grad_norm": 1.2591434717178345,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 27400
+    },
+    {
+      "epoch": 180.32894736842104,
+      "grad_norm": 1.4151883125305176,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 27410
+    },
+    {
+      "epoch": 180.39473684210526,
+      "grad_norm": 1.132414698600769,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 27420
+    },
+    {
+      "epoch": 180.46052631578948,
+      "grad_norm": 1.3552273511886597,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 27430
+    },
+    {
+      "epoch": 180.52631578947367,
+      "grad_norm": 0.8924891352653503,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 27440
+    },
+    {
+      "epoch": 180.5921052631579,
+      "grad_norm": 1.676880955696106,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 27450
+    },
+    {
+      "epoch": 180.6578947368421,
+      "grad_norm": 1.1529208421707153,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 27460
+    },
+    {
+      "epoch": 180.72368421052633,
+      "grad_norm": 1.1190521717071533,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 27470
+    },
+    {
+      "epoch": 180.78947368421052,
+      "grad_norm": 1.293173909187317,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 27480
+    },
+    {
+      "epoch": 180.85526315789474,
+      "grad_norm": 0.9071879386901855,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 27490
+    },
+    {
+      "epoch": 180.92105263157896,
+      "grad_norm": 1.2806898355484009,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 27500
+    },
+    {
+      "epoch": 180.98684210526315,
+      "grad_norm": 0.9975317120552063,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 27510
+    },
+    {
+      "epoch": 181.05263157894737,
+      "grad_norm": 1.2514346837997437,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 27520
+    },
+    {
+      "epoch": 181.1184210526316,
+      "grad_norm": 1.361411452293396,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 27530
+    },
+    {
+      "epoch": 181.18421052631578,
+      "grad_norm": 0.9159877896308899,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 27540
+    },
+    {
+      "epoch": 181.25,
+      "grad_norm": 1.0003042221069336,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 27550
+    },
+    {
+      "epoch": 181.31578947368422,
+      "grad_norm": 1.133141040802002,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 27560
+    },
+    {
+      "epoch": 181.3815789473684,
+      "grad_norm": 1.0461962223052979,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 27570
+    },
+    {
+      "epoch": 181.44736842105263,
+      "grad_norm": 1.3540297746658325,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 27580
+    },
+    {
+      "epoch": 181.51315789473685,
+      "grad_norm": 1.3331037759780884,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 27590
+    },
+    {
+      "epoch": 181.57894736842104,
+      "grad_norm": 1.4690834283828735,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 27600
+    },
+    {
+      "epoch": 181.64473684210526,
+      "grad_norm": 1.3080083131790161,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 27610
+    },
+    {
+      "epoch": 181.71052631578948,
+      "grad_norm": 1.1954478025436401,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 27620
+    },
+    {
+      "epoch": 181.77631578947367,
+      "grad_norm": 1.5837913751602173,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 27630
+    },
+    {
+      "epoch": 181.8421052631579,
+      "grad_norm": 1.9183199405670166,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 27640
+    },
+    {
+      "epoch": 181.9078947368421,
+      "grad_norm": 1.468712568283081,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 27650
+    },
+    {
+      "epoch": 181.97368421052633,
+      "grad_norm": 1.183106780052185,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 27660
+    },
+    {
+      "epoch": 182.03947368421052,
+      "grad_norm": 1.315909743309021,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 27670
+    },
+    {
+      "epoch": 182.10526315789474,
+      "grad_norm": 1.3399372100830078,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 27680
+    },
+    {
+      "epoch": 182.17105263157896,
+      "grad_norm": 1.3758108615875244,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 27690
+    },
+    {
+      "epoch": 182.23684210526315,
+      "grad_norm": 1.372221827507019,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 27700
+    },
+    {
+      "epoch": 182.30263157894737,
+      "grad_norm": 1.1618051528930664,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 27710
+    },
+    {
+      "epoch": 182.3684210526316,
+      "grad_norm": 1.5792875289916992,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 27720
+    },
+    {
+      "epoch": 182.43421052631578,
+      "grad_norm": 1.4284778833389282,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 27730
+    },
+    {
+      "epoch": 182.5,
+      "grad_norm": 1.1605561971664429,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 27740
+    },
+    {
+      "epoch": 182.56578947368422,
+      "grad_norm": 1.394383192062378,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 27750
+    },
+    {
+      "epoch": 182.6315789473684,
+      "grad_norm": 1.2640798091888428,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 27760
+    },
+    {
+      "epoch": 182.69736842105263,
+      "grad_norm": 1.0631877183914185,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 27770
+    },
+    {
+      "epoch": 182.76315789473685,
+      "grad_norm": 1.3253459930419922,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 27780
+    },
+    {
+      "epoch": 182.82894736842104,
+      "grad_norm": 1.4501514434814453,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 27790
+    },
+    {
+      "epoch": 182.89473684210526,
+      "grad_norm": 1.629292607307434,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 27800
+    },
+    {
+      "epoch": 182.96052631578948,
+      "grad_norm": 1.373126745223999,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 27810
+    },
+    {
+      "epoch": 183.02631578947367,
+      "grad_norm": 1.4406814575195312,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 27820
+    },
+    {
+      "epoch": 183.0921052631579,
+      "grad_norm": 1.3722494840621948,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 27830
+    },
+    {
+      "epoch": 183.1578947368421,
+      "grad_norm": 1.8163440227508545,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 27840
+    },
+    {
+      "epoch": 183.22368421052633,
+      "grad_norm": 2.5917444229125977,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 27850
+    },
+    {
+      "epoch": 183.28947368421052,
+      "grad_norm": 1.976516604423523,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 27860
+    },
+    {
+      "epoch": 183.35526315789474,
+      "grad_norm": 1.48430335521698,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 27870
+    },
+    {
+      "epoch": 183.42105263157896,
+      "grad_norm": 1.908106803894043,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 27880
+    },
+    {
+      "epoch": 183.48684210526315,
+      "grad_norm": 1.8086661100387573,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 27890
+    },
+    {
+      "epoch": 183.55263157894737,
+      "grad_norm": 1.7461923360824585,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 27900
+    },
+    {
+      "epoch": 183.6184210526316,
+      "grad_norm": 1.1997524499893188,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 27910
+    },
+    {
+      "epoch": 183.68421052631578,
+      "grad_norm": 1.2785133123397827,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 27920
+    },
+    {
+      "epoch": 183.75,
+      "grad_norm": 0.9899089932441711,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 27930
+    },
+    {
+      "epoch": 183.81578947368422,
+      "grad_norm": 1.36931574344635,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 27940
+    },
+    {
+      "epoch": 183.8815789473684,
+      "grad_norm": 1.0563100576400757,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 27950
+    },
+    {
+      "epoch": 183.94736842105263,
+      "grad_norm": 1.2459518909454346,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 27960
+    },
+    {
+      "epoch": 184.01315789473685,
+      "grad_norm": 1.7021054029464722,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 27970
+    },
+    {
+      "epoch": 184.07894736842104,
+      "grad_norm": 1.1068536043167114,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 27980
+    },
+    {
+      "epoch": 184.14473684210526,
+      "grad_norm": 1.299325704574585,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 27990
+    },
+    {
+      "epoch": 184.21052631578948,
+      "grad_norm": 1.3241454362869263,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 28000
+    },
+    {
+      "epoch": 184.27631578947367,
+      "grad_norm": 1.4273960590362549,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 28010
+    },
+    {
+      "epoch": 184.3421052631579,
+      "grad_norm": 1.1758840084075928,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 28020
+    },
+    {
+      "epoch": 184.4078947368421,
+      "grad_norm": 1.7336351871490479,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 28030
+    },
+    {
+      "epoch": 184.47368421052633,
+      "grad_norm": 1.4646220207214355,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 28040
+    },
+    {
+      "epoch": 184.53947368421052,
+      "grad_norm": 1.5751539468765259,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 28050
+    },
+    {
+      "epoch": 184.60526315789474,
+      "grad_norm": 1.675756812095642,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 28060
+    },
+    {
+      "epoch": 184.67105263157896,
+      "grad_norm": 1.7213208675384521,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 28070
+    },
+    {
+      "epoch": 184.73684210526315,
+      "grad_norm": 1.4388047456741333,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 28080
+    },
+    {
+      "epoch": 184.80263157894737,
+      "grad_norm": 1.40472412109375,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 28090
+    },
+    {
+      "epoch": 184.8684210526316,
+      "grad_norm": 1.187379240989685,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 28100
+    },
+    {
+      "epoch": 184.93421052631578,
+      "grad_norm": 1.5341135263442993,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 28110
+    },
+    {
+      "epoch": 185.0,
+      "grad_norm": 1.2194042205810547,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 28120
+    },
+    {
+      "epoch": 185.06578947368422,
+      "grad_norm": 1.183134913444519,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 28130
+    },
+    {
+      "epoch": 185.1315789473684,
+      "grad_norm": 1.4038760662078857,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 28140
+    },
+    {
+      "epoch": 185.19736842105263,
+      "grad_norm": 1.3915314674377441,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 28150
+    },
+    {
+      "epoch": 185.26315789473685,
+      "grad_norm": 1.3534201383590698,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 28160
+    },
+    {
+      "epoch": 185.32894736842104,
+      "grad_norm": 1.2670302391052246,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 28170
+    },
+    {
+      "epoch": 185.39473684210526,
+      "grad_norm": 1.6847999095916748,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 28180
+    },
+    {
+      "epoch": 185.46052631578948,
+      "grad_norm": 1.430661678314209,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 28190
+    },
+    {
+      "epoch": 185.52631578947367,
+      "grad_norm": 1.600559949874878,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 28200
+    },
+    {
+      "epoch": 185.5921052631579,
+      "grad_norm": 1.3401246070861816,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 28210
+    },
+    {
+      "epoch": 185.6578947368421,
+      "grad_norm": 1.1347683668136597,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 28220
+    },
+    {
+      "epoch": 185.72368421052633,
+      "grad_norm": 1.1868791580200195,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 28230
+    },
+    {
+      "epoch": 185.78947368421052,
+      "grad_norm": 1.1938930749893188,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 28240
+    },
+    {
+      "epoch": 185.85526315789474,
+      "grad_norm": 0.9007596969604492,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 28250
+    },
+    {
+      "epoch": 185.92105263157896,
+      "grad_norm": 1.0700163841247559,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 28260
+    },
+    {
+      "epoch": 185.98684210526315,
+      "grad_norm": 1.3188432455062866,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 28270
+    },
+    {
+      "epoch": 186.05263157894737,
+      "grad_norm": 1.2520577907562256,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 28280
+    },
+    {
+      "epoch": 186.1184210526316,
+      "grad_norm": 1.2639751434326172,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 28290
+    },
+    {
+      "epoch": 186.18421052631578,
+      "grad_norm": 1.2425426244735718,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 28300
+    },
+    {
+      "epoch": 186.25,
+      "grad_norm": 1.1979525089263916,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 28310
+    },
+    {
+      "epoch": 186.31578947368422,
+      "grad_norm": 1.25445556640625,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 28320
+    },
+    {
+      "epoch": 186.3815789473684,
+      "grad_norm": 1.0609337091445923,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 28330
+    },
+    {
+      "epoch": 186.44736842105263,
+      "grad_norm": 0.9719675183296204,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 28340
+    },
+    {
+      "epoch": 186.51315789473685,
+      "grad_norm": 1.2194287776947021,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 28350
+    },
+    {
+      "epoch": 186.57894736842104,
+      "grad_norm": 1.2803702354431152,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 28360
+    },
+    {
+      "epoch": 186.64473684210526,
+      "grad_norm": 1.472983717918396,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 28370
+    },
+    {
+      "epoch": 186.71052631578948,
+      "grad_norm": 0.7579768896102905,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 28380
+    },
+    {
+      "epoch": 186.77631578947367,
+      "grad_norm": 1.4877924919128418,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 28390
+    },
+    {
+      "epoch": 186.8421052631579,
+      "grad_norm": 1.1992541551589966,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 28400
+    },
+    {
+      "epoch": 186.9078947368421,
+      "grad_norm": 1.5354152917861938,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 28410
+    },
+    {
+      "epoch": 186.97368421052633,
+      "grad_norm": 1.009834885597229,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 28420
+    },
+    {
+      "epoch": 187.03947368421052,
+      "grad_norm": 1.4068772792816162,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 28430
+    },
+    {
+      "epoch": 187.10526315789474,
+      "grad_norm": 1.7721872329711914,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 28440
+    },
+    {
+      "epoch": 187.17105263157896,
+      "grad_norm": 1.3178023099899292,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 28450
+    },
+    {
+      "epoch": 187.23684210526315,
+      "grad_norm": 2.538278341293335,
+      "learning_rate": 0.0001,
+      "loss": 0.0277,
+      "step": 28460
+    },
+    {
+      "epoch": 187.30263157894737,
+      "grad_norm": 2.6935033798217773,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 28470
+    },
+    {
+      "epoch": 187.3684210526316,
+      "grad_norm": 2.583784341812134,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 28480
+    },
+    {
+      "epoch": 187.43421052631578,
+      "grad_norm": 2.7853147983551025,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 28490
+    },
+    {
+      "epoch": 187.5,
+      "grad_norm": 1.8979226350784302,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 28500
+    },
+    {
+      "epoch": 187.56578947368422,
+      "grad_norm": 1.951745867729187,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 28510
+    },
+    {
+      "epoch": 187.6315789473684,
+      "grad_norm": 1.744847059249878,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 28520
+    },
+    {
+      "epoch": 187.69736842105263,
+      "grad_norm": 1.6811963319778442,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 28530
+    },
+    {
+      "epoch": 187.76315789473685,
+      "grad_norm": 1.182780146598816,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 28540
+    },
+    {
+      "epoch": 187.82894736842104,
+      "grad_norm": 1.3196076154708862,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 28550
+    },
+    {
+      "epoch": 187.89473684210526,
+      "grad_norm": 1.376051425933838,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 28560
+    },
+    {
+      "epoch": 187.96052631578948,
+      "grad_norm": 1.3644719123840332,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 28570
+    },
+    {
+      "epoch": 188.02631578947367,
+      "grad_norm": 1.3114805221557617,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 28580
+    },
+    {
+      "epoch": 188.0921052631579,
+      "grad_norm": 1.1968399286270142,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 28590
+    },
+    {
+      "epoch": 188.1578947368421,
+      "grad_norm": 1.2592320442199707,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 28600
+    },
+    {
+      "epoch": 188.22368421052633,
+      "grad_norm": 1.2056187391281128,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 28610
+    },
+    {
+      "epoch": 188.28947368421052,
+      "grad_norm": 1.101921558380127,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 28620
+    },
+    {
+      "epoch": 188.35526315789474,
+      "grad_norm": 1.2965214252471924,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 28630
+    },
+    {
+      "epoch": 188.42105263157896,
+      "grad_norm": 1.5886129140853882,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 28640
+    },
+    {
+      "epoch": 188.48684210526315,
+      "grad_norm": 1.6472376585006714,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 28650
+    },
+    {
+      "epoch": 188.55263157894737,
+      "grad_norm": 1.4192408323287964,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 28660
+    },
+    {
+      "epoch": 188.6184210526316,
+      "grad_norm": 1.955345630645752,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 28670
+    },
+    {
+      "epoch": 188.68421052631578,
+      "grad_norm": 1.7373226881027222,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 28680
+    },
+    {
+      "epoch": 188.75,
+      "grad_norm": 1.6695311069488525,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 28690
+    },
+    {
+      "epoch": 188.81578947368422,
+      "grad_norm": 1.6247828006744385,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 28700
+    },
+    {
+      "epoch": 188.8815789473684,
+      "grad_norm": 1.587103009223938,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 28710
+    },
+    {
+      "epoch": 188.94736842105263,
+      "grad_norm": 1.3761903047561646,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 28720
+    },
+    {
+      "epoch": 189.01315789473685,
+      "grad_norm": 1.054154872894287,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 28730
+    },
+    {
+      "epoch": 189.07894736842104,
+      "grad_norm": 1.7570017576217651,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 28740
+    },
+    {
+      "epoch": 189.14473684210526,
+      "grad_norm": 1.302839994430542,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 28750
+    },
+    {
+      "epoch": 189.21052631578948,
+      "grad_norm": 1.2320873737335205,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 28760
+    },
+    {
+      "epoch": 189.27631578947367,
+      "grad_norm": 1.2856324911117554,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 28770
+    },
+    {
+      "epoch": 189.3421052631579,
+      "grad_norm": 1.2495955228805542,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 28780
+    },
+    {
+      "epoch": 189.4078947368421,
+      "grad_norm": 1.2113772630691528,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 28790
+    },
+    {
+      "epoch": 189.47368421052633,
+      "grad_norm": 1.1812059879302979,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 28800
+    },
+    {
+      "epoch": 189.53947368421052,
+      "grad_norm": 1.3910290002822876,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 28810
+    },
+    {
+      "epoch": 189.60526315789474,
+      "grad_norm": 1.076119303703308,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 28820
+    },
+    {
+      "epoch": 189.67105263157896,
+      "grad_norm": 2.222041606903076,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 28830
+    },
+    {
+      "epoch": 189.73684210526315,
+      "grad_norm": 2.265124559402466,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 28840
+    },
+    {
+      "epoch": 189.80263157894737,
+      "grad_norm": 1.6033040285110474,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 28850
+    },
+    {
+      "epoch": 189.8684210526316,
+      "grad_norm": 1.4938453435897827,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 28860
+    },
+    {
+      "epoch": 189.93421052631578,
+      "grad_norm": 1.3724675178527832,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 28870
+    },
+    {
+      "epoch": 190.0,
+      "grad_norm": 1.3057665824890137,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 28880
+    },
+    {
+      "epoch": 190.06578947368422,
+      "grad_norm": 1.1445728540420532,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 28890
+    },
+    {
+      "epoch": 190.1315789473684,
+      "grad_norm": 1.2155309915542603,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 28900
+    },
+    {
+      "epoch": 190.19736842105263,
+      "grad_norm": 1.1960302591323853,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 28910
+    },
+    {
+      "epoch": 190.26315789473685,
+      "grad_norm": 1.3511428833007812,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 28920
+    },
+    {
+      "epoch": 190.32894736842104,
+      "grad_norm": 0.9308839440345764,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 28930
+    },
+    {
+      "epoch": 190.39473684210526,
+      "grad_norm": 1.409009575843811,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 28940
+    },
+    {
+      "epoch": 190.46052631578948,
+      "grad_norm": 1.1859124898910522,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 28950
+    },
+    {
+      "epoch": 190.52631578947367,
+      "grad_norm": 1.6667457818984985,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 28960
+    },
+    {
+      "epoch": 190.5921052631579,
+      "grad_norm": 1.8634475469589233,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 28970
+    },
+    {
+      "epoch": 190.6578947368421,
+      "grad_norm": 1.645556092262268,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 28980
+    },
+    {
+      "epoch": 190.72368421052633,
+      "grad_norm": 1.3310197591781616,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 28990
+    },
+    {
+      "epoch": 190.78947368421052,
+      "grad_norm": 1.2974885702133179,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 29000
+    },
+    {
+      "epoch": 190.85526315789474,
+      "grad_norm": 1.270050048828125,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 29010
+    },
+    {
+      "epoch": 190.92105263157896,
+      "grad_norm": 1.1336300373077393,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 29020
+    },
+    {
+      "epoch": 190.98684210526315,
+      "grad_norm": 1.3560301065444946,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 29030
+    },
+    {
+      "epoch": 191.05263157894737,
+      "grad_norm": 1.405979871749878,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 29040
+    },
+    {
+      "epoch": 191.1184210526316,
+      "grad_norm": 1.3937962055206299,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 29050
+    },
+    {
+      "epoch": 191.18421052631578,
+      "grad_norm": 1.3739968538284302,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 29060
+    },
+    {
+      "epoch": 191.25,
+      "grad_norm": 1.259650707244873,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 29070
+    },
+    {
+      "epoch": 191.31578947368422,
+      "grad_norm": 0.9018769860267639,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 29080
+    },
+    {
+      "epoch": 191.3815789473684,
+      "grad_norm": 1.1630486249923706,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 29090
+    },
+    {
+      "epoch": 191.44736842105263,
+      "grad_norm": 1.2646576166152954,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 29100
+    },
+    {
+      "epoch": 191.51315789473685,
+      "grad_norm": 1.3670384883880615,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 29110
+    },
+    {
+      "epoch": 191.57894736842104,
+      "grad_norm": 1.4089142084121704,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 29120
+    },
+    {
+      "epoch": 191.64473684210526,
+      "grad_norm": 1.2056618928909302,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 29130
+    },
+    {
+      "epoch": 191.71052631578948,
+      "grad_norm": 1.2704377174377441,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 29140
+    },
+    {
+      "epoch": 191.77631578947367,
+      "grad_norm": 1.3765251636505127,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 29150
+    },
+    {
+      "epoch": 191.8421052631579,
+      "grad_norm": 1.2344194650650024,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 29160
+    },
+    {
+      "epoch": 191.9078947368421,
+      "grad_norm": 1.1715292930603027,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 29170
+    },
+    {
+      "epoch": 191.97368421052633,
+      "grad_norm": 1.3913319110870361,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 29180
+    },
+    {
+      "epoch": 192.03947368421052,
+      "grad_norm": 1.4549134969711304,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 29190
+    },
+    {
+      "epoch": 192.10526315789474,
+      "grad_norm": 1.2978013753890991,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 29200
+    },
+    {
+      "epoch": 192.17105263157896,
+      "grad_norm": 1.2289888858795166,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 29210
+    },
+    {
+      "epoch": 192.23684210526315,
+      "grad_norm": 0.9477553367614746,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 29220
+    },
+    {
+      "epoch": 192.30263157894737,
+      "grad_norm": 1.459433674812317,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 29230
+    },
+    {
+      "epoch": 192.3684210526316,
+      "grad_norm": 1.5326049327850342,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 29240
+    },
+    {
+      "epoch": 192.43421052631578,
+      "grad_norm": 1.2914258241653442,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 29250
+    },
+    {
+      "epoch": 192.5,
+      "grad_norm": 1.216827630996704,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 29260
+    },
+    {
+      "epoch": 192.56578947368422,
+      "grad_norm": 1.4195771217346191,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 29270
+    },
+    {
+      "epoch": 192.6315789473684,
+      "grad_norm": 1.1989006996154785,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 29280
+    },
+    {
+      "epoch": 192.69736842105263,
+      "grad_norm": 1.2557868957519531,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 29290
+    },
+    {
+      "epoch": 192.76315789473685,
+      "grad_norm": 1.556070327758789,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 29300
+    },
+    {
+      "epoch": 192.82894736842104,
+      "grad_norm": 1.3581067323684692,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 29310
+    },
+    {
+      "epoch": 192.89473684210526,
+      "grad_norm": 1.4777802228927612,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 29320
+    },
+    {
+      "epoch": 192.96052631578948,
+      "grad_norm": 1.00076162815094,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 29330
+    },
+    {
+      "epoch": 193.02631578947367,
+      "grad_norm": 1.448732852935791,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 29340
+    },
+    {
+      "epoch": 193.0921052631579,
+      "grad_norm": 1.2730473279953003,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 29350
+    },
+    {
+      "epoch": 193.1578947368421,
+      "grad_norm": 1.256740689277649,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 29360
+    },
+    {
+      "epoch": 193.22368421052633,
+      "grad_norm": 1.0726568698883057,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 29370
+    },
+    {
+      "epoch": 193.28947368421052,
+      "grad_norm": 0.9188181757926941,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 29380
+    },
+    {
+      "epoch": 193.35526315789474,
+      "grad_norm": 1.2872679233551025,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 29390
+    },
+    {
+      "epoch": 193.42105263157896,
+      "grad_norm": 1.520761251449585,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 29400
+    },
+    {
+      "epoch": 193.48684210526315,
+      "grad_norm": 1.1832773685455322,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 29410
+    },
+    {
+      "epoch": 193.55263157894737,
+      "grad_norm": 1.4867194890975952,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 29420
+    },
+    {
+      "epoch": 193.6184210526316,
+      "grad_norm": 1.7297718524932861,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 29430
+    },
+    {
+      "epoch": 193.68421052631578,
+      "grad_norm": 1.251755952835083,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 29440
+    },
+    {
+      "epoch": 193.75,
+      "grad_norm": 1.3701705932617188,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 29450
+    },
+    {
+      "epoch": 193.81578947368422,
+      "grad_norm": 1.7457464933395386,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 29460
+    },
+    {
+      "epoch": 193.8815789473684,
+      "grad_norm": 1.126453161239624,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 29470
+    },
+    {
+      "epoch": 193.94736842105263,
+      "grad_norm": 1.2443026304244995,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 29480
+    },
+    {
+      "epoch": 194.01315789473685,
+      "grad_norm": 1.2029883861541748,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 29490
+    },
+    {
+      "epoch": 194.07894736842104,
+      "grad_norm": 1.449203610420227,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 29500
+    },
+    {
+      "epoch": 194.14473684210526,
+      "grad_norm": 1.0315684080123901,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 29510
+    },
+    {
+      "epoch": 194.21052631578948,
+      "grad_norm": 1.3804000616073608,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 29520
+    },
+    {
+      "epoch": 194.27631578947367,
+      "grad_norm": 1.527992844581604,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 29530
+    },
+    {
+      "epoch": 194.3421052631579,
+      "grad_norm": 1.5879637002944946,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 29540
+    },
+    {
+      "epoch": 194.4078947368421,
+      "grad_norm": 1.9236878156661987,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 29550
+    },
+    {
+      "epoch": 194.47368421052633,
+      "grad_norm": 1.539232611656189,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 29560
+    },
+    {
+      "epoch": 194.53947368421052,
+      "grad_norm": 1.3699829578399658,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 29570
+    },
+    {
+      "epoch": 194.60526315789474,
+      "grad_norm": 1.3466382026672363,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 29580
+    },
+    {
+      "epoch": 194.67105263157896,
+      "grad_norm": 1.6446095705032349,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 29590
+    },
+    {
+      "epoch": 194.73684210526315,
+      "grad_norm": 1.2877060174942017,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 29600
+    },
+    {
+      "epoch": 194.80263157894737,
+      "grad_norm": 1.1062644720077515,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 29610
+    },
+    {
+      "epoch": 194.8684210526316,
+      "grad_norm": 0.854597270488739,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 29620
+    },
+    {
+      "epoch": 194.93421052631578,
+      "grad_norm": 1.1438567638397217,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 29630
+    },
+    {
+      "epoch": 195.0,
+      "grad_norm": 1.058065414428711,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 29640
+    },
+    {
+      "epoch": 195.06578947368422,
+      "grad_norm": 1.3158515691757202,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 29650
+    },
+    {
+      "epoch": 195.1315789473684,
+      "grad_norm": 1.3402299880981445,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 29660
+    },
+    {
+      "epoch": 195.19736842105263,
+      "grad_norm": 1.5896971225738525,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 29670
+    },
+    {
+      "epoch": 195.26315789473685,
+      "grad_norm": 1.113895297050476,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 29680
+    },
+    {
+      "epoch": 195.32894736842104,
+      "grad_norm": 0.8616934418678284,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 29690
+    },
+    {
+      "epoch": 195.39473684210526,
+      "grad_norm": 1.0356990098953247,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 29700
+    },
+    {
+      "epoch": 195.46052631578948,
+      "grad_norm": 1.1201450824737549,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 29710
+    },
+    {
+      "epoch": 195.52631578947367,
+      "grad_norm": 1.386657476425171,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 29720
+    },
+    {
+      "epoch": 195.5921052631579,
+      "grad_norm": 0.8346795439720154,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 29730
+    },
+    {
+      "epoch": 195.6578947368421,
+      "grad_norm": 1.4521796703338623,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 29740
+    },
+    {
+      "epoch": 195.72368421052633,
+      "grad_norm": 1.4133926630020142,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 29750
+    },
+    {
+      "epoch": 195.78947368421052,
+      "grad_norm": 0.9600488543510437,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 29760
+    },
+    {
+      "epoch": 195.85526315789474,
+      "grad_norm": 1.1314250230789185,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 29770
+    },
+    {
+      "epoch": 195.92105263157896,
+      "grad_norm": 0.8754208087921143,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 29780
+    },
+    {
+      "epoch": 195.98684210526315,
+      "grad_norm": 1.3597506284713745,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 29790
+    },
+    {
+      "epoch": 196.05263157894737,
+      "grad_norm": 1.1357085704803467,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 29800
+    },
+    {
+      "epoch": 196.1184210526316,
+      "grad_norm": 1.1614354848861694,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 29810
+    },
+    {
+      "epoch": 196.18421052631578,
+      "grad_norm": 1.2745589017868042,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 29820
+    },
+    {
+      "epoch": 196.25,
+      "grad_norm": 1.1572297811508179,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 29830
+    },
+    {
+      "epoch": 196.31578947368422,
+      "grad_norm": 0.7867621183395386,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 29840
+    },
+    {
+      "epoch": 196.3815789473684,
+      "grad_norm": 1.308093786239624,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 29850
+    },
+    {
+      "epoch": 196.44736842105263,
+      "grad_norm": 1.2609182596206665,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 29860
+    },
+    {
+      "epoch": 196.51315789473685,
+      "grad_norm": 1.3445231914520264,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 29870
+    },
+    {
+      "epoch": 196.57894736842104,
+      "grad_norm": 1.0773708820343018,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 29880
+    },
+    {
+      "epoch": 196.64473684210526,
+      "grad_norm": 0.9694660902023315,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 29890
+    },
+    {
+      "epoch": 196.71052631578948,
+      "grad_norm": 1.11964750289917,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 29900
+    },
+    {
+      "epoch": 196.77631578947367,
+      "grad_norm": 1.1886194944381714,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 29910
+    },
+    {
+      "epoch": 196.8421052631579,
+      "grad_norm": 1.4369953870773315,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 29920
+    },
+    {
+      "epoch": 196.9078947368421,
+      "grad_norm": 0.9446222186088562,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 29930
+    },
+    {
+      "epoch": 196.97368421052633,
+      "grad_norm": 1.308263897895813,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 29940
+    },
+    {
+      "epoch": 197.03947368421052,
+      "grad_norm": 1.479591965675354,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 29950
+    },
+    {
+      "epoch": 197.10526315789474,
+      "grad_norm": 1.261594295501709,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 29960
+    },
+    {
+      "epoch": 197.17105263157896,
+      "grad_norm": 1.597878098487854,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 29970
+    },
+    {
+      "epoch": 197.23684210526315,
+      "grad_norm": 0.9023445248603821,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 29980
+    },
+    {
+      "epoch": 197.30263157894737,
+      "grad_norm": 1.7900540828704834,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 29990
+    },
+    {
+      "epoch": 197.3684210526316,
+      "grad_norm": 1.0138733386993408,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 30000
+    },
+    {
+      "epoch": 197.43421052631578,
+      "grad_norm": 1.617836594581604,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 30010
+    },
+    {
+      "epoch": 197.5,
+      "grad_norm": 1.0943124294281006,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 30020
+    },
+    {
+      "epoch": 197.56578947368422,
+      "grad_norm": 1.1068415641784668,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 30030
+    },
+    {
+      "epoch": 197.6315789473684,
+      "grad_norm": 1.248450756072998,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 30040
+    },
+    {
+      "epoch": 197.69736842105263,
+      "grad_norm": 1.4986473321914673,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 30050
+    },
+    {
+      "epoch": 197.76315789473685,
+      "grad_norm": 1.2006663084030151,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 30060
+    },
+    {
+      "epoch": 197.82894736842104,
+      "grad_norm": 1.0944031476974487,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 30070
+    },
+    {
+      "epoch": 197.89473684210526,
+      "grad_norm": 1.6953439712524414,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 30080
+    },
+    {
+      "epoch": 197.96052631578948,
+      "grad_norm": 1.1064839363098145,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 30090
+    },
+    {
+      "epoch": 198.02631578947367,
+      "grad_norm": 2.015153646469116,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 30100
+    },
+    {
+      "epoch": 198.0921052631579,
+      "grad_norm": 1.8545198440551758,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 30110
+    },
+    {
+      "epoch": 198.1578947368421,
+      "grad_norm": 1.5696916580200195,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 30120
+    },
+    {
+      "epoch": 198.22368421052633,
+      "grad_norm": 1.4251947402954102,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 30130
+    },
+    {
+      "epoch": 198.28947368421052,
+      "grad_norm": 1.1599140167236328,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 30140
+    },
+    {
+      "epoch": 198.35526315789474,
+      "grad_norm": 1.4993690252304077,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 30150
+    },
+    {
+      "epoch": 198.42105263157896,
+      "grad_norm": 1.4265055656433105,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 30160
+    },
+    {
+      "epoch": 198.48684210526315,
+      "grad_norm": 1.101048231124878,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 30170
+    },
+    {
+      "epoch": 198.55263157894737,
+      "grad_norm": 1.5096710920333862,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 30180
+    },
+    {
+      "epoch": 198.6184210526316,
+      "grad_norm": 1.190572738647461,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 30190
+    },
+    {
+      "epoch": 198.68421052631578,
+      "grad_norm": 1.591253399848938,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 30200
+    },
+    {
+      "epoch": 198.75,
+      "grad_norm": 1.0739487409591675,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 30210
+    },
+    {
+      "epoch": 198.81578947368422,
+      "grad_norm": 1.2973365783691406,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 30220
+    },
+    {
+      "epoch": 198.8815789473684,
+      "grad_norm": 1.2122310400009155,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 30230
+    },
+    {
+      "epoch": 198.94736842105263,
+      "grad_norm": 1.2195587158203125,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 30240
+    },
+    {
+      "epoch": 199.01315789473685,
+      "grad_norm": 1.0152171850204468,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 30250
+    },
+    {
+      "epoch": 199.07894736842104,
+      "grad_norm": 1.404311180114746,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 30260
+    },
+    {
+      "epoch": 199.14473684210526,
+      "grad_norm": 1.4140353202819824,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 30270
+    },
+    {
+      "epoch": 199.21052631578948,
+      "grad_norm": 1.4403727054595947,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 30280
+    },
+    {
+      "epoch": 199.27631578947367,
+      "grad_norm": 1.9959533214569092,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 30290
+    },
+    {
+      "epoch": 199.3421052631579,
+      "grad_norm": 1.6239718198776245,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 30300
+    },
+    {
+      "epoch": 199.4078947368421,
+      "grad_norm": 1.3269397020339966,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 30310
+    },
+    {
+      "epoch": 199.47368421052633,
+      "grad_norm": 1.6701898574829102,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 30320
+    },
+    {
+      "epoch": 199.53947368421052,
+      "grad_norm": 1.5344353914260864,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 30330
+    },
+    {
+      "epoch": 199.60526315789474,
+      "grad_norm": 1.4121617078781128,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 30340
+    },
+    {
+      "epoch": 199.67105263157896,
+      "grad_norm": 0.9731596112251282,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 30350
+    },
+    {
+      "epoch": 199.73684210526315,
+      "grad_norm": 1.1787141561508179,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 30360
+    },
+    {
+      "epoch": 199.80263157894737,
+      "grad_norm": 1.0008739233016968,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 30370
+    },
+    {
+      "epoch": 199.8684210526316,
+      "grad_norm": 1.2544138431549072,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 30380
+    },
+    {
+      "epoch": 199.93421052631578,
+      "grad_norm": 1.1031585931777954,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 30390
+    },
+    {
+      "epoch": 200.0,
+      "grad_norm": 1.5565379858016968,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 30400
+    },
+    {
+      "epoch": 200.06578947368422,
+      "grad_norm": 1.0128217935562134,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 30410
+    },
+    {
+      "epoch": 200.1315789473684,
+      "grad_norm": 0.9004884362220764,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 30420
+    },
+    {
+      "epoch": 200.19736842105263,
+      "grad_norm": 0.7593161463737488,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 30430
+    },
+    {
+      "epoch": 200.26315789473685,
+      "grad_norm": 1.4337648153305054,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 30440
+    },
+    {
+      "epoch": 200.32894736842104,
+      "grad_norm": 1.2471925020217896,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 30450
+    },
+    {
+      "epoch": 200.39473684210526,
+      "grad_norm": 1.1293731927871704,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 30460
+    },
+    {
+      "epoch": 200.46052631578948,
+      "grad_norm": 0.8459559679031372,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 30470
+    },
+    {
+      "epoch": 200.52631578947367,
+      "grad_norm": 1.5371299982070923,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 30480
+    },
+    {
+      "epoch": 200.5921052631579,
+      "grad_norm": 1.3424179553985596,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 30490
+    },
+    {
+      "epoch": 200.6578947368421,
+      "grad_norm": 1.5373995304107666,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 30500
+    },
+    {
+      "epoch": 200.72368421052633,
+      "grad_norm": 0.9894046783447266,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 30510
+    },
+    {
+      "epoch": 200.78947368421052,
+      "grad_norm": 1.7582042217254639,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 30520
+    },
+    {
+      "epoch": 200.85526315789474,
+      "grad_norm": 1.3425977230072021,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 30530
+    },
+    {
+      "epoch": 200.92105263157896,
+      "grad_norm": 1.3770979642868042,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 30540
+    },
+    {
+      "epoch": 200.98684210526315,
+      "grad_norm": 1.40471351146698,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 30550
+    },
+    {
+      "epoch": 201.05263157894737,
+      "grad_norm": 1.337066411972046,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 30560
+    },
+    {
+      "epoch": 201.1184210526316,
+      "grad_norm": 1.1256475448608398,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 30570
+    },
+    {
+      "epoch": 201.18421052631578,
+      "grad_norm": 1.3861485719680786,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 30580
+    },
+    {
+      "epoch": 201.25,
+      "grad_norm": 1.1584820747375488,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 30590
+    },
+    {
+      "epoch": 201.31578947368422,
+      "grad_norm": 1.3375623226165771,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 30600
+    },
+    {
+      "epoch": 201.3815789473684,
+      "grad_norm": 1.2848438024520874,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 30610
+    },
+    {
+      "epoch": 201.44736842105263,
+      "grad_norm": 1.2044715881347656,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 30620
+    },
+    {
+      "epoch": 201.51315789473685,
+      "grad_norm": 1.1717559099197388,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 30630
+    },
+    {
+      "epoch": 201.57894736842104,
+      "grad_norm": 1.0431238412857056,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 30640
+    },
+    {
+      "epoch": 201.64473684210526,
+      "grad_norm": 1.0847781896591187,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 30650
+    },
+    {
+      "epoch": 201.71052631578948,
+      "grad_norm": 1.1109511852264404,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 30660
+    },
+    {
+      "epoch": 201.77631578947367,
+      "grad_norm": 1.094912052154541,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 30670
+    },
+    {
+      "epoch": 201.8421052631579,
+      "grad_norm": 1.0548181533813477,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 30680
+    },
+    {
+      "epoch": 201.9078947368421,
+      "grad_norm": 1.3336670398712158,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 30690
+    },
+    {
+      "epoch": 201.97368421052633,
+      "grad_norm": 1.0279159545898438,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 30700
+    },
+    {
+      "epoch": 202.03947368421052,
+      "grad_norm": 0.7135113477706909,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 30710
+    },
+    {
+      "epoch": 202.10526315789474,
+      "grad_norm": 1.0755541324615479,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 30720
+    },
+    {
+      "epoch": 202.17105263157896,
+      "grad_norm": 1.3156121969223022,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 30730
+    },
+    {
+      "epoch": 202.23684210526315,
+      "grad_norm": 1.4101308584213257,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 30740
+    },
+    {
+      "epoch": 202.30263157894737,
+      "grad_norm": 1.3764532804489136,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 30750
+    },
+    {
+      "epoch": 202.3684210526316,
+      "grad_norm": 0.882384181022644,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 30760
+    },
+    {
+      "epoch": 202.43421052631578,
+      "grad_norm": 1.6236560344696045,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 30770
+    },
+    {
+      "epoch": 202.5,
+      "grad_norm": 1.4815741777420044,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 30780
+    },
+    {
+      "epoch": 202.56578947368422,
+      "grad_norm": 1.2555485963821411,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 30790
+    },
+    {
+      "epoch": 202.6315789473684,
+      "grad_norm": 1.3122458457946777,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 30800
+    },
+    {
+      "epoch": 202.69736842105263,
+      "grad_norm": 1.042381763458252,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 30810
+    },
+    {
+      "epoch": 202.76315789473685,
+      "grad_norm": 0.9814920425415039,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 30820
+    },
+    {
+      "epoch": 202.82894736842104,
+      "grad_norm": 1.0357002019882202,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 30830
+    },
+    {
+      "epoch": 202.89473684210526,
+      "grad_norm": 1.281749963760376,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 30840
+    },
+    {
+      "epoch": 202.96052631578948,
+      "grad_norm": 1.321096658706665,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 30850
+    },
+    {
+      "epoch": 203.02631578947367,
+      "grad_norm": 1.1207501888275146,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 30860
+    },
+    {
+      "epoch": 203.0921052631579,
+      "grad_norm": 1.5637404918670654,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 30870
+    },
+    {
+      "epoch": 203.1578947368421,
+      "grad_norm": 1.2370407581329346,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 30880
+    },
+    {
+      "epoch": 203.22368421052633,
+      "grad_norm": 1.0421314239501953,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 30890
+    },
+    {
+      "epoch": 203.28947368421052,
+      "grad_norm": 1.0301634073257446,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 30900
+    },
+    {
+      "epoch": 203.35526315789474,
+      "grad_norm": 1.2774720191955566,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 30910
+    },
+    {
+      "epoch": 203.42105263157896,
+      "grad_norm": 1.2703475952148438,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 30920
+    },
+    {
+      "epoch": 203.48684210526315,
+      "grad_norm": 1.1837942600250244,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 30930
+    },
+    {
+      "epoch": 203.55263157894737,
+      "grad_norm": 1.0560357570648193,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 30940
+    },
+    {
+      "epoch": 203.6184210526316,
+      "grad_norm": 1.2957569360733032,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 30950
+    },
+    {
+      "epoch": 203.68421052631578,
+      "grad_norm": 1.4275071620941162,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 30960
+    },
+    {
+      "epoch": 203.75,
+      "grad_norm": 1.513159155845642,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 30970
+    },
+    {
+      "epoch": 203.81578947368422,
+      "grad_norm": 1.6335196495056152,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 30980
+    },
+    {
+      "epoch": 203.8815789473684,
+      "grad_norm": 1.5469692945480347,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 30990
+    },
+    {
+      "epoch": 203.94736842105263,
+      "grad_norm": 1.5226917266845703,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 31000
+    },
+    {
+      "epoch": 204.01315789473685,
+      "grad_norm": 1.3117252588272095,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 31010
+    },
+    {
+      "epoch": 204.07894736842104,
+      "grad_norm": 1.800711989402771,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 31020
+    },
+    {
+      "epoch": 204.14473684210526,
+      "grad_norm": 1.6700514554977417,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 31030
+    },
+    {
+      "epoch": 204.21052631578948,
+      "grad_norm": 1.702384352684021,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 31040
+    },
+    {
+      "epoch": 204.27631578947367,
+      "grad_norm": 1.678499698638916,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 31050
+    },
+    {
+      "epoch": 204.3421052631579,
+      "grad_norm": 1.4093444347381592,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 31060
+    },
+    {
+      "epoch": 204.4078947368421,
+      "grad_norm": 1.5333293676376343,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 31070
+    },
+    {
+      "epoch": 204.47368421052633,
+      "grad_norm": 1.2159759998321533,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 31080
+    },
+    {
+      "epoch": 204.53947368421052,
+      "grad_norm": 1.358657956123352,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 31090
+    },
+    {
+      "epoch": 204.60526315789474,
+      "grad_norm": 1.1089929342269897,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 31100
+    },
+    {
+      "epoch": 204.67105263157896,
+      "grad_norm": 1.3180047273635864,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 31110
+    },
+    {
+      "epoch": 204.73684210526315,
+      "grad_norm": 1.2154042720794678,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 31120
+    },
+    {
+      "epoch": 204.80263157894737,
+      "grad_norm": 1.4112086296081543,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 31130
+    },
+    {
+      "epoch": 204.8684210526316,
+      "grad_norm": 1.432087779045105,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 31140
+    },
+    {
+      "epoch": 204.93421052631578,
+      "grad_norm": 1.3358913660049438,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 31150
+    },
+    {
+      "epoch": 205.0,
+      "grad_norm": 1.435907006263733,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 31160
+    },
+    {
+      "epoch": 205.06578947368422,
+      "grad_norm": 1.0711824893951416,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 31170
+    },
+    {
+      "epoch": 205.1315789473684,
+      "grad_norm": 1.3709452152252197,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 31180
+    },
+    {
+      "epoch": 205.19736842105263,
+      "grad_norm": 1.1924278736114502,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 31190
+    },
+    {
+      "epoch": 205.26315789473685,
+      "grad_norm": 1.3899904489517212,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 31200
+    },
+    {
+      "epoch": 205.32894736842104,
+      "grad_norm": 1.3233460187911987,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 31210
+    },
+    {
+      "epoch": 205.39473684210526,
+      "grad_norm": 1.4931349754333496,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 31220
+    },
+    {
+      "epoch": 205.46052631578948,
+      "grad_norm": 1.203282117843628,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 31230
+    },
+    {
+      "epoch": 205.52631578947367,
+      "grad_norm": 1.3920016288757324,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 31240
+    },
+    {
+      "epoch": 205.5921052631579,
+      "grad_norm": 1.077797770500183,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 31250
+    },
+    {
+      "epoch": 205.6578947368421,
+      "grad_norm": 1.7095680236816406,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 31260
+    },
+    {
+      "epoch": 205.72368421052633,
+      "grad_norm": 0.7705028057098389,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 31270
+    },
+    {
+      "epoch": 205.78947368421052,
+      "grad_norm": 1.1185240745544434,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 31280
+    },
+    {
+      "epoch": 205.85526315789474,
+      "grad_norm": 1.0305986404418945,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 31290
+    },
+    {
+      "epoch": 205.92105263157896,
+      "grad_norm": 1.5906319618225098,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 31300
+    },
+    {
+      "epoch": 205.98684210526315,
+      "grad_norm": 1.2245663404464722,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 31310
+    },
+    {
+      "epoch": 206.05263157894737,
+      "grad_norm": 1.706349492073059,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 31320
+    },
+    {
+      "epoch": 206.1184210526316,
+      "grad_norm": 1.1370997428894043,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 31330
+    },
+    {
+      "epoch": 206.18421052631578,
+      "grad_norm": 0.6681815385818481,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 31340
+    },
+    {
+      "epoch": 206.25,
+      "grad_norm": 1.397417664527893,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 31350
+    },
+    {
+      "epoch": 206.31578947368422,
+      "grad_norm": 1.0024604797363281,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 31360
+    },
+    {
+      "epoch": 206.3815789473684,
+      "grad_norm": 1.329154372215271,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 31370
+    },
+    {
+      "epoch": 206.44736842105263,
+      "grad_norm": 1.4420346021652222,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 31380
+    },
+    {
+      "epoch": 206.51315789473685,
+      "grad_norm": 1.2937798500061035,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 31390
+    },
+    {
+      "epoch": 206.57894736842104,
+      "grad_norm": 1.2045661211013794,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 31400
+    },
+    {
+      "epoch": 206.64473684210526,
+      "grad_norm": 1.0258898735046387,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 31410
+    },
+    {
+      "epoch": 206.71052631578948,
+      "grad_norm": 1.1112213134765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 31420
+    },
+    {
+      "epoch": 206.77631578947367,
+      "grad_norm": 1.2808059453964233,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 31430
+    },
+    {
+      "epoch": 206.8421052631579,
+      "grad_norm": 1.4980825185775757,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 31440
+    },
+    {
+      "epoch": 206.9078947368421,
+      "grad_norm": 1.1103272438049316,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 31450
+    },
+    {
+      "epoch": 206.97368421052633,
+      "grad_norm": 0.9099360704421997,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 31460
+    },
+    {
+      "epoch": 207.03947368421052,
+      "grad_norm": 1.1088165044784546,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 31470
+    },
+    {
+      "epoch": 207.10526315789474,
+      "grad_norm": 1.3863370418548584,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 31480
+    },
+    {
+      "epoch": 207.17105263157896,
+      "grad_norm": 1.522480845451355,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 31490
+    },
+    {
+      "epoch": 207.23684210526315,
+      "grad_norm": 1.1153115034103394,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 31500
+    },
+    {
+      "epoch": 207.30263157894737,
+      "grad_norm": 1.211844563484192,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 31510
+    },
+    {
+      "epoch": 207.3684210526316,
+      "grad_norm": 1.3280556201934814,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 31520
+    },
+    {
+      "epoch": 207.43421052631578,
+      "grad_norm": 1.232966423034668,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 31530
+    },
+    {
+      "epoch": 207.5,
+      "grad_norm": 1.3021212816238403,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 31540
+    },
+    {
+      "epoch": 207.56578947368422,
+      "grad_norm": 1.4837563037872314,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 31550
+    },
+    {
+      "epoch": 207.6315789473684,
+      "grad_norm": 1.122427225112915,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 31560
+    },
+    {
+      "epoch": 207.69736842105263,
+      "grad_norm": 1.358843207359314,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 31570
+    },
+    {
+      "epoch": 207.76315789473685,
+      "grad_norm": 1.1522997617721558,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 31580
+    },
+    {
+      "epoch": 207.82894736842104,
+      "grad_norm": 1.1149085760116577,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 31590
+    },
+    {
+      "epoch": 207.89473684210526,
+      "grad_norm": 1.3184139728546143,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 31600
+    },
+    {
+      "epoch": 207.96052631578948,
+      "grad_norm": 1.1251814365386963,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 31610
+    },
+    {
+      "epoch": 208.02631578947367,
+      "grad_norm": 1.4511667490005493,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 31620
+    },
+    {
+      "epoch": 208.0921052631579,
+      "grad_norm": 1.4698028564453125,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 31630
+    },
+    {
+      "epoch": 208.1578947368421,
+      "grad_norm": 1.1822978258132935,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 31640
+    },
+    {
+      "epoch": 208.22368421052633,
+      "grad_norm": 1.1715060472488403,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 31650
+    },
+    {
+      "epoch": 208.28947368421052,
+      "grad_norm": 1.4276723861694336,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 31660
+    },
+    {
+      "epoch": 208.35526315789474,
+      "grad_norm": 1.4519996643066406,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 31670
+    },
+    {
+      "epoch": 208.42105263157896,
+      "grad_norm": 1.0336174964904785,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 31680
+    },
+    {
+      "epoch": 208.48684210526315,
+      "grad_norm": 1.6115700006484985,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 31690
+    },
+    {
+      "epoch": 208.55263157894737,
+      "grad_norm": 1.6710377931594849,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 31700
+    },
+    {
+      "epoch": 208.6184210526316,
+      "grad_norm": 1.5397769212722778,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 31710
+    },
+    {
+      "epoch": 208.68421052631578,
+      "grad_norm": 1.5007388591766357,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 31720
+    },
+    {
+      "epoch": 208.75,
+      "grad_norm": 1.3674216270446777,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 31730
+    },
+    {
+      "epoch": 208.81578947368422,
+      "grad_norm": 2.207770347595215,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 31740
+    },
+    {
+      "epoch": 208.8815789473684,
+      "grad_norm": 1.5573480129241943,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 31750
+    },
+    {
+      "epoch": 208.94736842105263,
+      "grad_norm": 1.5399261713027954,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 31760
+    },
+    {
+      "epoch": 209.01315789473685,
+      "grad_norm": 1.6872302293777466,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 31770
+    },
+    {
+      "epoch": 209.07894736842104,
+      "grad_norm": 1.3373833894729614,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 31780
+    },
+    {
+      "epoch": 209.14473684210526,
+      "grad_norm": 1.6845489740371704,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 31790
+    },
+    {
+      "epoch": 209.21052631578948,
+      "grad_norm": 1.3818607330322266,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 31800
+    },
+    {
+      "epoch": 209.27631578947367,
+      "grad_norm": 1.3296219110488892,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 31810
+    },
+    {
+      "epoch": 209.3421052631579,
+      "grad_norm": 1.1584928035736084,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 31820
+    },
+    {
+      "epoch": 209.4078947368421,
+      "grad_norm": 1.3986483812332153,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 31830
+    },
+    {
+      "epoch": 209.47368421052633,
+      "grad_norm": 1.7007300853729248,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 31840
+    },
+    {
+      "epoch": 209.53947368421052,
+      "grad_norm": 0.916429877281189,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 31850
+    },
+    {
+      "epoch": 209.60526315789474,
+      "grad_norm": 1.1400072574615479,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 31860
+    },
+    {
+      "epoch": 209.67105263157896,
+      "grad_norm": 1.3716983795166016,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 31870
+    },
+    {
+      "epoch": 209.73684210526315,
+      "grad_norm": 0.9862516522407532,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 31880
+    },
+    {
+      "epoch": 209.80263157894737,
+      "grad_norm": 1.1696069240570068,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 31890
+    },
+    {
+      "epoch": 209.8684210526316,
+      "grad_norm": 1.4392577409744263,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 31900
+    },
+    {
+      "epoch": 209.93421052631578,
+      "grad_norm": 2.1823062896728516,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 31910
+    },
+    {
+      "epoch": 210.0,
+      "grad_norm": 1.6324290037155151,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 31920
+    },
+    {
+      "epoch": 210.06578947368422,
+      "grad_norm": 1.2188938856124878,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 31930
+    },
+    {
+      "epoch": 210.1315789473684,
+      "grad_norm": 1.401517391204834,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 31940
+    },
+    {
+      "epoch": 210.19736842105263,
+      "grad_norm": 1.3221532106399536,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 31950
+    },
+    {
+      "epoch": 210.26315789473685,
+      "grad_norm": 1.5217102766036987,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 31960
+    },
+    {
+      "epoch": 210.32894736842104,
+      "grad_norm": 1.4652286767959595,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 31970
+    },
+    {
+      "epoch": 210.39473684210526,
+      "grad_norm": 1.2757086753845215,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 31980
+    },
+    {
+      "epoch": 210.46052631578948,
+      "grad_norm": 1.227782130241394,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 31990
+    },
+    {
+      "epoch": 210.52631578947367,
+      "grad_norm": 1.3850603103637695,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 32000
+    },
+    {
+      "epoch": 210.5921052631579,
+      "grad_norm": 1.2859461307525635,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 32010
+    },
+    {
+      "epoch": 210.6578947368421,
+      "grad_norm": 1.3917819261550903,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 32020
+    },
+    {
+      "epoch": 210.72368421052633,
+      "grad_norm": 1.4318441152572632,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 32030
+    },
+    {
+      "epoch": 210.78947368421052,
+      "grad_norm": 1.5002471208572388,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 32040
+    },
+    {
+      "epoch": 210.85526315789474,
+      "grad_norm": 1.2311240434646606,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 32050
+    },
+    {
+      "epoch": 210.92105263157896,
+      "grad_norm": 1.293976902961731,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 32060
+    },
+    {
+      "epoch": 210.98684210526315,
+      "grad_norm": 1.1462241411209106,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 32070
+    },
+    {
+      "epoch": 211.05263157894737,
+      "grad_norm": 1.0642205476760864,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 32080
+    },
+    {
+      "epoch": 211.1184210526316,
+      "grad_norm": 1.0781315565109253,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 32090
+    },
+    {
+      "epoch": 211.18421052631578,
+      "grad_norm": 1.2331435680389404,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 32100
+    },
+    {
+      "epoch": 211.25,
+      "grad_norm": 1.129051685333252,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 32110
+    },
+    {
+      "epoch": 211.31578947368422,
+      "grad_norm": 1.301344871520996,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 32120
+    },
+    {
+      "epoch": 211.3815789473684,
+      "grad_norm": 1.6704833507537842,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 32130
+    },
+    {
+      "epoch": 211.44736842105263,
+      "grad_norm": 1.4150813817977905,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 32140
+    },
+    {
+      "epoch": 211.51315789473685,
+      "grad_norm": 1.3065874576568604,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 32150
+    },
+    {
+      "epoch": 211.57894736842104,
+      "grad_norm": 1.0689506530761719,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 32160
+    },
+    {
+      "epoch": 211.64473684210526,
+      "grad_norm": 1.220293641090393,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 32170
+    },
+    {
+      "epoch": 211.71052631578948,
+      "grad_norm": 1.4087119102478027,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 32180
+    },
+    {
+      "epoch": 211.77631578947367,
+      "grad_norm": 1.3689287900924683,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 32190
+    },
+    {
+      "epoch": 211.8421052631579,
+      "grad_norm": 1.4669215679168701,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 32200
+    },
+    {
+      "epoch": 211.9078947368421,
+      "grad_norm": 1.2780801057815552,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 32210
+    },
+    {
+      "epoch": 211.97368421052633,
+      "grad_norm": 1.0613449811935425,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 32220
+    },
+    {
+      "epoch": 212.03947368421052,
+      "grad_norm": 1.206862211227417,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 32230
+    },
+    {
+      "epoch": 212.10526315789474,
+      "grad_norm": 1.3805917501449585,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 32240
+    },
+    {
+      "epoch": 212.17105263157896,
+      "grad_norm": 1.1278748512268066,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 32250
+    },
+    {
+      "epoch": 212.23684210526315,
+      "grad_norm": 1.1782323122024536,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 32260
+    },
+    {
+      "epoch": 212.30263157894737,
+      "grad_norm": 1.0774714946746826,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 32270
+    },
+    {
+      "epoch": 212.3684210526316,
+      "grad_norm": 1.1292855739593506,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 32280
+    },
+    {
+      "epoch": 212.43421052631578,
+      "grad_norm": 1.4095853567123413,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 32290
+    },
+    {
+      "epoch": 212.5,
+      "grad_norm": 1.196883201599121,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 32300
+    },
+    {
+      "epoch": 212.56578947368422,
+      "grad_norm": 1.2091253995895386,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 32310
+    },
+    {
+      "epoch": 212.6315789473684,
+      "grad_norm": 1.3275787830352783,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 32320
+    },
+    {
+      "epoch": 212.69736842105263,
+      "grad_norm": 1.531229019165039,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 32330
+    },
+    {
+      "epoch": 212.76315789473685,
+      "grad_norm": 1.4241570234298706,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 32340
+    },
+    {
+      "epoch": 212.82894736842104,
+      "grad_norm": 0.965205192565918,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 32350
+    },
+    {
+      "epoch": 212.89473684210526,
+      "grad_norm": 1.515239953994751,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 32360
+    },
+    {
+      "epoch": 212.96052631578948,
+      "grad_norm": 1.5352081060409546,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 32370
+    },
+    {
+      "epoch": 213.02631578947367,
+      "grad_norm": 1.2615565061569214,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 32380
+    },
+    {
+      "epoch": 213.0921052631579,
+      "grad_norm": 1.6691360473632812,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 32390
+    },
+    {
+      "epoch": 213.1578947368421,
+      "grad_norm": 1.4845948219299316,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 32400
+    },
+    {
+      "epoch": 213.22368421052633,
+      "grad_norm": 1.1609926223754883,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 32410
+    },
+    {
+      "epoch": 213.28947368421052,
+      "grad_norm": 1.4472464323043823,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 32420
+    },
+    {
+      "epoch": 213.35526315789474,
+      "grad_norm": 1.1727287769317627,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 32430
+    },
+    {
+      "epoch": 213.42105263157896,
+      "grad_norm": 1.0012319087982178,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 32440
+    },
+    {
+      "epoch": 213.48684210526315,
+      "grad_norm": 1.4010659456253052,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 32450
+    },
+    {
+      "epoch": 213.55263157894737,
+      "grad_norm": 1.6938822269439697,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 32460
+    },
+    {
+      "epoch": 213.6184210526316,
+      "grad_norm": 1.7426364421844482,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 32470
+    },
+    {
+      "epoch": 213.68421052631578,
+      "grad_norm": 1.147688627243042,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 32480
+    },
+    {
+      "epoch": 213.75,
+      "grad_norm": 1.278310775756836,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 32490
+    },
+    {
+      "epoch": 213.81578947368422,
+      "grad_norm": 1.183099627494812,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 32500
+    },
+    {
+      "epoch": 213.8815789473684,
+      "grad_norm": 1.738754391670227,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 32510
+    },
+    {
+      "epoch": 213.94736842105263,
+      "grad_norm": 1.2291311025619507,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 32520
+    },
+    {
+      "epoch": 214.01315789473685,
+      "grad_norm": 1.1750850677490234,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 32530
+    },
+    {
+      "epoch": 214.07894736842104,
+      "grad_norm": 1.2301994562149048,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 32540
+    },
+    {
+      "epoch": 214.14473684210526,
+      "grad_norm": 1.2301639318466187,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 32550
+    },
+    {
+      "epoch": 214.21052631578948,
+      "grad_norm": 1.4040155410766602,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 32560
+    },
+    {
+      "epoch": 214.27631578947367,
+      "grad_norm": 1.4484095573425293,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 32570
+    },
+    {
+      "epoch": 214.3421052631579,
+      "grad_norm": 1.0612156391143799,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 32580
+    },
+    {
+      "epoch": 214.4078947368421,
+      "grad_norm": 1.1924538612365723,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 32590
+    },
+    {
+      "epoch": 214.47368421052633,
+      "grad_norm": 1.3676021099090576,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 32600
+    },
+    {
+      "epoch": 214.53947368421052,
+      "grad_norm": 1.155792474746704,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 32610
+    },
+    {
+      "epoch": 214.60526315789474,
+      "grad_norm": 1.2802067995071411,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 32620
+    },
+    {
+      "epoch": 214.67105263157896,
+      "grad_norm": 1.1130114793777466,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 32630
+    },
+    {
+      "epoch": 214.73684210526315,
+      "grad_norm": 1.3467772006988525,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 32640
+    },
+    {
+      "epoch": 214.80263157894737,
+      "grad_norm": 1.1575422286987305,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 32650
+    },
+    {
+      "epoch": 214.8684210526316,
+      "grad_norm": 1.0883058309555054,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 32660
+    },
+    {
+      "epoch": 214.93421052631578,
+      "grad_norm": 1.3685483932495117,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 32670
+    },
+    {
+      "epoch": 215.0,
+      "grad_norm": 1.452791690826416,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 32680
+    },
+    {
+      "epoch": 215.06578947368422,
+      "grad_norm": 1.2236331701278687,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 32690
+    },
+    {
+      "epoch": 215.1315789473684,
+      "grad_norm": 1.1579644680023193,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 32700
+    },
+    {
+      "epoch": 215.19736842105263,
+      "grad_norm": 1.2398250102996826,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 32710
+    },
+    {
+      "epoch": 215.26315789473685,
+      "grad_norm": 0.9351480603218079,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 32720
+    },
+    {
+      "epoch": 215.32894736842104,
+      "grad_norm": 0.8759018778800964,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 32730
+    },
+    {
+      "epoch": 215.39473684210526,
+      "grad_norm": 1.077101230621338,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 32740
+    },
+    {
+      "epoch": 215.46052631578948,
+      "grad_norm": 0.8699200749397278,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 32750
+    },
+    {
+      "epoch": 215.52631578947367,
+      "grad_norm": 1.2081657648086548,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 32760
+    },
+    {
+      "epoch": 215.5921052631579,
+      "grad_norm": 1.218841791152954,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 32770
+    },
+    {
+      "epoch": 215.6578947368421,
+      "grad_norm": 1.0985344648361206,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 32780
+    },
+    {
+      "epoch": 215.72368421052633,
+      "grad_norm": 1.242873191833496,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 32790
+    },
+    {
+      "epoch": 215.78947368421052,
+      "grad_norm": 0.956916868686676,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 32800
+    },
+    {
+      "epoch": 215.85526315789474,
+      "grad_norm": 0.9822515249252319,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 32810
+    },
+    {
+      "epoch": 215.92105263157896,
+      "grad_norm": 0.8483900427818298,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 32820
+    },
+    {
+      "epoch": 215.98684210526315,
+      "grad_norm": 1.0709314346313477,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 32830
+    },
+    {
+      "epoch": 216.05263157894737,
+      "grad_norm": 1.0873550176620483,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 32840
+    },
+    {
+      "epoch": 216.1184210526316,
+      "grad_norm": 1.20504891872406,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 32850
+    },
+    {
+      "epoch": 216.18421052631578,
+      "grad_norm": 1.1691912412643433,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 32860
+    },
+    {
+      "epoch": 216.25,
+      "grad_norm": 1.352332592010498,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 32870
+    },
+    {
+      "epoch": 216.31578947368422,
+      "grad_norm": 0.8410949110984802,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 32880
+    },
+    {
+      "epoch": 216.3815789473684,
+      "grad_norm": 1.0081835985183716,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 32890
+    },
+    {
+      "epoch": 216.44736842105263,
+      "grad_norm": 0.9765644073486328,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 32900
+    },
+    {
+      "epoch": 216.51315789473685,
+      "grad_norm": 1.4321998357772827,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 32910
+    },
+    {
+      "epoch": 216.57894736842104,
+      "grad_norm": 1.1078122854232788,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 32920
+    },
+    {
+      "epoch": 216.64473684210526,
+      "grad_norm": 1.004533290863037,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 32930
+    },
+    {
+      "epoch": 216.71052631578948,
+      "grad_norm": 1.0383466482162476,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 32940
+    },
+    {
+      "epoch": 216.77631578947367,
+      "grad_norm": 1.178542137145996,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 32950
+    },
+    {
+      "epoch": 216.8421052631579,
+      "grad_norm": 1.2944022417068481,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 32960
+    },
+    {
+      "epoch": 216.9078947368421,
+      "grad_norm": 1.0292662382125854,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 32970
+    },
+    {
+      "epoch": 216.97368421052633,
+      "grad_norm": 1.4745246171951294,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 32980
+    },
+    {
+      "epoch": 217.03947368421052,
+      "grad_norm": 1.0737231969833374,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 32990
+    },
+    {
+      "epoch": 217.10526315789474,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 33000
+    },
+    {
+      "epoch": 217.17105263157896,
+      "grad_norm": 1.142307162284851,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 33010
+    },
+    {
+      "epoch": 217.23684210526315,
+      "grad_norm": 0.9450036883354187,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 33020
+    },
+    {
+      "epoch": 217.30263157894737,
+      "grad_norm": 0.9609930515289307,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 33030
+    },
+    {
+      "epoch": 217.3684210526316,
+      "grad_norm": 0.949567437171936,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 33040
+    },
+    {
+      "epoch": 217.43421052631578,
+      "grad_norm": 1.1803269386291504,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 33050
+    },
+    {
+      "epoch": 217.5,
+      "grad_norm": 1.2131317853927612,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 33060
+    },
+    {
+      "epoch": 217.56578947368422,
+      "grad_norm": 1.3640658855438232,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 33070
+    },
+    {
+      "epoch": 217.6315789473684,
+      "grad_norm": 1.1110533475875854,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 33080
+    },
+    {
+      "epoch": 217.69736842105263,
+      "grad_norm": 1.4728652238845825,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 33090
+    },
+    {
+      "epoch": 217.76315789473685,
+      "grad_norm": 1.1803345680236816,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 33100
+    },
+    {
+      "epoch": 217.82894736842104,
+      "grad_norm": 1.0455313920974731,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 33110
+    },
+    {
+      "epoch": 217.89473684210526,
+      "grad_norm": 1.290387749671936,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 33120
+    },
+    {
+      "epoch": 217.96052631578948,
+      "grad_norm": 1.0073187351226807,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 33130
+    },
+    {
+      "epoch": 218.02631578947367,
+      "grad_norm": 1.0683584213256836,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 33140
+    },
+    {
+      "epoch": 218.0921052631579,
+      "grad_norm": 1.0977116823196411,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 33150
+    },
+    {
+      "epoch": 218.1578947368421,
+      "grad_norm": 1.2289358377456665,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 33160
+    },
+    {
+      "epoch": 218.22368421052633,
+      "grad_norm": 1.4504176378250122,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 33170
+    },
+    {
+      "epoch": 218.28947368421052,
+      "grad_norm": 1.3620048761367798,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 33180
+    },
+    {
+      "epoch": 218.35526315789474,
+      "grad_norm": 1.0863347053527832,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 33190
+    },
+    {
+      "epoch": 218.42105263157896,
+      "grad_norm": 1.5031284093856812,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 33200
+    },
+    {
+      "epoch": 218.48684210526315,
+      "grad_norm": 1.6940463781356812,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 33210
+    },
+    {
+      "epoch": 218.55263157894737,
+      "grad_norm": 1.5782625675201416,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 33220
+    },
+    {
+      "epoch": 218.6184210526316,
+      "grad_norm": 1.5026166439056396,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 33230
+    },
+    {
+      "epoch": 218.68421052631578,
+      "grad_norm": 0.9451717734336853,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 33240
+    },
+    {
+      "epoch": 218.75,
+      "grad_norm": 1.007091999053955,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 33250
+    },
+    {
+      "epoch": 218.81578947368422,
+      "grad_norm": 1.0861047506332397,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 33260
+    },
+    {
+      "epoch": 218.8815789473684,
+      "grad_norm": 1.3030095100402832,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 33270
+    },
+    {
+      "epoch": 218.94736842105263,
+      "grad_norm": 1.0647660493850708,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 33280
+    },
+    {
+      "epoch": 219.01315789473685,
+      "grad_norm": 1.160729169845581,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 33290
+    },
+    {
+      "epoch": 219.07894736842104,
+      "grad_norm": 0.992527425289154,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 33300
+    },
+    {
+      "epoch": 219.14473684210526,
+      "grad_norm": 1.6091036796569824,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 33310
+    },
+    {
+      "epoch": 219.21052631578948,
+      "grad_norm": 1.16289222240448,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 33320
+    },
+    {
+      "epoch": 219.27631578947367,
+      "grad_norm": 0.95487380027771,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 33330
+    },
+    {
+      "epoch": 219.3421052631579,
+      "grad_norm": 0.9959771633148193,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 33340
+    },
+    {
+      "epoch": 219.4078947368421,
+      "grad_norm": 1.1076931953430176,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 33350
+    },
+    {
+      "epoch": 219.47368421052633,
+      "grad_norm": 1.2252377271652222,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 33360
+    },
+    {
+      "epoch": 219.53947368421052,
+      "grad_norm": 1.2380799055099487,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 33370
+    },
+    {
+      "epoch": 219.60526315789474,
+      "grad_norm": 1.33864164352417,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 33380
+    },
+    {
+      "epoch": 219.67105263157896,
+      "grad_norm": 1.1309316158294678,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 33390
+    },
+    {
+      "epoch": 219.73684210526315,
+      "grad_norm": 1.1230326890945435,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 33400
+    },
+    {
+      "epoch": 219.80263157894737,
+      "grad_norm": 1.306256890296936,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 33410
+    },
+    {
+      "epoch": 219.8684210526316,
+      "grad_norm": 1.3912127017974854,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 33420
+    },
+    {
+      "epoch": 219.93421052631578,
+      "grad_norm": 1.529556155204773,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 33430
+    },
+    {
+      "epoch": 220.0,
+      "grad_norm": 1.9190083742141724,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 33440
+    },
+    {
+      "epoch": 220.06578947368422,
+      "grad_norm": 1.1662191152572632,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 33450
+    },
+    {
+      "epoch": 220.1315789473684,
+      "grad_norm": 1.6552345752716064,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 33460
+    },
+    {
+      "epoch": 220.19736842105263,
+      "grad_norm": 1.3389320373535156,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 33470
+    },
+    {
+      "epoch": 220.26315789473685,
+      "grad_norm": 1.5310899019241333,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 33480
+    },
+    {
+      "epoch": 220.32894736842104,
+      "grad_norm": 1.621925950050354,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 33490
+    },
+    {
+      "epoch": 220.39473684210526,
+      "grad_norm": 1.3250995874404907,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 33500
+    },
+    {
+      "epoch": 220.46052631578948,
+      "grad_norm": 1.3257274627685547,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 33510
+    },
+    {
+      "epoch": 220.52631578947367,
+      "grad_norm": 1.333977460861206,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 33520
+    },
+    {
+      "epoch": 220.5921052631579,
+      "grad_norm": 1.4061793088912964,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 33530
+    },
+    {
+      "epoch": 220.6578947368421,
+      "grad_norm": 1.223670244216919,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 33540
+    },
+    {
+      "epoch": 220.72368421052633,
+      "grad_norm": 1.1718958616256714,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 33550
+    },
+    {
+      "epoch": 220.78947368421052,
+      "grad_norm": 1.171622395515442,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 33560
+    },
+    {
+      "epoch": 220.85526315789474,
+      "grad_norm": 1.121756911277771,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 33570
+    },
+    {
+      "epoch": 220.92105263157896,
+      "grad_norm": 1.4940133094787598,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 33580
+    },
+    {
+      "epoch": 220.98684210526315,
+      "grad_norm": 0.976639449596405,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 33590
+    },
+    {
+      "epoch": 221.05263157894737,
+      "grad_norm": 1.2112866640090942,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 33600
+    },
+    {
+      "epoch": 221.1184210526316,
+      "grad_norm": 1.3244744539260864,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 33610
+    },
+    {
+      "epoch": 221.18421052631578,
+      "grad_norm": 1.2948193550109863,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 33620
+    },
+    {
+      "epoch": 221.25,
+      "grad_norm": 1.39680016040802,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 33630
+    },
+    {
+      "epoch": 221.31578947368422,
+      "grad_norm": 1.0296803712844849,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 33640
+    },
+    {
+      "epoch": 221.3815789473684,
+      "grad_norm": 1.1246095895767212,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 33650
+    },
+    {
+      "epoch": 221.44736842105263,
+      "grad_norm": 1.0090177059173584,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 33660
+    },
+    {
+      "epoch": 221.51315789473685,
+      "grad_norm": 1.1224722862243652,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 33670
+    },
+    {
+      "epoch": 221.57894736842104,
+      "grad_norm": 1.6427732706069946,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 33680
+    },
+    {
+      "epoch": 221.64473684210526,
+      "grad_norm": 1.3435227870941162,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 33690
+    },
+    {
+      "epoch": 221.71052631578948,
+      "grad_norm": 1.7769924402236938,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 33700
+    },
+    {
+      "epoch": 221.77631578947367,
+      "grad_norm": 1.2814626693725586,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 33710
+    },
+    {
+      "epoch": 221.8421052631579,
+      "grad_norm": 1.3737503290176392,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 33720
+    },
+    {
+      "epoch": 221.9078947368421,
+      "grad_norm": 1.2347252368927002,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 33730
+    },
+    {
+      "epoch": 221.97368421052633,
+      "grad_norm": 1.3277086019515991,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 33740
+    },
+    {
+      "epoch": 222.03947368421052,
+      "grad_norm": 1.7966270446777344,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 33750
+    },
+    {
+      "epoch": 222.10526315789474,
+      "grad_norm": 1.3284343481063843,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 33760
+    },
+    {
+      "epoch": 222.17105263157896,
+      "grad_norm": 1.6279863119125366,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 33770
+    },
+    {
+      "epoch": 222.23684210526315,
+      "grad_norm": 1.5999658107757568,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 33780
+    },
+    {
+      "epoch": 222.30263157894737,
+      "grad_norm": 1.5054606199264526,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 33790
+    },
+    {
+      "epoch": 222.3684210526316,
+      "grad_norm": 1.4686840772628784,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 33800
+    },
+    {
+      "epoch": 222.43421052631578,
+      "grad_norm": 1.1565569639205933,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 33810
+    },
+    {
+      "epoch": 222.5,
+      "grad_norm": 1.396141529083252,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 33820
+    },
+    {
+      "epoch": 222.56578947368422,
+      "grad_norm": 1.1285852193832397,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 33830
+    },
+    {
+      "epoch": 222.6315789473684,
+      "grad_norm": 1.093670129776001,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 33840
+    },
+    {
+      "epoch": 222.69736842105263,
+      "grad_norm": 1.443837285041809,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 33850
+    },
+    {
+      "epoch": 222.76315789473685,
+      "grad_norm": 1.3673397302627563,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 33860
+    },
+    {
+      "epoch": 222.82894736842104,
+      "grad_norm": 1.2065106630325317,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 33870
+    },
+    {
+      "epoch": 222.89473684210526,
+      "grad_norm": 1.3784736394882202,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 33880
+    },
+    {
+      "epoch": 222.96052631578948,
+      "grad_norm": 1.6203807592391968,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 33890
+    },
+    {
+      "epoch": 223.02631578947367,
+      "grad_norm": 1.314103126525879,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 33900
+    },
+    {
+      "epoch": 223.0921052631579,
+      "grad_norm": 1.6348367929458618,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 33910
+    },
+    {
+      "epoch": 223.1578947368421,
+      "grad_norm": 1.20957350730896,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 33920
+    },
+    {
+      "epoch": 223.22368421052633,
+      "grad_norm": 1.1228580474853516,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 33930
+    },
+    {
+      "epoch": 223.28947368421052,
+      "grad_norm": 1.4203588962554932,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 33940
+    },
+    {
+      "epoch": 223.35526315789474,
+      "grad_norm": 1.2360961437225342,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 33950
+    },
+    {
+      "epoch": 223.42105263157896,
+      "grad_norm": 0.9158451557159424,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 33960
+    },
+    {
+      "epoch": 223.48684210526315,
+      "grad_norm": 1.6221871376037598,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 33970
+    },
+    {
+      "epoch": 223.55263157894737,
+      "grad_norm": 1.559161901473999,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 33980
+    },
+    {
+      "epoch": 223.6184210526316,
+      "grad_norm": 1.876190423965454,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 33990
+    },
+    {
+      "epoch": 223.68421052631578,
+      "grad_norm": 1.5894197225570679,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 34000
+    },
+    {
+      "epoch": 223.75,
+      "grad_norm": 1.2699931859970093,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 34010
+    },
+    {
+      "epoch": 223.81578947368422,
+      "grad_norm": 1.1223138570785522,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 34020
+    },
+    {
+      "epoch": 223.8815789473684,
+      "grad_norm": 1.2991082668304443,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 34030
+    },
+    {
+      "epoch": 223.94736842105263,
+      "grad_norm": 1.1831358671188354,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 34040
+    },
+    {
+      "epoch": 224.01315789473685,
+      "grad_norm": 0.8384124636650085,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 34050
+    },
+    {
+      "epoch": 224.07894736842104,
+      "grad_norm": 1.0878987312316895,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 34060
+    },
+    {
+      "epoch": 224.14473684210526,
+      "grad_norm": 1.440589189529419,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 34070
+    },
+    {
+      "epoch": 224.21052631578948,
+      "grad_norm": 1.6876184940338135,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 34080
+    },
+    {
+      "epoch": 224.27631578947367,
+      "grad_norm": 1.424138069152832,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 34090
+    },
+    {
+      "epoch": 224.3421052631579,
+      "grad_norm": 1.2443312406539917,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 34100
+    },
+    {
+      "epoch": 224.4078947368421,
+      "grad_norm": 1.3643524646759033,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 34110
+    },
+    {
+      "epoch": 224.47368421052633,
+      "grad_norm": 1.1311956644058228,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 34120
+    },
+    {
+      "epoch": 224.53947368421052,
+      "grad_norm": 1.0207782983779907,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 34130
+    },
+    {
+      "epoch": 224.60526315789474,
+      "grad_norm": 1.3237141370773315,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 34140
+    },
+    {
+      "epoch": 224.67105263157896,
+      "grad_norm": 1.5647715330123901,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 34150
+    },
+    {
+      "epoch": 224.73684210526315,
+      "grad_norm": 1.4168190956115723,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 34160
+    },
+    {
+      "epoch": 224.80263157894737,
+      "grad_norm": 1.3588078022003174,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 34170
+    },
+    {
+      "epoch": 224.8684210526316,
+      "grad_norm": 1.0193902254104614,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 34180
+    },
+    {
+      "epoch": 224.93421052631578,
+      "grad_norm": 1.3121767044067383,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 34190
+    },
+    {
+      "epoch": 225.0,
+      "grad_norm": 1.4055311679840088,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 34200
+    },
+    {
+      "epoch": 225.06578947368422,
+      "grad_norm": 1.3310362100601196,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 34210
+    },
+    {
+      "epoch": 225.1315789473684,
+      "grad_norm": 1.2854124307632446,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 34220
+    },
+    {
+      "epoch": 225.19736842105263,
+      "grad_norm": 1.1616575717926025,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 34230
+    },
+    {
+      "epoch": 225.26315789473685,
+      "grad_norm": 1.6716482639312744,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 34240
+    },
+    {
+      "epoch": 225.32894736842104,
+      "grad_norm": 1.0361127853393555,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 34250
+    },
+    {
+      "epoch": 225.39473684210526,
+      "grad_norm": 1.1146401166915894,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 34260
+    },
+    {
+      "epoch": 225.46052631578948,
+      "grad_norm": 1.2672135829925537,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 34270
+    },
+    {
+      "epoch": 225.52631578947367,
+      "grad_norm": 1.557548999786377,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 34280
+    },
+    {
+      "epoch": 225.5921052631579,
+      "grad_norm": 1.0384529829025269,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 34290
+    },
+    {
+      "epoch": 225.6578947368421,
+      "grad_norm": 1.3350151777267456,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 34300
+    },
+    {
+      "epoch": 225.72368421052633,
+      "grad_norm": 1.3474798202514648,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 34310
+    },
+    {
+      "epoch": 225.78947368421052,
+      "grad_norm": 0.9546133875846863,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 34320
+    },
+    {
+      "epoch": 225.85526315789474,
+      "grad_norm": 1.325697660446167,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 34330
+    },
+    {
+      "epoch": 225.92105263157896,
+      "grad_norm": 1.2987803220748901,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 34340
+    },
+    {
+      "epoch": 225.98684210526315,
+      "grad_norm": 1.4725877046585083,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 34350
+    },
+    {
+      "epoch": 226.05263157894737,
+      "grad_norm": 1.2885562181472778,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 34360
+    },
+    {
+      "epoch": 226.1184210526316,
+      "grad_norm": 1.1747262477874756,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 34370
+    },
+    {
+      "epoch": 226.18421052631578,
+      "grad_norm": 1.6619282960891724,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 34380
+    },
+    {
+      "epoch": 226.25,
+      "grad_norm": 1.142620325088501,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 34390
+    },
+    {
+      "epoch": 226.31578947368422,
+      "grad_norm": 1.11537766456604,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 34400
+    },
+    {
+      "epoch": 226.3815789473684,
+      "grad_norm": 0.9072920083999634,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 34410
+    },
+    {
+      "epoch": 226.44736842105263,
+      "grad_norm": 0.9555834531784058,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 34420
+    },
+    {
+      "epoch": 226.51315789473685,
+      "grad_norm": 1.3261579275131226,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 34430
+    },
+    {
+      "epoch": 226.57894736842104,
+      "grad_norm": 1.6103168725967407,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 34440
+    },
+    {
+      "epoch": 226.64473684210526,
+      "grad_norm": 1.1931551694869995,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 34450
+    },
+    {
+      "epoch": 226.71052631578948,
+      "grad_norm": 1.633232593536377,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 34460
+    },
+    {
+      "epoch": 226.77631578947367,
+      "grad_norm": 1.4821163415908813,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 34470
+    },
+    {
+      "epoch": 226.8421052631579,
+      "grad_norm": 0.9790869951248169,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 34480
+    },
+    {
+      "epoch": 226.9078947368421,
+      "grad_norm": 0.8639764785766602,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 34490
+    },
+    {
+      "epoch": 226.97368421052633,
+      "grad_norm": 1.0887365341186523,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 34500
+    },
+    {
+      "epoch": 227.03947368421052,
+      "grad_norm": 0.8437522649765015,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 34510
+    },
+    {
+      "epoch": 227.10526315789474,
+      "grad_norm": 1.2919409275054932,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 34520
+    },
+    {
+      "epoch": 227.17105263157896,
+      "grad_norm": 1.2690752744674683,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 34530
+    },
+    {
+      "epoch": 227.23684210526315,
+      "grad_norm": 1.1717513799667358,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 34540
+    },
+    {
+      "epoch": 227.30263157894737,
+      "grad_norm": 1.1742141246795654,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 34550
+    },
+    {
+      "epoch": 227.3684210526316,
+      "grad_norm": 1.2285854816436768,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 34560
+    },
+    {
+      "epoch": 227.43421052631578,
+      "grad_norm": 1.6287882328033447,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 34570
+    },
+    {
+      "epoch": 227.5,
+      "grad_norm": 1.4671905040740967,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 34580
+    },
+    {
+      "epoch": 227.56578947368422,
+      "grad_norm": 1.2717926502227783,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 34590
+    },
+    {
+      "epoch": 227.6315789473684,
+      "grad_norm": 1.2138713598251343,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 34600
+    },
+    {
+      "epoch": 227.69736842105263,
+      "grad_norm": 1.219657301902771,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 34610
+    },
+    {
+      "epoch": 227.76315789473685,
+      "grad_norm": 1.2700283527374268,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 34620
+    },
+    {
+      "epoch": 227.82894736842104,
+      "grad_norm": 1.2568398714065552,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 34630
+    },
+    {
+      "epoch": 227.89473684210526,
+      "grad_norm": 1.308393120765686,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 34640
+    },
+    {
+      "epoch": 227.96052631578948,
+      "grad_norm": 1.6060470342636108,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 34650
+    },
+    {
+      "epoch": 228.02631578947367,
+      "grad_norm": 1.4891085624694824,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 34660
+    },
+    {
+      "epoch": 228.0921052631579,
+      "grad_norm": 1.2525702714920044,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 34670
+    },
+    {
+      "epoch": 228.1578947368421,
+      "grad_norm": 1.2235273122787476,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 34680
+    },
+    {
+      "epoch": 228.22368421052633,
+      "grad_norm": 1.2164841890335083,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 34690
+    },
+    {
+      "epoch": 228.28947368421052,
+      "grad_norm": 1.487396001815796,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 34700
+    },
+    {
+      "epoch": 228.35526315789474,
+      "grad_norm": 1.3770133256912231,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 34710
+    },
+    {
+      "epoch": 228.42105263157896,
+      "grad_norm": 1.5073363780975342,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 34720
+    },
+    {
+      "epoch": 228.48684210526315,
+      "grad_norm": 1.1958956718444824,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 34730
+    },
+    {
+      "epoch": 228.55263157894737,
+      "grad_norm": 1.1365365982055664,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 34740
+    },
+    {
+      "epoch": 228.6184210526316,
+      "grad_norm": 0.9599331021308899,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 34750
+    },
+    {
+      "epoch": 228.68421052631578,
+      "grad_norm": 0.7286871671676636,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 34760
+    },
+    {
+      "epoch": 228.75,
+      "grad_norm": 0.8963000774383545,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 34770
+    },
+    {
+      "epoch": 228.81578947368422,
+      "grad_norm": 1.069594144821167,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 34780
+    },
+    {
+      "epoch": 228.8815789473684,
+      "grad_norm": 1.107754111289978,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 34790
+    },
+    {
+      "epoch": 228.94736842105263,
+      "grad_norm": 0.8711170554161072,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 34800
+    },
+    {
+      "epoch": 229.01315789473685,
+      "grad_norm": 1.2165173292160034,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 34810
+    },
+    {
+      "epoch": 229.07894736842104,
+      "grad_norm": 1.3709280490875244,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 34820
+    },
+    {
+      "epoch": 229.14473684210526,
+      "grad_norm": 0.9676075577735901,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 34830
+    },
+    {
+      "epoch": 229.21052631578948,
+      "grad_norm": 1.192239761352539,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 34840
+    },
+    {
+      "epoch": 229.27631578947367,
+      "grad_norm": 1.1368480920791626,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 34850
+    },
+    {
+      "epoch": 229.3421052631579,
+      "grad_norm": 1.0366135835647583,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 34860
+    },
+    {
+      "epoch": 229.4078947368421,
+      "grad_norm": 1.52436101436615,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 34870
+    },
+    {
+      "epoch": 229.47368421052633,
+      "grad_norm": 1.2649922370910645,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 34880
+    },
+    {
+      "epoch": 229.53947368421052,
+      "grad_norm": 1.4526771306991577,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 34890
+    },
+    {
+      "epoch": 229.60526315789474,
+      "grad_norm": 1.317568063735962,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 34900
+    },
+    {
+      "epoch": 229.67105263157896,
+      "grad_norm": 1.5620092153549194,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 34910
+    },
+    {
+      "epoch": 229.73684210526315,
+      "grad_norm": 1.2714239358901978,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 34920
+    },
+    {
+      "epoch": 229.80263157894737,
+      "grad_norm": 1.5597683191299438,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 34930
+    },
+    {
+      "epoch": 229.8684210526316,
+      "grad_norm": 1.3051207065582275,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 34940
+    },
+    {
+      "epoch": 229.93421052631578,
+      "grad_norm": 1.688822865486145,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 34950
+    },
+    {
+      "epoch": 230.0,
+      "grad_norm": 1.4721719026565552,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 34960
+    },
+    {
+      "epoch": 230.06578947368422,
+      "grad_norm": 1.6833611726760864,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 34970
+    },
+    {
+      "epoch": 230.1315789473684,
+      "grad_norm": 1.749559998512268,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 34980
+    },
+    {
+      "epoch": 230.19736842105263,
+      "grad_norm": 1.0270739793777466,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 34990
+    },
+    {
+      "epoch": 230.26315789473685,
+      "grad_norm": 1.5380287170410156,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 35000
+    },
+    {
+      "epoch": 230.32894736842104,
+      "grad_norm": 1.2859152555465698,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 35010
+    },
+    {
+      "epoch": 230.39473684210526,
+      "grad_norm": 1.2448034286499023,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 35020
+    },
+    {
+      "epoch": 230.46052631578948,
+      "grad_norm": 1.1096408367156982,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 35030
+    },
+    {
+      "epoch": 230.52631578947367,
+      "grad_norm": 1.2876746654510498,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 35040
+    },
+    {
+      "epoch": 230.5921052631579,
+      "grad_norm": 1.4161434173583984,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 35050
+    },
+    {
+      "epoch": 230.6578947368421,
+      "grad_norm": 1.1465400457382202,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 35060
+    },
+    {
+      "epoch": 230.72368421052633,
+      "grad_norm": 1.4166468381881714,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 35070
+    },
+    {
+      "epoch": 230.78947368421052,
+      "grad_norm": 1.3646936416625977,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 35080
+    },
+    {
+      "epoch": 230.85526315789474,
+      "grad_norm": 1.3088558912277222,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 35090
+    },
+    {
+      "epoch": 230.92105263157896,
+      "grad_norm": 1.1793044805526733,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 35100
+    },
+    {
+      "epoch": 230.98684210526315,
+      "grad_norm": 1.1064832210540771,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 35110
+    },
+    {
+      "epoch": 231.05263157894737,
+      "grad_norm": 1.2353724241256714,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 35120
+    },
+    {
+      "epoch": 231.1184210526316,
+      "grad_norm": 1.4452154636383057,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 35130
+    },
+    {
+      "epoch": 231.18421052631578,
+      "grad_norm": 1.1226993799209595,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 35140
+    },
+    {
+      "epoch": 231.25,
+      "grad_norm": 1.0637484788894653,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 35150
+    },
+    {
+      "epoch": 231.31578947368422,
+      "grad_norm": 0.9938138723373413,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 35160
+    },
+    {
+      "epoch": 231.3815789473684,
+      "grad_norm": 1.5435380935668945,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 35170
+    },
+    {
+      "epoch": 231.44736842105263,
+      "grad_norm": 1.042462944984436,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 35180
+    },
+    {
+      "epoch": 231.51315789473685,
+      "grad_norm": 1.0510119199752808,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 35190
+    },
+    {
+      "epoch": 231.57894736842104,
+      "grad_norm": 1.3831288814544678,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 35200
+    },
+    {
+      "epoch": 231.64473684210526,
+      "grad_norm": 1.462022066116333,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 35210
+    },
+    {
+      "epoch": 231.71052631578948,
+      "grad_norm": 1.0927327871322632,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 35220
+    },
+    {
+      "epoch": 231.77631578947367,
+      "grad_norm": 1.1333094835281372,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 35230
+    },
+    {
+      "epoch": 231.8421052631579,
+      "grad_norm": 1.2732223272323608,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 35240
+    },
+    {
+      "epoch": 231.9078947368421,
+      "grad_norm": 1.2944512367248535,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 35250
+    },
+    {
+      "epoch": 231.97368421052633,
+      "grad_norm": 1.2028173208236694,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 35260
+    },
+    {
+      "epoch": 232.03947368421052,
+      "grad_norm": 1.0721412897109985,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 35270
+    },
+    {
+      "epoch": 232.10526315789474,
+      "grad_norm": 1.3471068143844604,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 35280
+    },
+    {
+      "epoch": 232.17105263157896,
+      "grad_norm": 1.085255742073059,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 35290
+    },
+    {
+      "epoch": 232.23684210526315,
+      "grad_norm": 1.335943579673767,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 35300
+    },
+    {
+      "epoch": 232.30263157894737,
+      "grad_norm": 1.0885100364685059,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 35310
+    },
+    {
+      "epoch": 232.3684210526316,
+      "grad_norm": 1.2813364267349243,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 35320
+    },
+    {
+      "epoch": 232.43421052631578,
+      "grad_norm": 1.1324126720428467,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 35330
+    },
+    {
+      "epoch": 232.5,
+      "grad_norm": 1.5118268728256226,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 35340
+    },
+    {
+      "epoch": 232.56578947368422,
+      "grad_norm": 1.3967148065567017,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 35350
+    },
+    {
+      "epoch": 232.6315789473684,
+      "grad_norm": 1.3626654148101807,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 35360
+    },
+    {
+      "epoch": 232.69736842105263,
+      "grad_norm": 0.9588110446929932,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 35370
+    },
+    {
+      "epoch": 232.76315789473685,
+      "grad_norm": 1.7593961954116821,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 35380
+    },
+    {
+      "epoch": 232.82894736842104,
+      "grad_norm": 1.4406646490097046,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 35390
+    },
+    {
+      "epoch": 232.89473684210526,
+      "grad_norm": 1.0357986688613892,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 35400
+    },
+    {
+      "epoch": 232.96052631578948,
+      "grad_norm": 1.051256775856018,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 35410
+    },
+    {
+      "epoch": 233.02631578947367,
+      "grad_norm": 1.338333249092102,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 35420
+    },
+    {
+      "epoch": 233.0921052631579,
+      "grad_norm": 1.0235182046890259,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 35430
+    },
+    {
+      "epoch": 233.1578947368421,
+      "grad_norm": 1.265726923942566,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 35440
+    },
+    {
+      "epoch": 233.22368421052633,
+      "grad_norm": 1.3721272945404053,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 35450
+    },
+    {
+      "epoch": 233.28947368421052,
+      "grad_norm": 1.0401647090911865,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 35460
+    },
+    {
+      "epoch": 233.35526315789474,
+      "grad_norm": 1.3844105005264282,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 35470
+    },
+    {
+      "epoch": 233.42105263157896,
+      "grad_norm": 1.977495551109314,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 35480
+    },
+    {
+      "epoch": 233.48684210526315,
+      "grad_norm": 1.13108229637146,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 35490
+    },
+    {
+      "epoch": 233.55263157894737,
+      "grad_norm": 1.4336754083633423,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 35500
+    },
+    {
+      "epoch": 233.6184210526316,
+      "grad_norm": 1.2272390127182007,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 35510
+    },
+    {
+      "epoch": 233.68421052631578,
+      "grad_norm": 0.978657066822052,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 35520
+    },
+    {
+      "epoch": 233.75,
+      "grad_norm": 1.2305833101272583,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 35530
+    },
+    {
+      "epoch": 233.81578947368422,
+      "grad_norm": 1.1648582220077515,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 35540
+    },
+    {
+      "epoch": 233.8815789473684,
+      "grad_norm": 1.3043948411941528,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 35550
+    },
+    {
+      "epoch": 233.94736842105263,
+      "grad_norm": 1.4503754377365112,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 35560
+    },
+    {
+      "epoch": 234.01315789473685,
+      "grad_norm": 1.3145157098770142,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 35570
+    },
+    {
+      "epoch": 234.07894736842104,
+      "grad_norm": 0.9834286570549011,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 35580
+    },
+    {
+      "epoch": 234.14473684210526,
+      "grad_norm": 1.5456626415252686,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 35590
+    },
+    {
+      "epoch": 234.21052631578948,
+      "grad_norm": 1.1707838773727417,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 35600
+    },
+    {
+      "epoch": 234.27631578947367,
+      "grad_norm": 1.4479657411575317,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 35610
+    },
+    {
+      "epoch": 234.3421052631579,
+      "grad_norm": 1.352433204650879,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 35620
+    },
+    {
+      "epoch": 234.4078947368421,
+      "grad_norm": 1.1182633638381958,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 35630
+    },
+    {
+      "epoch": 234.47368421052633,
+      "grad_norm": 0.9117094278335571,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 35640
+    },
+    {
+      "epoch": 234.53947368421052,
+      "grad_norm": 1.345800757408142,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 35650
+    },
+    {
+      "epoch": 234.60526315789474,
+      "grad_norm": 1.145963430404663,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 35660
+    },
+    {
+      "epoch": 234.67105263157896,
+      "grad_norm": 0.9094061851501465,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 35670
+    },
+    {
+      "epoch": 234.73684210526315,
+      "grad_norm": 1.2878849506378174,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 35680
+    },
+    {
+      "epoch": 234.80263157894737,
+      "grad_norm": 1.4541486501693726,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 35690
+    },
+    {
+      "epoch": 234.8684210526316,
+      "grad_norm": 0.971204400062561,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 35700
+    },
+    {
+      "epoch": 234.93421052631578,
+      "grad_norm": 1.1860767602920532,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 35710
+    },
+    {
+      "epoch": 235.0,
+      "grad_norm": 1.2985316514968872,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 35720
+    },
+    {
+      "epoch": 235.06578947368422,
+      "grad_norm": 1.0510177612304688,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 35730
+    },
+    {
+      "epoch": 235.1315789473684,
+      "grad_norm": 0.9717345833778381,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 35740
+    },
+    {
+      "epoch": 235.19736842105263,
+      "grad_norm": 1.131279468536377,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 35750
+    },
+    {
+      "epoch": 235.26315789473685,
+      "grad_norm": 1.148485541343689,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 35760
+    },
+    {
+      "epoch": 235.32894736842104,
+      "grad_norm": 0.8427010774612427,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 35770
+    },
+    {
+      "epoch": 235.39473684210526,
+      "grad_norm": 0.973970890045166,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 35780
+    },
+    {
+      "epoch": 235.46052631578948,
+      "grad_norm": 0.9726606607437134,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 35790
+    },
+    {
+      "epoch": 235.52631578947367,
+      "grad_norm": 0.849587082862854,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 35800
+    },
+    {
+      "epoch": 235.5921052631579,
+      "grad_norm": 1.053238868713379,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 35810
+    },
+    {
+      "epoch": 235.6578947368421,
+      "grad_norm": 1.2742220163345337,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 35820
+    },
+    {
+      "epoch": 235.72368421052633,
+      "grad_norm": 1.3794277906417847,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 35830
+    },
+    {
+      "epoch": 235.78947368421052,
+      "grad_norm": 1.2692276239395142,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 35840
+    },
+    {
+      "epoch": 235.85526315789474,
+      "grad_norm": 1.1622906923294067,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 35850
+    },
+    {
+      "epoch": 235.92105263157896,
+      "grad_norm": 1.1229708194732666,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 35860
+    },
+    {
+      "epoch": 235.98684210526315,
+      "grad_norm": 1.1721322536468506,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 35870
+    },
+    {
+      "epoch": 236.05263157894737,
+      "grad_norm": 1.2149195671081543,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 35880
+    },
+    {
+      "epoch": 236.1184210526316,
+      "grad_norm": 1.552794337272644,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 35890
+    },
+    {
+      "epoch": 236.18421052631578,
+      "grad_norm": 0.9946601986885071,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 35900
+    },
+    {
+      "epoch": 236.25,
+      "grad_norm": 1.1157256364822388,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 35910
+    },
+    {
+      "epoch": 236.31578947368422,
+      "grad_norm": 1.2444145679473877,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 35920
+    },
+    {
+      "epoch": 236.3815789473684,
+      "grad_norm": 1.2110340595245361,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 35930
+    },
+    {
+      "epoch": 236.44736842105263,
+      "grad_norm": 1.4335511922836304,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 35940
+    },
+    {
+      "epoch": 236.51315789473685,
+      "grad_norm": 1.3083791732788086,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 35950
+    },
+    {
+      "epoch": 236.57894736842104,
+      "grad_norm": 0.8935396075248718,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 35960
+    },
+    {
+      "epoch": 236.64473684210526,
+      "grad_norm": 0.9433857798576355,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 35970
+    },
+    {
+      "epoch": 236.71052631578948,
+      "grad_norm": 1.1531612873077393,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 35980
+    },
+    {
+      "epoch": 236.77631578947367,
+      "grad_norm": 1.431472659111023,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 35990
+    },
+    {
+      "epoch": 236.8421052631579,
+      "grad_norm": 1.6425496339797974,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 36000
+    },
+    {
+      "epoch": 236.9078947368421,
+      "grad_norm": 1.0566476583480835,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 36010
+    },
+    {
+      "epoch": 236.97368421052633,
+      "grad_norm": 0.8310641050338745,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 36020
+    },
+    {
+      "epoch": 237.03947368421052,
+      "grad_norm": 1.0086852312088013,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 36030
+    },
+    {
+      "epoch": 237.10526315789474,
+      "grad_norm": 1.44770085811615,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 36040
+    },
+    {
+      "epoch": 237.17105263157896,
+      "grad_norm": 1.002258539199829,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 36050
+    },
+    {
+      "epoch": 237.23684210526315,
+      "grad_norm": 1.0949816703796387,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 36060
+    },
+    {
+      "epoch": 237.30263157894737,
+      "grad_norm": 0.9487596750259399,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 36070
+    },
+    {
+      "epoch": 237.3684210526316,
+      "grad_norm": 0.8451818227767944,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 36080
+    },
+    {
+      "epoch": 237.43421052631578,
+      "grad_norm": 1.455590844154358,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 36090
+    },
+    {
+      "epoch": 237.5,
+      "grad_norm": 1.2920894622802734,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 36100
+    },
+    {
+      "epoch": 237.56578947368422,
+      "grad_norm": 1.3782984018325806,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 36110
+    },
+    {
+      "epoch": 237.6315789473684,
+      "grad_norm": 1.0426079034805298,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 36120
+    },
+    {
+      "epoch": 237.69736842105263,
+      "grad_norm": 1.128311276435852,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 36130
+    },
+    {
+      "epoch": 237.76315789473685,
+      "grad_norm": 1.0771710872650146,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 36140
+    },
+    {
+      "epoch": 237.82894736842104,
+      "grad_norm": 0.8696498274803162,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 36150
+    },
+    {
+      "epoch": 237.89473684210526,
+      "grad_norm": 1.1533684730529785,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 36160
+    },
+    {
+      "epoch": 237.96052631578948,
+      "grad_norm": 1.2097066640853882,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 36170
+    },
+    {
+      "epoch": 238.02631578947367,
+      "grad_norm": 1.1316962242126465,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 36180
+    },
+    {
+      "epoch": 238.0921052631579,
+      "grad_norm": 1.1402578353881836,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 36190
+    },
+    {
+      "epoch": 238.1578947368421,
+      "grad_norm": 1.0534335374832153,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 36200
+    },
+    {
+      "epoch": 238.22368421052633,
+      "grad_norm": 1.2206014394760132,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 36210
+    },
+    {
+      "epoch": 238.28947368421052,
+      "grad_norm": 0.9274782538414001,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 36220
+    },
+    {
+      "epoch": 238.35526315789474,
+      "grad_norm": 1.1802715063095093,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 36230
+    },
+    {
+      "epoch": 238.42105263157896,
+      "grad_norm": 1.3021156787872314,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 36240
+    },
+    {
+      "epoch": 238.48684210526315,
+      "grad_norm": 0.9791660904884338,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 36250
+    },
+    {
+      "epoch": 238.55263157894737,
+      "grad_norm": 1.458195686340332,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 36260
+    },
+    {
+      "epoch": 238.6184210526316,
+      "grad_norm": 1.2013226747512817,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 36270
+    },
+    {
+      "epoch": 238.68421052631578,
+      "grad_norm": 0.897367000579834,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 36280
+    },
+    {
+      "epoch": 238.75,
+      "grad_norm": 1.430230736732483,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 36290
+    },
+    {
+      "epoch": 238.81578947368422,
+      "grad_norm": 1.2003045082092285,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 36300
+    },
+    {
+      "epoch": 238.8815789473684,
+      "grad_norm": 0.8546904921531677,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 36310
+    },
+    {
+      "epoch": 238.94736842105263,
+      "grad_norm": 0.9247255325317383,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 36320
+    },
+    {
+      "epoch": 239.01315789473685,
+      "grad_norm": 1.1530381441116333,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 36330
+    },
+    {
+      "epoch": 239.07894736842104,
+      "grad_norm": 1.2787343263626099,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 36340
+    },
+    {
+      "epoch": 239.14473684210526,
+      "grad_norm": 1.2188760042190552,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 36350
+    },
+    {
+      "epoch": 239.21052631578948,
+      "grad_norm": 1.1100471019744873,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 36360
+    },
+    {
+      "epoch": 239.27631578947367,
+      "grad_norm": 0.8758446574211121,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 36370
+    },
+    {
+      "epoch": 239.3421052631579,
+      "grad_norm": 0.9716992974281311,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 36380
+    },
+    {
+      "epoch": 239.4078947368421,
+      "grad_norm": 1.394768476486206,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 36390
+    },
+    {
+      "epoch": 239.47368421052633,
+      "grad_norm": 1.2768748998641968,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 36400
+    },
+    {
+      "epoch": 239.53947368421052,
+      "grad_norm": 1.1910803318023682,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 36410
+    },
+    {
+      "epoch": 239.60526315789474,
+      "grad_norm": 1.2653837203979492,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 36420
+    },
+    {
+      "epoch": 239.67105263157896,
+      "grad_norm": 1.348326325416565,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 36430
+    },
+    {
+      "epoch": 239.73684210526315,
+      "grad_norm": 1.2382081747055054,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 36440
+    },
+    {
+      "epoch": 239.80263157894737,
+      "grad_norm": 1.3600709438323975,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 36450
+    },
+    {
+      "epoch": 239.8684210526316,
+      "grad_norm": 0.9798856377601624,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 36460
+    },
+    {
+      "epoch": 239.93421052631578,
+      "grad_norm": 0.7852506637573242,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 36470
+    },
+    {
+      "epoch": 240.0,
+      "grad_norm": 1.0045945644378662,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 36480
+    },
+    {
+      "epoch": 240.06578947368422,
+      "grad_norm": 0.8587341904640198,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 36490
+    },
+    {
+      "epoch": 240.1315789473684,
+      "grad_norm": 1.4854038953781128,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 36500
+    },
+    {
+      "epoch": 240.19736842105263,
+      "grad_norm": 1.3243857622146606,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 36510
+    },
+    {
+      "epoch": 240.26315789473685,
+      "grad_norm": 1.048046588897705,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 36520
+    },
+    {
+      "epoch": 240.32894736842104,
+      "grad_norm": 1.2763868570327759,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 36530
+    },
+    {
+      "epoch": 240.39473684210526,
+      "grad_norm": 0.910820722579956,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 36540
+    },
+    {
+      "epoch": 240.46052631578948,
+      "grad_norm": 1.2709863185882568,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 36550
+    },
+    {
+      "epoch": 240.52631578947367,
+      "grad_norm": 1.4138668775558472,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 36560
+    },
+    {
+      "epoch": 240.5921052631579,
+      "grad_norm": 1.7599332332611084,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 36570
+    },
+    {
+      "epoch": 240.6578947368421,
+      "grad_norm": 1.1151326894760132,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 36580
+    },
+    {
+      "epoch": 240.72368421052633,
+      "grad_norm": 0.9767569899559021,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 36590
+    },
+    {
+      "epoch": 240.78947368421052,
+      "grad_norm": 0.9586136341094971,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 36600
+    },
+    {
+      "epoch": 240.85526315789474,
+      "grad_norm": 0.9484444856643677,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 36610
+    },
+    {
+      "epoch": 240.92105263157896,
+      "grad_norm": 1.5435513257980347,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 36620
+    },
+    {
+      "epoch": 240.98684210526315,
+      "grad_norm": 1.2928390502929688,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 36630
+    },
+    {
+      "epoch": 241.05263157894737,
+      "grad_norm": 1.3825304508209229,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 36640
+    },
+    {
+      "epoch": 241.1184210526316,
+      "grad_norm": 1.3325434923171997,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 36650
+    },
+    {
+      "epoch": 241.18421052631578,
+      "grad_norm": 1.4383955001831055,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 36660
+    },
+    {
+      "epoch": 241.25,
+      "grad_norm": 1.0420336723327637,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 36670
+    },
+    {
+      "epoch": 241.31578947368422,
+      "grad_norm": 1.0319572687149048,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 36680
+    },
+    {
+      "epoch": 241.3815789473684,
+      "grad_norm": 1.3533285856246948,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 36690
+    },
+    {
+      "epoch": 241.44736842105263,
+      "grad_norm": 1.112142562866211,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 36700
+    },
+    {
+      "epoch": 241.51315789473685,
+      "grad_norm": 1.3508294820785522,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 36710
+    },
+    {
+      "epoch": 241.57894736842104,
+      "grad_norm": 0.8377172946929932,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 36720
+    },
+    {
+      "epoch": 241.64473684210526,
+      "grad_norm": 1.2829350233078003,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 36730
+    },
+    {
+      "epoch": 241.71052631578948,
+      "grad_norm": 1.1019694805145264,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 36740
+    },
+    {
+      "epoch": 241.77631578947367,
+      "grad_norm": 1.6284654140472412,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 36750
+    },
+    {
+      "epoch": 241.8421052631579,
+      "grad_norm": 1.409332036972046,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 36760
+    },
+    {
+      "epoch": 241.9078947368421,
+      "grad_norm": 1.5580686330795288,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 36770
+    },
+    {
+      "epoch": 241.97368421052633,
+      "grad_norm": 1.2454417943954468,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 36780
+    },
+    {
+      "epoch": 242.03947368421052,
+      "grad_norm": 1.194043755531311,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 36790
+    },
+    {
+      "epoch": 242.10526315789474,
+      "grad_norm": 1.0807819366455078,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 36800
+    },
+    {
+      "epoch": 242.17105263157896,
+      "grad_norm": 1.1862661838531494,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 36810
+    },
+    {
+      "epoch": 242.23684210526315,
+      "grad_norm": 1.3441433906555176,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 36820
+    },
+    {
+      "epoch": 242.30263157894737,
+      "grad_norm": 1.0832732915878296,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 36830
+    },
+    {
+      "epoch": 242.3684210526316,
+      "grad_norm": 0.8166968822479248,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 36840
+    },
+    {
+      "epoch": 242.43421052631578,
+      "grad_norm": 1.1842600107192993,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 36850
+    },
+    {
+      "epoch": 242.5,
+      "grad_norm": 1.0731406211853027,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 36860
+    },
+    {
+      "epoch": 242.56578947368422,
+      "grad_norm": 1.1145586967468262,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 36870
+    },
+    {
+      "epoch": 242.6315789473684,
+      "grad_norm": 1.2087843418121338,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 36880
+    },
+    {
+      "epoch": 242.69736842105263,
+      "grad_norm": 0.9884072542190552,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 36890
+    },
+    {
+      "epoch": 242.76315789473685,
+      "grad_norm": 1.011466383934021,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 36900
+    },
+    {
+      "epoch": 242.82894736842104,
+      "grad_norm": 0.9709348678588867,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 36910
+    },
+    {
+      "epoch": 242.89473684210526,
+      "grad_norm": 1.2150779962539673,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 36920
+    },
+    {
+      "epoch": 242.96052631578948,
+      "grad_norm": 1.3461240530014038,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 36930
+    },
+    {
+      "epoch": 243.02631578947367,
+      "grad_norm": 1.4247745275497437,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 36940
+    },
+    {
+      "epoch": 243.0921052631579,
+      "grad_norm": 1.213616132736206,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 36950
+    },
+    {
+      "epoch": 243.1578947368421,
+      "grad_norm": 1.3313192129135132,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 36960
+    },
+    {
+      "epoch": 243.22368421052633,
+      "grad_norm": 1.5116862058639526,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 36970
+    },
+    {
+      "epoch": 243.28947368421052,
+      "grad_norm": 1.200487494468689,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 36980
+    },
+    {
+      "epoch": 243.35526315789474,
+      "grad_norm": 1.5582853555679321,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 36990
+    },
+    {
+      "epoch": 243.42105263157896,
+      "grad_norm": 1.1975919008255005,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 37000
+    },
+    {
+      "epoch": 243.48684210526315,
+      "grad_norm": 1.3097623586654663,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 37010
+    },
+    {
+      "epoch": 243.55263157894737,
+      "grad_norm": 1.28339421749115,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 37020
+    },
+    {
+      "epoch": 243.6184210526316,
+      "grad_norm": 1.0238935947418213,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 37030
+    },
+    {
+      "epoch": 243.68421052631578,
+      "grad_norm": 1.203718900680542,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 37040
+    },
+    {
+      "epoch": 243.75,
+      "grad_norm": 1.2539606094360352,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 37050
+    },
+    {
+      "epoch": 243.81578947368422,
+      "grad_norm": 1.581852674484253,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 37060
+    },
+    {
+      "epoch": 243.8815789473684,
+      "grad_norm": 0.8200912475585938,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 37070
+    },
+    {
+      "epoch": 243.94736842105263,
+      "grad_norm": 1.1216977834701538,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 37080
+    },
+    {
+      "epoch": 244.01315789473685,
+      "grad_norm": 1.295255422592163,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 37090
+    },
+    {
+      "epoch": 244.07894736842104,
+      "grad_norm": 0.9966356158256531,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 37100
+    },
+    {
+      "epoch": 244.14473684210526,
+      "grad_norm": 1.278610110282898,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 37110
+    },
+    {
+      "epoch": 244.21052631578948,
+      "grad_norm": 1.1381927728652954,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 37120
+    },
+    {
+      "epoch": 244.27631578947367,
+      "grad_norm": 1.3754022121429443,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 37130
+    },
+    {
+      "epoch": 244.3421052631579,
+      "grad_norm": 1.0089901685714722,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 37140
+    },
+    {
+      "epoch": 244.4078947368421,
+      "grad_norm": 1.4417517185211182,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 37150
+    },
+    {
+      "epoch": 244.47368421052633,
+      "grad_norm": 1.4486968517303467,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 37160
+    },
+    {
+      "epoch": 244.53947368421052,
+      "grad_norm": 1.428544044494629,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 37170
+    },
+    {
+      "epoch": 244.60526315789474,
+      "grad_norm": 1.239670753479004,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 37180
+    },
+    {
+      "epoch": 244.67105263157896,
+      "grad_norm": 0.8541073203086853,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 37190
+    },
+    {
+      "epoch": 244.73684210526315,
+      "grad_norm": 1.2242367267608643,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 37200
+    },
+    {
+      "epoch": 244.80263157894737,
+      "grad_norm": 1.1044442653656006,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 37210
+    },
+    {
+      "epoch": 244.8684210526316,
+      "grad_norm": 1.4338637590408325,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 37220
+    },
+    {
+      "epoch": 244.93421052631578,
+      "grad_norm": 1.1796600818634033,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 37230
+    },
+    {
+      "epoch": 245.0,
+      "grad_norm": 1.4116781949996948,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 37240
+    },
+    {
+      "epoch": 245.06578947368422,
+      "grad_norm": 1.2402573823928833,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 37250
+    },
+    {
+      "epoch": 245.1315789473684,
+      "grad_norm": 1.3560236692428589,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 37260
+    },
+    {
+      "epoch": 245.19736842105263,
+      "grad_norm": 1.0826292037963867,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 37270
+    },
+    {
+      "epoch": 245.26315789473685,
+      "grad_norm": 1.4319058656692505,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 37280
+    },
+    {
+      "epoch": 245.32894736842104,
+      "grad_norm": 1.4913933277130127,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 37290
+    },
+    {
+      "epoch": 245.39473684210526,
+      "grad_norm": 1.6130610704421997,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 37300
+    },
+    {
+      "epoch": 245.46052631578948,
+      "grad_norm": 1.419812560081482,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 37310
+    },
+    {
+      "epoch": 245.52631578947367,
+      "grad_norm": 1.1114962100982666,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 37320
+    },
+    {
+      "epoch": 245.5921052631579,
+      "grad_norm": 0.8215615749359131,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 37330
+    },
+    {
+      "epoch": 245.6578947368421,
+      "grad_norm": 0.9287834763526917,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 37340
+    },
+    {
+      "epoch": 245.72368421052633,
+      "grad_norm": 1.5063482522964478,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 37350
+    },
+    {
+      "epoch": 245.78947368421052,
+      "grad_norm": 1.3184797763824463,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 37360
+    },
+    {
+      "epoch": 245.85526315789474,
+      "grad_norm": 1.147472858428955,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 37370
+    },
+    {
+      "epoch": 245.92105263157896,
+      "grad_norm": 1.3581057786941528,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 37380
+    },
+    {
+      "epoch": 245.98684210526315,
+      "grad_norm": 0.9130412340164185,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 37390
+    },
+    {
+      "epoch": 246.05263157894737,
+      "grad_norm": 1.0670521259307861,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 37400
+    },
+    {
+      "epoch": 246.1184210526316,
+      "grad_norm": 1.1035175323486328,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 37410
+    },
+    {
+      "epoch": 246.18421052631578,
+      "grad_norm": 1.24159574508667,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 37420
+    },
+    {
+      "epoch": 246.25,
+      "grad_norm": 1.571832537651062,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 37430
+    },
+    {
+      "epoch": 246.31578947368422,
+      "grad_norm": 1.1885900497436523,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 37440
+    },
+    {
+      "epoch": 246.3815789473684,
+      "grad_norm": 1.5651880502700806,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 37450
+    },
+    {
+      "epoch": 246.44736842105263,
+      "grad_norm": 1.3197070360183716,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 37460
+    },
+    {
+      "epoch": 246.51315789473685,
+      "grad_norm": 1.3817650079727173,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 37470
+    },
+    {
+      "epoch": 246.57894736842104,
+      "grad_norm": 1.5466119050979614,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 37480
+    },
+    {
+      "epoch": 246.64473684210526,
+      "grad_norm": 1.2215481996536255,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 37490
+    },
+    {
+      "epoch": 246.71052631578948,
+      "grad_norm": 1.4967329502105713,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 37500
+    },
+    {
+      "epoch": 246.77631578947367,
+      "grad_norm": 1.4192287921905518,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 37510
+    },
+    {
+      "epoch": 246.8421052631579,
+      "grad_norm": 1.3284549713134766,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 37520
+    },
+    {
+      "epoch": 246.9078947368421,
+      "grad_norm": 1.3947001695632935,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 37530
+    },
+    {
+      "epoch": 246.97368421052633,
+      "grad_norm": 1.2742048501968384,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 37540
+    },
+    {
+      "epoch": 247.03947368421052,
+      "grad_norm": 1.200553059577942,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 37550
+    },
+    {
+      "epoch": 247.10526315789474,
+      "grad_norm": 1.2947498559951782,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 37560
+    },
+    {
+      "epoch": 247.17105263157896,
+      "grad_norm": 1.950449824333191,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 37570
+    },
+    {
+      "epoch": 247.23684210526315,
+      "grad_norm": 1.7037091255187988,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 37580
+    },
+    {
+      "epoch": 247.30263157894737,
+      "grad_norm": 1.2691845893859863,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 37590
+    },
+    {
+      "epoch": 247.3684210526316,
+      "grad_norm": 0.9007494449615479,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 37600
+    },
+    {
+      "epoch": 247.43421052631578,
+      "grad_norm": 1.374837040901184,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 37610
+    },
+    {
+      "epoch": 247.5,
+      "grad_norm": 1.1083370447158813,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 37620
+    },
+    {
+      "epoch": 247.56578947368422,
+      "grad_norm": 1.2770500183105469,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 37630
+    },
+    {
+      "epoch": 247.6315789473684,
+      "grad_norm": 1.4671261310577393,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 37640
+    },
+    {
+      "epoch": 247.69736842105263,
+      "grad_norm": 1.6228381395339966,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 37650
+    },
+    {
+      "epoch": 247.76315789473685,
+      "grad_norm": 1.6250126361846924,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 37660
+    },
+    {
+      "epoch": 247.82894736842104,
+      "grad_norm": 1.4353737831115723,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 37670
+    },
+    {
+      "epoch": 247.89473684210526,
+      "grad_norm": 1.6463098526000977,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 37680
+    },
+    {
+      "epoch": 247.96052631578948,
+      "grad_norm": 1.2910631895065308,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 37690
+    },
+    {
+      "epoch": 248.02631578947367,
+      "grad_norm": 1.0633682012557983,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 37700
+    },
+    {
+      "epoch": 248.0921052631579,
+      "grad_norm": 1.6399755477905273,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 37710
+    },
+    {
+      "epoch": 248.1578947368421,
+      "grad_norm": 1.4523167610168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 37720
+    },
+    {
+      "epoch": 248.22368421052633,
+      "grad_norm": 1.307390570640564,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 37730
+    },
+    {
+      "epoch": 248.28947368421052,
+      "grad_norm": 1.414101243019104,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 37740
+    },
+    {
+      "epoch": 248.35526315789474,
+      "grad_norm": 1.1803747415542603,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 37750
+    },
+    {
+      "epoch": 248.42105263157896,
+      "grad_norm": 1.416965365409851,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 37760
+    },
+    {
+      "epoch": 248.48684210526315,
+      "grad_norm": 1.0817677974700928,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 37770
+    },
+    {
+      "epoch": 248.55263157894737,
+      "grad_norm": 1.3911010026931763,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 37780
+    },
+    {
+      "epoch": 248.6184210526316,
+      "grad_norm": 1.0438708066940308,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 37790
+    },
+    {
+      "epoch": 248.68421052631578,
+      "grad_norm": 1.1672842502593994,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 37800
+    },
+    {
+      "epoch": 248.75,
+      "grad_norm": 1.5514910221099854,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 37810
+    },
+    {
+      "epoch": 248.81578947368422,
+      "grad_norm": 1.686729073524475,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 37820
+    },
+    {
+      "epoch": 248.8815789473684,
+      "grad_norm": 1.720362901687622,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 37830
+    },
+    {
+      "epoch": 248.94736842105263,
+      "grad_norm": 1.4286376237869263,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 37840
+    },
+    {
+      "epoch": 249.01315789473685,
+      "grad_norm": 1.447860836982727,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 37850
+    },
+    {
+      "epoch": 249.07894736842104,
+      "grad_norm": 1.2636222839355469,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 37860
+    },
+    {
+      "epoch": 249.14473684210526,
+      "grad_norm": 1.2671912908554077,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 37870
+    },
+    {
+      "epoch": 249.21052631578948,
+      "grad_norm": 0.9959325194358826,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 37880
+    },
+    {
+      "epoch": 249.27631578947367,
+      "grad_norm": 1.2429472208023071,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 37890
+    },
+    {
+      "epoch": 249.3421052631579,
+      "grad_norm": 1.1575770378112793,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 37900
+    },
+    {
+      "epoch": 249.4078947368421,
+      "grad_norm": 1.075370192527771,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 37910
+    },
+    {
+      "epoch": 249.47368421052633,
+      "grad_norm": 1.1879488229751587,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 37920
+    },
+    {
+      "epoch": 249.53947368421052,
+      "grad_norm": 1.0791507959365845,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 37930
+    },
+    {
+      "epoch": 249.60526315789474,
+      "grad_norm": 0.9675527811050415,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 37940
+    },
+    {
+      "epoch": 249.67105263157896,
+      "grad_norm": 1.2315634489059448,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 37950
+    },
+    {
+      "epoch": 249.73684210526315,
+      "grad_norm": 0.8122895956039429,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 37960
+    },
+    {
+      "epoch": 249.80263157894737,
+      "grad_norm": 1.209394931793213,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 37970
+    },
+    {
+      "epoch": 249.8684210526316,
+      "grad_norm": 1.001671314239502,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 37980
+    },
+    {
+      "epoch": 249.93421052631578,
+      "grad_norm": 1.5136141777038574,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 37990
+    },
+    {
+      "epoch": 250.0,
+      "grad_norm": 1.4685696363449097,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 38000
+    },
+    {
+      "epoch": 250.06578947368422,
+      "grad_norm": 1.1837599277496338,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 38010
+    },
+    {
+      "epoch": 250.1315789473684,
+      "grad_norm": 1.018049955368042,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 38020
+    },
+    {
+      "epoch": 250.19736842105263,
+      "grad_norm": 1.0847623348236084,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 38030
+    },
+    {
+      "epoch": 250.26315789473685,
+      "grad_norm": 1.4358677864074707,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 38040
+    },
+    {
+      "epoch": 250.32894736842104,
+      "grad_norm": 1.3127168416976929,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 38050
+    },
+    {
+      "epoch": 250.39473684210526,
+      "grad_norm": 1.6243581771850586,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 38060
+    },
+    {
+      "epoch": 250.46052631578948,
+      "grad_norm": 1.3248577117919922,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 38070
+    },
+    {
+      "epoch": 250.52631578947367,
+      "grad_norm": 1.2250198125839233,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 38080
+    },
+    {
+      "epoch": 250.5921052631579,
+      "grad_norm": 1.238546371459961,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 38090
+    },
+    {
+      "epoch": 250.6578947368421,
+      "grad_norm": 1.3515560626983643,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 38100
+    },
+    {
+      "epoch": 250.72368421052633,
+      "grad_norm": 1.272128701210022,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 38110
+    },
+    {
+      "epoch": 250.78947368421052,
+      "grad_norm": 1.1292731761932373,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 38120
+    },
+    {
+      "epoch": 250.85526315789474,
+      "grad_norm": 1.018415927886963,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 38130
+    },
+    {
+      "epoch": 250.92105263157896,
+      "grad_norm": 1.367689609527588,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 38140
+    },
+    {
+      "epoch": 250.98684210526315,
+      "grad_norm": 1.3254410028457642,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 38150
+    },
+    {
+      "epoch": 251.05263157894737,
+      "grad_norm": 1.6456955671310425,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 38160
+    },
+    {
+      "epoch": 251.1184210526316,
+      "grad_norm": 1.1170157194137573,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 38170
+    },
+    {
+      "epoch": 251.18421052631578,
+      "grad_norm": 0.998716413974762,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 38180
+    },
+    {
+      "epoch": 251.25,
+      "grad_norm": 0.9729195833206177,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 38190
+    },
+    {
+      "epoch": 251.31578947368422,
+      "grad_norm": 1.274336576461792,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 38200
+    },
+    {
+      "epoch": 251.3815789473684,
+      "grad_norm": 1.0042622089385986,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 38210
+    },
+    {
+      "epoch": 251.44736842105263,
+      "grad_norm": 1.3611550331115723,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 38220
+    },
+    {
+      "epoch": 251.51315789473685,
+      "grad_norm": 1.4068595170974731,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 38230
+    },
+    {
+      "epoch": 251.57894736842104,
+      "grad_norm": 1.0429598093032837,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 38240
+    },
+    {
+      "epoch": 251.64473684210526,
+      "grad_norm": 1.2321966886520386,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 38250
+    },
+    {
+      "epoch": 251.71052631578948,
+      "grad_norm": 1.5043584108352661,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 38260
+    },
+    {
+      "epoch": 251.77631578947367,
+      "grad_norm": 1.3396488428115845,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 38270
+    },
+    {
+      "epoch": 251.8421052631579,
+      "grad_norm": 1.1610597372055054,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 38280
+    },
+    {
+      "epoch": 251.9078947368421,
+      "grad_norm": 1.2713470458984375,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 38290
+    },
+    {
+      "epoch": 251.97368421052633,
+      "grad_norm": 1.6690809726715088,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 38300
+    },
+    {
+      "epoch": 252.03947368421052,
+      "grad_norm": 1.3469371795654297,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 38310
+    },
+    {
+      "epoch": 252.10526315789474,
+      "grad_norm": 1.1202424764633179,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 38320
+    },
+    {
+      "epoch": 252.17105263157896,
+      "grad_norm": 1.3286770582199097,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 38330
+    },
+    {
+      "epoch": 252.23684210526315,
+      "grad_norm": 1.2388663291931152,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 38340
+    },
+    {
+      "epoch": 252.30263157894737,
+      "grad_norm": 1.1364609003067017,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 38350
+    },
+    {
+      "epoch": 252.3684210526316,
+      "grad_norm": 1.43731689453125,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 38360
+    },
+    {
+      "epoch": 252.43421052631578,
+      "grad_norm": 1.337084412574768,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 38370
+    },
+    {
+      "epoch": 252.5,
+      "grad_norm": 1.568783164024353,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 38380
+    },
+    {
+      "epoch": 252.56578947368422,
+      "grad_norm": 1.5150007009506226,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 38390
+    },
+    {
+      "epoch": 252.6315789473684,
+      "grad_norm": 1.7503875494003296,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 38400
+    },
+    {
+      "epoch": 252.69736842105263,
+      "grad_norm": 1.123337745666504,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 38410
+    },
+    {
+      "epoch": 252.76315789473685,
+      "grad_norm": 1.359995722770691,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 38420
+    },
+    {
+      "epoch": 252.82894736842104,
+      "grad_norm": 1.1792511940002441,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 38430
+    },
+    {
+      "epoch": 252.89473684210526,
+      "grad_norm": 1.2069591283798218,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 38440
+    },
+    {
+      "epoch": 252.96052631578948,
+      "grad_norm": 1.2496885061264038,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 38450
+    },
+    {
+      "epoch": 253.02631578947367,
+      "grad_norm": 1.0140652656555176,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 38460
+    },
+    {
+      "epoch": 253.0921052631579,
+      "grad_norm": 1.251288652420044,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 38470
+    },
+    {
+      "epoch": 253.1578947368421,
+      "grad_norm": 1.283728837966919,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 38480
+    },
+    {
+      "epoch": 253.22368421052633,
+      "grad_norm": 1.3904523849487305,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 38490
+    },
+    {
+      "epoch": 253.28947368421052,
+      "grad_norm": 1.363517165184021,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 38500
+    },
+    {
+      "epoch": 253.35526315789474,
+      "grad_norm": 1.120867133140564,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 38510
+    },
+    {
+      "epoch": 253.42105263157896,
+      "grad_norm": 1.360049843788147,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 38520
+    },
+    {
+      "epoch": 253.48684210526315,
+      "grad_norm": 1.39319908618927,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 38530
+    },
+    {
+      "epoch": 253.55263157894737,
+      "grad_norm": 1.0926111936569214,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 38540
+    },
+    {
+      "epoch": 253.6184210526316,
+      "grad_norm": 1.4008203744888306,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 38550
+    },
+    {
+      "epoch": 253.68421052631578,
+      "grad_norm": 0.974804699420929,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 38560
+    },
+    {
+      "epoch": 253.75,
+      "grad_norm": 1.2252269983291626,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 38570
+    },
+    {
+      "epoch": 253.81578947368422,
+      "grad_norm": 1.2093294858932495,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 38580
+    },
+    {
+      "epoch": 253.8815789473684,
+      "grad_norm": 1.3294168710708618,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 38590
+    },
+    {
+      "epoch": 253.94736842105263,
+      "grad_norm": 1.3876254558563232,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 38600
+    },
+    {
+      "epoch": 254.01315789473685,
+      "grad_norm": 1.2935510873794556,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 38610
+    },
+    {
+      "epoch": 254.07894736842104,
+      "grad_norm": 1.1402019262313843,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 38620
+    },
+    {
+      "epoch": 254.14473684210526,
+      "grad_norm": 1.167262315750122,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 38630
+    },
+    {
+      "epoch": 254.21052631578948,
+      "grad_norm": 1.2734203338623047,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 38640
+    },
+    {
+      "epoch": 254.27631578947367,
+      "grad_norm": 1.4504673480987549,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 38650
+    },
+    {
+      "epoch": 254.3421052631579,
+      "grad_norm": 1.499895691871643,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 38660
+    },
+    {
+      "epoch": 254.4078947368421,
+      "grad_norm": 0.8160152435302734,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 38670
+    },
+    {
+      "epoch": 254.47368421052633,
+      "grad_norm": 1.31825852394104,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 38680
+    },
+    {
+      "epoch": 254.53947368421052,
+      "grad_norm": 1.0252249240875244,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 38690
+    },
+    {
+      "epoch": 254.60526315789474,
+      "grad_norm": 1.2557110786437988,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 38700
+    },
+    {
+      "epoch": 254.67105263157896,
+      "grad_norm": 1.0297420024871826,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 38710
+    },
+    {
+      "epoch": 254.73684210526315,
+      "grad_norm": 1.1971931457519531,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 38720
+    },
+    {
+      "epoch": 254.80263157894737,
+      "grad_norm": 0.9345195889472961,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 38730
+    },
+    {
+      "epoch": 254.8684210526316,
+      "grad_norm": 1.0931565761566162,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 38740
+    },
+    {
+      "epoch": 254.93421052631578,
+      "grad_norm": 1.1553313732147217,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 38750
+    },
+    {
+      "epoch": 255.0,
+      "grad_norm": 0.8764203786849976,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 38760
+    },
+    {
+      "epoch": 255.06578947368422,
+      "grad_norm": 1.2318146228790283,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 38770
+    },
+    {
+      "epoch": 255.1315789473684,
+      "grad_norm": 1.1166127920150757,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 38780
+    },
+    {
+      "epoch": 255.19736842105263,
+      "grad_norm": 1.0156936645507812,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 38790
+    },
+    {
+      "epoch": 255.26315789473685,
+      "grad_norm": 1.1941726207733154,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 38800
+    },
+    {
+      "epoch": 255.32894736842104,
+      "grad_norm": 1.557681679725647,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 38810
+    },
+    {
+      "epoch": 255.39473684210526,
+      "grad_norm": 1.3124456405639648,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 38820
+    },
+    {
+      "epoch": 255.46052631578948,
+      "grad_norm": 1.4480106830596924,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 38830
+    },
+    {
+      "epoch": 255.52631578947367,
+      "grad_norm": 1.780182957649231,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 38840
+    },
+    {
+      "epoch": 255.5921052631579,
+      "grad_norm": 1.6105772256851196,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 38850
+    },
+    {
+      "epoch": 255.6578947368421,
+      "grad_norm": 1.3613028526306152,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 38860
+    },
+    {
+      "epoch": 255.72368421052633,
+      "grad_norm": 1.5420035123825073,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 38870
+    },
+    {
+      "epoch": 255.78947368421052,
+      "grad_norm": 2.7019877433776855,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 38880
+    },
+    {
+      "epoch": 255.85526315789474,
+      "grad_norm": 2.009016752243042,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 38890
+    },
+    {
+      "epoch": 255.92105263157896,
+      "grad_norm": 1.6684277057647705,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 38900
+    },
+    {
+      "epoch": 255.98684210526315,
+      "grad_norm": 1.6645160913467407,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 38910
+    },
+    {
+      "epoch": 256.05263157894734,
+      "grad_norm": 1.6908752918243408,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 38920
+    },
+    {
+      "epoch": 256.11842105263156,
+      "grad_norm": 1.4804624319076538,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 38930
+    },
+    {
+      "epoch": 256.1842105263158,
+      "grad_norm": 1.5433467626571655,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 38940
+    },
+    {
+      "epoch": 256.25,
+      "grad_norm": 1.4467060565948486,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 38950
+    },
+    {
+      "epoch": 256.3157894736842,
+      "grad_norm": 1.5117400884628296,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 38960
+    },
+    {
+      "epoch": 256.38157894736844,
+      "grad_norm": 1.5240405797958374,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 38970
+    },
+    {
+      "epoch": 256.44736842105266,
+      "grad_norm": 1.3311848640441895,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 38980
+    },
+    {
+      "epoch": 256.5131578947368,
+      "grad_norm": 1.3353846073150635,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 38990
+    },
+    {
+      "epoch": 256.57894736842104,
+      "grad_norm": 2.797309398651123,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 39000
+    },
+    {
+      "epoch": 256.64473684210526,
+      "grad_norm": 2.06365966796875,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 39010
+    },
+    {
+      "epoch": 256.7105263157895,
+      "grad_norm": 2.192547082901001,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 39020
+    },
+    {
+      "epoch": 256.7763157894737,
+      "grad_norm": 1.957229733467102,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 39030
+    },
+    {
+      "epoch": 256.8421052631579,
+      "grad_norm": 1.7908846139907837,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 39040
+    },
+    {
+      "epoch": 256.9078947368421,
+      "grad_norm": 1.4980082511901855,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 39050
+    },
+    {
+      "epoch": 256.9736842105263,
+      "grad_norm": 1.5401355028152466,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 39060
+    },
+    {
+      "epoch": 257.0394736842105,
+      "grad_norm": 1.2455133199691772,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 39070
+    },
+    {
+      "epoch": 257.10526315789474,
+      "grad_norm": 1.4379152059555054,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 39080
+    },
+    {
+      "epoch": 257.17105263157896,
+      "grad_norm": 1.4791769981384277,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 39090
+    },
+    {
+      "epoch": 257.2368421052632,
+      "grad_norm": 1.1306533813476562,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 39100
+    },
+    {
+      "epoch": 257.30263157894734,
+      "grad_norm": 1.0871785879135132,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 39110
+    },
+    {
+      "epoch": 257.36842105263156,
+      "grad_norm": 1.423475980758667,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 39120
+    },
+    {
+      "epoch": 257.4342105263158,
+      "grad_norm": 1.7465128898620605,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 39130
+    },
+    {
+      "epoch": 257.5,
+      "grad_norm": 1.4371193647384644,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 39140
+    },
+    {
+      "epoch": 257.5657894736842,
+      "grad_norm": 1.4257088899612427,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 39150
+    },
+    {
+      "epoch": 257.63157894736844,
+      "grad_norm": 1.1062278747558594,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 39160
+    },
+    {
+      "epoch": 257.69736842105266,
+      "grad_norm": 0.7491597533226013,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 39170
+    },
+    {
+      "epoch": 257.7631578947368,
+      "grad_norm": 1.2030974626541138,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 39180
+    },
+    {
+      "epoch": 257.82894736842104,
+      "grad_norm": 0.9825753569602966,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 39190
+    },
+    {
+      "epoch": 257.89473684210526,
+      "grad_norm": 1.3896033763885498,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 39200
+    },
+    {
+      "epoch": 257.9605263157895,
+      "grad_norm": 1.0874238014221191,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 39210
+    },
+    {
+      "epoch": 258.0263157894737,
+      "grad_norm": 1.1241742372512817,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 39220
+    },
+    {
+      "epoch": 258.0921052631579,
+      "grad_norm": 1.2342971563339233,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 39230
+    },
+    {
+      "epoch": 258.1578947368421,
+      "grad_norm": 1.249884843826294,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 39240
+    },
+    {
+      "epoch": 258.2236842105263,
+      "grad_norm": 1.0342882871627808,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 39250
+    },
+    {
+      "epoch": 258.2894736842105,
+      "grad_norm": 1.2886340618133545,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 39260
+    },
+    {
+      "epoch": 258.35526315789474,
+      "grad_norm": 1.3061116933822632,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 39270
+    },
+    {
+      "epoch": 258.42105263157896,
+      "grad_norm": 1.4507635831832886,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 39280
+    },
+    {
+      "epoch": 258.4868421052632,
+      "grad_norm": 1.4641185998916626,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 39290
+    },
+    {
+      "epoch": 258.55263157894734,
+      "grad_norm": 1.3404735326766968,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 39300
+    },
+    {
+      "epoch": 258.61842105263156,
+      "grad_norm": 0.9776154160499573,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 39310
+    },
+    {
+      "epoch": 258.6842105263158,
+      "grad_norm": 1.0288668870925903,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 39320
+    },
+    {
+      "epoch": 258.75,
+      "grad_norm": 1.1790403127670288,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 39330
+    },
+    {
+      "epoch": 258.8157894736842,
+      "grad_norm": 0.8931066989898682,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 39340
+    },
+    {
+      "epoch": 258.88157894736844,
+      "grad_norm": 1.307145118713379,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 39350
+    },
+    {
+      "epoch": 258.94736842105266,
+      "grad_norm": 1.0468474626541138,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 39360
+    },
+    {
+      "epoch": 259.0131578947368,
+      "grad_norm": 0.9798098206520081,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 39370
+    },
+    {
+      "epoch": 259.07894736842104,
+      "grad_norm": 0.8931979537010193,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 39380
+    },
+    {
+      "epoch": 259.14473684210526,
+      "grad_norm": 1.3130055665969849,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 39390
+    },
+    {
+      "epoch": 259.2105263157895,
+      "grad_norm": 1.2557275295257568,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 39400
+    },
+    {
+      "epoch": 259.2763157894737,
+      "grad_norm": 1.2544786930084229,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 39410
+    },
+    {
+      "epoch": 259.3421052631579,
+      "grad_norm": 1.079939365386963,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 39420
+    },
+    {
+      "epoch": 259.4078947368421,
+      "grad_norm": 1.3079148530960083,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 39430
+    },
+    {
+      "epoch": 259.4736842105263,
+      "grad_norm": 1.2881779670715332,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 39440
+    },
+    {
+      "epoch": 259.5394736842105,
+      "grad_norm": 1.4759607315063477,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 39450
+    },
+    {
+      "epoch": 259.60526315789474,
+      "grad_norm": 1.1630066633224487,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 39460
+    },
+    {
+      "epoch": 259.67105263157896,
+      "grad_norm": 1.894482970237732,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 39470
+    },
+    {
+      "epoch": 259.7368421052632,
+      "grad_norm": 1.7164617776870728,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 39480
+    },
+    {
+      "epoch": 259.80263157894734,
+      "grad_norm": 1.365029215812683,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 39490
+    },
+    {
+      "epoch": 259.86842105263156,
+      "grad_norm": 1.2268500328063965,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 39500
+    },
+    {
+      "epoch": 259.9342105263158,
+      "grad_norm": 1.182262897491455,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 39510
+    },
+    {
+      "epoch": 260.0,
+      "grad_norm": 1.3532487154006958,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 39520
+    },
+    {
+      "epoch": 260.0657894736842,
+      "grad_norm": 1.1759036779403687,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 39530
+    },
+    {
+      "epoch": 260.13157894736844,
+      "grad_norm": 1.0273208618164062,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 39540
+    },
+    {
+      "epoch": 260.19736842105266,
+      "grad_norm": 1.1280847787857056,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 39550
+    },
+    {
+      "epoch": 260.2631578947368,
+      "grad_norm": 1.3776977062225342,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 39560
+    },
+    {
+      "epoch": 260.32894736842104,
+      "grad_norm": 1.1340792179107666,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 39570
+    },
+    {
+      "epoch": 260.39473684210526,
+      "grad_norm": 1.010206699371338,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 39580
+    },
+    {
+      "epoch": 260.4605263157895,
+      "grad_norm": 1.4387060403823853,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 39590
+    },
+    {
+      "epoch": 260.5263157894737,
+      "grad_norm": 1.1363633871078491,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 39600
+    },
+    {
+      "epoch": 260.5921052631579,
+      "grad_norm": 1.204533338546753,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 39610
+    },
+    {
+      "epoch": 260.6578947368421,
+      "grad_norm": 1.3235585689544678,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 39620
+    },
+    {
+      "epoch": 260.7236842105263,
+      "grad_norm": 1.2263247966766357,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 39630
+    },
+    {
+      "epoch": 260.7894736842105,
+      "grad_norm": 1.1299937963485718,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 39640
+    },
+    {
+      "epoch": 260.85526315789474,
+      "grad_norm": 1.2160582542419434,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 39650
+    },
+    {
+      "epoch": 260.92105263157896,
+      "grad_norm": 1.1547380685806274,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 39660
+    },
+    {
+      "epoch": 260.9868421052632,
+      "grad_norm": 1.1193982362747192,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 39670
+    },
+    {
+      "epoch": 261.05263157894734,
+      "grad_norm": 1.039002776145935,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 39680
+    },
+    {
+      "epoch": 261.11842105263156,
+      "grad_norm": 1.2692370414733887,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 39690
+    },
+    {
+      "epoch": 261.1842105263158,
+      "grad_norm": 1.4166964292526245,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 39700
+    },
+    {
+      "epoch": 261.25,
+      "grad_norm": 1.2513854503631592,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 39710
+    },
+    {
+      "epoch": 261.3157894736842,
+      "grad_norm": 1.340706467628479,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 39720
+    },
+    {
+      "epoch": 261.38157894736844,
+      "grad_norm": 1.2143534421920776,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 39730
+    },
+    {
+      "epoch": 261.44736842105266,
+      "grad_norm": 1.6067289113998413,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 39740
+    },
+    {
+      "epoch": 261.5131578947368,
+      "grad_norm": 1.3668274879455566,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 39750
+    },
+    {
+      "epoch": 261.57894736842104,
+      "grad_norm": 0.8968744874000549,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 39760
+    },
+    {
+      "epoch": 261.64473684210526,
+      "grad_norm": 1.5481125116348267,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 39770
+    },
+    {
+      "epoch": 261.7105263157895,
+      "grad_norm": 1.5846647024154663,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 39780
+    },
+    {
+      "epoch": 261.7763157894737,
+      "grad_norm": 0.9274570345878601,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 39790
+    },
+    {
+      "epoch": 261.8421052631579,
+      "grad_norm": 1.1757594347000122,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 39800
+    },
+    {
+      "epoch": 261.9078947368421,
+      "grad_norm": 1.5246598720550537,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 39810
+    },
+    {
+      "epoch": 261.9736842105263,
+      "grad_norm": 1.5944026708602905,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 39820
+    },
+    {
+      "epoch": 262.0394736842105,
+      "grad_norm": 1.3532087802886963,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 39830
+    },
+    {
+      "epoch": 262.10526315789474,
+      "grad_norm": 1.1372880935668945,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 39840
+    },
+    {
+      "epoch": 262.17105263157896,
+      "grad_norm": 1.1915911436080933,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 39850
+    },
+    {
+      "epoch": 262.2368421052632,
+      "grad_norm": 1.1271971464157104,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 39860
+    },
+    {
+      "epoch": 262.30263157894734,
+      "grad_norm": 1.3422695398330688,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 39870
+    },
+    {
+      "epoch": 262.36842105263156,
+      "grad_norm": 1.7832303047180176,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 39880
+    },
+    {
+      "epoch": 262.4342105263158,
+      "grad_norm": 1.4639548063278198,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 39890
+    },
+    {
+      "epoch": 262.5,
+      "grad_norm": 1.2754369974136353,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 39900
+    },
+    {
+      "epoch": 262.5657894736842,
+      "grad_norm": 1.5328514575958252,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 39910
+    },
+    {
+      "epoch": 262.63157894736844,
+      "grad_norm": 1.320682168006897,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 39920
+    },
+    {
+      "epoch": 262.69736842105266,
+      "grad_norm": 0.9552030563354492,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 39930
+    },
+    {
+      "epoch": 262.7631578947368,
+      "grad_norm": 1.2544821500778198,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 39940
+    },
+    {
+      "epoch": 262.82894736842104,
+      "grad_norm": 1.4101918935775757,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 39950
+    },
+    {
+      "epoch": 262.89473684210526,
+      "grad_norm": 1.5442532300949097,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 39960
+    },
+    {
+      "epoch": 262.9605263157895,
+      "grad_norm": 1.2423205375671387,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 39970
+    },
+    {
+      "epoch": 263.0263157894737,
+      "grad_norm": 1.0011227130889893,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 39980
+    },
+    {
+      "epoch": 263.0921052631579,
+      "grad_norm": 1.0351630449295044,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 39990
+    },
+    {
+      "epoch": 263.1578947368421,
+      "grad_norm": 1.29893159866333,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 40000
+    },
+    {
+      "epoch": 263.2236842105263,
+      "grad_norm": 1.5265520811080933,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 40010
+    },
+    {
+      "epoch": 263.2894736842105,
+      "grad_norm": 1.2051048278808594,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 40020
+    },
+    {
+      "epoch": 263.35526315789474,
+      "grad_norm": 1.576757788658142,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 40030
+    },
+    {
+      "epoch": 263.42105263157896,
+      "grad_norm": 1.4314970970153809,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 40040
+    },
+    {
+      "epoch": 263.4868421052632,
+      "grad_norm": 1.0734888315200806,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 40050
+    },
+    {
+      "epoch": 263.55263157894734,
+      "grad_norm": 1.0169308185577393,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 40060
+    },
+    {
+      "epoch": 263.61842105263156,
+      "grad_norm": 1.0590006113052368,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 40070
+    },
+    {
+      "epoch": 263.6842105263158,
+      "grad_norm": 1.1199071407318115,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 40080
+    },
+    {
+      "epoch": 263.75,
+      "grad_norm": 0.9885268211364746,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 40090
+    },
+    {
+      "epoch": 263.8157894736842,
+      "grad_norm": 0.7480038404464722,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 40100
+    },
+    {
+      "epoch": 263.88157894736844,
+      "grad_norm": 0.7771443128585815,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 40110
+    },
+    {
+      "epoch": 263.94736842105266,
+      "grad_norm": 1.1133038997650146,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 40120
+    },
+    {
+      "epoch": 264.0131578947368,
+      "grad_norm": 0.8933395147323608,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 40130
+    },
+    {
+      "epoch": 264.07894736842104,
+      "grad_norm": 1.2622970342636108,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 40140
+    },
+    {
+      "epoch": 264.14473684210526,
+      "grad_norm": 1.2373851537704468,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 40150
+    },
+    {
+      "epoch": 264.2105263157895,
+      "grad_norm": 1.3317362070083618,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 40160
+    },
+    {
+      "epoch": 264.2763157894737,
+      "grad_norm": 1.274640440940857,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 40170
+    },
+    {
+      "epoch": 264.3421052631579,
+      "grad_norm": 0.8638913035392761,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 40180
+    },
+    {
+      "epoch": 264.4078947368421,
+      "grad_norm": 1.1487349271774292,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 40190
+    },
+    {
+      "epoch": 264.4736842105263,
+      "grad_norm": 1.4805721044540405,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 40200
+    },
+    {
+      "epoch": 264.5394736842105,
+      "grad_norm": 1.180604338645935,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 40210
+    },
+    {
+      "epoch": 264.60526315789474,
+      "grad_norm": 1.2854849100112915,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 40220
+    },
+    {
+      "epoch": 264.67105263157896,
+      "grad_norm": 1.2706739902496338,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 40230
+    },
+    {
+      "epoch": 264.7368421052632,
+      "grad_norm": 1.0163190364837646,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 40240
+    },
+    {
+      "epoch": 264.80263157894734,
+      "grad_norm": 1.3690663576126099,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 40250
+    },
+    {
+      "epoch": 264.86842105263156,
+      "grad_norm": 0.8744670748710632,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 40260
+    },
+    {
+      "epoch": 264.9342105263158,
+      "grad_norm": 1.1227748394012451,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 40270
+    },
+    {
+      "epoch": 265.0,
+      "grad_norm": 1.1143121719360352,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 40280
+    },
+    {
+      "epoch": 265.0657894736842,
+      "grad_norm": 1.5000947713851929,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 40290
+    },
+    {
+      "epoch": 265.13157894736844,
+      "grad_norm": 1.3426282405853271,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 40300
+    },
+    {
+      "epoch": 265.19736842105266,
+      "grad_norm": 0.9148724675178528,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 40310
+    },
+    {
+      "epoch": 265.2631578947368,
+      "grad_norm": 0.9080812335014343,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 40320
+    },
+    {
+      "epoch": 265.32894736842104,
+      "grad_norm": 0.8644049167633057,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 40330
+    },
+    {
+      "epoch": 265.39473684210526,
+      "grad_norm": 0.9863361120223999,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 40340
+    },
+    {
+      "epoch": 265.4605263157895,
+      "grad_norm": 1.1581717729568481,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 40350
+    },
+    {
+      "epoch": 265.5263157894737,
+      "grad_norm": 0.8015313148498535,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 40360
+    },
+    {
+      "epoch": 265.5921052631579,
+      "grad_norm": 0.9356069564819336,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 40370
+    },
+    {
+      "epoch": 265.6578947368421,
+      "grad_norm": 1.1475307941436768,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 40380
+    },
+    {
+      "epoch": 265.7236842105263,
+      "grad_norm": 1.0415571928024292,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 40390
+    },
+    {
+      "epoch": 265.7894736842105,
+      "grad_norm": 1.4499433040618896,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 40400
+    },
+    {
+      "epoch": 265.85526315789474,
+      "grad_norm": 1.0016824007034302,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 40410
+    },
+    {
+      "epoch": 265.92105263157896,
+      "grad_norm": 1.426163911819458,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 40420
+    },
+    {
+      "epoch": 265.9868421052632,
+      "grad_norm": 1.2102915048599243,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 40430
+    },
+    {
+      "epoch": 266.05263157894734,
+      "grad_norm": 1.2748351097106934,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 40440
+    },
+    {
+      "epoch": 266.11842105263156,
+      "grad_norm": 1.0148237943649292,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 40450
+    },
+    {
+      "epoch": 266.1842105263158,
+      "grad_norm": 1.6188982725143433,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 40460
+    },
+    {
+      "epoch": 266.25,
+      "grad_norm": 1.2381527423858643,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 40470
+    },
+    {
+      "epoch": 266.3157894736842,
+      "grad_norm": 1.1073635816574097,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 40480
+    },
+    {
+      "epoch": 266.38157894736844,
+      "grad_norm": 1.1717376708984375,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 40490
+    },
+    {
+      "epoch": 266.44736842105266,
+      "grad_norm": 1.3967905044555664,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 40500
+    },
+    {
+      "epoch": 266.5131578947368,
+      "grad_norm": 0.78745037317276,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 40510
+    },
+    {
+      "epoch": 266.57894736842104,
+      "grad_norm": 1.2975760698318481,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 40520
+    },
+    {
+      "epoch": 266.64473684210526,
+      "grad_norm": 1.0333595275878906,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 40530
+    },
+    {
+      "epoch": 266.7105263157895,
+      "grad_norm": 1.132156252861023,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 40540
+    },
+    {
+      "epoch": 266.7763157894737,
+      "grad_norm": 1.1081570386886597,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 40550
+    },
+    {
+      "epoch": 266.8421052631579,
+      "grad_norm": 1.6667684316635132,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 40560
+    },
+    {
+      "epoch": 266.9078947368421,
+      "grad_norm": 1.6296488046646118,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 40570
+    },
+    {
+      "epoch": 266.9736842105263,
+      "grad_norm": 1.4970145225524902,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 40580
+    },
+    {
+      "epoch": 267.0394736842105,
+      "grad_norm": 1.2086564302444458,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 40590
+    },
+    {
+      "epoch": 267.10526315789474,
+      "grad_norm": 1.0567036867141724,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 40600
+    },
+    {
+      "epoch": 267.17105263157896,
+      "grad_norm": 1.2320477962493896,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 40610
+    },
+    {
+      "epoch": 267.2368421052632,
+      "grad_norm": 1.1581894159317017,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 40620
+    },
+    {
+      "epoch": 267.30263157894734,
+      "grad_norm": 1.2620066404342651,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 40630
+    },
+    {
+      "epoch": 267.36842105263156,
+      "grad_norm": 1.0989741086959839,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 40640
+    },
+    {
+      "epoch": 267.4342105263158,
+      "grad_norm": 1.0343804359436035,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 40650
+    },
+    {
+      "epoch": 267.5,
+      "grad_norm": 1.1863844394683838,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 40660
+    },
+    {
+      "epoch": 267.5657894736842,
+      "grad_norm": 1.5389338731765747,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 40670
+    },
+    {
+      "epoch": 267.63157894736844,
+      "grad_norm": 0.8726987242698669,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 40680
+    },
+    {
+      "epoch": 267.69736842105266,
+      "grad_norm": 0.9518386721611023,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 40690
+    },
+    {
+      "epoch": 267.7631578947368,
+      "grad_norm": 0.9239757657051086,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 40700
+    },
+    {
+      "epoch": 267.82894736842104,
+      "grad_norm": 0.800014078617096,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 40710
+    },
+    {
+      "epoch": 267.89473684210526,
+      "grad_norm": 1.1381514072418213,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 40720
+    },
+    {
+      "epoch": 267.9605263157895,
+      "grad_norm": 1.2737356424331665,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 40730
+    },
+    {
+      "epoch": 268.0263157894737,
+      "grad_norm": 1.2936004400253296,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 40740
+    },
+    {
+      "epoch": 268.0921052631579,
+      "grad_norm": 1.2660059928894043,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 40750
+    },
+    {
+      "epoch": 268.1578947368421,
+      "grad_norm": 1.2411798238754272,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 40760
+    },
+    {
+      "epoch": 268.2236842105263,
+      "grad_norm": 0.9876865744590759,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 40770
+    },
+    {
+      "epoch": 268.2894736842105,
+      "grad_norm": 1.146823525428772,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 40780
+    },
+    {
+      "epoch": 268.35526315789474,
+      "grad_norm": 1.0858898162841797,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 40790
+    },
+    {
+      "epoch": 268.42105263157896,
+      "grad_norm": 1.208295464515686,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 40800
+    },
+    {
+      "epoch": 268.4868421052632,
+      "grad_norm": 1.1481982469558716,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 40810
+    },
+    {
+      "epoch": 268.55263157894734,
+      "grad_norm": 1.2117033004760742,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 40820
+    },
+    {
+      "epoch": 268.61842105263156,
+      "grad_norm": 1.550671100616455,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 40830
+    },
+    {
+      "epoch": 268.6842105263158,
+      "grad_norm": 1.131887435913086,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 40840
+    },
+    {
+      "epoch": 268.75,
+      "grad_norm": 1.248721957206726,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 40850
+    },
+    {
+      "epoch": 268.8157894736842,
+      "grad_norm": 1.0527727603912354,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 40860
+    },
+    {
+      "epoch": 268.88157894736844,
+      "grad_norm": 1.243199348449707,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 40870
+    },
+    {
+      "epoch": 268.94736842105266,
+      "grad_norm": 1.0433778762817383,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 40880
+    },
+    {
+      "epoch": 269.0131578947368,
+      "grad_norm": 1.5813816785812378,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 40890
+    },
+    {
+      "epoch": 269.07894736842104,
+      "grad_norm": 1.3357555866241455,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 40900
+    },
+    {
+      "epoch": 269.14473684210526,
+      "grad_norm": 1.4037662744522095,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 40910
+    },
+    {
+      "epoch": 269.2105263157895,
+      "grad_norm": 1.459270715713501,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 40920
+    },
+    {
+      "epoch": 269.2763157894737,
+      "grad_norm": 1.2540459632873535,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 40930
+    },
+    {
+      "epoch": 269.3421052631579,
+      "grad_norm": 0.7612592577934265,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 40940
+    },
+    {
+      "epoch": 269.4078947368421,
+      "grad_norm": 1.0705971717834473,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 40950
+    },
+    {
+      "epoch": 269.4736842105263,
+      "grad_norm": 1.2538421154022217,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 40960
+    },
+    {
+      "epoch": 269.5394736842105,
+      "grad_norm": 1.143507719039917,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 40970
+    },
+    {
+      "epoch": 269.60526315789474,
+      "grad_norm": 1.3670626878738403,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 40980
+    },
+    {
+      "epoch": 269.67105263157896,
+      "grad_norm": 1.126387596130371,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 40990
+    },
+    {
+      "epoch": 269.7368421052632,
+      "grad_norm": 1.3656985759735107,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 41000
+    },
+    {
+      "epoch": 269.80263157894734,
+      "grad_norm": 1.4171274900436401,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 41010
+    },
+    {
+      "epoch": 269.86842105263156,
+      "grad_norm": 1.4547640085220337,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 41020
+    },
+    {
+      "epoch": 269.9342105263158,
+      "grad_norm": 1.1633445024490356,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 41030
+    },
+    {
+      "epoch": 270.0,
+      "grad_norm": 1.2674797773361206,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 41040
+    },
+    {
+      "epoch": 270.0657894736842,
+      "grad_norm": 1.2385079860687256,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 41050
+    },
+    {
+      "epoch": 270.13157894736844,
+      "grad_norm": 1.0688042640686035,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 41060
+    },
+    {
+      "epoch": 270.19736842105266,
+      "grad_norm": 1.8457244634628296,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 41070
+    },
+    {
+      "epoch": 270.2631578947368,
+      "grad_norm": 1.4692785739898682,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 41080
+    },
+    {
+      "epoch": 270.32894736842104,
+      "grad_norm": 1.0676112174987793,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 41090
+    },
+    {
+      "epoch": 270.39473684210526,
+      "grad_norm": 1.3465666770935059,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 41100
+    },
+    {
+      "epoch": 270.4605263157895,
+      "grad_norm": 1.012242317199707,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 41110
+    },
+    {
+      "epoch": 270.5263157894737,
+      "grad_norm": 1.1431010961532593,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 41120
+    },
+    {
+      "epoch": 270.5921052631579,
+      "grad_norm": 1.2994465827941895,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 41130
+    },
+    {
+      "epoch": 270.6578947368421,
+      "grad_norm": 1.2632924318313599,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 41140
+    },
+    {
+      "epoch": 270.7236842105263,
+      "grad_norm": 1.0502218008041382,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 41150
+    },
+    {
+      "epoch": 270.7894736842105,
+      "grad_norm": 1.1967225074768066,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 41160
+    },
+    {
+      "epoch": 270.85526315789474,
+      "grad_norm": 1.095524787902832,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 41170
+    },
+    {
+      "epoch": 270.92105263157896,
+      "grad_norm": 1.2695966958999634,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 41180
+    },
+    {
+      "epoch": 270.9868421052632,
+      "grad_norm": 1.425173282623291,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 41190
+    },
+    {
+      "epoch": 271.05263157894734,
+      "grad_norm": 0.9129456877708435,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 41200
+    },
+    {
+      "epoch": 271.11842105263156,
+      "grad_norm": 0.9223254919052124,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 41210
+    },
+    {
+      "epoch": 271.1842105263158,
+      "grad_norm": 1.0759285688400269,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 41220
+    },
+    {
+      "epoch": 271.25,
+      "grad_norm": 1.4797776937484741,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 41230
+    },
+    {
+      "epoch": 271.3157894736842,
+      "grad_norm": 1.368739128112793,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 41240
+    },
+    {
+      "epoch": 271.38157894736844,
+      "grad_norm": 1.1188558340072632,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 41250
+    },
+    {
+      "epoch": 271.44736842105266,
+      "grad_norm": 1.3473429679870605,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 41260
+    },
+    {
+      "epoch": 271.5131578947368,
+      "grad_norm": 1.3430105447769165,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 41270
+    },
+    {
+      "epoch": 271.57894736842104,
+      "grad_norm": 1.654737949371338,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 41280
+    },
+    {
+      "epoch": 271.64473684210526,
+      "grad_norm": 1.832235336303711,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 41290
+    },
+    {
+      "epoch": 271.7105263157895,
+      "grad_norm": 1.3615492582321167,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 41300
+    },
+    {
+      "epoch": 271.7763157894737,
+      "grad_norm": 1.5935784578323364,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 41310
+    },
+    {
+      "epoch": 271.8421052631579,
+      "grad_norm": 1.6195662021636963,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 41320
+    },
+    {
+      "epoch": 271.9078947368421,
+      "grad_norm": 1.3322649002075195,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 41330
+    },
+    {
+      "epoch": 271.9736842105263,
+      "grad_norm": 1.2479629516601562,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 41340
+    },
+    {
+      "epoch": 272.0394736842105,
+      "grad_norm": 1.4976459741592407,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 41350
+    },
+    {
+      "epoch": 272.10526315789474,
+      "grad_norm": 1.2839045524597168,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 41360
+    },
+    {
+      "epoch": 272.17105263157896,
+      "grad_norm": 1.2758049964904785,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 41370
+    },
+    {
+      "epoch": 272.2368421052632,
+      "grad_norm": 1.352516531944275,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 41380
+    },
+    {
+      "epoch": 272.30263157894734,
+      "grad_norm": 1.0206512212753296,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 41390
+    },
+    {
+      "epoch": 272.36842105263156,
+      "grad_norm": 1.2636960744857788,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 41400
+    },
+    {
+      "epoch": 272.4342105263158,
+      "grad_norm": 1.058984398841858,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 41410
+    },
+    {
+      "epoch": 272.5,
+      "grad_norm": 1.2583855390548706,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 41420
+    },
+    {
+      "epoch": 272.5657894736842,
+      "grad_norm": 1.1971688270568848,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 41430
+    },
+    {
+      "epoch": 272.63157894736844,
+      "grad_norm": 1.403083324432373,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 41440
+    },
+    {
+      "epoch": 272.69736842105266,
+      "grad_norm": 1.0412906408309937,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 41450
+    },
+    {
+      "epoch": 272.7631578947368,
+      "grad_norm": 1.2968218326568604,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 41460
+    },
+    {
+      "epoch": 272.82894736842104,
+      "grad_norm": 1.4638139009475708,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 41470
+    },
+    {
+      "epoch": 272.89473684210526,
+      "grad_norm": 1.3034776449203491,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 41480
+    },
+    {
+      "epoch": 272.9605263157895,
+      "grad_norm": 1.0589145421981812,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 41490
+    },
+    {
+      "epoch": 273.0263157894737,
+      "grad_norm": 0.905729353427887,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 41500
+    },
+    {
+      "epoch": 273.0921052631579,
+      "grad_norm": 1.093269944190979,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 41510
+    },
+    {
+      "epoch": 273.1578947368421,
+      "grad_norm": 1.5292916297912598,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 41520
+    },
+    {
+      "epoch": 273.2236842105263,
+      "grad_norm": 1.5569329261779785,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 41530
+    },
+    {
+      "epoch": 273.2894736842105,
+      "grad_norm": 1.620352029800415,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 41540
+    },
+    {
+      "epoch": 273.35526315789474,
+      "grad_norm": 1.3029897212982178,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 41550
+    },
+    {
+      "epoch": 273.42105263157896,
+      "grad_norm": 1.1392931938171387,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 41560
+    },
+    {
+      "epoch": 273.4868421052632,
+      "grad_norm": 1.3479526042938232,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 41570
+    },
+    {
+      "epoch": 273.55263157894734,
+      "grad_norm": 1.3470922708511353,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 41580
+    },
+    {
+      "epoch": 273.61842105263156,
+      "grad_norm": 1.2211447954177856,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 41590
+    },
+    {
+      "epoch": 273.6842105263158,
+      "grad_norm": 1.1938384771347046,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 41600
+    },
+    {
+      "epoch": 273.75,
+      "grad_norm": 1.1164087057113647,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 41610
+    },
+    {
+      "epoch": 273.8157894736842,
+      "grad_norm": 1.1596921682357788,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 41620
+    },
+    {
+      "epoch": 273.88157894736844,
+      "grad_norm": 0.9100044369697571,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 41630
+    },
+    {
+      "epoch": 273.94736842105266,
+      "grad_norm": 0.9539213180541992,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 41640
+    },
+    {
+      "epoch": 274.0131578947368,
+      "grad_norm": 0.9775730967521667,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 41650
+    },
+    {
+      "epoch": 274.07894736842104,
+      "grad_norm": 1.1786597967147827,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 41660
+    },
+    {
+      "epoch": 274.14473684210526,
+      "grad_norm": 0.8824886679649353,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 41670
+    },
+    {
+      "epoch": 274.2105263157895,
+      "grad_norm": 1.173654556274414,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 41680
+    },
+    {
+      "epoch": 274.2763157894737,
+      "grad_norm": 1.509282112121582,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 41690
+    },
+    {
+      "epoch": 274.3421052631579,
+      "grad_norm": 1.3084309101104736,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 41700
+    },
+    {
+      "epoch": 274.4078947368421,
+      "grad_norm": 1.2768287658691406,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 41710
+    },
+    {
+      "epoch": 274.4736842105263,
+      "grad_norm": 1.142025351524353,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 41720
+    },
+    {
+      "epoch": 274.5394736842105,
+      "grad_norm": 1.13783597946167,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 41730
+    },
+    {
+      "epoch": 274.60526315789474,
+      "grad_norm": 0.8221064805984497,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 41740
+    },
+    {
+      "epoch": 274.67105263157896,
+      "grad_norm": 1.6410062313079834,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 41750
+    },
+    {
+      "epoch": 274.7368421052632,
+      "grad_norm": 1.4255626201629639,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 41760
+    },
+    {
+      "epoch": 274.80263157894734,
+      "grad_norm": 1.1363221406936646,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 41770
+    },
+    {
+      "epoch": 274.86842105263156,
+      "grad_norm": 1.3313068151474,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 41780
+    },
+    {
+      "epoch": 274.9342105263158,
+      "grad_norm": 1.4202282428741455,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 41790
+    },
+    {
+      "epoch": 275.0,
+      "grad_norm": 1.3749666213989258,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 41800
+    },
+    {
+      "epoch": 275.0657894736842,
+      "grad_norm": 1.2333338260650635,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 41810
+    },
+    {
+      "epoch": 275.13157894736844,
+      "grad_norm": 1.221083164215088,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 41820
+    },
+    {
+      "epoch": 275.19736842105266,
+      "grad_norm": 0.8739851713180542,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 41830
+    },
+    {
+      "epoch": 275.2631578947368,
+      "grad_norm": 1.3110480308532715,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 41840
+    },
+    {
+      "epoch": 275.32894736842104,
+      "grad_norm": 0.8947159647941589,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 41850
+    },
+    {
+      "epoch": 275.39473684210526,
+      "grad_norm": 1.2646197080612183,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 41860
+    },
+    {
+      "epoch": 275.4605263157895,
+      "grad_norm": 1.0885008573532104,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 41870
+    },
+    {
+      "epoch": 275.5263157894737,
+      "grad_norm": 1.4110907316207886,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 41880
+    },
+    {
+      "epoch": 275.5921052631579,
+      "grad_norm": 1.312502145767212,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 41890
+    },
+    {
+      "epoch": 275.6578947368421,
+      "grad_norm": 1.3466676473617554,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 41900
+    },
+    {
+      "epoch": 275.7236842105263,
+      "grad_norm": 1.6430773735046387,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 41910
+    },
+    {
+      "epoch": 275.7894736842105,
+      "grad_norm": 1.3257335424423218,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 41920
+    },
+    {
+      "epoch": 275.85526315789474,
+      "grad_norm": 1.3992869853973389,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 41930
+    },
+    {
+      "epoch": 275.92105263157896,
+      "grad_norm": 1.3640083074569702,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 41940
+    },
+    {
+      "epoch": 275.9868421052632,
+      "grad_norm": 1.2238960266113281,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 41950
+    },
+    {
+      "epoch": 276.05263157894734,
+      "grad_norm": 0.9982905983924866,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 41960
+    },
+    {
+      "epoch": 276.11842105263156,
+      "grad_norm": 1.0933977365493774,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 41970
+    },
+    {
+      "epoch": 276.1842105263158,
+      "grad_norm": 1.1937322616577148,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 41980
+    },
+    {
+      "epoch": 276.25,
+      "grad_norm": 1.1956714391708374,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 41990
+    },
+    {
+      "epoch": 276.3157894736842,
+      "grad_norm": 1.256412386894226,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 42000
+    },
+    {
+      "epoch": 276.38157894736844,
+      "grad_norm": 1.035623550415039,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 42010
+    },
+    {
+      "epoch": 276.44736842105266,
+      "grad_norm": 1.2642019987106323,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 42020
+    },
+    {
+      "epoch": 276.5131578947368,
+      "grad_norm": 1.7125762701034546,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 42030
+    },
+    {
+      "epoch": 276.57894736842104,
+      "grad_norm": 1.3358557224273682,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 42040
+    },
+    {
+      "epoch": 276.64473684210526,
+      "grad_norm": 1.3947197198867798,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 42050
+    },
+    {
+      "epoch": 276.7105263157895,
+      "grad_norm": 1.3491345643997192,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 42060
+    },
+    {
+      "epoch": 276.7763157894737,
+      "grad_norm": 1.3073524236679077,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 42070
+    },
+    {
+      "epoch": 276.8421052631579,
+      "grad_norm": 1.403847098350525,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 42080
+    },
+    {
+      "epoch": 276.9078947368421,
+      "grad_norm": 1.1999281644821167,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 42090
+    },
+    {
+      "epoch": 276.9736842105263,
+      "grad_norm": 1.2854633331298828,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 42100
+    },
+    {
+      "epoch": 277.0394736842105,
+      "grad_norm": 1.1524097919464111,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 42110
+    },
+    {
+      "epoch": 277.10526315789474,
+      "grad_norm": 1.1198937892913818,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 42120
+    },
+    {
+      "epoch": 277.17105263157896,
+      "grad_norm": 0.9820287823677063,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 42130
+    },
+    {
+      "epoch": 277.2368421052632,
+      "grad_norm": 0.7458430528640747,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 42140
+    },
+    {
+      "epoch": 277.30263157894734,
+      "grad_norm": 1.2689908742904663,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 42150
+    },
+    {
+      "epoch": 277.36842105263156,
+      "grad_norm": 1.062583088874817,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 42160
+    },
+    {
+      "epoch": 277.4342105263158,
+      "grad_norm": 0.82270747423172,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 42170
+    },
+    {
+      "epoch": 277.5,
+      "grad_norm": 0.8221926093101501,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 42180
+    },
+    {
+      "epoch": 277.5657894736842,
+      "grad_norm": 0.8966209888458252,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 42190
+    },
+    {
+      "epoch": 277.63157894736844,
+      "grad_norm": 1.6077913045883179,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 42200
+    },
+    {
+      "epoch": 277.69736842105266,
+      "grad_norm": 0.9948518872261047,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 42210
+    },
+    {
+      "epoch": 277.7631578947368,
+      "grad_norm": 1.1815922260284424,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 42220
+    },
+    {
+      "epoch": 277.82894736842104,
+      "grad_norm": 1.0743731260299683,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 42230
+    },
+    {
+      "epoch": 277.89473684210526,
+      "grad_norm": 0.9307351112365723,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 42240
+    },
+    {
+      "epoch": 277.9605263157895,
+      "grad_norm": 1.2313666343688965,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 42250
+    },
+    {
+      "epoch": 278.0263157894737,
+      "grad_norm": 1.1681599617004395,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 42260
+    },
+    {
+      "epoch": 278.0921052631579,
+      "grad_norm": 1.4541995525360107,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 42270
+    },
+    {
+      "epoch": 278.1578947368421,
+      "grad_norm": 1.107507586479187,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 42280
+    },
+    {
+      "epoch": 278.2236842105263,
+      "grad_norm": 0.9326557517051697,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 42290
+    },
+    {
+      "epoch": 278.2894736842105,
+      "grad_norm": 1.512211799621582,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 42300
+    },
+    {
+      "epoch": 278.35526315789474,
+      "grad_norm": 1.3133347034454346,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 42310
+    },
+    {
+      "epoch": 278.42105263157896,
+      "grad_norm": 1.0264031887054443,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 42320
+    },
+    {
+      "epoch": 278.4868421052632,
+      "grad_norm": 1.1678309440612793,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 42330
+    },
+    {
+      "epoch": 278.55263157894734,
+      "grad_norm": 0.7482702732086182,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 42340
+    },
+    {
+      "epoch": 278.61842105263156,
+      "grad_norm": 1.0552853345870972,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 42350
+    },
+    {
+      "epoch": 278.6842105263158,
+      "grad_norm": 1.301750659942627,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 42360
+    },
+    {
+      "epoch": 278.75,
+      "grad_norm": 0.8871774673461914,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 42370
+    },
+    {
+      "epoch": 278.8157894736842,
+      "grad_norm": 0.9362806677818298,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 42380
+    },
+    {
+      "epoch": 278.88157894736844,
+      "grad_norm": 0.6715094447135925,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 42390
+    },
+    {
+      "epoch": 278.94736842105266,
+      "grad_norm": 1.3032416105270386,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 42400
+    },
+    {
+      "epoch": 279.0131578947368,
+      "grad_norm": 1.2517324686050415,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 42410
+    },
+    {
+      "epoch": 279.07894736842104,
+      "grad_norm": 1.0311671495437622,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 42420
+    },
+    {
+      "epoch": 279.14473684210526,
+      "grad_norm": 0.9779248237609863,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 42430
+    },
+    {
+      "epoch": 279.2105263157895,
+      "grad_norm": 1.322513461112976,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 42440
+    },
+    {
+      "epoch": 279.2763157894737,
+      "grad_norm": 1.117714285850525,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 42450
+    },
+    {
+      "epoch": 279.3421052631579,
+      "grad_norm": 1.0339843034744263,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 42460
+    },
+    {
+      "epoch": 279.4078947368421,
+      "grad_norm": 1.3660122156143188,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 42470
+    },
+    {
+      "epoch": 279.4736842105263,
+      "grad_norm": 1.191132664680481,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 42480
+    },
+    {
+      "epoch": 279.5394736842105,
+      "grad_norm": 1.476515293121338,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 42490
+    },
+    {
+      "epoch": 279.60526315789474,
+      "grad_norm": 1.0748845338821411,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 42500
+    },
+    {
+      "epoch": 279.67105263157896,
+      "grad_norm": 1.480494499206543,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 42510
+    },
+    {
+      "epoch": 279.7368421052632,
+      "grad_norm": 1.1140382289886475,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 42520
+    },
+    {
+      "epoch": 279.80263157894734,
+      "grad_norm": 1.4050570726394653,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 42530
+    },
+    {
+      "epoch": 279.86842105263156,
+      "grad_norm": 1.3111275434494019,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 42540
+    },
+    {
+      "epoch": 279.9342105263158,
+      "grad_norm": 1.2099816799163818,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 42550
+    },
+    {
+      "epoch": 280.0,
+      "grad_norm": 1.2732938528060913,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 42560
+    },
+    {
+      "epoch": 280.0657894736842,
+      "grad_norm": 1.4480763673782349,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 42570
+    },
+    {
+      "epoch": 280.13157894736844,
+      "grad_norm": 1.1580476760864258,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 42580
+    },
+    {
+      "epoch": 280.19736842105266,
+      "grad_norm": 1.4821127653121948,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 42590
+    },
+    {
+      "epoch": 280.2631578947368,
+      "grad_norm": 1.0352067947387695,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 42600
+    },
+    {
+      "epoch": 280.32894736842104,
+      "grad_norm": 1.536171793937683,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 42610
+    },
+    {
+      "epoch": 280.39473684210526,
+      "grad_norm": 1.3197144269943237,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 42620
+    },
+    {
+      "epoch": 280.4605263157895,
+      "grad_norm": 1.0145988464355469,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 42630
+    },
+    {
+      "epoch": 280.5263157894737,
+      "grad_norm": 1.5612324476242065,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 42640
+    },
+    {
+      "epoch": 280.5921052631579,
+      "grad_norm": 1.013548493385315,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 42650
+    },
+    {
+      "epoch": 280.6578947368421,
+      "grad_norm": 1.0804232358932495,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 42660
+    },
+    {
+      "epoch": 280.7236842105263,
+      "grad_norm": 1.183570146560669,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 42670
+    },
+    {
+      "epoch": 280.7894736842105,
+      "grad_norm": 1.551816463470459,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 42680
+    },
+    {
+      "epoch": 280.85526315789474,
+      "grad_norm": 1.5829601287841797,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 42690
+    },
+    {
+      "epoch": 280.92105263157896,
+      "grad_norm": 1.6313118934631348,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 42700
+    },
+    {
+      "epoch": 280.9868421052632,
+      "grad_norm": 0.820428192615509,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 42710
+    },
+    {
+      "epoch": 281.05263157894734,
+      "grad_norm": 1.293191909790039,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 42720
+    },
+    {
+      "epoch": 281.11842105263156,
+      "grad_norm": 1.3593038320541382,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 42730
+    },
+    {
+      "epoch": 281.1842105263158,
+      "grad_norm": 1.3805264234542847,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 42740
+    },
+    {
+      "epoch": 281.25,
+      "grad_norm": 1.3258028030395508,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 42750
+    },
+    {
+      "epoch": 281.3157894736842,
+      "grad_norm": 1.2791032791137695,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 42760
+    },
+    {
+      "epoch": 281.38157894736844,
+      "grad_norm": 1.3307384252548218,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 42770
+    },
+    {
+      "epoch": 281.44736842105266,
+      "grad_norm": 1.1572662591934204,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 42780
+    },
+    {
+      "epoch": 281.5131578947368,
+      "grad_norm": 1.5701806545257568,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 42790
+    },
+    {
+      "epoch": 281.57894736842104,
+      "grad_norm": 1.2336055040359497,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 42800
+    },
+    {
+      "epoch": 281.64473684210526,
+      "grad_norm": 1.0266544818878174,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 42810
+    },
+    {
+      "epoch": 281.7105263157895,
+      "grad_norm": 0.840146541595459,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 42820
+    },
+    {
+      "epoch": 281.7763157894737,
+      "grad_norm": 1.1825062036514282,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 42830
+    },
+    {
+      "epoch": 281.8421052631579,
+      "grad_norm": 1.0842541456222534,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 42840
+    },
+    {
+      "epoch": 281.9078947368421,
+      "grad_norm": 1.5062053203582764,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 42850
+    },
+    {
+      "epoch": 281.9736842105263,
+      "grad_norm": 0.8112286925315857,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 42860
+    },
+    {
+      "epoch": 282.0394736842105,
+      "grad_norm": 1.1789262294769287,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 42870
+    },
+    {
+      "epoch": 282.10526315789474,
+      "grad_norm": 1.375529170036316,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 42880
+    },
+    {
+      "epoch": 282.17105263157896,
+      "grad_norm": 1.330922245979309,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 42890
+    },
+    {
+      "epoch": 282.2368421052632,
+      "grad_norm": 1.3827651739120483,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 42900
+    },
+    {
+      "epoch": 282.30263157894734,
+      "grad_norm": 1.2131527662277222,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 42910
+    },
+    {
+      "epoch": 282.36842105263156,
+      "grad_norm": 1.3317664861679077,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 42920
+    },
+    {
+      "epoch": 282.4342105263158,
+      "grad_norm": 1.1144113540649414,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 42930
+    },
+    {
+      "epoch": 282.5,
+      "grad_norm": 1.6057426929473877,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 42940
+    },
+    {
+      "epoch": 282.5657894736842,
+      "grad_norm": 1.3409185409545898,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 42950
+    },
+    {
+      "epoch": 282.63157894736844,
+      "grad_norm": 1.5420650243759155,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 42960
+    },
+    {
+      "epoch": 282.69736842105266,
+      "grad_norm": 1.1373865604400635,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 42970
+    },
+    {
+      "epoch": 282.7631578947368,
+      "grad_norm": 1.2429486513137817,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 42980
+    },
+    {
+      "epoch": 282.82894736842104,
+      "grad_norm": 0.9992790222167969,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 42990
+    },
+    {
+      "epoch": 282.89473684210526,
+      "grad_norm": 1.142056941986084,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 43000
+    },
+    {
+      "epoch": 282.9605263157895,
+      "grad_norm": 1.0192174911499023,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 43010
+    },
+    {
+      "epoch": 283.0263157894737,
+      "grad_norm": 0.9924356937408447,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 43020
+    },
+    {
+      "epoch": 283.0921052631579,
+      "grad_norm": 0.8189639449119568,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 43030
+    },
+    {
+      "epoch": 283.1578947368421,
+      "grad_norm": 0.9445114731788635,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 43040
+    },
+    {
+      "epoch": 283.2236842105263,
+      "grad_norm": 1.3871409893035889,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 43050
+    },
+    {
+      "epoch": 283.2894736842105,
+      "grad_norm": 1.086130142211914,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 43060
+    },
+    {
+      "epoch": 283.35526315789474,
+      "grad_norm": 0.9964144825935364,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 43070
+    },
+    {
+      "epoch": 283.42105263157896,
+      "grad_norm": 1.0563546419143677,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 43080
+    },
+    {
+      "epoch": 283.4868421052632,
+      "grad_norm": 1.270878553390503,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 43090
+    },
+    {
+      "epoch": 283.55263157894734,
+      "grad_norm": 1.3381757736206055,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 43100
+    },
+    {
+      "epoch": 283.61842105263156,
+      "grad_norm": 1.1857870817184448,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 43110
+    },
+    {
+      "epoch": 283.6842105263158,
+      "grad_norm": 1.2086889743804932,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 43120
+    },
+    {
+      "epoch": 283.75,
+      "grad_norm": 1.3116366863250732,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 43130
+    },
+    {
+      "epoch": 283.8157894736842,
+      "grad_norm": 1.1488587856292725,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 43140
+    },
+    {
+      "epoch": 283.88157894736844,
+      "grad_norm": 0.9865978360176086,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 43150
+    },
+    {
+      "epoch": 283.94736842105266,
+      "grad_norm": 0.9599931240081787,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 43160
+    },
+    {
+      "epoch": 284.0131578947368,
+      "grad_norm": 1.4896612167358398,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 43170
+    },
+    {
+      "epoch": 284.07894736842104,
+      "grad_norm": 1.335207223892212,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 43180
+    },
+    {
+      "epoch": 284.14473684210526,
+      "grad_norm": 0.7765811085700989,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 43190
+    },
+    {
+      "epoch": 284.2105263157895,
+      "grad_norm": 1.1684138774871826,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 43200
+    },
+    {
+      "epoch": 284.2763157894737,
+      "grad_norm": 1.3540507555007935,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 43210
+    },
+    {
+      "epoch": 284.3421052631579,
+      "grad_norm": 1.5758388042449951,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 43220
+    },
+    {
+      "epoch": 284.4078947368421,
+      "grad_norm": 1.411137580871582,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 43230
+    },
+    {
+      "epoch": 284.4736842105263,
+      "grad_norm": 1.1263060569763184,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 43240
+    },
+    {
+      "epoch": 284.5394736842105,
+      "grad_norm": 1.4823040962219238,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 43250
+    },
+    {
+      "epoch": 284.60526315789474,
+      "grad_norm": 1.3825323581695557,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 43260
+    },
+    {
+      "epoch": 284.67105263157896,
+      "grad_norm": 1.1096550226211548,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 43270
+    },
+    {
+      "epoch": 284.7368421052632,
+      "grad_norm": 1.0073033571243286,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 43280
+    },
+    {
+      "epoch": 284.80263157894734,
+      "grad_norm": 1.08073091506958,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 43290
+    },
+    {
+      "epoch": 284.86842105263156,
+      "grad_norm": 0.6875160336494446,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 43300
+    },
+    {
+      "epoch": 284.9342105263158,
+      "grad_norm": 1.0610051155090332,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 43310
+    },
+    {
+      "epoch": 285.0,
+      "grad_norm": 0.9582310318946838,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 43320
+    },
+    {
+      "epoch": 285.0657894736842,
+      "grad_norm": 0.8970026969909668,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 43330
+    },
+    {
+      "epoch": 285.13157894736844,
+      "grad_norm": 0.8817126750946045,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 43340
+    },
+    {
+      "epoch": 285.19736842105266,
+      "grad_norm": 1.2000654935836792,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 43350
+    },
+    {
+      "epoch": 285.2631578947368,
+      "grad_norm": 1.3868666887283325,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 43360
+    },
+    {
+      "epoch": 285.32894736842104,
+      "grad_norm": 0.9346180558204651,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 43370
+    },
+    {
+      "epoch": 285.39473684210526,
+      "grad_norm": 0.9020205736160278,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 43380
+    },
+    {
+      "epoch": 285.4605263157895,
+      "grad_norm": 1.231022834777832,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 43390
+    },
+    {
+      "epoch": 285.5263157894737,
+      "grad_norm": 1.107666015625,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 43400
+    },
+    {
+      "epoch": 285.5921052631579,
+      "grad_norm": 1.5066663026809692,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 43410
+    },
+    {
+      "epoch": 285.6578947368421,
+      "grad_norm": 1.7655541896820068,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 43420
+    },
+    {
+      "epoch": 285.7236842105263,
+      "grad_norm": 1.2582757472991943,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 43430
+    },
+    {
+      "epoch": 285.7894736842105,
+      "grad_norm": 1.1627318859100342,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 43440
+    },
+    {
+      "epoch": 285.85526315789474,
+      "grad_norm": 1.3102083206176758,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 43450
+    },
+    {
+      "epoch": 285.92105263157896,
+      "grad_norm": 1.2415940761566162,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 43460
+    },
+    {
+      "epoch": 285.9868421052632,
+      "grad_norm": 1.494537353515625,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 43470
+    },
+    {
+      "epoch": 286.05263157894734,
+      "grad_norm": 1.4216185808181763,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 43480
+    },
+    {
+      "epoch": 286.11842105263156,
+      "grad_norm": 1.2979694604873657,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 43490
+    },
+    {
+      "epoch": 286.1842105263158,
+      "grad_norm": 0.9436559677124023,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 43500
+    },
+    {
+      "epoch": 286.25,
+      "grad_norm": 1.34527587890625,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 43510
+    },
+    {
+      "epoch": 286.3157894736842,
+      "grad_norm": 1.5451651811599731,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 43520
+    },
+    {
+      "epoch": 286.38157894736844,
+      "grad_norm": 1.5682357549667358,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 43530
+    },
+    {
+      "epoch": 286.44736842105266,
+      "grad_norm": 1.3602068424224854,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 43540
+    },
+    {
+      "epoch": 286.5131578947368,
+      "grad_norm": 1.146419644355774,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 43550
+    },
+    {
+      "epoch": 286.57894736842104,
+      "grad_norm": 0.9245599508285522,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 43560
+    },
+    {
+      "epoch": 286.64473684210526,
+      "grad_norm": 0.9519632458686829,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 43570
+    },
+    {
+      "epoch": 286.7105263157895,
+      "grad_norm": 1.0367894172668457,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 43580
+    },
+    {
+      "epoch": 286.7763157894737,
+      "grad_norm": 1.3567640781402588,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 43590
+    },
+    {
+      "epoch": 286.8421052631579,
+      "grad_norm": 1.3016269207000732,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 43600
+    },
+    {
+      "epoch": 286.9078947368421,
+      "grad_norm": 1.4324815273284912,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 43610
+    },
+    {
+      "epoch": 286.9736842105263,
+      "grad_norm": 1.0870546102523804,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 43620
+    },
+    {
+      "epoch": 287.0394736842105,
+      "grad_norm": 1.5067437887191772,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 43630
+    },
+    {
+      "epoch": 287.10526315789474,
+      "grad_norm": 1.3528972864151,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 43640
+    },
+    {
+      "epoch": 287.17105263157896,
+      "grad_norm": 1.3971104621887207,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 43650
+    },
+    {
+      "epoch": 287.2368421052632,
+      "grad_norm": 1.1645212173461914,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 43660
+    },
+    {
+      "epoch": 287.30263157894734,
+      "grad_norm": 1.295596718788147,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 43670
+    },
+    {
+      "epoch": 287.36842105263156,
+      "grad_norm": 0.900801420211792,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 43680
+    },
+    {
+      "epoch": 287.4342105263158,
+      "grad_norm": 1.128415584564209,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 43690
+    },
+    {
+      "epoch": 287.5,
+      "grad_norm": 1.3298012018203735,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 43700
+    },
+    {
+      "epoch": 287.5657894736842,
+      "grad_norm": 1.2503217458724976,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 43710
+    },
+    {
+      "epoch": 287.63157894736844,
+      "grad_norm": 1.2108162641525269,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 43720
+    },
+    {
+      "epoch": 287.69736842105266,
+      "grad_norm": 1.3603051900863647,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 43730
+    },
+    {
+      "epoch": 287.7631578947368,
+      "grad_norm": 1.156879186630249,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 43740
+    },
+    {
+      "epoch": 287.82894736842104,
+      "grad_norm": 1.0096931457519531,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 43750
+    },
+    {
+      "epoch": 287.89473684210526,
+      "grad_norm": 1.3088841438293457,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 43760
+    },
+    {
+      "epoch": 287.9605263157895,
+      "grad_norm": 1.609463095664978,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 43770
+    },
+    {
+      "epoch": 288.0263157894737,
+      "grad_norm": 1.2793583869934082,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 43780
+    },
+    {
+      "epoch": 288.0921052631579,
+      "grad_norm": 1.3340468406677246,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 43790
+    },
+    {
+      "epoch": 288.1578947368421,
+      "grad_norm": 1.108888030052185,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 43800
+    },
+    {
+      "epoch": 288.2236842105263,
+      "grad_norm": 1.1814451217651367,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 43810
+    },
+    {
+      "epoch": 288.2894736842105,
+      "grad_norm": 1.0461862087249756,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 43820
+    },
+    {
+      "epoch": 288.35526315789474,
+      "grad_norm": 1.1520538330078125,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 43830
+    },
+    {
+      "epoch": 288.42105263157896,
+      "grad_norm": 1.4166773557662964,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 43840
+    },
+    {
+      "epoch": 288.4868421052632,
+      "grad_norm": 1.0651955604553223,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 43850
+    },
+    {
+      "epoch": 288.55263157894734,
+      "grad_norm": 1.5009607076644897,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 43860
+    },
+    {
+      "epoch": 288.61842105263156,
+      "grad_norm": 1.2824265956878662,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 43870
+    },
+    {
+      "epoch": 288.6842105263158,
+      "grad_norm": 1.1175551414489746,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 43880
+    },
+    {
+      "epoch": 288.75,
+      "grad_norm": 1.3193415403366089,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 43890
+    },
+    {
+      "epoch": 288.8157894736842,
+      "grad_norm": 1.3196364641189575,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 43900
+    },
+    {
+      "epoch": 288.88157894736844,
+      "grad_norm": 1.0883409976959229,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 43910
+    },
+    {
+      "epoch": 288.94736842105266,
+      "grad_norm": 1.2597954273223877,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 43920
+    },
+    {
+      "epoch": 289.0131578947368,
+      "grad_norm": 1.32917058467865,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 43930
+    },
+    {
+      "epoch": 289.07894736842104,
+      "grad_norm": 1.2906899452209473,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 43940
+    },
+    {
+      "epoch": 289.14473684210526,
+      "grad_norm": 0.9421364068984985,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 43950
+    },
+    {
+      "epoch": 289.2105263157895,
+      "grad_norm": 0.7365257143974304,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 43960
+    },
+    {
+      "epoch": 289.2763157894737,
+      "grad_norm": 1.0555346012115479,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 43970
+    },
+    {
+      "epoch": 289.3421052631579,
+      "grad_norm": 1.3130031824111938,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 43980
+    },
+    {
+      "epoch": 289.4078947368421,
+      "grad_norm": 1.474686622619629,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 43990
+    },
+    {
+      "epoch": 289.4736842105263,
+      "grad_norm": 1.148776650428772,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 44000
+    },
+    {
+      "epoch": 289.5394736842105,
+      "grad_norm": 1.4202920198440552,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 44010
+    },
+    {
+      "epoch": 289.60526315789474,
+      "grad_norm": 1.267279863357544,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 44020
+    },
+    {
+      "epoch": 289.67105263157896,
+      "grad_norm": 1.1646653413772583,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 44030
+    },
+    {
+      "epoch": 289.7368421052632,
+      "grad_norm": 1.241349458694458,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 44040
+    },
+    {
+      "epoch": 289.80263157894734,
+      "grad_norm": 1.201326847076416,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 44050
+    },
+    {
+      "epoch": 289.86842105263156,
+      "grad_norm": 1.1308850049972534,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 44060
+    },
+    {
+      "epoch": 289.9342105263158,
+      "grad_norm": 1.6807210445404053,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 44070
+    },
+    {
+      "epoch": 290.0,
+      "grad_norm": 1.081008791923523,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 44080
+    },
+    {
+      "epoch": 290.0657894736842,
+      "grad_norm": 1.1021829843521118,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 44090
+    },
+    {
+      "epoch": 290.13157894736844,
+      "grad_norm": 1.4114763736724854,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 44100
+    },
+    {
+      "epoch": 290.19736842105266,
+      "grad_norm": 1.2354848384857178,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 44110
+    },
+    {
+      "epoch": 290.2631578947368,
+      "grad_norm": 1.4642871618270874,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 44120
+    },
+    {
+      "epoch": 290.32894736842104,
+      "grad_norm": 0.9256936311721802,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 44130
+    },
+    {
+      "epoch": 290.39473684210526,
+      "grad_norm": 1.1769204139709473,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 44140
+    },
+    {
+      "epoch": 290.4605263157895,
+      "grad_norm": 1.3560692071914673,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 44150
+    },
+    {
+      "epoch": 290.5263157894737,
+      "grad_norm": 1.3157851696014404,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 44160
+    },
+    {
+      "epoch": 290.5921052631579,
+      "grad_norm": 1.5221748352050781,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 44170
+    },
+    {
+      "epoch": 290.6578947368421,
+      "grad_norm": 1.3193734884262085,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 44180
+    },
+    {
+      "epoch": 290.7236842105263,
+      "grad_norm": 1.5841879844665527,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 44190
+    },
+    {
+      "epoch": 290.7894736842105,
+      "grad_norm": 1.5025423765182495,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 44200
+    },
+    {
+      "epoch": 290.85526315789474,
+      "grad_norm": 1.6240133047103882,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 44210
+    },
+    {
+      "epoch": 290.92105263157896,
+      "grad_norm": 1.4948194026947021,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 44220
+    },
+    {
+      "epoch": 290.9868421052632,
+      "grad_norm": 1.317069411277771,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 44230
+    },
+    {
+      "epoch": 291.05263157894734,
+      "grad_norm": 1.513387680053711,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 44240
+    },
+    {
+      "epoch": 291.11842105263156,
+      "grad_norm": 1.4927419424057007,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 44250
+    },
+    {
+      "epoch": 291.1842105263158,
+      "grad_norm": 1.4577031135559082,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 44260
+    },
+    {
+      "epoch": 291.25,
+      "grad_norm": 1.673728585243225,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 44270
+    },
+    {
+      "epoch": 291.3157894736842,
+      "grad_norm": 0.9547315835952759,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 44280
+    },
+    {
+      "epoch": 291.38157894736844,
+      "grad_norm": 0.90570068359375,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 44290
+    },
+    {
+      "epoch": 291.44736842105266,
+      "grad_norm": 1.1997649669647217,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 44300
+    },
+    {
+      "epoch": 291.5131578947368,
+      "grad_norm": 1.1620087623596191,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 44310
+    },
+    {
+      "epoch": 291.57894736842104,
+      "grad_norm": 1.410526990890503,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 44320
+    },
+    {
+      "epoch": 291.64473684210526,
+      "grad_norm": 1.3183605670928955,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 44330
+    },
+    {
+      "epoch": 291.7105263157895,
+      "grad_norm": 1.7681739330291748,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 44340
+    },
+    {
+      "epoch": 291.7763157894737,
+      "grad_norm": 1.3306128978729248,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 44350
+    },
+    {
+      "epoch": 291.8421052631579,
+      "grad_norm": 1.658753514289856,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 44360
+    },
+    {
+      "epoch": 291.9078947368421,
+      "grad_norm": 1.2005608081817627,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 44370
+    },
+    {
+      "epoch": 291.9736842105263,
+      "grad_norm": 1.5707415342330933,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 44380
+    },
+    {
+      "epoch": 292.0394736842105,
+      "grad_norm": 1.211821436882019,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 44390
+    },
+    {
+      "epoch": 292.10526315789474,
+      "grad_norm": 1.5584156513214111,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 44400
+    },
+    {
+      "epoch": 292.17105263157896,
+      "grad_norm": 1.2665433883666992,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 44410
+    },
+    {
+      "epoch": 292.2368421052632,
+      "grad_norm": 0.9688684344291687,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 44420
+    },
+    {
+      "epoch": 292.30263157894734,
+      "grad_norm": 1.0552864074707031,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 44430
+    },
+    {
+      "epoch": 292.36842105263156,
+      "grad_norm": 1.3883039951324463,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 44440
+    },
+    {
+      "epoch": 292.4342105263158,
+      "grad_norm": 1.1752902269363403,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 44450
+    },
+    {
+      "epoch": 292.5,
+      "grad_norm": 1.1989336013793945,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 44460
+    },
+    {
+      "epoch": 292.5657894736842,
+      "grad_norm": 1.2750636339187622,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 44470
+    },
+    {
+      "epoch": 292.63157894736844,
+      "grad_norm": 1.1853424310684204,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 44480
+    },
+    {
+      "epoch": 292.69736842105266,
+      "grad_norm": 0.9358903765678406,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 44490
+    },
+    {
+      "epoch": 292.7631578947368,
+      "grad_norm": 1.2150110006332397,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 44500
+    },
+    {
+      "epoch": 292.82894736842104,
+      "grad_norm": 1.1464403867721558,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 44510
+    },
+    {
+      "epoch": 292.89473684210526,
+      "grad_norm": 1.5787771940231323,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 44520
+    },
+    {
+      "epoch": 292.9605263157895,
+      "grad_norm": 1.234574794769287,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 44530
+    },
+    {
+      "epoch": 293.0263157894737,
+      "grad_norm": 1.4504281282424927,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 44540
+    },
+    {
+      "epoch": 293.0921052631579,
+      "grad_norm": 1.2790708541870117,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 44550
+    },
+    {
+      "epoch": 293.1578947368421,
+      "grad_norm": 1.0297752618789673,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 44560
+    },
+    {
+      "epoch": 293.2236842105263,
+      "grad_norm": 1.4982763528823853,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 44570
+    },
+    {
+      "epoch": 293.2894736842105,
+      "grad_norm": 1.4405145645141602,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 44580
+    },
+    {
+      "epoch": 293.35526315789474,
+      "grad_norm": 1.618911862373352,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 44590
+    },
+    {
+      "epoch": 293.42105263157896,
+      "grad_norm": 1.1812571287155151,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 44600
+    },
+    {
+      "epoch": 293.4868421052632,
+      "grad_norm": 1.3874218463897705,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 44610
+    },
+    {
+      "epoch": 293.55263157894734,
+      "grad_norm": 1.1860712766647339,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 44620
+    },
+    {
+      "epoch": 293.61842105263156,
+      "grad_norm": 1.5251539945602417,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 44630
+    },
+    {
+      "epoch": 293.6842105263158,
+      "grad_norm": 1.2755794525146484,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 44640
+    },
+    {
+      "epoch": 293.75,
+      "grad_norm": 1.8280200958251953,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 44650
+    },
+    {
+      "epoch": 293.8157894736842,
+      "grad_norm": 1.7811310291290283,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 44660
+    },
+    {
+      "epoch": 293.88157894736844,
+      "grad_norm": 1.8284488916397095,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 44670
+    },
+    {
+      "epoch": 293.94736842105266,
+      "grad_norm": 1.3257440328598022,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 44680
+    },
+    {
+      "epoch": 294.0131578947368,
+      "grad_norm": 1.3897215127944946,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 44690
+    },
+    {
+      "epoch": 294.07894736842104,
+      "grad_norm": 1.2009508609771729,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 44700
+    },
+    {
+      "epoch": 294.14473684210526,
+      "grad_norm": 1.5008352994918823,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 44710
+    },
+    {
+      "epoch": 294.2105263157895,
+      "grad_norm": 1.0818425416946411,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 44720
+    },
+    {
+      "epoch": 294.2763157894737,
+      "grad_norm": 1.3108577728271484,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 44730
+    },
+    {
+      "epoch": 294.3421052631579,
+      "grad_norm": 1.39351224899292,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 44740
+    },
+    {
+      "epoch": 294.4078947368421,
+      "grad_norm": 1.2173326015472412,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 44750
+    },
+    {
+      "epoch": 294.4736842105263,
+      "grad_norm": 1.1741549968719482,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 44760
+    },
+    {
+      "epoch": 294.5394736842105,
+      "grad_norm": 1.123326301574707,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 44770
+    },
+    {
+      "epoch": 294.60526315789474,
+      "grad_norm": 1.1277564764022827,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 44780
+    },
+    {
+      "epoch": 294.67105263157896,
+      "grad_norm": 0.9613653421401978,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 44790
+    },
+    {
+      "epoch": 294.7368421052632,
+      "grad_norm": 0.9356263279914856,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 44800
+    },
+    {
+      "epoch": 294.80263157894734,
+      "grad_norm": 0.9894914031028748,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 44810
+    },
+    {
+      "epoch": 294.86842105263156,
+      "grad_norm": 1.1962043046951294,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 44820
+    },
+    {
+      "epoch": 294.9342105263158,
+      "grad_norm": 1.2527235746383667,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 44830
+    },
+    {
+      "epoch": 295.0,
+      "grad_norm": 1.2737168073654175,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 44840
+    },
+    {
+      "epoch": 295.0657894736842,
+      "grad_norm": 1.6003625392913818,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 44850
+    },
+    {
+      "epoch": 295.13157894736844,
+      "grad_norm": 1.2156645059585571,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 44860
+    },
+    {
+      "epoch": 295.19736842105266,
+      "grad_norm": 1.3037317991256714,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 44870
+    },
+    {
+      "epoch": 295.2631578947368,
+      "grad_norm": 1.2797876596450806,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 44880
+    },
+    {
+      "epoch": 295.32894736842104,
+      "grad_norm": 0.9522796273231506,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 44890
+    },
+    {
+      "epoch": 295.39473684210526,
+      "grad_norm": 0.9512258172035217,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 44900
+    },
+    {
+      "epoch": 295.4605263157895,
+      "grad_norm": 0.9569043517112732,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 44910
+    },
+    {
+      "epoch": 295.5263157894737,
+      "grad_norm": 0.8486437797546387,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 44920
+    },
+    {
+      "epoch": 295.5921052631579,
+      "grad_norm": 0.7993052005767822,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 44930
+    },
+    {
+      "epoch": 295.6578947368421,
+      "grad_norm": 0.8292239308357239,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 44940
+    },
+    {
+      "epoch": 295.7236842105263,
+      "grad_norm": 0.9200776219367981,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 44950
+    },
+    {
+      "epoch": 295.7894736842105,
+      "grad_norm": 1.134240984916687,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 44960
+    },
+    {
+      "epoch": 295.85526315789474,
+      "grad_norm": 1.2464627027511597,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 44970
+    },
+    {
+      "epoch": 295.92105263157896,
+      "grad_norm": 1.2314313650131226,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 44980
+    },
+    {
+      "epoch": 295.9868421052632,
+      "grad_norm": 1.2230592966079712,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 44990
+    },
+    {
+      "epoch": 296.05263157894734,
+      "grad_norm": 1.0356382131576538,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 45000
+    },
+    {
+      "epoch": 296.11842105263156,
+      "grad_norm": 1.4570183753967285,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 45010
+    },
+    {
+      "epoch": 296.1842105263158,
+      "grad_norm": 0.9928843975067139,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 45020
+    },
+    {
+      "epoch": 296.25,
+      "grad_norm": 1.1260870695114136,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 45030
+    },
+    {
+      "epoch": 296.3157894736842,
+      "grad_norm": 1.1443604230880737,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 45040
+    },
+    {
+      "epoch": 296.38157894736844,
+      "grad_norm": 1.091615915298462,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 45050
+    },
+    {
+      "epoch": 296.44736842105266,
+      "grad_norm": 0.9866099953651428,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45060
+    },
+    {
+      "epoch": 296.5131578947368,
+      "grad_norm": 1.350821852684021,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 45070
+    },
+    {
+      "epoch": 296.57894736842104,
+      "grad_norm": 0.8730394244194031,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 45080
+    },
+    {
+      "epoch": 296.64473684210526,
+      "grad_norm": 1.3271095752716064,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 45090
+    },
+    {
+      "epoch": 296.7105263157895,
+      "grad_norm": 1.1160370111465454,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 45100
+    },
+    {
+      "epoch": 296.7763157894737,
+      "grad_norm": 1.0997018814086914,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 45110
+    },
+    {
+      "epoch": 296.8421052631579,
+      "grad_norm": 1.1898871660232544,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 45120
+    },
+    {
+      "epoch": 296.9078947368421,
+      "grad_norm": 0.9266688227653503,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 45130
+    },
+    {
+      "epoch": 296.9736842105263,
+      "grad_norm": 1.0709036588668823,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 45140
+    },
+    {
+      "epoch": 297.0394736842105,
+      "grad_norm": 1.0864067077636719,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 45150
+    },
+    {
+      "epoch": 297.10526315789474,
+      "grad_norm": 0.7684114575386047,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 45160
+    },
+    {
+      "epoch": 297.17105263157896,
+      "grad_norm": 1.0278398990631104,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 45170
+    },
+    {
+      "epoch": 297.2368421052632,
+      "grad_norm": 1.1107127666473389,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 45180
+    },
+    {
+      "epoch": 297.30263157894734,
+      "grad_norm": 0.9575175046920776,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 45190
+    },
+    {
+      "epoch": 297.36842105263156,
+      "grad_norm": 1.2704232931137085,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 45200
+    },
+    {
+      "epoch": 297.4342105263158,
+      "grad_norm": 1.383664608001709,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 45210
+    },
+    {
+      "epoch": 297.5,
+      "grad_norm": 0.8386159539222717,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 45220
+    },
+    {
+      "epoch": 297.5657894736842,
+      "grad_norm": 1.1953688859939575,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 45230
+    },
+    {
+      "epoch": 297.63157894736844,
+      "grad_norm": 1.0901776552200317,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 45240
+    },
+    {
+      "epoch": 297.69736842105266,
+      "grad_norm": 1.1900759935379028,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 45250
+    },
+    {
+      "epoch": 297.7631578947368,
+      "grad_norm": 0.9617878794670105,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 45260
+    },
+    {
+      "epoch": 297.82894736842104,
+      "grad_norm": 1.275898814201355,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 45270
+    },
+    {
+      "epoch": 297.89473684210526,
+      "grad_norm": 1.2279614210128784,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 45280
+    },
+    {
+      "epoch": 297.9605263157895,
+      "grad_norm": 1.4091862440109253,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 45290
+    },
+    {
+      "epoch": 298.0263157894737,
+      "grad_norm": 1.1640609502792358,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 45300
+    },
+    {
+      "epoch": 298.0921052631579,
+      "grad_norm": 1.000942587852478,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45310
+    },
+    {
+      "epoch": 298.1578947368421,
+      "grad_norm": 0.9416703581809998,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 45320
+    },
+    {
+      "epoch": 298.2236842105263,
+      "grad_norm": 1.3636903762817383,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 45330
+    },
+    {
+      "epoch": 298.2894736842105,
+      "grad_norm": 0.9758198261260986,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 45340
+    },
+    {
+      "epoch": 298.35526315789474,
+      "grad_norm": 1.266970157623291,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 45350
+    },
+    {
+      "epoch": 298.42105263157896,
+      "grad_norm": 1.5186119079589844,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 45360
+    },
+    {
+      "epoch": 298.4868421052632,
+      "grad_norm": 1.2090457677841187,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 45370
+    },
+    {
+      "epoch": 298.55263157894734,
+      "grad_norm": 1.673222541809082,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 45380
+    },
+    {
+      "epoch": 298.61842105263156,
+      "grad_norm": 1.3236820697784424,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 45390
+    },
+    {
+      "epoch": 298.6842105263158,
+      "grad_norm": 1.2713639736175537,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 45400
+    },
+    {
+      "epoch": 298.75,
+      "grad_norm": 1.2878004312515259,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 45410
+    },
+    {
+      "epoch": 298.8157894736842,
+      "grad_norm": 1.013006567955017,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 45420
+    },
+    {
+      "epoch": 298.88157894736844,
+      "grad_norm": 1.3307374715805054,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 45430
+    },
+    {
+      "epoch": 298.94736842105266,
+      "grad_norm": 1.2865517139434814,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45440
+    },
+    {
+      "epoch": 299.0131578947368,
+      "grad_norm": 1.2010281085968018,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 45450
+    },
+    {
+      "epoch": 299.07894736842104,
+      "grad_norm": 1.3164724111557007,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 45460
+    },
+    {
+      "epoch": 299.14473684210526,
+      "grad_norm": 1.2685626745224,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 45470
+    },
+    {
+      "epoch": 299.2105263157895,
+      "grad_norm": 1.0890898704528809,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 45480
+    },
+    {
+      "epoch": 299.2763157894737,
+      "grad_norm": 1.3427873849868774,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 45490
+    },
+    {
+      "epoch": 299.3421052631579,
+      "grad_norm": 1.1876996755599976,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 45500
+    },
+    {
+      "epoch": 299.4078947368421,
+      "grad_norm": 1.2783021926879883,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 45510
+    },
+    {
+      "epoch": 299.4736842105263,
+      "grad_norm": 1.2588651180267334,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45520
+    },
+    {
+      "epoch": 299.5394736842105,
+      "grad_norm": 1.005475640296936,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 45530
+    },
+    {
+      "epoch": 299.60526315789474,
+      "grad_norm": 1.3194289207458496,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 45540
+    },
+    {
+      "epoch": 299.67105263157896,
+      "grad_norm": 1.593381643295288,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 45550
+    },
+    {
+      "epoch": 299.7368421052632,
+      "grad_norm": 1.1744670867919922,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 45560
+    },
+    {
+      "epoch": 299.80263157894734,
+      "grad_norm": 1.151694893836975,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 45570
+    },
+    {
+      "epoch": 299.86842105263156,
+      "grad_norm": 0.9926699995994568,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 45580
+    },
+    {
+      "epoch": 299.9342105263158,
+      "grad_norm": 1.3312798738479614,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 45590
+    },
+    {
+      "epoch": 300.0,
+      "grad_norm": 1.242369294166565,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 45600
+    },
+    {
+      "epoch": 300.0657894736842,
+      "grad_norm": 1.139743447303772,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 45610
+    },
+    {
+      "epoch": 300.13157894736844,
+      "grad_norm": 1.255859375,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 45620
+    },
+    {
+      "epoch": 300.19736842105266,
+      "grad_norm": 1.1239902973175049,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 45630
+    },
+    {
+      "epoch": 300.2631578947368,
+      "grad_norm": 1.4652752876281738,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 45640
+    },
+    {
+      "epoch": 300.32894736842104,
+      "grad_norm": 1.098900556564331,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 45650
+    },
+    {
+      "epoch": 300.39473684210526,
+      "grad_norm": 0.6829715371131897,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45660
+    },
+    {
+      "epoch": 300.4605263157895,
+      "grad_norm": 1.10642409324646,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45670
+    },
+    {
+      "epoch": 300.5263157894737,
+      "grad_norm": 0.9976691007614136,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45680
+    },
+    {
+      "epoch": 300.5921052631579,
+      "grad_norm": 1.3558682203292847,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 45690
+    },
+    {
+      "epoch": 300.6578947368421,
+      "grad_norm": 1.131054162979126,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 45700
+    },
+    {
+      "epoch": 300.7236842105263,
+      "grad_norm": 1.3741534948349,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 45710
+    },
+    {
+      "epoch": 300.7894736842105,
+      "grad_norm": 1.3326232433319092,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 45720
+    },
+    {
+      "epoch": 300.85526315789474,
+      "grad_norm": 1.5174059867858887,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 45730
+    },
+    {
+      "epoch": 300.92105263157896,
+      "grad_norm": 1.343550443649292,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 45740
+    },
+    {
+      "epoch": 300.9868421052632,
+      "grad_norm": 1.3771088123321533,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 45750
+    },
+    {
+      "epoch": 301.05263157894734,
+      "grad_norm": 1.3173880577087402,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 45760
+    },
+    {
+      "epoch": 301.11842105263156,
+      "grad_norm": 1.4637528657913208,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 45770
+    },
+    {
+      "epoch": 301.1842105263158,
+      "grad_norm": 1.0429658889770508,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 45780
+    },
+    {
+      "epoch": 301.25,
+      "grad_norm": 1.3570419549942017,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 45790
+    },
+    {
+      "epoch": 301.3157894736842,
+      "grad_norm": 1.0001505613327026,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 45800
+    },
+    {
+      "epoch": 301.38157894736844,
+      "grad_norm": 1.192205786705017,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45810
+    },
+    {
+      "epoch": 301.44736842105266,
+      "grad_norm": 1.4170939922332764,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 45820
+    },
+    {
+      "epoch": 301.5131578947368,
+      "grad_norm": 1.2274196147918701,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45830
+    },
+    {
+      "epoch": 301.57894736842104,
+      "grad_norm": 1.2040157318115234,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 45840
+    },
+    {
+      "epoch": 301.64473684210526,
+      "grad_norm": 1.2414194345474243,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 45850
+    },
+    {
+      "epoch": 301.7105263157895,
+      "grad_norm": 1.185573935508728,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 45860
+    },
+    {
+      "epoch": 301.7763157894737,
+      "grad_norm": 1.1964563131332397,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 45870
+    },
+    {
+      "epoch": 301.8421052631579,
+      "grad_norm": 1.363934874534607,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 45880
+    },
+    {
+      "epoch": 301.9078947368421,
+      "grad_norm": 1.0584746599197388,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 45890
+    },
+    {
+      "epoch": 301.9736842105263,
+      "grad_norm": 1.0964704751968384,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 45900
+    },
+    {
+      "epoch": 302.0394736842105,
+      "grad_norm": 0.9291341304779053,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45910
+    },
+    {
+      "epoch": 302.10526315789474,
+      "grad_norm": 1.3490403890609741,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 45920
+    },
+    {
+      "epoch": 302.17105263157896,
+      "grad_norm": 1.4308061599731445,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 45930
+    },
+    {
+      "epoch": 302.2368421052632,
+      "grad_norm": 1.307144045829773,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 45940
+    },
+    {
+      "epoch": 302.30263157894734,
+      "grad_norm": 1.7374738454818726,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 45950
+    },
+    {
+      "epoch": 302.36842105263156,
+      "grad_norm": 1.5400711297988892,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 45960
+    },
+    {
+      "epoch": 302.4342105263158,
+      "grad_norm": 1.6926573514938354,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 45970
+    },
+    {
+      "epoch": 302.5,
+      "grad_norm": 1.3912023305892944,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 45980
+    },
+    {
+      "epoch": 302.5657894736842,
+      "grad_norm": 1.3487814664840698,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 45990
+    },
+    {
+      "epoch": 302.63157894736844,
+      "grad_norm": 1.3603285551071167,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 46000
+    },
+    {
+      "epoch": 302.69736842105266,
+      "grad_norm": 1.107670545578003,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 46010
+    },
+    {
+      "epoch": 302.7631578947368,
+      "grad_norm": 1.2520265579223633,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 46020
+    },
+    {
+      "epoch": 302.82894736842104,
+      "grad_norm": 1.2443839311599731,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 46030
+    },
+    {
+      "epoch": 302.89473684210526,
+      "grad_norm": 1.4623485803604126,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 46040
+    },
+    {
+      "epoch": 302.9605263157895,
+      "grad_norm": 1.1879698038101196,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 46050
+    },
+    {
+      "epoch": 303.0263157894737,
+      "grad_norm": 1.3571995496749878,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 46060
+    },
+    {
+      "epoch": 303.0921052631579,
+      "grad_norm": 1.199619174003601,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 46070
+    },
+    {
+      "epoch": 303.1578947368421,
+      "grad_norm": 1.4226540327072144,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 46080
+    },
+    {
+      "epoch": 303.2236842105263,
+      "grad_norm": 1.490116834640503,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 46090
+    },
+    {
+      "epoch": 303.2894736842105,
+      "grad_norm": 1.325772762298584,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 46100
+    },
+    {
+      "epoch": 303.35526315789474,
+      "grad_norm": 1.2829171419143677,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 46110
+    },
+    {
+      "epoch": 303.42105263157896,
+      "grad_norm": 1.164736032485962,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 46120
+    },
+    {
+      "epoch": 303.4868421052632,
+      "grad_norm": 1.147537350654602,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 46130
+    },
+    {
+      "epoch": 303.55263157894734,
+      "grad_norm": 0.7918049693107605,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 46140
+    },
+    {
+      "epoch": 303.61842105263156,
+      "grad_norm": 1.1674093008041382,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 46150
+    },
+    {
+      "epoch": 303.6842105263158,
+      "grad_norm": 1.2786973714828491,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 46160
+    },
+    {
+      "epoch": 303.75,
+      "grad_norm": 1.233464241027832,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 46170
+    },
+    {
+      "epoch": 303.8157894736842,
+      "grad_norm": 1.2651668787002563,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 46180
+    },
+    {
+      "epoch": 303.88157894736844,
+      "grad_norm": 1.4438832998275757,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 46190
+    },
+    {
+      "epoch": 303.94736842105266,
+      "grad_norm": 1.30561363697052,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 46200
+    },
+    {
+      "epoch": 304.0131578947368,
+      "grad_norm": 1.2133570909500122,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 46210
+    },
+    {
+      "epoch": 304.07894736842104,
+      "grad_norm": 1.1664997339248657,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 46220
+    },
+    {
+      "epoch": 304.14473684210526,
+      "grad_norm": 0.8901975750923157,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 46230
+    },
+    {
+      "epoch": 304.2105263157895,
+      "grad_norm": 0.9484691619873047,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 46240
+    },
+    {
+      "epoch": 304.2763157894737,
+      "grad_norm": 1.0902068614959717,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 46250
+    },
+    {
+      "epoch": 304.3421052631579,
+      "grad_norm": 0.908757209777832,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 46260
+    },
+    {
+      "epoch": 304.4078947368421,
+      "grad_norm": 1.406864047050476,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 46270
+    },
+    {
+      "epoch": 304.4736842105263,
+      "grad_norm": 1.625902533531189,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 46280
+    },
+    {
+      "epoch": 304.5394736842105,
+      "grad_norm": 1.1790637969970703,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 46290
+    },
+    {
+      "epoch": 304.60526315789474,
+      "grad_norm": 1.3175780773162842,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 46300
+    },
+    {
+      "epoch": 304.67105263157896,
+      "grad_norm": 1.1840614080429077,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 46310
+    },
+    {
+      "epoch": 304.7368421052632,
+      "grad_norm": 1.3935436010360718,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 46320
+    },
+    {
+      "epoch": 304.80263157894734,
+      "grad_norm": 1.0257351398468018,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 46330
+    },
+    {
+      "epoch": 304.86842105263156,
+      "grad_norm": 0.9773437976837158,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 46340
+    },
+    {
+      "epoch": 304.9342105263158,
+      "grad_norm": 1.1705121994018555,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 46350
+    },
+    {
+      "epoch": 305.0,
+      "grad_norm": 1.569278359413147,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 46360
+    },
+    {
+      "epoch": 305.0657894736842,
+      "grad_norm": 1.1362944841384888,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 46370
+    },
+    {
+      "epoch": 305.13157894736844,
+      "grad_norm": 0.9014955759048462,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 46380
+    },
+    {
+      "epoch": 305.19736842105266,
+      "grad_norm": 1.377353549003601,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 46390
+    },
+    {
+      "epoch": 305.2631578947368,
+      "grad_norm": 1.3454692363739014,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 46400
+    },
+    {
+      "epoch": 305.32894736842104,
+      "grad_norm": 1.1798051595687866,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 46410
+    },
+    {
+      "epoch": 305.39473684210526,
+      "grad_norm": 1.0870208740234375,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 46420
+    },
+    {
+      "epoch": 305.4605263157895,
+      "grad_norm": 0.8006429672241211,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 46430
+    },
+    {
+      "epoch": 305.5263157894737,
+      "grad_norm": 0.890498161315918,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 46440
+    },
+    {
+      "epoch": 305.5921052631579,
+      "grad_norm": 1.2902061939239502,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 46450
+    },
+    {
+      "epoch": 305.6578947368421,
+      "grad_norm": 1.146141529083252,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 46460
+    },
+    {
+      "epoch": 305.7236842105263,
+      "grad_norm": 0.9566381573677063,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 46470
+    },
+    {
+      "epoch": 305.7894736842105,
+      "grad_norm": 1.5867373943328857,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 46480
+    },
+    {
+      "epoch": 305.85526315789474,
+      "grad_norm": 1.2137548923492432,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 46490
+    },
+    {
+      "epoch": 305.92105263157896,
+      "grad_norm": 1.305253505706787,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 46500
+    },
+    {
+      "epoch": 305.9868421052632,
+      "grad_norm": 0.8301891088485718,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 46510
+    },
+    {
+      "epoch": 306.05263157894734,
+      "grad_norm": 0.9847671389579773,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 46520
+    },
+    {
+      "epoch": 306.11842105263156,
+      "grad_norm": 1.592354655265808,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 46530
+    },
+    {
+      "epoch": 306.1842105263158,
+      "grad_norm": 1.2078450918197632,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 46540
+    },
+    {
+      "epoch": 306.25,
+      "grad_norm": 1.000321865081787,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 46550
+    },
+    {
+      "epoch": 306.3157894736842,
+      "grad_norm": 1.207289695739746,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 46560
+    },
+    {
+      "epoch": 306.38157894736844,
+      "grad_norm": 1.2162929773330688,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 46570
+    },
+    {
+      "epoch": 306.44736842105266,
+      "grad_norm": 1.147230863571167,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 46580
+    },
+    {
+      "epoch": 306.5131578947368,
+      "grad_norm": 1.2170518636703491,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 46590
+    },
+    {
+      "epoch": 306.57894736842104,
+      "grad_norm": 1.1582554578781128,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 46600
+    },
+    {
+      "epoch": 306.64473684210526,
+      "grad_norm": 1.2563729286193848,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 46610
+    },
+    {
+      "epoch": 306.7105263157895,
+      "grad_norm": 0.9959396123886108,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 46620
+    },
+    {
+      "epoch": 306.7763157894737,
+      "grad_norm": 1.2448713779449463,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 46630
+    },
+    {
+      "epoch": 306.8421052631579,
+      "grad_norm": 1.0271637439727783,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 46640
+    },
+    {
+      "epoch": 306.9078947368421,
+      "grad_norm": 1.0369125604629517,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 46650
+    },
+    {
+      "epoch": 306.9736842105263,
+      "grad_norm": 1.0116463899612427,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 46660
+    },
+    {
+      "epoch": 307.0394736842105,
+      "grad_norm": 1.3489338159561157,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 46670
+    },
+    {
+      "epoch": 307.10526315789474,
+      "grad_norm": 1.214111328125,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 46680
+    },
+    {
+      "epoch": 307.17105263157896,
+      "grad_norm": 1.4091652631759644,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 46690
+    },
+    {
+      "epoch": 307.2368421052632,
+      "grad_norm": 1.3401622772216797,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 46700
+    },
+    {
+      "epoch": 307.30263157894734,
+      "grad_norm": 1.0717833042144775,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 46710
+    },
+    {
+      "epoch": 307.36842105263156,
+      "grad_norm": 1.215564489364624,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 46720
+    },
+    {
+      "epoch": 307.4342105263158,
+      "grad_norm": 1.126434326171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 46730
+    },
+    {
+      "epoch": 307.5,
+      "grad_norm": 1.1707587242126465,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 46740
+    },
+    {
+      "epoch": 307.5657894736842,
+      "grad_norm": 1.0138065814971924,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 46750
+    },
+    {
+      "epoch": 307.63157894736844,
+      "grad_norm": 0.8410609364509583,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 46760
+    },
+    {
+      "epoch": 307.69736842105266,
+      "grad_norm": 1.0997475385665894,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 46770
+    },
+    {
+      "epoch": 307.7631578947368,
+      "grad_norm": 1.0100196599960327,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 46780
+    },
+    {
+      "epoch": 307.82894736842104,
+      "grad_norm": 1.0084043741226196,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 46790
+    },
+    {
+      "epoch": 307.89473684210526,
+      "grad_norm": 1.1131081581115723,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 46800
+    },
+    {
+      "epoch": 307.9605263157895,
+      "grad_norm": 0.966968297958374,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 46810
+    },
+    {
+      "epoch": 308.0263157894737,
+      "grad_norm": 0.7994092106819153,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 46820
+    },
+    {
+      "epoch": 308.0921052631579,
+      "grad_norm": 1.2726207971572876,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 46830
+    },
+    {
+      "epoch": 308.1578947368421,
+      "grad_norm": 1.2597211599349976,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 46840
+    },
+    {
+      "epoch": 308.2236842105263,
+      "grad_norm": 1.6768962144851685,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 46850
+    },
+    {
+      "epoch": 308.2894736842105,
+      "grad_norm": 1.068867564201355,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 46860
+    },
+    {
+      "epoch": 308.35526315789474,
+      "grad_norm": 1.2445541620254517,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 46870
+    },
+    {
+      "epoch": 308.42105263157896,
+      "grad_norm": 1.1465497016906738,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 46880
+    },
+    {
+      "epoch": 308.4868421052632,
+      "grad_norm": 1.4472042322158813,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 46890
+    },
+    {
+      "epoch": 308.55263157894734,
+      "grad_norm": 1.262213945388794,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 46900
+    },
+    {
+      "epoch": 308.61842105263156,
+      "grad_norm": 1.1717816591262817,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 46910
+    },
+    {
+      "epoch": 308.6842105263158,
+      "grad_norm": 0.995729923248291,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 46920
+    },
+    {
+      "epoch": 308.75,
+      "grad_norm": 1.394285798072815,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 46930
+    },
+    {
+      "epoch": 308.8157894736842,
+      "grad_norm": 1.8521760702133179,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 46940
+    },
+    {
+      "epoch": 308.88157894736844,
+      "grad_norm": 1.5286895036697388,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 46950
+    },
+    {
+      "epoch": 308.94736842105266,
+      "grad_norm": 0.9966525435447693,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 46960
+    },
+    {
+      "epoch": 309.0131578947368,
+      "grad_norm": 1.0583655834197998,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 46970
+    },
+    {
+      "epoch": 309.07894736842104,
+      "grad_norm": 1.0200626850128174,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 46980
+    },
+    {
+      "epoch": 309.14473684210526,
+      "grad_norm": 1.2817143201828003,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 46990
+    },
+    {
+      "epoch": 309.2105263157895,
+      "grad_norm": 1.4597275257110596,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 47000
+    },
+    {
+      "epoch": 309.2763157894737,
+      "grad_norm": 1.026604413986206,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 47010
+    },
+    {
+      "epoch": 309.3421052631579,
+      "grad_norm": 1.1650899648666382,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 47020
+    },
+    {
+      "epoch": 309.4078947368421,
+      "grad_norm": 1.0153883695602417,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 47030
+    },
+    {
+      "epoch": 309.4736842105263,
+      "grad_norm": 1.2171595096588135,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 47040
+    },
+    {
+      "epoch": 309.5394736842105,
+      "grad_norm": 1.2337496280670166,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 47050
+    },
+    {
+      "epoch": 309.60526315789474,
+      "grad_norm": 1.2742234468460083,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 47060
+    },
+    {
+      "epoch": 309.67105263157896,
+      "grad_norm": 1.307140588760376,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 47070
+    },
+    {
+      "epoch": 309.7368421052632,
+      "grad_norm": 0.6933286190032959,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 47080
+    },
+    {
+      "epoch": 309.80263157894734,
+      "grad_norm": 1.448660135269165,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 47090
+    },
+    {
+      "epoch": 309.86842105263156,
+      "grad_norm": 1.2095212936401367,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 47100
+    },
+    {
+      "epoch": 309.9342105263158,
+      "grad_norm": 1.179511308670044,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 47110
+    },
+    {
+      "epoch": 310.0,
+      "grad_norm": 1.0663738250732422,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 47120
+    },
+    {
+      "epoch": 310.0657894736842,
+      "grad_norm": 1.395041823387146,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 47130
+    },
+    {
+      "epoch": 310.13157894736844,
+      "grad_norm": 1.5173497200012207,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 47140
+    },
+    {
+      "epoch": 310.19736842105266,
+      "grad_norm": 1.4677331447601318,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 47150
+    },
+    {
+      "epoch": 310.2631578947368,
+      "grad_norm": 1.263916015625,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 47160
+    },
+    {
+      "epoch": 310.32894736842104,
+      "grad_norm": 1.796497106552124,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 47170
+    },
+    {
+      "epoch": 310.39473684210526,
+      "grad_norm": 1.4208660125732422,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 47180
+    },
+    {
+      "epoch": 310.4605263157895,
+      "grad_norm": 1.3126176595687866,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 47190
+    },
+    {
+      "epoch": 310.5263157894737,
+      "grad_norm": 1.1969044208526611,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 47200
+    },
+    {
+      "epoch": 310.5921052631579,
+      "grad_norm": 1.3965052366256714,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 47210
+    },
+    {
+      "epoch": 310.6578947368421,
+      "grad_norm": 1.1293995380401611,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 47220
+    },
+    {
+      "epoch": 310.7236842105263,
+      "grad_norm": 1.213976263999939,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 47230
+    },
+    {
+      "epoch": 310.7894736842105,
+      "grad_norm": 1.2718894481658936,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 47240
+    },
+    {
+      "epoch": 310.85526315789474,
+      "grad_norm": 1.3019602298736572,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 47250
+    },
+    {
+      "epoch": 310.92105263157896,
+      "grad_norm": 1.1459777355194092,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 47260
+    },
+    {
+      "epoch": 310.9868421052632,
+      "grad_norm": 0.9818292856216431,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 47270
+    },
+    {
+      "epoch": 311.05263157894734,
+      "grad_norm": 1.271238088607788,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 47280
+    },
+    {
+      "epoch": 311.11842105263156,
+      "grad_norm": 1.5601221323013306,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 47290
+    },
+    {
+      "epoch": 311.1842105263158,
+      "grad_norm": 1.2876839637756348,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 47300
+    },
+    {
+      "epoch": 311.25,
+      "grad_norm": 1.2727855443954468,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 47310
+    },
+    {
+      "epoch": 311.3157894736842,
+      "grad_norm": 1.0672112703323364,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 47320
+    },
+    {
+      "epoch": 311.38157894736844,
+      "grad_norm": 1.3671296834945679,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 47330
+    },
+    {
+      "epoch": 311.44736842105266,
+      "grad_norm": 0.953478991985321,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 47340
+    },
+    {
+      "epoch": 311.5131578947368,
+      "grad_norm": 1.3494404554367065,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 47350
+    },
+    {
+      "epoch": 311.57894736842104,
+      "grad_norm": 1.256812572479248,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 47360
+    },
+    {
+      "epoch": 311.64473684210526,
+      "grad_norm": 1.1076347827911377,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 47370
+    },
+    {
+      "epoch": 311.7105263157895,
+      "grad_norm": 1.1264309883117676,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 47380
+    },
+    {
+      "epoch": 311.7763157894737,
+      "grad_norm": 1.082269310951233,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 47390
+    },
+    {
+      "epoch": 311.8421052631579,
+      "grad_norm": 1.2496287822723389,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 47400
+    },
+    {
+      "epoch": 311.9078947368421,
+      "grad_norm": 1.2073965072631836,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 47410
+    },
+    {
+      "epoch": 311.9736842105263,
+      "grad_norm": 1.4305161237716675,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 47420
+    },
+    {
+      "epoch": 312.0394736842105,
+      "grad_norm": 1.231873631477356,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 47430
+    },
+    {
+      "epoch": 312.10526315789474,
+      "grad_norm": 1.3474899530410767,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 47440
+    },
+    {
+      "epoch": 312.17105263157896,
+      "grad_norm": 1.301870346069336,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 47450
+    },
+    {
+      "epoch": 312.2368421052632,
+      "grad_norm": 1.3354082107543945,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 47460
+    },
+    {
+      "epoch": 312.30263157894734,
+      "grad_norm": 1.112229347229004,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 47470
+    },
+    {
+      "epoch": 312.36842105263156,
+      "grad_norm": 0.764408528804779,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 47480
+    },
+    {
+      "epoch": 312.4342105263158,
+      "grad_norm": 0.8579666018486023,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 47490
+    },
+    {
+      "epoch": 312.5,
+      "grad_norm": 0.7732484340667725,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 47500
+    },
+    {
+      "epoch": 312.5657894736842,
+      "grad_norm": 1.0402549505233765,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 47510
+    },
+    {
+      "epoch": 312.63157894736844,
+      "grad_norm": 0.876788318157196,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 47520
+    },
+    {
+      "epoch": 312.69736842105266,
+      "grad_norm": 0.8869333267211914,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 47530
+    },
+    {
+      "epoch": 312.7631578947368,
+      "grad_norm": 1.4594961404800415,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 47540
+    },
+    {
+      "epoch": 312.82894736842104,
+      "grad_norm": 1.2416234016418457,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 47550
+    },
+    {
+      "epoch": 312.89473684210526,
+      "grad_norm": 1.632511854171753,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 47560
+    },
+    {
+      "epoch": 312.9605263157895,
+      "grad_norm": 1.8223305940628052,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 47570
+    },
+    {
+      "epoch": 313.0263157894737,
+      "grad_norm": 1.4376325607299805,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 47580
+    },
+    {
+      "epoch": 313.0921052631579,
+      "grad_norm": 1.3378567695617676,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 47590
+    },
+    {
+      "epoch": 313.1578947368421,
+      "grad_norm": 1.1414756774902344,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 47600
+    },
+    {
+      "epoch": 313.2236842105263,
+      "grad_norm": 1.2612454891204834,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 47610
+    },
+    {
+      "epoch": 313.2894736842105,
+      "grad_norm": 1.2293567657470703,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 47620
+    },
+    {
+      "epoch": 313.35526315789474,
+      "grad_norm": 1.0517131090164185,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 47630
+    },
+    {
+      "epoch": 313.42105263157896,
+      "grad_norm": 1.222216248512268,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 47640
+    },
+    {
+      "epoch": 313.4868421052632,
+      "grad_norm": 0.9757271409034729,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 47650
+    },
+    {
+      "epoch": 313.55263157894734,
+      "grad_norm": 0.8784149289131165,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 47660
+    },
+    {
+      "epoch": 313.61842105263156,
+      "grad_norm": 1.1013492345809937,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 47670
+    },
+    {
+      "epoch": 313.6842105263158,
+      "grad_norm": 1.3613831996917725,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 47680
+    },
+    {
+      "epoch": 313.75,
+      "grad_norm": 1.004358172416687,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 47690
+    },
+    {
+      "epoch": 313.8157894736842,
+      "grad_norm": 1.2566474676132202,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 47700
+    },
+    {
+      "epoch": 313.88157894736844,
+      "grad_norm": 1.362387776374817,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 47710
+    },
+    {
+      "epoch": 313.94736842105266,
+      "grad_norm": 0.9462046027183533,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 47720
+    },
+    {
+      "epoch": 314.0131578947368,
+      "grad_norm": 1.5367804765701294,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 47730
+    },
+    {
+      "epoch": 314.07894736842104,
+      "grad_norm": 1.370335340499878,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 47740
+    },
+    {
+      "epoch": 314.14473684210526,
+      "grad_norm": 1.0723284482955933,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 47750
+    },
+    {
+      "epoch": 314.2105263157895,
+      "grad_norm": 1.293366551399231,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 47760
+    },
+    {
+      "epoch": 314.2763157894737,
+      "grad_norm": 1.2416729927062988,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 47770
+    },
+    {
+      "epoch": 314.3421052631579,
+      "grad_norm": 0.9720730781555176,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 47780
+    },
+    {
+      "epoch": 314.4078947368421,
+      "grad_norm": 1.1090688705444336,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 47790
+    },
+    {
+      "epoch": 314.4736842105263,
+      "grad_norm": 1.4181932210922241,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 47800
+    },
+    {
+      "epoch": 314.5394736842105,
+      "grad_norm": 1.1214467287063599,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 47810
+    },
+    {
+      "epoch": 314.60526315789474,
+      "grad_norm": 1.126190185546875,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 47820
+    },
+    {
+      "epoch": 314.67105263157896,
+      "grad_norm": 1.3774467706680298,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 47830
+    },
+    {
+      "epoch": 314.7368421052632,
+      "grad_norm": 1.2156517505645752,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 47840
+    },
+    {
+      "epoch": 314.80263157894734,
+      "grad_norm": 0.9228918552398682,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 47850
+    },
+    {
+      "epoch": 314.86842105263156,
+      "grad_norm": 1.2994784116744995,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 47860
+    },
+    {
+      "epoch": 314.9342105263158,
+      "grad_norm": 1.2928920984268188,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 47870
+    },
+    {
+      "epoch": 315.0,
+      "grad_norm": 0.9638012647628784,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 47880
+    },
+    {
+      "epoch": 315.0657894736842,
+      "grad_norm": 0.971041738986969,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 47890
+    },
+    {
+      "epoch": 315.13157894736844,
+      "grad_norm": 0.9184520244598389,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 47900
+    },
+    {
+      "epoch": 315.19736842105266,
+      "grad_norm": 0.8455377221107483,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 47910
+    },
+    {
+      "epoch": 315.2631578947368,
+      "grad_norm": 0.7738654017448425,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 47920
+    },
+    {
+      "epoch": 315.32894736842104,
+      "grad_norm": 1.2264034748077393,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 47930
+    },
+    {
+      "epoch": 315.39473684210526,
+      "grad_norm": 0.7875370979309082,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 47940
+    },
+    {
+      "epoch": 315.4605263157895,
+      "grad_norm": 0.7856130599975586,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 47950
+    },
+    {
+      "epoch": 315.5263157894737,
+      "grad_norm": 1.2851800918579102,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 47960
+    },
+    {
+      "epoch": 315.5921052631579,
+      "grad_norm": 1.3538479804992676,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 47970
+    },
+    {
+      "epoch": 315.6578947368421,
+      "grad_norm": 1.427662968635559,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 47980
+    },
+    {
+      "epoch": 315.7236842105263,
+      "grad_norm": 1.2748855352401733,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 47990
+    },
+    {
+      "epoch": 315.7894736842105,
+      "grad_norm": 1.1658332347869873,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 48000
+    },
+    {
+      "epoch": 315.85526315789474,
+      "grad_norm": 1.1847903728485107,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 48010
+    },
+    {
+      "epoch": 315.92105263157896,
+      "grad_norm": 1.2371987104415894,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 48020
+    },
+    {
+      "epoch": 315.9868421052632,
+      "grad_norm": 1.256600260734558,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 48030
+    },
+    {
+      "epoch": 316.05263157894734,
+      "grad_norm": 1.1105140447616577,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 48040
+    },
+    {
+      "epoch": 316.11842105263156,
+      "grad_norm": 1.4505999088287354,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 48050
+    },
+    {
+      "epoch": 316.1842105263158,
+      "grad_norm": 1.5613057613372803,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 48060
+    },
+    {
+      "epoch": 316.25,
+      "grad_norm": 1.2023935317993164,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 48070
+    },
+    {
+      "epoch": 316.3157894736842,
+      "grad_norm": 1.123512864112854,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 48080
+    },
+    {
+      "epoch": 316.38157894736844,
+      "grad_norm": 1.049616813659668,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 48090
+    },
+    {
+      "epoch": 316.44736842105266,
+      "grad_norm": 0.9086410403251648,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 48100
+    },
+    {
+      "epoch": 316.5131578947368,
+      "grad_norm": 0.9956292510032654,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 48110
+    },
+    {
+      "epoch": 316.57894736842104,
+      "grad_norm": 1.0076509714126587,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 48120
+    },
+    {
+      "epoch": 316.64473684210526,
+      "grad_norm": 0.9078097343444824,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 48130
+    },
+    {
+      "epoch": 316.7105263157895,
+      "grad_norm": 1.2351521253585815,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 48140
+    },
+    {
+      "epoch": 316.7763157894737,
+      "grad_norm": 0.9075007438659668,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 48150
+    },
+    {
+      "epoch": 316.8421052631579,
+      "grad_norm": 1.0333648920059204,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 48160
+    },
+    {
+      "epoch": 316.9078947368421,
+      "grad_norm": 1.42146635055542,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 48170
+    },
+    {
+      "epoch": 316.9736842105263,
+      "grad_norm": 1.4142286777496338,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 48180
+    },
+    {
+      "epoch": 317.0394736842105,
+      "grad_norm": 1.693477749824524,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 48190
+    },
+    {
+      "epoch": 317.10526315789474,
+      "grad_norm": 1.183012843132019,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 48200
+    },
+    {
+      "epoch": 317.17105263157896,
+      "grad_norm": 1.2209391593933105,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 48210
+    },
+    {
+      "epoch": 317.2368421052632,
+      "grad_norm": 1.294472098350525,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 48220
+    },
+    {
+      "epoch": 317.30263157894734,
+      "grad_norm": 1.6887290477752686,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 48230
+    },
+    {
+      "epoch": 317.36842105263156,
+      "grad_norm": 1.9167300462722778,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 48240
+    },
+    {
+      "epoch": 317.4342105263158,
+      "grad_norm": 1.2060270309448242,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 48250
+    },
+    {
+      "epoch": 317.5,
+      "grad_norm": 0.9500623941421509,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 48260
+    },
+    {
+      "epoch": 317.5657894736842,
+      "grad_norm": 1.399133563041687,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 48270
+    },
+    {
+      "epoch": 317.63157894736844,
+      "grad_norm": 1.2392226457595825,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 48280
+    },
+    {
+      "epoch": 317.69736842105266,
+      "grad_norm": 1.2007087469100952,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 48290
+    },
+    {
+      "epoch": 317.7631578947368,
+      "grad_norm": 1.4812746047973633,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 48300
+    },
+    {
+      "epoch": 317.82894736842104,
+      "grad_norm": 1.3363618850708008,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 48310
+    },
+    {
+      "epoch": 317.89473684210526,
+      "grad_norm": 1.3857090473175049,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 48320
+    },
+    {
+      "epoch": 317.9605263157895,
+      "grad_norm": 1.5482040643692017,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 48330
+    },
+    {
+      "epoch": 318.0263157894737,
+      "grad_norm": 1.3046433925628662,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 48340
+    },
+    {
+      "epoch": 318.0921052631579,
+      "grad_norm": 1.4874904155731201,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 48350
+    },
+    {
+      "epoch": 318.1578947368421,
+      "grad_norm": 1.252055048942566,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 48360
+    },
+    {
+      "epoch": 318.2236842105263,
+      "grad_norm": 1.297682762145996,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 48370
+    },
+    {
+      "epoch": 318.2894736842105,
+      "grad_norm": 1.7662724256515503,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 48380
+    },
+    {
+      "epoch": 318.35526315789474,
+      "grad_norm": 1.5304863452911377,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 48390
+    },
+    {
+      "epoch": 318.42105263157896,
+      "grad_norm": 1.3960685729980469,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 48400
+    },
+    {
+      "epoch": 318.4868421052632,
+      "grad_norm": 1.506137490272522,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 48410
+    },
+    {
+      "epoch": 318.55263157894734,
+      "grad_norm": 1.1885483264923096,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 48420
+    },
+    {
+      "epoch": 318.61842105263156,
+      "grad_norm": 1.2131962776184082,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 48430
+    },
+    {
+      "epoch": 318.6842105263158,
+      "grad_norm": 1.2580586671829224,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 48440
+    },
+    {
+      "epoch": 318.75,
+      "grad_norm": 1.2965505123138428,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 48450
+    },
+    {
+      "epoch": 318.8157894736842,
+      "grad_norm": 1.243948221206665,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 48460
+    },
+    {
+      "epoch": 318.88157894736844,
+      "grad_norm": 0.9847695231437683,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 48470
+    },
+    {
+      "epoch": 318.94736842105266,
+      "grad_norm": 1.0571852922439575,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 48480
+    },
+    {
+      "epoch": 319.0131578947368,
+      "grad_norm": 0.8944369554519653,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 48490
+    },
+    {
+      "epoch": 319.07894736842104,
+      "grad_norm": 1.3923020362854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 48500
+    },
+    {
+      "epoch": 319.14473684210526,
+      "grad_norm": 1.0978920459747314,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 48510
+    },
+    {
+      "epoch": 319.2105263157895,
+      "grad_norm": 1.073713779449463,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 48520
+    },
+    {
+      "epoch": 319.2763157894737,
+      "grad_norm": 0.8668067455291748,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 48530
+    },
+    {
+      "epoch": 319.3421052631579,
+      "grad_norm": 0.8067622184753418,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 48540
+    },
+    {
+      "epoch": 319.4078947368421,
+      "grad_norm": 0.9731525182723999,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 48550
+    },
+    {
+      "epoch": 319.4736842105263,
+      "grad_norm": 0.9085482358932495,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 48560
+    },
+    {
+      "epoch": 319.5394736842105,
+      "grad_norm": 1.2935781478881836,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 48570
+    },
+    {
+      "epoch": 319.60526315789474,
+      "grad_norm": 2.343533515930176,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 48580
+    },
+    {
+      "epoch": 319.67105263157896,
+      "grad_norm": 1.5465039014816284,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 48590
+    },
+    {
+      "epoch": 319.7368421052632,
+      "grad_norm": 1.2827335596084595,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 48600
+    },
+    {
+      "epoch": 319.80263157894734,
+      "grad_norm": 1.5653749704360962,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 48610
+    },
+    {
+      "epoch": 319.86842105263156,
+      "grad_norm": 1.6465096473693848,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 48620
+    },
+    {
+      "epoch": 319.9342105263158,
+      "grad_norm": 1.4881956577301025,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 48630
+    },
+    {
+      "epoch": 320.0,
+      "grad_norm": 1.5080645084381104,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 48640
+    },
+    {
+      "epoch": 320.0657894736842,
+      "grad_norm": 1.516532063484192,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 48650
+    },
+    {
+      "epoch": 320.13157894736844,
+      "grad_norm": 1.2939443588256836,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 48660
+    },
+    {
+      "epoch": 320.19736842105266,
+      "grad_norm": 1.2543706893920898,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 48670
+    },
+    {
+      "epoch": 320.2631578947368,
+      "grad_norm": 1.0210703611373901,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 48680
+    },
+    {
+      "epoch": 320.32894736842104,
+      "grad_norm": 1.2187316417694092,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 48690
+    },
+    {
+      "epoch": 320.39473684210526,
+      "grad_norm": 0.8297109007835388,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 48700
+    },
+    {
+      "epoch": 320.4605263157895,
+      "grad_norm": 1.0933237075805664,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 48710
+    },
+    {
+      "epoch": 320.5263157894737,
+      "grad_norm": 0.9736988544464111,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 48720
+    },
+    {
+      "epoch": 320.5921052631579,
+      "grad_norm": 1.174633502960205,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 48730
+    },
+    {
+      "epoch": 320.6578947368421,
+      "grad_norm": 1.419814109802246,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 48740
+    },
+    {
+      "epoch": 320.7236842105263,
+      "grad_norm": 0.9784946441650391,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 48750
+    },
+    {
+      "epoch": 320.7894736842105,
+      "grad_norm": 0.8846175670623779,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 48760
+    },
+    {
+      "epoch": 320.85526315789474,
+      "grad_norm": 1.217024326324463,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 48770
+    },
+    {
+      "epoch": 320.92105263157896,
+      "grad_norm": 1.0269222259521484,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 48780
+    },
+    {
+      "epoch": 320.9868421052632,
+      "grad_norm": 1.176324486732483,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 48790
+    },
+    {
+      "epoch": 321.05263157894734,
+      "grad_norm": 1.1904447078704834,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 48800
+    },
+    {
+      "epoch": 321.11842105263156,
+      "grad_norm": 1.0044549703598022,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 48810
+    },
+    {
+      "epoch": 321.1842105263158,
+      "grad_norm": 1.2409372329711914,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 48820
+    },
+    {
+      "epoch": 321.25,
+      "grad_norm": 1.3968594074249268,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 48830
+    },
+    {
+      "epoch": 321.3157894736842,
+      "grad_norm": 1.2159334421157837,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 48840
+    },
+    {
+      "epoch": 321.38157894736844,
+      "grad_norm": 1.4838895797729492,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 48850
+    },
+    {
+      "epoch": 321.44736842105266,
+      "grad_norm": 1.58181893825531,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 48860
+    },
+    {
+      "epoch": 321.5131578947368,
+      "grad_norm": 1.1558905839920044,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 48870
+    },
+    {
+      "epoch": 321.57894736842104,
+      "grad_norm": 1.1674673557281494,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 48880
+    },
+    {
+      "epoch": 321.64473684210526,
+      "grad_norm": 1.2923530340194702,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 48890
+    },
+    {
+      "epoch": 321.7105263157895,
+      "grad_norm": 1.4258642196655273,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 48900
+    },
+    {
+      "epoch": 321.7763157894737,
+      "grad_norm": 1.1080553531646729,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 48910
+    },
+    {
+      "epoch": 321.8421052631579,
+      "grad_norm": 1.5684802532196045,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 48920
+    },
+    {
+      "epoch": 321.9078947368421,
+      "grad_norm": 1.5006873607635498,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 48930
+    },
+    {
+      "epoch": 321.9736842105263,
+      "grad_norm": 1.164925217628479,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 48940
+    },
+    {
+      "epoch": 322.0394736842105,
+      "grad_norm": 1.4386217594146729,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 48950
+    },
+    {
+      "epoch": 322.10526315789474,
+      "grad_norm": 1.5981786251068115,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 48960
+    },
+    {
+      "epoch": 322.17105263157896,
+      "grad_norm": 1.1611731052398682,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 48970
+    },
+    {
+      "epoch": 322.2368421052632,
+      "grad_norm": 1.0494667291641235,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 48980
+    },
+    {
+      "epoch": 322.30263157894734,
+      "grad_norm": 1.397523045539856,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 48990
+    },
+    {
+      "epoch": 322.36842105263156,
+      "grad_norm": 1.5347926616668701,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 49000
+    },
+    {
+      "epoch": 322.4342105263158,
+      "grad_norm": 1.1124234199523926,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 49010
+    },
+    {
+      "epoch": 322.5,
+      "grad_norm": 1.2991044521331787,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 49020
+    },
+    {
+      "epoch": 322.5657894736842,
+      "grad_norm": 1.0345944166183472,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 49030
+    },
+    {
+      "epoch": 322.63157894736844,
+      "grad_norm": 1.2403799295425415,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 49040
+    },
+    {
+      "epoch": 322.69736842105266,
+      "grad_norm": 1.202883005142212,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 49050
+    },
+    {
+      "epoch": 322.7631578947368,
+      "grad_norm": 1.2251741886138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 49060
+    },
+    {
+      "epoch": 322.82894736842104,
+      "grad_norm": 1.258322834968567,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 49070
+    },
+    {
+      "epoch": 322.89473684210526,
+      "grad_norm": 1.1949845552444458,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 49080
+    },
+    {
+      "epoch": 322.9605263157895,
+      "grad_norm": 1.3184500932693481,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 49090
+    },
+    {
+      "epoch": 323.0263157894737,
+      "grad_norm": 1.1563595533370972,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49100
+    },
+    {
+      "epoch": 323.0921052631579,
+      "grad_norm": 1.0812554359436035,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 49110
+    },
+    {
+      "epoch": 323.1578947368421,
+      "grad_norm": 1.1537151336669922,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49120
+    },
+    {
+      "epoch": 323.2236842105263,
+      "grad_norm": 1.4239085912704468,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49130
+    },
+    {
+      "epoch": 323.2894736842105,
+      "grad_norm": 1.167175531387329,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 49140
+    },
+    {
+      "epoch": 323.35526315789474,
+      "grad_norm": 1.0651171207427979,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 49150
+    },
+    {
+      "epoch": 323.42105263157896,
+      "grad_norm": 1.1401432752609253,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49160
+    },
+    {
+      "epoch": 323.4868421052632,
+      "grad_norm": 1.0716900825500488,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 49170
+    },
+    {
+      "epoch": 323.55263157894734,
+      "grad_norm": 0.9888905882835388,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 49180
+    },
+    {
+      "epoch": 323.61842105263156,
+      "grad_norm": 1.1560784578323364,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 49190
+    },
+    {
+      "epoch": 323.6842105263158,
+      "grad_norm": 1.1108367443084717,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 49200
+    },
+    {
+      "epoch": 323.75,
+      "grad_norm": 0.7860233783721924,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 49210
+    },
+    {
+      "epoch": 323.8157894736842,
+      "grad_norm": 1.1012498140335083,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 49220
+    },
+    {
+      "epoch": 323.88157894736844,
+      "grad_norm": 0.9723058938980103,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 49230
+    },
+    {
+      "epoch": 323.94736842105266,
+      "grad_norm": 1.2430423498153687,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 49240
+    },
+    {
+      "epoch": 324.0131578947368,
+      "grad_norm": 1.3332782983779907,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49250
+    },
+    {
+      "epoch": 324.07894736842104,
+      "grad_norm": 1.341118335723877,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 49260
+    },
+    {
+      "epoch": 324.14473684210526,
+      "grad_norm": 1.2712905406951904,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 49270
+    },
+    {
+      "epoch": 324.2105263157895,
+      "grad_norm": 0.7532151937484741,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 49280
+    },
+    {
+      "epoch": 324.2763157894737,
+      "grad_norm": 1.4031445980072021,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 49290
+    },
+    {
+      "epoch": 324.3421052631579,
+      "grad_norm": 0.9208416938781738,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 49300
+    },
+    {
+      "epoch": 324.4078947368421,
+      "grad_norm": 1.2247257232666016,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 49310
+    },
+    {
+      "epoch": 324.4736842105263,
+      "grad_norm": 1.0223255157470703,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49320
+    },
+    {
+      "epoch": 324.5394736842105,
+      "grad_norm": 1.1954952478408813,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 49330
+    },
+    {
+      "epoch": 324.60526315789474,
+      "grad_norm": 1.2614434957504272,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 49340
+    },
+    {
+      "epoch": 324.67105263157896,
+      "grad_norm": 1.0313371419906616,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 49350
+    },
+    {
+      "epoch": 324.7368421052632,
+      "grad_norm": 1.0749729871749878,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 49360
+    },
+    {
+      "epoch": 324.80263157894734,
+      "grad_norm": 1.2034530639648438,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 49370
+    },
+    {
+      "epoch": 324.86842105263156,
+      "grad_norm": 1.1916738748550415,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 49380
+    },
+    {
+      "epoch": 324.9342105263158,
+      "grad_norm": 0.9107323288917542,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 49390
+    },
+    {
+      "epoch": 325.0,
+      "grad_norm": 1.1662636995315552,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 49400
+    },
+    {
+      "epoch": 325.0657894736842,
+      "grad_norm": 1.0498578548431396,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 49410
+    },
+    {
+      "epoch": 325.13157894736844,
+      "grad_norm": 1.193319320678711,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 49420
+    },
+    {
+      "epoch": 325.19736842105266,
+      "grad_norm": 1.2099950313568115,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 49430
+    },
+    {
+      "epoch": 325.2631578947368,
+      "grad_norm": 1.0245522260665894,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 49440
+    },
+    {
+      "epoch": 325.32894736842104,
+      "grad_norm": 1.201566219329834,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 49450
+    },
+    {
+      "epoch": 325.39473684210526,
+      "grad_norm": 0.9498182535171509,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 49460
+    },
+    {
+      "epoch": 325.4605263157895,
+      "grad_norm": 1.081298828125,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 49470
+    },
+    {
+      "epoch": 325.5263157894737,
+      "grad_norm": 1.2874219417572021,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 49480
+    },
+    {
+      "epoch": 325.5921052631579,
+      "grad_norm": 1.1696557998657227,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 49490
+    },
+    {
+      "epoch": 325.6578947368421,
+      "grad_norm": 1.0769000053405762,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 49500
+    },
+    {
+      "epoch": 325.7236842105263,
+      "grad_norm": 1.134009838104248,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 49510
+    },
+    {
+      "epoch": 325.7894736842105,
+      "grad_norm": 0.9424777626991272,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 49520
+    },
+    {
+      "epoch": 325.85526315789474,
+      "grad_norm": 1.0407861471176147,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49530
+    },
+    {
+      "epoch": 325.92105263157896,
+      "grad_norm": 0.9889567494392395,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 49540
+    },
+    {
+      "epoch": 325.9868421052632,
+      "grad_norm": 0.6618396043777466,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 49550
+    },
+    {
+      "epoch": 326.05263157894734,
+      "grad_norm": 1.2362364530563354,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 49560
+    },
+    {
+      "epoch": 326.11842105263156,
+      "grad_norm": 1.2794060707092285,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 49570
+    },
+    {
+      "epoch": 326.1842105263158,
+      "grad_norm": 1.1880918741226196,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 49580
+    },
+    {
+      "epoch": 326.25,
+      "grad_norm": 1.0328410863876343,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 49590
+    },
+    {
+      "epoch": 326.3157894736842,
+      "grad_norm": 1.3989026546478271,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49600
+    },
+    {
+      "epoch": 326.38157894736844,
+      "grad_norm": 0.8209923505783081,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 49610
+    },
+    {
+      "epoch": 326.44736842105266,
+      "grad_norm": 1.04630446434021,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 49620
+    },
+    {
+      "epoch": 326.5131578947368,
+      "grad_norm": 1.4371641874313354,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 49630
+    },
+    {
+      "epoch": 326.57894736842104,
+      "grad_norm": 1.1267870664596558,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 49640
+    },
+    {
+      "epoch": 326.64473684210526,
+      "grad_norm": 1.2246888875961304,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 49650
+    },
+    {
+      "epoch": 326.7105263157895,
+      "grad_norm": 1.1389400959014893,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 49660
+    },
+    {
+      "epoch": 326.7763157894737,
+      "grad_norm": 0.806943953037262,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49670
+    },
+    {
+      "epoch": 326.8421052631579,
+      "grad_norm": 1.1246274709701538,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 49680
+    },
+    {
+      "epoch": 326.9078947368421,
+      "grad_norm": 1.3414881229400635,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 49690
+    },
+    {
+      "epoch": 326.9736842105263,
+      "grad_norm": 0.9751403331756592,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 49700
+    },
+    {
+      "epoch": 327.0394736842105,
+      "grad_norm": 0.9873465299606323,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 49710
+    },
+    {
+      "epoch": 327.10526315789474,
+      "grad_norm": 1.0995386838912964,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 49720
+    },
+    {
+      "epoch": 327.17105263157896,
+      "grad_norm": 1.18962824344635,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 49730
+    },
+    {
+      "epoch": 327.2368421052632,
+      "grad_norm": 1.1961159706115723,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 49740
+    },
+    {
+      "epoch": 327.30263157894734,
+      "grad_norm": 1.5299499034881592,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 49750
+    },
+    {
+      "epoch": 327.36842105263156,
+      "grad_norm": 1.2997653484344482,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 49760
+    },
+    {
+      "epoch": 327.4342105263158,
+      "grad_norm": 1.19481360912323,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 49770
+    },
+    {
+      "epoch": 327.5,
+      "grad_norm": 1.1865720748901367,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 49780
+    },
+    {
+      "epoch": 327.5657894736842,
+      "grad_norm": 1.3562146425247192,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 49790
+    },
+    {
+      "epoch": 327.63157894736844,
+      "grad_norm": 1.2696094512939453,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 49800
+    },
+    {
+      "epoch": 327.69736842105266,
+      "grad_norm": 1.2589432001113892,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 49810
+    },
+    {
+      "epoch": 327.7631578947368,
+      "grad_norm": 0.9365496039390564,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 49820
+    },
+    {
+      "epoch": 327.82894736842104,
+      "grad_norm": 1.6220346689224243,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 49830
+    },
+    {
+      "epoch": 327.89473684210526,
+      "grad_norm": 1.4582602977752686,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 49840
+    },
+    {
+      "epoch": 327.9605263157895,
+      "grad_norm": 1.0291483402252197,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 49850
+    },
+    {
+      "epoch": 328.0263157894737,
+      "grad_norm": 1.1116678714752197,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 49860
+    },
+    {
+      "epoch": 328.0921052631579,
+      "grad_norm": 1.533398985862732,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 49870
+    },
+    {
+      "epoch": 328.1578947368421,
+      "grad_norm": 1.562609076499939,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 49880
+    },
+    {
+      "epoch": 328.2236842105263,
+      "grad_norm": 1.3016607761383057,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 49890
+    },
+    {
+      "epoch": 328.2894736842105,
+      "grad_norm": 1.1702256202697754,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 49900
+    },
+    {
+      "epoch": 328.35526315789474,
+      "grad_norm": 1.1809632778167725,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 49910
+    },
+    {
+      "epoch": 328.42105263157896,
+      "grad_norm": 1.109625220298767,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 49920
+    },
+    {
+      "epoch": 328.4868421052632,
+      "grad_norm": 1.376266360282898,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 49930
+    },
+    {
+      "epoch": 328.55263157894734,
+      "grad_norm": 1.2670068740844727,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 49940
+    },
+    {
+      "epoch": 328.61842105263156,
+      "grad_norm": 1.1520689725875854,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 49950
+    },
+    {
+      "epoch": 328.6842105263158,
+      "grad_norm": 1.171880841255188,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 49960
+    },
+    {
+      "epoch": 328.75,
+      "grad_norm": 1.2454252243041992,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 49970
+    },
+    {
+      "epoch": 328.8157894736842,
+      "grad_norm": 0.9792435169219971,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 49980
+    },
+    {
+      "epoch": 328.88157894736844,
+      "grad_norm": 1.4465330839157104,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 49990
+    },
+    {
+      "epoch": 328.94736842105266,
+      "grad_norm": 1.2347135543823242,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 50000
+    },
+    {
+      "epoch": 329.0131578947368,
+      "grad_norm": 1.15986967086792,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 50010
+    },
+    {
+      "epoch": 329.07894736842104,
+      "grad_norm": 1.2300969362258911,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 50020
+    },
+    {
+      "epoch": 329.14473684210526,
+      "grad_norm": 0.9923980832099915,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 50030
+    },
+    {
+      "epoch": 329.2105263157895,
+      "grad_norm": 1.0011688470840454,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50040
+    },
+    {
+      "epoch": 329.2763157894737,
+      "grad_norm": 0.9253886938095093,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 50050
+    },
+    {
+      "epoch": 329.3421052631579,
+      "grad_norm": 0.8614893555641174,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 50060
+    },
+    {
+      "epoch": 329.4078947368421,
+      "grad_norm": 1.411759614944458,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 50070
+    },
+    {
+      "epoch": 329.4736842105263,
+      "grad_norm": 0.7994163632392883,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 50080
+    },
+    {
+      "epoch": 329.5394736842105,
+      "grad_norm": 0.9142425060272217,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 50090
+    },
+    {
+      "epoch": 329.60526315789474,
+      "grad_norm": 0.7740334272384644,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 50100
+    },
+    {
+      "epoch": 329.67105263157896,
+      "grad_norm": 0.847666323184967,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 50110
+    },
+    {
+      "epoch": 329.7368421052632,
+      "grad_norm": 1.1491421461105347,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 50120
+    },
+    {
+      "epoch": 329.80263157894734,
+      "grad_norm": 0.9505961537361145,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 50130
+    },
+    {
+      "epoch": 329.86842105263156,
+      "grad_norm": 1.0270600318908691,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 50140
+    },
+    {
+      "epoch": 329.9342105263158,
+      "grad_norm": 1.1158539056777954,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 50150
+    },
+    {
+      "epoch": 330.0,
+      "grad_norm": 1.6066099405288696,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 50160
+    },
+    {
+      "epoch": 330.0657894736842,
+      "grad_norm": 1.3317687511444092,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 50170
+    },
+    {
+      "epoch": 330.13157894736844,
+      "grad_norm": 1.1555328369140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 50180
+    },
+    {
+      "epoch": 330.19736842105266,
+      "grad_norm": 1.1379915475845337,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50190
+    },
+    {
+      "epoch": 330.2631578947368,
+      "grad_norm": 1.0150611400604248,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 50200
+    },
+    {
+      "epoch": 330.32894736842104,
+      "grad_norm": 1.422431468963623,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 50210
+    },
+    {
+      "epoch": 330.39473684210526,
+      "grad_norm": 1.1463091373443604,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 50220
+    },
+    {
+      "epoch": 330.4605263157895,
+      "grad_norm": 1.2713172435760498,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 50230
+    },
+    {
+      "epoch": 330.5263157894737,
+      "grad_norm": 1.2657397985458374,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 50240
+    },
+    {
+      "epoch": 330.5921052631579,
+      "grad_norm": 1.109223484992981,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 50250
+    },
+    {
+      "epoch": 330.6578947368421,
+      "grad_norm": 1.1030226945877075,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 50260
+    },
+    {
+      "epoch": 330.7236842105263,
+      "grad_norm": 1.2392481565475464,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 50270
+    },
+    {
+      "epoch": 330.7894736842105,
+      "grad_norm": 1.1051968336105347,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 50280
+    },
+    {
+      "epoch": 330.85526315789474,
+      "grad_norm": 1.110648274421692,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50290
+    },
+    {
+      "epoch": 330.92105263157896,
+      "grad_norm": 1.3679182529449463,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 50300
+    },
+    {
+      "epoch": 330.9868421052632,
+      "grad_norm": 1.1558316946029663,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 50310
+    },
+    {
+      "epoch": 331.05263157894734,
+      "grad_norm": 0.9238816499710083,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 50320
+    },
+    {
+      "epoch": 331.11842105263156,
+      "grad_norm": 1.363661766052246,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 50330
+    },
+    {
+      "epoch": 331.1842105263158,
+      "grad_norm": 1.5912078619003296,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 50340
+    },
+    {
+      "epoch": 331.25,
+      "grad_norm": 1.1807664632797241,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 50350
+    },
+    {
+      "epoch": 331.3157894736842,
+      "grad_norm": 1.0620604753494263,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 50360
+    },
+    {
+      "epoch": 331.38157894736844,
+      "grad_norm": 1.0566785335540771,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 50370
+    },
+    {
+      "epoch": 331.44736842105266,
+      "grad_norm": 0.9269898533821106,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 50380
+    },
+    {
+      "epoch": 331.5131578947368,
+      "grad_norm": 1.0795518159866333,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 50390
+    },
+    {
+      "epoch": 331.57894736842104,
+      "grad_norm": 1.1941287517547607,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 50400
+    },
+    {
+      "epoch": 331.64473684210526,
+      "grad_norm": 1.3070482015609741,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 50410
+    },
+    {
+      "epoch": 331.7105263157895,
+      "grad_norm": 1.1180983781814575,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 50420
+    },
+    {
+      "epoch": 331.7763157894737,
+      "grad_norm": 1.0409196615219116,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 50430
+    },
+    {
+      "epoch": 331.8421052631579,
+      "grad_norm": 1.5609703063964844,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 50440
+    },
+    {
+      "epoch": 331.9078947368421,
+      "grad_norm": 1.2838364839553833,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 50450
+    },
+    {
+      "epoch": 331.9736842105263,
+      "grad_norm": 1.4142061471939087,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 50460
+    },
+    {
+      "epoch": 332.0394736842105,
+      "grad_norm": 1.0563913583755493,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 50470
+    },
+    {
+      "epoch": 332.10526315789474,
+      "grad_norm": 1.106583833694458,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 50480
+    },
+    {
+      "epoch": 332.17105263157896,
+      "grad_norm": 1.0735349655151367,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 50490
+    },
+    {
+      "epoch": 332.2368421052632,
+      "grad_norm": 1.5495939254760742,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 50500
+    },
+    {
+      "epoch": 332.30263157894734,
+      "grad_norm": 1.1184154748916626,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 50510
+    },
+    {
+      "epoch": 332.36842105263156,
+      "grad_norm": 1.4668242931365967,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50520
+    },
+    {
+      "epoch": 332.4342105263158,
+      "grad_norm": 1.1980876922607422,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 50530
+    },
+    {
+      "epoch": 332.5,
+      "grad_norm": 1.3636940717697144,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 50540
+    },
+    {
+      "epoch": 332.5657894736842,
+      "grad_norm": 1.2938543558120728,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 50550
+    },
+    {
+      "epoch": 332.63157894736844,
+      "grad_norm": 1.1288648843765259,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 50560
+    },
+    {
+      "epoch": 332.69736842105266,
+      "grad_norm": 1.0192151069641113,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50570
+    },
+    {
+      "epoch": 332.7631578947368,
+      "grad_norm": 1.2933300733566284,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 50580
+    },
+    {
+      "epoch": 332.82894736842104,
+      "grad_norm": 1.2441681623458862,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 50590
+    },
+    {
+      "epoch": 332.89473684210526,
+      "grad_norm": 1.2795456647872925,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 50600
+    },
+    {
+      "epoch": 332.9605263157895,
+      "grad_norm": 1.3279870748519897,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 50610
+    },
+    {
+      "epoch": 333.0263157894737,
+      "grad_norm": 1.3241667747497559,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 50620
+    },
+    {
+      "epoch": 333.0921052631579,
+      "grad_norm": 0.7530394196510315,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 50630
+    },
+    {
+      "epoch": 333.1578947368421,
+      "grad_norm": 1.23948073387146,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 50640
+    },
+    {
+      "epoch": 333.2236842105263,
+      "grad_norm": 1.1512643098831177,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 50650
+    },
+    {
+      "epoch": 333.2894736842105,
+      "grad_norm": 1.1450495719909668,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 50660
+    },
+    {
+      "epoch": 333.35526315789474,
+      "grad_norm": 1.0570194721221924,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50670
+    },
+    {
+      "epoch": 333.42105263157896,
+      "grad_norm": 1.017793893814087,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 50680
+    },
+    {
+      "epoch": 333.4868421052632,
+      "grad_norm": 1.2568644285202026,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 50690
+    },
+    {
+      "epoch": 333.55263157894734,
+      "grad_norm": 1.0966501235961914,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50700
+    },
+    {
+      "epoch": 333.61842105263156,
+      "grad_norm": 1.404511570930481,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 50710
+    },
+    {
+      "epoch": 333.6842105263158,
+      "grad_norm": 0.9019613265991211,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 50720
+    },
+    {
+      "epoch": 333.75,
+      "grad_norm": 1.206984519958496,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 50730
+    },
+    {
+      "epoch": 333.8157894736842,
+      "grad_norm": 0.9531241059303284,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 50740
+    },
+    {
+      "epoch": 333.88157894736844,
+      "grad_norm": 1.3552753925323486,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50750
+    },
+    {
+      "epoch": 333.94736842105266,
+      "grad_norm": 1.5873048305511475,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 50760
+    },
+    {
+      "epoch": 334.0131578947368,
+      "grad_norm": 1.065086007118225,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 50770
+    },
+    {
+      "epoch": 334.07894736842104,
+      "grad_norm": 1.4782254695892334,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 50780
+    },
+    {
+      "epoch": 334.14473684210526,
+      "grad_norm": 1.0018125772476196,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 50790
+    },
+    {
+      "epoch": 334.2105263157895,
+      "grad_norm": 1.2389479875564575,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 50800
+    },
+    {
+      "epoch": 334.2763157894737,
+      "grad_norm": 1.1885042190551758,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 50810
+    },
+    {
+      "epoch": 334.3421052631579,
+      "grad_norm": 0.9912749528884888,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 50820
+    },
+    {
+      "epoch": 334.4078947368421,
+      "grad_norm": 1.206180214881897,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 50830
+    },
+    {
+      "epoch": 334.4736842105263,
+      "grad_norm": 1.087846279144287,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 50840
+    },
+    {
+      "epoch": 334.5394736842105,
+      "grad_norm": 1.4583332538604736,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 50850
+    },
+    {
+      "epoch": 334.60526315789474,
+      "grad_norm": 1.2611576318740845,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50860
+    },
+    {
+      "epoch": 334.67105263157896,
+      "grad_norm": 1.2593576908111572,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 50870
+    },
+    {
+      "epoch": 334.7368421052632,
+      "grad_norm": 0.8674020171165466,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 50880
+    },
+    {
+      "epoch": 334.80263157894734,
+      "grad_norm": 0.9973218441009521,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 50890
+    },
+    {
+      "epoch": 334.86842105263156,
+      "grad_norm": 1.1739544868469238,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 50900
+    },
+    {
+      "epoch": 334.9342105263158,
+      "grad_norm": 1.2401018142700195,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 50910
+    },
+    {
+      "epoch": 335.0,
+      "grad_norm": 1.03706955909729,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 50920
+    },
+    {
+      "epoch": 335.0657894736842,
+      "grad_norm": 1.213478684425354,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 50930
+    },
+    {
+      "epoch": 335.13157894736844,
+      "grad_norm": 1.3878175020217896,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 50940
+    },
+    {
+      "epoch": 335.19736842105266,
+      "grad_norm": 1.3380154371261597,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 50950
+    },
+    {
+      "epoch": 335.2631578947368,
+      "grad_norm": 0.9307578802108765,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 50960
+    },
+    {
+      "epoch": 335.32894736842104,
+      "grad_norm": 1.1519443988800049,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 50970
+    },
+    {
+      "epoch": 335.39473684210526,
+      "grad_norm": 1.3810474872589111,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 50980
+    },
+    {
+      "epoch": 335.4605263157895,
+      "grad_norm": 1.1275713443756104,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 50990
+    },
+    {
+      "epoch": 335.5263157894737,
+      "grad_norm": 1.1235930919647217,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 51000
+    },
+    {
+      "epoch": 335.5921052631579,
+      "grad_norm": 1.6423934698104858,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 51010
+    },
+    {
+      "epoch": 335.6578947368421,
+      "grad_norm": 1.104622721672058,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 51020
+    },
+    {
+      "epoch": 335.7236842105263,
+      "grad_norm": 1.2324247360229492,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 51030
+    },
+    {
+      "epoch": 335.7894736842105,
+      "grad_norm": 1.1801762580871582,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 51040
+    },
+    {
+      "epoch": 335.85526315789474,
+      "grad_norm": 1.164038896560669,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 51050
+    },
+    {
+      "epoch": 335.92105263157896,
+      "grad_norm": 0.7976166009902954,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 51060
+    },
+    {
+      "epoch": 335.9868421052632,
+      "grad_norm": 1.0072071552276611,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 51070
+    },
+    {
+      "epoch": 336.05263157894734,
+      "grad_norm": 0.8578206896781921,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 51080
+    },
+    {
+      "epoch": 336.11842105263156,
+      "grad_norm": 1.0572422742843628,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 51090
+    },
+    {
+      "epoch": 336.1842105263158,
+      "grad_norm": 0.9827244281768799,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 51100
+    },
+    {
+      "epoch": 336.25,
+      "grad_norm": 1.35405695438385,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 51110
+    },
+    {
+      "epoch": 336.3157894736842,
+      "grad_norm": 0.9394972920417786,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 51120
+    },
+    {
+      "epoch": 336.38157894736844,
+      "grad_norm": 1.3125892877578735,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 51130
+    },
+    {
+      "epoch": 336.44736842105266,
+      "grad_norm": 1.182879090309143,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 51140
+    },
+    {
+      "epoch": 336.5131578947368,
+      "grad_norm": 0.9829082489013672,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 51150
+    },
+    {
+      "epoch": 336.57894736842104,
+      "grad_norm": 1.1734791994094849,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 51160
+    },
+    {
+      "epoch": 336.64473684210526,
+      "grad_norm": 1.2387040853500366,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 51170
+    },
+    {
+      "epoch": 336.7105263157895,
+      "grad_norm": 1.2995988130569458,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 51180
+    },
+    {
+      "epoch": 336.7763157894737,
+      "grad_norm": 1.099161982536316,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 51190
+    },
+    {
+      "epoch": 336.8421052631579,
+      "grad_norm": 1.1866588592529297,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 51200
+    },
+    {
+      "epoch": 336.9078947368421,
+      "grad_norm": 1.2611371278762817,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 51210
+    },
+    {
+      "epoch": 336.9736842105263,
+      "grad_norm": 1.6526949405670166,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 51220
+    },
+    {
+      "epoch": 337.0394736842105,
+      "grad_norm": 1.461340069770813,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 51230
+    },
+    {
+      "epoch": 337.10526315789474,
+      "grad_norm": 1.127563238143921,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 51240
+    },
+    {
+      "epoch": 337.17105263157896,
+      "grad_norm": 1.2652572393417358,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 51250
+    },
+    {
+      "epoch": 337.2368421052632,
+      "grad_norm": 1.25266695022583,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 51260
+    },
+    {
+      "epoch": 337.30263157894734,
+      "grad_norm": 1.2490793466567993,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 51270
+    },
+    {
+      "epoch": 337.36842105263156,
+      "grad_norm": 1.2546192407608032,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 51280
+    },
+    {
+      "epoch": 337.4342105263158,
+      "grad_norm": 1.2390152215957642,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 51290
+    },
+    {
+      "epoch": 337.5,
+      "grad_norm": 1.2711284160614014,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 51300
+    },
+    {
+      "epoch": 337.5657894736842,
+      "grad_norm": 1.480649471282959,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 51310
+    },
+    {
+      "epoch": 337.63157894736844,
+      "grad_norm": 1.043129563331604,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 51320
+    },
+    {
+      "epoch": 337.69736842105266,
+      "grad_norm": 0.9065507054328918,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 51330
+    },
+    {
+      "epoch": 337.7631578947368,
+      "grad_norm": 1.0650672912597656,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 51340
+    },
+    {
+      "epoch": 337.82894736842104,
+      "grad_norm": 1.24941086769104,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 51350
+    },
+    {
+      "epoch": 337.89473684210526,
+      "grad_norm": 1.2355936765670776,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 51360
+    },
+    {
+      "epoch": 337.9605263157895,
+      "grad_norm": 1.2188667058944702,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 51370
+    },
+    {
+      "epoch": 338.0263157894737,
+      "grad_norm": 1.136441707611084,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 51380
+    },
+    {
+      "epoch": 338.0921052631579,
+      "grad_norm": 1.5337886810302734,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 51390
+    },
+    {
+      "epoch": 338.1578947368421,
+      "grad_norm": 1.4767624139785767,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 51400
+    },
+    {
+      "epoch": 338.2236842105263,
+      "grad_norm": 1.1894986629486084,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 51410
+    },
+    {
+      "epoch": 338.2894736842105,
+      "grad_norm": 1.316213846206665,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 51420
+    },
+    {
+      "epoch": 338.35526315789474,
+      "grad_norm": 1.614834189414978,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 51430
+    },
+    {
+      "epoch": 338.42105263157896,
+      "grad_norm": 1.146628737449646,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 51440
+    },
+    {
+      "epoch": 338.4868421052632,
+      "grad_norm": 1.2770928144454956,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 51450
+    },
+    {
+      "epoch": 338.55263157894734,
+      "grad_norm": 1.2232598066329956,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 51460
+    },
+    {
+      "epoch": 338.61842105263156,
+      "grad_norm": 1.278428554534912,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 51470
+    },
+    {
+      "epoch": 338.6842105263158,
+      "grad_norm": 1.1535056829452515,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 51480
+    },
+    {
+      "epoch": 338.75,
+      "grad_norm": 0.9330645203590393,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 51490
+    },
+    {
+      "epoch": 338.8157894736842,
+      "grad_norm": 1.2083619832992554,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 51500
+    },
+    {
+      "epoch": 338.88157894736844,
+      "grad_norm": 1.0455271005630493,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 51510
+    },
+    {
+      "epoch": 338.94736842105266,
+      "grad_norm": 1.051611065864563,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 51520
+    },
+    {
+      "epoch": 339.0131578947368,
+      "grad_norm": 0.9898776412010193,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 51530
+    },
+    {
+      "epoch": 339.07894736842104,
+      "grad_norm": 1.3422266244888306,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 51540
+    },
+    {
+      "epoch": 339.14473684210526,
+      "grad_norm": 0.925194501876831,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 51550
+    },
+    {
+      "epoch": 339.2105263157895,
+      "grad_norm": 1.2911876440048218,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 51560
+    },
+    {
+      "epoch": 339.2763157894737,
+      "grad_norm": 1.5111730098724365,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 51570
+    },
+    {
+      "epoch": 339.3421052631579,
+      "grad_norm": 1.3134963512420654,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 51580
+    },
+    {
+      "epoch": 339.4078947368421,
+      "grad_norm": 1.3850778341293335,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 51590
+    },
+    {
+      "epoch": 339.4736842105263,
+      "grad_norm": 0.9667593836784363,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 51600
+    },
+    {
+      "epoch": 339.5394736842105,
+      "grad_norm": 1.3687058687210083,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 51610
+    },
+    {
+      "epoch": 339.60526315789474,
+      "grad_norm": 1.0629100799560547,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 51620
+    },
+    {
+      "epoch": 339.67105263157896,
+      "grad_norm": 0.9805474877357483,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 51630
+    },
+    {
+      "epoch": 339.7368421052632,
+      "grad_norm": 1.029873251914978,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 51640
+    },
+    {
+      "epoch": 339.80263157894734,
+      "grad_norm": 1.1182074546813965,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 51650
+    },
+    {
+      "epoch": 339.86842105263156,
+      "grad_norm": 0.8829177618026733,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 51660
+    },
+    {
+      "epoch": 339.9342105263158,
+      "grad_norm": 1.0989736318588257,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 51670
+    },
+    {
+      "epoch": 340.0,
+      "grad_norm": 1.5837504863739014,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 51680
+    },
+    {
+      "epoch": 340.0657894736842,
+      "grad_norm": 1.524735689163208,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 51690
+    },
+    {
+      "epoch": 340.13157894736844,
+      "grad_norm": 1.6917763948440552,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 51700
+    },
+    {
+      "epoch": 340.19736842105266,
+      "grad_norm": 1.0942375659942627,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 51710
+    },
+    {
+      "epoch": 340.2631578947368,
+      "grad_norm": 1.0912152528762817,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 51720
+    },
+    {
+      "epoch": 340.32894736842104,
+      "grad_norm": 0.9950433373451233,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 51730
+    },
+    {
+      "epoch": 340.39473684210526,
+      "grad_norm": 1.1365388631820679,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 51740
+    },
+    {
+      "epoch": 340.4605263157895,
+      "grad_norm": 1.5053037405014038,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 51750
+    },
+    {
+      "epoch": 340.5263157894737,
+      "grad_norm": 1.1808347702026367,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 51760
+    },
+    {
+      "epoch": 340.5921052631579,
+      "grad_norm": 1.6994234323501587,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 51770
+    },
+    {
+      "epoch": 340.6578947368421,
+      "grad_norm": 1.4310306310653687,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 51780
+    },
+    {
+      "epoch": 340.7236842105263,
+      "grad_norm": 1.2134596109390259,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 51790
+    },
+    {
+      "epoch": 340.7894736842105,
+      "grad_norm": 1.7126917839050293,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 51800
+    },
+    {
+      "epoch": 340.85526315789474,
+      "grad_norm": 1.0184993743896484,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 51810
+    },
+    {
+      "epoch": 340.92105263157896,
+      "grad_norm": 1.3017053604125977,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 51820
+    },
+    {
+      "epoch": 340.9868421052632,
+      "grad_norm": 0.8897969126701355,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 51830
+    },
+    {
+      "epoch": 341.05263157894734,
+      "grad_norm": 1.1662821769714355,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 51840
+    },
+    {
+      "epoch": 341.11842105263156,
+      "grad_norm": 1.338837742805481,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 51850
+    },
+    {
+      "epoch": 341.1842105263158,
+      "grad_norm": 1.1382125616073608,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 51860
+    },
+    {
+      "epoch": 341.25,
+      "grad_norm": 0.9120957851409912,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 51870
+    },
+    {
+      "epoch": 341.3157894736842,
+      "grad_norm": 1.134007215499878,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 51880
+    },
+    {
+      "epoch": 341.38157894736844,
+      "grad_norm": 1.2043037414550781,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 51890
+    },
+    {
+      "epoch": 341.44736842105266,
+      "grad_norm": 1.6688247919082642,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 51900
+    },
+    {
+      "epoch": 341.5131578947368,
+      "grad_norm": 1.2863478660583496,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 51910
+    },
+    {
+      "epoch": 341.57894736842104,
+      "grad_norm": 1.4102164506912231,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 51920
+    },
+    {
+      "epoch": 341.64473684210526,
+      "grad_norm": 1.6692852973937988,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 51930
+    },
+    {
+      "epoch": 341.7105263157895,
+      "grad_norm": 1.386616826057434,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 51940
+    },
+    {
+      "epoch": 341.7763157894737,
+      "grad_norm": 1.1404385566711426,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 51950
+    },
+    {
+      "epoch": 341.8421052631579,
+      "grad_norm": 1.3173713684082031,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 51960
+    },
+    {
+      "epoch": 341.9078947368421,
+      "grad_norm": 1.272148847579956,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 51970
+    },
+    {
+      "epoch": 341.9736842105263,
+      "grad_norm": 1.2954585552215576,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 51980
+    },
+    {
+      "epoch": 342.0394736842105,
+      "grad_norm": 1.546271800994873,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 51990
+    },
+    {
+      "epoch": 342.10526315789474,
+      "grad_norm": 1.21526300907135,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 52000
+    },
+    {
+      "epoch": 342.17105263157896,
+      "grad_norm": 1.2381492853164673,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 52010
+    },
+    {
+      "epoch": 342.2368421052632,
+      "grad_norm": 1.5428966283798218,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 52020
+    },
+    {
+      "epoch": 342.30263157894734,
+      "grad_norm": 1.5430893898010254,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 52030
+    },
+    {
+      "epoch": 342.36842105263156,
+      "grad_norm": 0.9203903079032898,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 52040
+    },
+    {
+      "epoch": 342.4342105263158,
+      "grad_norm": 0.9490885734558105,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 52050
+    },
+    {
+      "epoch": 342.5,
+      "grad_norm": 0.8359211087226868,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 52060
+    },
+    {
+      "epoch": 342.5657894736842,
+      "grad_norm": 1.2397630214691162,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 52070
+    },
+    {
+      "epoch": 342.63157894736844,
+      "grad_norm": 1.1610674858093262,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 52080
+    },
+    {
+      "epoch": 342.69736842105266,
+      "grad_norm": 1.3798439502716064,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 52090
+    },
+    {
+      "epoch": 342.7631578947368,
+      "grad_norm": 1.1632894277572632,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 52100
+    },
+    {
+      "epoch": 342.82894736842104,
+      "grad_norm": 1.3754557371139526,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 52110
+    },
+    {
+      "epoch": 342.89473684210526,
+      "grad_norm": 1.087170124053955,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 52120
+    },
+    {
+      "epoch": 342.9605263157895,
+      "grad_norm": 0.9013633131980896,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 52130
+    },
+    {
+      "epoch": 343.0263157894737,
+      "grad_norm": 1.5039817094802856,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 52140
+    },
+    {
+      "epoch": 343.0921052631579,
+      "grad_norm": 1.4814598560333252,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 52150
+    },
+    {
+      "epoch": 343.1578947368421,
+      "grad_norm": 1.120323896408081,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 52160
+    },
+    {
+      "epoch": 343.2236842105263,
+      "grad_norm": 0.9336338639259338,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 52170
+    },
+    {
+      "epoch": 343.2894736842105,
+      "grad_norm": 0.7774641513824463,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 52180
+    },
+    {
+      "epoch": 343.35526315789474,
+      "grad_norm": 0.8035191297531128,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 52190
+    },
+    {
+      "epoch": 343.42105263157896,
+      "grad_norm": 0.9542168974876404,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 52200
+    },
+    {
+      "epoch": 343.4868421052632,
+      "grad_norm": 1.1261277198791504,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 52210
+    },
+    {
+      "epoch": 343.55263157894734,
+      "grad_norm": 1.2652689218521118,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 52220
+    },
+    {
+      "epoch": 343.61842105263156,
+      "grad_norm": 1.4694147109985352,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 52230
+    },
+    {
+      "epoch": 343.6842105263158,
+      "grad_norm": 0.8322300910949707,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 52240
+    },
+    {
+      "epoch": 343.75,
+      "grad_norm": 1.4689432382583618,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 52250
+    },
+    {
+      "epoch": 343.8157894736842,
+      "grad_norm": 1.446466088294983,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 52260
+    },
+    {
+      "epoch": 343.88157894736844,
+      "grad_norm": 1.2933214902877808,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 52270
+    },
+    {
+      "epoch": 343.94736842105266,
+      "grad_norm": 1.245603322982788,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 52280
+    },
+    {
+      "epoch": 344.0131578947368,
+      "grad_norm": 1.0786410570144653,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 52290
+    },
+    {
+      "epoch": 344.07894736842104,
+      "grad_norm": 1.2536273002624512,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 52300
+    },
+    {
+      "epoch": 344.14473684210526,
+      "grad_norm": 1.1757087707519531,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 52310
+    },
+    {
+      "epoch": 344.2105263157895,
+      "grad_norm": 1.397163987159729,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 52320
+    },
+    {
+      "epoch": 344.2763157894737,
+      "grad_norm": 1.2626746892929077,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 52330
+    },
+    {
+      "epoch": 344.3421052631579,
+      "grad_norm": 1.5072098970413208,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 52340
+    },
+    {
+      "epoch": 344.4078947368421,
+      "grad_norm": 1.0162240266799927,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 52350
+    },
+    {
+      "epoch": 344.4736842105263,
+      "grad_norm": 1.1723800897598267,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 52360
+    },
+    {
+      "epoch": 344.5394736842105,
+      "grad_norm": 1.722672939300537,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 52370
+    },
+    {
+      "epoch": 344.60526315789474,
+      "grad_norm": 1.4377808570861816,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 52380
+    },
+    {
+      "epoch": 344.67105263157896,
+      "grad_norm": 1.4362269639968872,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 52390
+    },
+    {
+      "epoch": 344.7368421052632,
+      "grad_norm": 1.1657408475875854,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 52400
+    },
+    {
+      "epoch": 344.80263157894734,
+      "grad_norm": 1.3093397617340088,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 52410
+    },
+    {
+      "epoch": 344.86842105263156,
+      "grad_norm": 1.1783092021942139,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 52420
+    },
+    {
+      "epoch": 344.9342105263158,
+      "grad_norm": 1.0397982597351074,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 52430
+    },
+    {
+      "epoch": 345.0,
+      "grad_norm": 1.1493473052978516,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 52440
+    },
+    {
+      "epoch": 345.0657894736842,
+      "grad_norm": 1.1874574422836304,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 52450
+    },
+    {
+      "epoch": 345.13157894736844,
+      "grad_norm": 1.2612534761428833,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 52460
+    },
+    {
+      "epoch": 345.19736842105266,
+      "grad_norm": 1.2384288311004639,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 52470
+    },
+    {
+      "epoch": 345.2631578947368,
+      "grad_norm": 1.2733972072601318,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 52480
+    },
+    {
+      "epoch": 345.32894736842104,
+      "grad_norm": 1.046722650527954,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 52490
+    },
+    {
+      "epoch": 345.39473684210526,
+      "grad_norm": 1.7517563104629517,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 52500
+    },
+    {
+      "epoch": 345.4605263157895,
+      "grad_norm": 1.5715398788452148,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 52510
+    },
+    {
+      "epoch": 345.5263157894737,
+      "grad_norm": 1.4876675605773926,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 52520
+    },
+    {
+      "epoch": 345.5921052631579,
+      "grad_norm": 1.2595746517181396,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 52530
+    },
+    {
+      "epoch": 345.6578947368421,
+      "grad_norm": 1.4354819059371948,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 52540
+    },
+    {
+      "epoch": 345.7236842105263,
+      "grad_norm": 1.2486364841461182,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 52550
+    },
+    {
+      "epoch": 345.7894736842105,
+      "grad_norm": 1.4385180473327637,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 52560
+    },
+    {
+      "epoch": 345.85526315789474,
+      "grad_norm": 1.4834891557693481,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 52570
+    },
+    {
+      "epoch": 345.92105263157896,
+      "grad_norm": 1.1149605512619019,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 52580
+    },
+    {
+      "epoch": 345.9868421052632,
+      "grad_norm": 1.0604519844055176,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 52590
+    },
+    {
+      "epoch": 346.05263157894734,
+      "grad_norm": 1.482948899269104,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 52600
+    },
+    {
+      "epoch": 346.11842105263156,
+      "grad_norm": 0.9433545470237732,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 52610
+    },
+    {
+      "epoch": 346.1842105263158,
+      "grad_norm": 1.2080388069152832,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 52620
+    },
+    {
+      "epoch": 346.25,
+      "grad_norm": 1.2017712593078613,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 52630
+    },
+    {
+      "epoch": 346.3157894736842,
+      "grad_norm": 1.3862873315811157,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 52640
+    },
+    {
+      "epoch": 346.38157894736844,
+      "grad_norm": 1.4891222715377808,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 52650
+    },
+    {
+      "epoch": 346.44736842105266,
+      "grad_norm": 1.356292724609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 52660
+    },
+    {
+      "epoch": 346.5131578947368,
+      "grad_norm": 1.215113639831543,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 52670
+    },
+    {
+      "epoch": 346.57894736842104,
+      "grad_norm": 1.551318883895874,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 52680
+    },
+    {
+      "epoch": 346.64473684210526,
+      "grad_norm": 1.6980267763137817,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 52690
+    },
+    {
+      "epoch": 346.7105263157895,
+      "grad_norm": 1.528921365737915,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 52700
+    },
+    {
+      "epoch": 346.7763157894737,
+      "grad_norm": 1.3663548231124878,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 52710
+    },
+    {
+      "epoch": 346.8421052631579,
+      "grad_norm": 1.3794395923614502,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 52720
+    },
+    {
+      "epoch": 346.9078947368421,
+      "grad_norm": 1.5531320571899414,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 52730
+    },
+    {
+      "epoch": 346.9736842105263,
+      "grad_norm": 1.017933964729309,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 52740
+    },
+    {
+      "epoch": 347.0394736842105,
+      "grad_norm": 1.3421839475631714,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 52750
+    },
+    {
+      "epoch": 347.10526315789474,
+      "grad_norm": 1.4590574502944946,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 52760
+    },
+    {
+      "epoch": 347.17105263157896,
+      "grad_norm": 1.309173345565796,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 52770
+    },
+    {
+      "epoch": 347.2368421052632,
+      "grad_norm": 1.1551851034164429,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 52780
+    },
+    {
+      "epoch": 347.30263157894734,
+      "grad_norm": 1.4142370223999023,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 52790
+    },
+    {
+      "epoch": 347.36842105263156,
+      "grad_norm": 1.2437849044799805,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 52800
+    },
+    {
+      "epoch": 347.4342105263158,
+      "grad_norm": 1.08604097366333,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 52810
+    },
+    {
+      "epoch": 347.5,
+      "grad_norm": 1.0768139362335205,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 52820
+    },
+    {
+      "epoch": 347.5657894736842,
+      "grad_norm": 1.507045865058899,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 52830
+    },
+    {
+      "epoch": 347.63157894736844,
+      "grad_norm": 1.1751208305358887,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 52840
+    },
+    {
+      "epoch": 347.69736842105266,
+      "grad_norm": 1.4803016185760498,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 52850
+    },
+    {
+      "epoch": 347.7631578947368,
+      "grad_norm": 1.3335362672805786,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 52860
+    },
+    {
+      "epoch": 347.82894736842104,
+      "grad_norm": 0.9887474179267883,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 52870
+    },
+    {
+      "epoch": 347.89473684210526,
+      "grad_norm": 1.2500858306884766,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 52880
+    },
+    {
+      "epoch": 347.9605263157895,
+      "grad_norm": 1.0655533075332642,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 52890
+    },
+    {
+      "epoch": 348.0263157894737,
+      "grad_norm": 1.797389268875122,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 52900
+    },
+    {
+      "epoch": 348.0921052631579,
+      "grad_norm": 1.0135571956634521,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 52910
+    },
+    {
+      "epoch": 348.1578947368421,
+      "grad_norm": 1.4118088483810425,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 52920
+    },
+    {
+      "epoch": 348.2236842105263,
+      "grad_norm": 1.5907139778137207,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 52930
+    },
+    {
+      "epoch": 348.2894736842105,
+      "grad_norm": 0.9017720818519592,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 52940
+    },
+    {
+      "epoch": 348.35526315789474,
+      "grad_norm": 1.0268571376800537,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 52950
+    },
+    {
+      "epoch": 348.42105263157896,
+      "grad_norm": 1.0724031925201416,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 52960
+    },
+    {
+      "epoch": 348.4868421052632,
+      "grad_norm": 0.9979811906814575,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 52970
+    },
+    {
+      "epoch": 348.55263157894734,
+      "grad_norm": 1.0473777055740356,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 52980
+    },
+    {
+      "epoch": 348.61842105263156,
+      "grad_norm": 1.4823224544525146,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 52990
+    },
+    {
+      "epoch": 348.6842105263158,
+      "grad_norm": 1.1118677854537964,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 53000
+    },
+    {
+      "epoch": 348.75,
+      "grad_norm": 1.2485218048095703,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 53010
+    },
+    {
+      "epoch": 348.8157894736842,
+      "grad_norm": 1.1463638544082642,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 53020
+    },
+    {
+      "epoch": 348.88157894736844,
+      "grad_norm": 0.8539924025535583,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 53030
+    },
+    {
+      "epoch": 348.94736842105266,
+      "grad_norm": 0.826951801776886,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 53040
+    },
+    {
+      "epoch": 349.0131578947368,
+      "grad_norm": 1.3441721200942993,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 53050
+    },
+    {
+      "epoch": 349.07894736842104,
+      "grad_norm": 1.0839155912399292,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 53060
+    },
+    {
+      "epoch": 349.14473684210526,
+      "grad_norm": 1.1949419975280762,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 53070
+    },
+    {
+      "epoch": 349.2105263157895,
+      "grad_norm": 0.8186034560203552,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 53080
+    },
+    {
+      "epoch": 349.2763157894737,
+      "grad_norm": 0.8655272126197815,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 53090
+    },
+    {
+      "epoch": 349.3421052631579,
+      "grad_norm": 0.7736495733261108,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 53100
+    },
+    {
+      "epoch": 349.4078947368421,
+      "grad_norm": 0.711251437664032,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 53110
+    },
+    {
+      "epoch": 349.4736842105263,
+      "grad_norm": 1.2453712224960327,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 53120
+    },
+    {
+      "epoch": 349.5394736842105,
+      "grad_norm": 1.0265164375305176,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 53130
+    },
+    {
+      "epoch": 349.60526315789474,
+      "grad_norm": 0.9887539744377136,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 53140
+    },
+    {
+      "epoch": 349.67105263157896,
+      "grad_norm": 0.9116423726081848,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 53150
+    },
+    {
+      "epoch": 349.7368421052632,
+      "grad_norm": 1.176967978477478,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 53160
+    },
+    {
+      "epoch": 349.80263157894734,
+      "grad_norm": 1.2313488721847534,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 53170
+    },
+    {
+      "epoch": 349.86842105263156,
+      "grad_norm": 0.7793358564376831,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 53180
+    },
+    {
+      "epoch": 349.9342105263158,
+      "grad_norm": 0.9506605863571167,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 53190
+    },
+    {
+      "epoch": 350.0,
+      "grad_norm": 0.9445679783821106,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 53200
+    },
+    {
+      "epoch": 350.0657894736842,
+      "grad_norm": 1.0125964879989624,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 53210
+    },
+    {
+      "epoch": 350.13157894736844,
+      "grad_norm": 1.2516371011734009,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 53220
+    },
+    {
+      "epoch": 350.19736842105266,
+      "grad_norm": 1.4000440835952759,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 53230
+    },
+    {
+      "epoch": 350.2631578947368,
+      "grad_norm": 1.0942318439483643,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 53240
+    },
+    {
+      "epoch": 350.32894736842104,
+      "grad_norm": 1.0864887237548828,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 53250
+    },
+    {
+      "epoch": 350.39473684210526,
+      "grad_norm": 1.2966852188110352,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 53260
+    },
+    {
+      "epoch": 350.4605263157895,
+      "grad_norm": 1.227264404296875,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 53270
+    },
+    {
+      "epoch": 350.5263157894737,
+      "grad_norm": 1.1729018688201904,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 53280
+    },
+    {
+      "epoch": 350.5921052631579,
+      "grad_norm": 0.7288147211074829,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 53290
+    },
+    {
+      "epoch": 350.6578947368421,
+      "grad_norm": 0.893334150314331,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 53300
+    },
+    {
+      "epoch": 350.7236842105263,
+      "grad_norm": 1.2939788103103638,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 53310
+    },
+    {
+      "epoch": 350.7894736842105,
+      "grad_norm": 0.9002676010131836,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 53320
+    },
+    {
+      "epoch": 350.85526315789474,
+      "grad_norm": 0.9034085869789124,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 53330
+    },
+    {
+      "epoch": 350.92105263157896,
+      "grad_norm": 1.2274662256240845,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 53340
+    },
+    {
+      "epoch": 350.9868421052632,
+      "grad_norm": 0.969467043876648,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 53350
+    },
+    {
+      "epoch": 351.05263157894734,
+      "grad_norm": 1.1215009689331055,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 53360
+    },
+    {
+      "epoch": 351.11842105263156,
+      "grad_norm": 1.4852532148361206,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 53370
+    },
+    {
+      "epoch": 351.1842105263158,
+      "grad_norm": 1.1253795623779297,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 53380
+    },
+    {
+      "epoch": 351.25,
+      "grad_norm": 0.8905407786369324,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 53390
+    },
+    {
+      "epoch": 351.3157894736842,
+      "grad_norm": 1.2344647645950317,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 53400
+    },
+    {
+      "epoch": 351.38157894736844,
+      "grad_norm": 0.9669228196144104,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 53410
+    },
+    {
+      "epoch": 351.44736842105266,
+      "grad_norm": 1.300685167312622,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 53420
+    },
+    {
+      "epoch": 351.5131578947368,
+      "grad_norm": 1.1283105611801147,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 53430
+    },
+    {
+      "epoch": 351.57894736842104,
+      "grad_norm": 1.390939474105835,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 53440
+    },
+    {
+      "epoch": 351.64473684210526,
+      "grad_norm": 1.187654972076416,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 53450
+    },
+    {
+      "epoch": 351.7105263157895,
+      "grad_norm": 0.9280898571014404,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 53460
+    },
+    {
+      "epoch": 351.7763157894737,
+      "grad_norm": 1.1642065048217773,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 53470
+    },
+    {
+      "epoch": 351.8421052631579,
+      "grad_norm": 1.3112080097198486,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 53480
+    },
+    {
+      "epoch": 351.9078947368421,
+      "grad_norm": 1.1810340881347656,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 53490
+    },
+    {
+      "epoch": 351.9736842105263,
+      "grad_norm": 1.1890909671783447,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 53500
+    },
+    {
+      "epoch": 352.0394736842105,
+      "grad_norm": 0.9350360035896301,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 53510
+    },
+    {
+      "epoch": 352.10526315789474,
+      "grad_norm": 1.0225274562835693,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 53520
+    },
+    {
+      "epoch": 352.17105263157896,
+      "grad_norm": 1.1939771175384521,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 53530
+    },
+    {
+      "epoch": 352.2368421052632,
+      "grad_norm": 0.9395789504051208,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 53540
+    },
+    {
+      "epoch": 352.30263157894734,
+      "grad_norm": 1.4980489015579224,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 53550
+    },
+    {
+      "epoch": 352.36842105263156,
+      "grad_norm": 1.4125310182571411,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 53560
+    },
+    {
+      "epoch": 352.4342105263158,
+      "grad_norm": 1.1465486288070679,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 53570
+    },
+    {
+      "epoch": 352.5,
+      "grad_norm": 1.5160516500473022,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 53580
+    },
+    {
+      "epoch": 352.5657894736842,
+      "grad_norm": 1.0091547966003418,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 53590
+    },
+    {
+      "epoch": 352.63157894736844,
+      "grad_norm": 1.6452295780181885,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 53600
+    },
+    {
+      "epoch": 352.69736842105266,
+      "grad_norm": 0.8045058846473694,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 53610
+    },
+    {
+      "epoch": 352.7631578947368,
+      "grad_norm": 0.9372885823249817,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 53620
+    },
+    {
+      "epoch": 352.82894736842104,
+      "grad_norm": 1.2400946617126465,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 53630
+    },
+    {
+      "epoch": 352.89473684210526,
+      "grad_norm": 1.2363208532333374,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 53640
+    },
+    {
+      "epoch": 352.9605263157895,
+      "grad_norm": 1.4583326578140259,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 53650
+    },
+    {
+      "epoch": 353.0263157894737,
+      "grad_norm": 1.4135735034942627,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 53660
+    },
+    {
+      "epoch": 353.0921052631579,
+      "grad_norm": 0.8179410099983215,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 53670
+    },
+    {
+      "epoch": 353.1578947368421,
+      "grad_norm": 1.1129777431488037,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 53680
+    },
+    {
+      "epoch": 353.2236842105263,
+      "grad_norm": 0.9328417778015137,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 53690
+    },
+    {
+      "epoch": 353.2894736842105,
+      "grad_norm": 1.0493443012237549,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 53700
+    },
+    {
+      "epoch": 353.35526315789474,
+      "grad_norm": 1.4180327653884888,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 53710
+    },
+    {
+      "epoch": 353.42105263157896,
+      "grad_norm": 1.10849130153656,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 53720
+    },
+    {
+      "epoch": 353.4868421052632,
+      "grad_norm": 0.9594805836677551,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 53730
+    },
+    {
+      "epoch": 353.55263157894734,
+      "grad_norm": 1.55135977268219,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 53740
+    },
+    {
+      "epoch": 353.61842105263156,
+      "grad_norm": 1.0522472858428955,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 53750
+    },
+    {
+      "epoch": 353.6842105263158,
+      "grad_norm": 1.2985424995422363,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 53760
+    },
+    {
+      "epoch": 353.75,
+      "grad_norm": 1.2076297998428345,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 53770
+    },
+    {
+      "epoch": 353.8157894736842,
+      "grad_norm": 1.4441065788269043,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 53780
+    },
+    {
+      "epoch": 353.88157894736844,
+      "grad_norm": 1.4848010540008545,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 53790
+    },
+    {
+      "epoch": 353.94736842105266,
+      "grad_norm": 1.2929956912994385,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 53800
+    },
+    {
+      "epoch": 354.0131578947368,
+      "grad_norm": 1.4430453777313232,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 53810
+    },
+    {
+      "epoch": 354.07894736842104,
+      "grad_norm": 1.5679028034210205,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 53820
+    },
+    {
+      "epoch": 354.14473684210526,
+      "grad_norm": 0.8929752111434937,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 53830
+    },
+    {
+      "epoch": 354.2105263157895,
+      "grad_norm": 1.3934887647628784,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 53840
+    },
+    {
+      "epoch": 354.2763157894737,
+      "grad_norm": 1.3150675296783447,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 53850
+    },
+    {
+      "epoch": 354.3421052631579,
+      "grad_norm": 1.1540673971176147,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 53860
+    },
+    {
+      "epoch": 354.4078947368421,
+      "grad_norm": 1.2295771837234497,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 53870
+    },
+    {
+      "epoch": 354.4736842105263,
+      "grad_norm": 1.2992318868637085,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 53880
+    },
+    {
+      "epoch": 354.5394736842105,
+      "grad_norm": 1.3661577701568604,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 53890
+    },
+    {
+      "epoch": 354.60526315789474,
+      "grad_norm": 1.475778579711914,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 53900
+    },
+    {
+      "epoch": 354.67105263157896,
+      "grad_norm": 1.3246492147445679,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 53910
+    },
+    {
+      "epoch": 354.7368421052632,
+      "grad_norm": 1.3882511854171753,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 53920
+    },
+    {
+      "epoch": 354.80263157894734,
+      "grad_norm": 0.9738257527351379,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 53930
+    },
+    {
+      "epoch": 354.86842105263156,
+      "grad_norm": 1.405836820602417,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 53940
+    },
+    {
+      "epoch": 354.9342105263158,
+      "grad_norm": 1.3983672857284546,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 53950
+    },
+    {
+      "epoch": 355.0,
+      "grad_norm": 1.1816673278808594,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 53960
+    },
+    {
+      "epoch": 355.0657894736842,
+      "grad_norm": 0.901597797870636,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 53970
+    },
+    {
+      "epoch": 355.13157894736844,
+      "grad_norm": 0.8046585321426392,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 53980
+    },
+    {
+      "epoch": 355.19736842105266,
+      "grad_norm": 0.8949022889137268,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 53990
+    },
+    {
+      "epoch": 355.2631578947368,
+      "grad_norm": 0.8433843851089478,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 54000
+    },
+    {
+      "epoch": 355.32894736842104,
+      "grad_norm": 1.3329582214355469,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 54010
+    },
+    {
+      "epoch": 355.39473684210526,
+      "grad_norm": 1.203683614730835,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 54020
+    },
+    {
+      "epoch": 355.4605263157895,
+      "grad_norm": 1.1506283283233643,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 54030
+    },
+    {
+      "epoch": 355.5263157894737,
+      "grad_norm": 1.4512953758239746,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 54040
+    },
+    {
+      "epoch": 355.5921052631579,
+      "grad_norm": 1.065973162651062,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 54050
+    },
+    {
+      "epoch": 355.6578947368421,
+      "grad_norm": 0.8735188245773315,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 54060
+    },
+    {
+      "epoch": 355.7236842105263,
+      "grad_norm": 1.3658928871154785,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 54070
+    },
+    {
+      "epoch": 355.7894736842105,
+      "grad_norm": 1.037252426147461,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 54080
+    },
+    {
+      "epoch": 355.85526315789474,
+      "grad_norm": 1.2813448905944824,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 54090
+    },
+    {
+      "epoch": 355.92105263157896,
+      "grad_norm": 1.114138126373291,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 54100
+    },
+    {
+      "epoch": 355.9868421052632,
+      "grad_norm": 1.242543339729309,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 54110
+    },
+    {
+      "epoch": 356.05263157894734,
+      "grad_norm": 1.8601607084274292,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 54120
+    },
+    {
+      "epoch": 356.11842105263156,
+      "grad_norm": 1.521649718284607,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 54130
+    },
+    {
+      "epoch": 356.1842105263158,
+      "grad_norm": 1.109571099281311,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 54140
+    },
+    {
+      "epoch": 356.25,
+      "grad_norm": 1.3010332584381104,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 54150
+    },
+    {
+      "epoch": 356.3157894736842,
+      "grad_norm": 1.2290462255477905,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 54160
+    },
+    {
+      "epoch": 356.38157894736844,
+      "grad_norm": 1.0519379377365112,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 54170
+    },
+    {
+      "epoch": 356.44736842105266,
+      "grad_norm": 1.036097764968872,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 54180
+    },
+    {
+      "epoch": 356.5131578947368,
+      "grad_norm": 1.120662808418274,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 54190
+    },
+    {
+      "epoch": 356.57894736842104,
+      "grad_norm": 0.9310142993927002,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 54200
+    },
+    {
+      "epoch": 356.64473684210526,
+      "grad_norm": 1.009954810142517,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 54210
+    },
+    {
+      "epoch": 356.7105263157895,
+      "grad_norm": 0.9205268621444702,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 54220
+    },
+    {
+      "epoch": 356.7763157894737,
+      "grad_norm": 1.1177328824996948,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 54230
+    },
+    {
+      "epoch": 356.8421052631579,
+      "grad_norm": 1.0612447261810303,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 54240
+    },
+    {
+      "epoch": 356.9078947368421,
+      "grad_norm": 0.7625839710235596,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 54250
+    },
+    {
+      "epoch": 356.9736842105263,
+      "grad_norm": 0.8989898562431335,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 54260
+    },
+    {
+      "epoch": 357.0394736842105,
+      "grad_norm": 1.3405532836914062,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 54270
+    },
+    {
+      "epoch": 357.10526315789474,
+      "grad_norm": 1.3939285278320312,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 54280
+    },
+    {
+      "epoch": 357.17105263157896,
+      "grad_norm": 1.1620155572891235,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 54290
+    },
+    {
+      "epoch": 357.2368421052632,
+      "grad_norm": 1.2777410745620728,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 54300
+    },
+    {
+      "epoch": 357.30263157894734,
+      "grad_norm": 1.271978735923767,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 54310
+    },
+    {
+      "epoch": 357.36842105263156,
+      "grad_norm": 1.060716152191162,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 54320
+    },
+    {
+      "epoch": 357.4342105263158,
+      "grad_norm": 1.1990467309951782,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 54330
+    },
+    {
+      "epoch": 357.5,
+      "grad_norm": 1.0680278539657593,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 54340
+    },
+    {
+      "epoch": 357.5657894736842,
+      "grad_norm": 1.2417722940444946,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 54350
+    },
+    {
+      "epoch": 357.63157894736844,
+      "grad_norm": 1.0456074476242065,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 54360
+    },
+    {
+      "epoch": 357.69736842105266,
+      "grad_norm": 1.1133095026016235,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 54370
+    },
+    {
+      "epoch": 357.7631578947368,
+      "grad_norm": 1.355271816253662,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 54380
+    },
+    {
+      "epoch": 357.82894736842104,
+      "grad_norm": 1.0156669616699219,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 54390
+    },
+    {
+      "epoch": 357.89473684210526,
+      "grad_norm": 1.571312427520752,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 54400
+    },
+    {
+      "epoch": 357.9605263157895,
+      "grad_norm": 1.2970470190048218,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 54410
+    },
+    {
+      "epoch": 358.0263157894737,
+      "grad_norm": 1.343517780303955,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 54420
+    },
+    {
+      "epoch": 358.0921052631579,
+      "grad_norm": 1.6380618810653687,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 54430
+    },
+    {
+      "epoch": 358.1578947368421,
+      "grad_norm": 1.1087614297866821,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 54440
+    },
+    {
+      "epoch": 358.2236842105263,
+      "grad_norm": 1.523401141166687,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 54450
+    },
+    {
+      "epoch": 358.2894736842105,
+      "grad_norm": 1.1688175201416016,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 54460
+    },
+    {
+      "epoch": 358.35526315789474,
+      "grad_norm": 1.3475173711776733,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 54470
+    },
+    {
+      "epoch": 358.42105263157896,
+      "grad_norm": 1.240882396697998,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 54480
+    },
+    {
+      "epoch": 358.4868421052632,
+      "grad_norm": 1.5280466079711914,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 54490
+    },
+    {
+      "epoch": 358.55263157894734,
+      "grad_norm": 1.313010334968567,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 54500
+    },
+    {
+      "epoch": 358.61842105263156,
+      "grad_norm": 1.4535280466079712,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 54510
+    },
+    {
+      "epoch": 358.6842105263158,
+      "grad_norm": 0.7499117255210876,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 54520
+    },
+    {
+      "epoch": 358.75,
+      "grad_norm": 1.1150604486465454,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 54530
+    },
+    {
+      "epoch": 358.8157894736842,
+      "grad_norm": 1.5090540647506714,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 54540
+    },
+    {
+      "epoch": 358.88157894736844,
+      "grad_norm": 1.2874553203582764,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 54550
+    },
+    {
+      "epoch": 358.94736842105266,
+      "grad_norm": 1.4591070413589478,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 54560
+    },
+    {
+      "epoch": 359.0131578947368,
+      "grad_norm": 1.5048354864120483,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 54570
+    },
+    {
+      "epoch": 359.07894736842104,
+      "grad_norm": 1.1533069610595703,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 54580
+    },
+    {
+      "epoch": 359.14473684210526,
+      "grad_norm": 1.287916660308838,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 54590
+    },
+    {
+      "epoch": 359.2105263157895,
+      "grad_norm": 1.1660511493682861,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 54600
+    },
+    {
+      "epoch": 359.2763157894737,
+      "grad_norm": 1.1380329132080078,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 54610
+    },
+    {
+      "epoch": 359.3421052631579,
+      "grad_norm": 1.1352312564849854,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 54620
+    },
+    {
+      "epoch": 359.4078947368421,
+      "grad_norm": 1.5034499168395996,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 54630
+    },
+    {
+      "epoch": 359.4736842105263,
+      "grad_norm": 1.6546825170516968,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 54640
+    },
+    {
+      "epoch": 359.5394736842105,
+      "grad_norm": 1.4479390382766724,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 54650
+    },
+    {
+      "epoch": 359.60526315789474,
+      "grad_norm": 1.404312014579773,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 54660
+    },
+    {
+      "epoch": 359.67105263157896,
+      "grad_norm": 1.507351040840149,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 54670
+    },
+    {
+      "epoch": 359.7368421052632,
+      "grad_norm": 1.2614494562149048,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 54680
+    },
+    {
+      "epoch": 359.80263157894734,
+      "grad_norm": 0.779187798500061,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 54690
+    },
+    {
+      "epoch": 359.86842105263156,
+      "grad_norm": 1.3693275451660156,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 54700
+    },
+    {
+      "epoch": 359.9342105263158,
+      "grad_norm": 1.4360358715057373,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 54710
+    },
+    {
+      "epoch": 360.0,
+      "grad_norm": 1.4990230798721313,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 54720
+    },
+    {
+      "epoch": 360.0657894736842,
+      "grad_norm": 1.2829705476760864,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 54730
+    },
+    {
+      "epoch": 360.13157894736844,
+      "grad_norm": 0.8983616232872009,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 54740
+    },
+    {
+      "epoch": 360.19736842105266,
+      "grad_norm": 1.4845367670059204,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 54750
+    },
+    {
+      "epoch": 360.2631578947368,
+      "grad_norm": 1.124158501625061,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 54760
+    },
+    {
+      "epoch": 360.32894736842104,
+      "grad_norm": 1.3822499513626099,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 54770
+    },
+    {
+      "epoch": 360.39473684210526,
+      "grad_norm": 1.2073116302490234,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 54780
+    },
+    {
+      "epoch": 360.4605263157895,
+      "grad_norm": 1.0493237972259521,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 54790
+    },
+    {
+      "epoch": 360.5263157894737,
+      "grad_norm": 1.4061884880065918,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 54800
+    },
+    {
+      "epoch": 360.5921052631579,
+      "grad_norm": 2.0674805641174316,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 54810
+    },
+    {
+      "epoch": 360.6578947368421,
+      "grad_norm": 1.6307419538497925,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 54820
+    },
+    {
+      "epoch": 360.7236842105263,
+      "grad_norm": 1.5388418436050415,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 54830
+    },
+    {
+      "epoch": 360.7894736842105,
+      "grad_norm": 1.1315034627914429,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 54840
+    },
+    {
+      "epoch": 360.85526315789474,
+      "grad_norm": 1.5357550382614136,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 54850
+    },
+    {
+      "epoch": 360.92105263157896,
+      "grad_norm": 1.019702672958374,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 54860
+    },
+    {
+      "epoch": 360.9868421052632,
+      "grad_norm": 1.2998785972595215,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 54870
+    },
+    {
+      "epoch": 361.05263157894734,
+      "grad_norm": 0.9051557183265686,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 54880
+    },
+    {
+      "epoch": 361.11842105263156,
+      "grad_norm": 1.275030493736267,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 54890
+    },
+    {
+      "epoch": 361.1842105263158,
+      "grad_norm": 0.9858371615409851,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 54900
+    },
+    {
+      "epoch": 361.25,
+      "grad_norm": 1.1971203088760376,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 54910
+    },
+    {
+      "epoch": 361.3157894736842,
+      "grad_norm": 1.2768568992614746,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 54920
+    },
+    {
+      "epoch": 361.38157894736844,
+      "grad_norm": 1.199203372001648,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 54930
+    },
+    {
+      "epoch": 361.44736842105266,
+      "grad_norm": 0.8393948078155518,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 54940
+    },
+    {
+      "epoch": 361.5131578947368,
+      "grad_norm": 1.0057693719863892,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 54950
+    },
+    {
+      "epoch": 361.57894736842104,
+      "grad_norm": 1.328454852104187,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 54960
+    },
+    {
+      "epoch": 361.64473684210526,
+      "grad_norm": 1.4820215702056885,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 54970
+    },
+    {
+      "epoch": 361.7105263157895,
+      "grad_norm": 1.2273774147033691,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 54980
+    },
+    {
+      "epoch": 361.7763157894737,
+      "grad_norm": 0.9895418882369995,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 54990
+    },
+    {
+      "epoch": 361.8421052631579,
+      "grad_norm": 1.2177479267120361,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 55000
+    },
+    {
+      "epoch": 361.9078947368421,
+      "grad_norm": 1.5505794286727905,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 55010
+    },
+    {
+      "epoch": 361.9736842105263,
+      "grad_norm": 1.2999932765960693,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 55020
+    },
+    {
+      "epoch": 362.0394736842105,
+      "grad_norm": 1.0745863914489746,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 55030
+    },
+    {
+      "epoch": 362.10526315789474,
+      "grad_norm": 1.1478854417800903,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 55040
+    },
+    {
+      "epoch": 362.17105263157896,
+      "grad_norm": 1.1435706615447998,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 55050
+    },
+    {
+      "epoch": 362.2368421052632,
+      "grad_norm": 1.1603302955627441,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 55060
+    },
+    {
+      "epoch": 362.30263157894734,
+      "grad_norm": 1.1007798910140991,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 55070
+    },
+    {
+      "epoch": 362.36842105263156,
+      "grad_norm": 0.8749071359634399,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 55080
+    },
+    {
+      "epoch": 362.4342105263158,
+      "grad_norm": 1.287734031677246,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 55090
+    },
+    {
+      "epoch": 362.5,
+      "grad_norm": 1.153018593788147,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 55100
+    },
+    {
+      "epoch": 362.5657894736842,
+      "grad_norm": 1.2814031839370728,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 55110
+    },
+    {
+      "epoch": 362.63157894736844,
+      "grad_norm": 1.211867094039917,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 55120
+    },
+    {
+      "epoch": 362.69736842105266,
+      "grad_norm": 1.1564773321151733,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 55130
+    },
+    {
+      "epoch": 362.7631578947368,
+      "grad_norm": 0.9832426309585571,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 55140
+    },
+    {
+      "epoch": 362.82894736842104,
+      "grad_norm": 1.05033278465271,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 55150
+    },
+    {
+      "epoch": 362.89473684210526,
+      "grad_norm": 1.2688597440719604,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 55160
+    },
+    {
+      "epoch": 362.9605263157895,
+      "grad_norm": 1.314858078956604,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 55170
+    },
+    {
+      "epoch": 363.0263157894737,
+      "grad_norm": 1.2743005752563477,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 55180
+    },
+    {
+      "epoch": 363.0921052631579,
+      "grad_norm": 1.3456010818481445,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 55190
+    },
+    {
+      "epoch": 363.1578947368421,
+      "grad_norm": 1.2974659204483032,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 55200
+    },
+    {
+      "epoch": 363.2236842105263,
+      "grad_norm": 1.0319002866744995,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 55210
+    },
+    {
+      "epoch": 363.2894736842105,
+      "grad_norm": 1.074663519859314,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 55220
+    },
+    {
+      "epoch": 363.35526315789474,
+      "grad_norm": 1.0586227178573608,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 55230
+    },
+    {
+      "epoch": 363.42105263157896,
+      "grad_norm": 1.1722004413604736,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 55240
+    },
+    {
+      "epoch": 363.4868421052632,
+      "grad_norm": 1.1225117444992065,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 55250
+    },
+    {
+      "epoch": 363.55263157894734,
+      "grad_norm": 0.9017395973205566,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 55260
+    },
+    {
+      "epoch": 363.61842105263156,
+      "grad_norm": 1.1716728210449219,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 55270
+    },
+    {
+      "epoch": 363.6842105263158,
+      "grad_norm": 0.8839274644851685,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 55280
+    },
+    {
+      "epoch": 363.75,
+      "grad_norm": 1.1430319547653198,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 55290
+    },
+    {
+      "epoch": 363.8157894736842,
+      "grad_norm": 1.2584179639816284,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 55300
+    },
+    {
+      "epoch": 363.88157894736844,
+      "grad_norm": 1.3413207530975342,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 55310
+    },
+    {
+      "epoch": 363.94736842105266,
+      "grad_norm": 1.4143167734146118,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 55320
+    },
+    {
+      "epoch": 364.0131578947368,
+      "grad_norm": 1.290185570716858,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 55330
+    },
+    {
+      "epoch": 364.07894736842104,
+      "grad_norm": 1.241997241973877,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 55340
+    },
+    {
+      "epoch": 364.14473684210526,
+      "grad_norm": 1.200785756111145,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 55350
+    },
+    {
+      "epoch": 364.2105263157895,
+      "grad_norm": 1.1258078813552856,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 55360
+    },
+    {
+      "epoch": 364.2763157894737,
+      "grad_norm": 1.1537657976150513,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 55370
+    },
+    {
+      "epoch": 364.3421052631579,
+      "grad_norm": 1.3146229982376099,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 55380
+    },
+    {
+      "epoch": 364.4078947368421,
+      "grad_norm": 1.3080146312713623,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 55390
+    },
+    {
+      "epoch": 364.4736842105263,
+      "grad_norm": 1.3526312112808228,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 55400
+    },
+    {
+      "epoch": 364.5394736842105,
+      "grad_norm": 1.6095331907272339,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 55410
+    },
+    {
+      "epoch": 364.60526315789474,
+      "grad_norm": 1.4046205282211304,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 55420
+    },
+    {
+      "epoch": 364.67105263157896,
+      "grad_norm": 1.0746979713439941,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 55430
+    },
+    {
+      "epoch": 364.7368421052632,
+      "grad_norm": 1.3545022010803223,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 55440
+    },
+    {
+      "epoch": 364.80263157894734,
+      "grad_norm": 1.36902916431427,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 55450
+    },
+    {
+      "epoch": 364.86842105263156,
+      "grad_norm": 1.1085587739944458,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 55460
+    },
+    {
+      "epoch": 364.9342105263158,
+      "grad_norm": 0.8094392418861389,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 55470
+    },
+    {
+      "epoch": 365.0,
+      "grad_norm": 0.8859060406684875,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 55480
+    },
+    {
+      "epoch": 365.0657894736842,
+      "grad_norm": 1.0482323169708252,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 55490
+    },
+    {
+      "epoch": 365.13157894736844,
+      "grad_norm": 1.1933088302612305,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 55500
+    },
+    {
+      "epoch": 365.19736842105266,
+      "grad_norm": 1.1254971027374268,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 55510
+    },
+    {
+      "epoch": 365.2631578947368,
+      "grad_norm": 1.3154945373535156,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 55520
+    },
+    {
+      "epoch": 365.32894736842104,
+      "grad_norm": 0.9752078056335449,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 55530
+    },
+    {
+      "epoch": 365.39473684210526,
+      "grad_norm": 1.297957420349121,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 55540
+    },
+    {
+      "epoch": 365.4605263157895,
+      "grad_norm": 0.6452957391738892,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 55550
+    },
+    {
+      "epoch": 365.5263157894737,
+      "grad_norm": 1.1209876537322998,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 55560
+    },
+    {
+      "epoch": 365.5921052631579,
+      "grad_norm": 1.1463710069656372,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 55570
+    },
+    {
+      "epoch": 365.6578947368421,
+      "grad_norm": 0.9191970825195312,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 55580
+    },
+    {
+      "epoch": 365.7236842105263,
+      "grad_norm": 1.2849448919296265,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 55590
+    },
+    {
+      "epoch": 365.7894736842105,
+      "grad_norm": 1.0626157522201538,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 55600
+    },
+    {
+      "epoch": 365.85526315789474,
+      "grad_norm": 1.1464340686798096,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 55610
+    },
+    {
+      "epoch": 365.92105263157896,
+      "grad_norm": 0.7038142681121826,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 55620
+    },
+    {
+      "epoch": 365.9868421052632,
+      "grad_norm": 1.4112557172775269,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 55630
+    },
+    {
+      "epoch": 366.05263157894734,
+      "grad_norm": 1.230036735534668,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 55640
+    },
+    {
+      "epoch": 366.11842105263156,
+      "grad_norm": 0.9362917542457581,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 55650
+    },
+    {
+      "epoch": 366.1842105263158,
+      "grad_norm": 1.2693713903427124,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 55660
+    },
+    {
+      "epoch": 366.25,
+      "grad_norm": 1.3503483533859253,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 55670
+    },
+    {
+      "epoch": 366.3157894736842,
+      "grad_norm": 1.1417988538742065,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 55680
+    },
+    {
+      "epoch": 366.38157894736844,
+      "grad_norm": 1.155261516571045,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 55690
+    },
+    {
+      "epoch": 366.44736842105266,
+      "grad_norm": 1.2800606489181519,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 55700
+    },
+    {
+      "epoch": 366.5131578947368,
+      "grad_norm": 1.3735575675964355,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 55710
+    },
+    {
+      "epoch": 366.57894736842104,
+      "grad_norm": 1.050938367843628,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 55720
+    },
+    {
+      "epoch": 366.64473684210526,
+      "grad_norm": 0.9390007257461548,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 55730
+    },
+    {
+      "epoch": 366.7105263157895,
+      "grad_norm": 1.1883047819137573,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 55740
+    },
+    {
+      "epoch": 366.7763157894737,
+      "grad_norm": 1.5340079069137573,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 55750
+    },
+    {
+      "epoch": 366.8421052631579,
+      "grad_norm": 1.2325025796890259,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 55760
+    },
+    {
+      "epoch": 366.9078947368421,
+      "grad_norm": 1.1925488710403442,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 55770
+    },
+    {
+      "epoch": 366.9736842105263,
+      "grad_norm": 1.3321561813354492,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 55780
+    },
+    {
+      "epoch": 367.0394736842105,
+      "grad_norm": 1.380933165550232,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 55790
+    },
+    {
+      "epoch": 367.10526315789474,
+      "grad_norm": 1.200136423110962,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 55800
+    },
+    {
+      "epoch": 367.17105263157896,
+      "grad_norm": 0.8746941685676575,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 55810
+    },
+    {
+      "epoch": 367.2368421052632,
+      "grad_norm": 1.1686458587646484,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 55820
+    },
+    {
+      "epoch": 367.30263157894734,
+      "grad_norm": 0.9047589302062988,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 55830
+    },
+    {
+      "epoch": 367.36842105263156,
+      "grad_norm": 1.1924694776535034,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 55840
+    },
+    {
+      "epoch": 367.4342105263158,
+      "grad_norm": 1.4154642820358276,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 55850
+    },
+    {
+      "epoch": 367.5,
+      "grad_norm": 1.3152343034744263,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 55860
+    },
+    {
+      "epoch": 367.5657894736842,
+      "grad_norm": 1.2152726650238037,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 55870
+    },
+    {
+      "epoch": 367.63157894736844,
+      "grad_norm": 1.0928616523742676,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 55880
+    },
+    {
+      "epoch": 367.69736842105266,
+      "grad_norm": 1.3509669303894043,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 55890
+    },
+    {
+      "epoch": 367.7631578947368,
+      "grad_norm": 1.1209146976470947,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 55900
+    },
+    {
+      "epoch": 367.82894736842104,
+      "grad_norm": 1.1768624782562256,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 55910
+    },
+    {
+      "epoch": 367.89473684210526,
+      "grad_norm": 1.249316930770874,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 55920
+    },
+    {
+      "epoch": 367.9605263157895,
+      "grad_norm": 1.1277846097946167,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 55930
+    },
+    {
+      "epoch": 368.0263157894737,
+      "grad_norm": 1.326911449432373,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 55940
+    },
+    {
+      "epoch": 368.0921052631579,
+      "grad_norm": 1.5124350786209106,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 55950
+    },
+    {
+      "epoch": 368.1578947368421,
+      "grad_norm": 0.921715259552002,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 55960
+    },
+    {
+      "epoch": 368.2236842105263,
+      "grad_norm": 1.4262442588806152,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 55970
+    },
+    {
+      "epoch": 368.2894736842105,
+      "grad_norm": 1.1071768999099731,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 55980
+    },
+    {
+      "epoch": 368.35526315789474,
+      "grad_norm": 1.1342467069625854,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 55990
+    },
+    {
+      "epoch": 368.42105263157896,
+      "grad_norm": 1.218801736831665,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 56000
+    },
+    {
+      "epoch": 368.4868421052632,
+      "grad_norm": 1.243741750717163,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 56010
+    },
+    {
+      "epoch": 368.55263157894734,
+      "grad_norm": 1.1212338209152222,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 56020
+    },
+    {
+      "epoch": 368.61842105263156,
+      "grad_norm": 1.3740099668502808,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 56030
+    },
+    {
+      "epoch": 368.6842105263158,
+      "grad_norm": 1.2437816858291626,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 56040
+    },
+    {
+      "epoch": 368.75,
+      "grad_norm": 1.3597102165222168,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 56050
+    },
+    {
+      "epoch": 368.8157894736842,
+      "grad_norm": 1.264675498008728,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 56060
+    },
+    {
+      "epoch": 368.88157894736844,
+      "grad_norm": 1.0519795417785645,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 56070
+    },
+    {
+      "epoch": 368.94736842105266,
+      "grad_norm": 1.288150429725647,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 56080
+    },
+    {
+      "epoch": 369.0131578947368,
+      "grad_norm": 1.3312660455703735,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 56090
+    },
+    {
+      "epoch": 369.07894736842104,
+      "grad_norm": 1.0909931659698486,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 56100
+    },
+    {
+      "epoch": 369.14473684210526,
+      "grad_norm": 1.1233609914779663,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 56110
+    },
+    {
+      "epoch": 369.2105263157895,
+      "grad_norm": 1.1134326457977295,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 56120
+    },
+    {
+      "epoch": 369.2763157894737,
+      "grad_norm": 1.4635008573532104,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 56130
+    },
+    {
+      "epoch": 369.3421052631579,
+      "grad_norm": 1.5184545516967773,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 56140
+    },
+    {
+      "epoch": 369.4078947368421,
+      "grad_norm": 1.1048532724380493,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 56150
+    },
+    {
+      "epoch": 369.4736842105263,
+      "grad_norm": 1.1498737335205078,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 56160
+    },
+    {
+      "epoch": 369.5394736842105,
+      "grad_norm": 1.2778600454330444,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 56170
+    },
+    {
+      "epoch": 369.60526315789474,
+      "grad_norm": 1.0318385362625122,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 56180
+    },
+    {
+      "epoch": 369.67105263157896,
+      "grad_norm": 0.9718679785728455,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 56190
+    },
+    {
+      "epoch": 369.7368421052632,
+      "grad_norm": 1.2061275243759155,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 56200
+    },
+    {
+      "epoch": 369.80263157894734,
+      "grad_norm": 1.2633416652679443,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 56210
+    },
+    {
+      "epoch": 369.86842105263156,
+      "grad_norm": 1.3707966804504395,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 56220
+    },
+    {
+      "epoch": 369.9342105263158,
+      "grad_norm": 1.2290523052215576,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 56230
+    },
+    {
+      "epoch": 370.0,
+      "grad_norm": 0.9901204109191895,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 56240
+    },
+    {
+      "epoch": 370.0657894736842,
+      "grad_norm": 1.1951624155044556,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 56250
+    },
+    {
+      "epoch": 370.13157894736844,
+      "grad_norm": 1.1654387712478638,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 56260
+    },
+    {
+      "epoch": 370.19736842105266,
+      "grad_norm": 1.5832988023757935,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 56270
+    },
+    {
+      "epoch": 370.2631578947368,
+      "grad_norm": 1.0225611925125122,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 56280
+    },
+    {
+      "epoch": 370.32894736842104,
+      "grad_norm": 1.5608348846435547,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 56290
+    },
+    {
+      "epoch": 370.39473684210526,
+      "grad_norm": 1.4639538526535034,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 56300
+    },
+    {
+      "epoch": 370.4605263157895,
+      "grad_norm": 1.4396363496780396,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 56310
+    },
+    {
+      "epoch": 370.5263157894737,
+      "grad_norm": 0.938585638999939,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 56320
+    },
+    {
+      "epoch": 370.5921052631579,
+      "grad_norm": 1.232274055480957,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 56330
+    },
+    {
+      "epoch": 370.6578947368421,
+      "grad_norm": 1.1240586042404175,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 56340
+    },
+    {
+      "epoch": 370.7236842105263,
+      "grad_norm": 1.164602279663086,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 56350
+    },
+    {
+      "epoch": 370.7894736842105,
+      "grad_norm": 1.4600279331207275,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 56360
+    },
+    {
+      "epoch": 370.85526315789474,
+      "grad_norm": 1.1497042179107666,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 56370
+    },
+    {
+      "epoch": 370.92105263157896,
+      "grad_norm": 1.1283025741577148,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 56380
+    },
+    {
+      "epoch": 370.9868421052632,
+      "grad_norm": 1.196718454360962,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 56390
+    },
+    {
+      "epoch": 371.05263157894734,
+      "grad_norm": 0.8738747239112854,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 56400
+    },
+    {
+      "epoch": 371.11842105263156,
+      "grad_norm": 1.2236226797103882,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 56410
+    },
+    {
+      "epoch": 371.1842105263158,
+      "grad_norm": 1.3389105796813965,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 56420
+    },
+    {
+      "epoch": 371.25,
+      "grad_norm": 1.1783345937728882,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 56430
+    },
+    {
+      "epoch": 371.3157894736842,
+      "grad_norm": 1.1184672117233276,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 56440
+    },
+    {
+      "epoch": 371.38157894736844,
+      "grad_norm": 1.3328872919082642,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 56450
+    },
+    {
+      "epoch": 371.44736842105266,
+      "grad_norm": 1.3304328918457031,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 56460
+    },
+    {
+      "epoch": 371.5131578947368,
+      "grad_norm": 1.4431946277618408,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 56470
+    },
+    {
+      "epoch": 371.57894736842104,
+      "grad_norm": 1.3655774593353271,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 56480
+    },
+    {
+      "epoch": 371.64473684210526,
+      "grad_norm": 0.9467471837997437,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 56490
+    },
+    {
+      "epoch": 371.7105263157895,
+      "grad_norm": 1.276974081993103,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 56500
+    },
+    {
+      "epoch": 371.7763157894737,
+      "grad_norm": 1.1539723873138428,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 56510
+    },
+    {
+      "epoch": 371.8421052631579,
+      "grad_norm": 1.0478674173355103,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 56520
+    },
+    {
+      "epoch": 371.9078947368421,
+      "grad_norm": 1.2943617105484009,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 56530
+    },
+    {
+      "epoch": 371.9736842105263,
+      "grad_norm": 1.0241856575012207,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 56540
+    },
+    {
+      "epoch": 372.0394736842105,
+      "grad_norm": 1.068509817123413,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 56550
+    },
+    {
+      "epoch": 372.10526315789474,
+      "grad_norm": 1.422802209854126,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 56560
+    },
+    {
+      "epoch": 372.17105263157896,
+      "grad_norm": 1.3704060316085815,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 56570
+    },
+    {
+      "epoch": 372.2368421052632,
+      "grad_norm": 1.3845621347427368,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 56580
+    },
+    {
+      "epoch": 372.30263157894734,
+      "grad_norm": 1.221064805984497,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 56590
+    },
+    {
+      "epoch": 372.36842105263156,
+      "grad_norm": 0.9670616388320923,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 56600
+    },
+    {
+      "epoch": 372.4342105263158,
+      "grad_norm": 1.3151499032974243,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 56610
+    },
+    {
+      "epoch": 372.5,
+      "grad_norm": 1.210457682609558,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 56620
+    },
+    {
+      "epoch": 372.5657894736842,
+      "grad_norm": 1.1957846879959106,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 56630
+    },
+    {
+      "epoch": 372.63157894736844,
+      "grad_norm": 0.8912261724472046,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 56640
+    },
+    {
+      "epoch": 372.69736842105266,
+      "grad_norm": 1.1011767387390137,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 56650
+    },
+    {
+      "epoch": 372.7631578947368,
+      "grad_norm": 1.0120614767074585,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 56660
+    },
+    {
+      "epoch": 372.82894736842104,
+      "grad_norm": 1.5126779079437256,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 56670
+    },
+    {
+      "epoch": 372.89473684210526,
+      "grad_norm": 1.7110366821289062,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 56680
+    },
+    {
+      "epoch": 372.9605263157895,
+      "grad_norm": 1.3137553930282593,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 56690
+    },
+    {
+      "epoch": 373.0263157894737,
+      "grad_norm": 1.3267024755477905,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 56700
+    },
+    {
+      "epoch": 373.0921052631579,
+      "grad_norm": 1.0713841915130615,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 56710
+    },
+    {
+      "epoch": 373.1578947368421,
+      "grad_norm": 1.2255971431732178,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 56720
+    },
+    {
+      "epoch": 373.2236842105263,
+      "grad_norm": 1.1243081092834473,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 56730
+    },
+    {
+      "epoch": 373.2894736842105,
+      "grad_norm": 0.909152090549469,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 56740
+    },
+    {
+      "epoch": 373.35526315789474,
+      "grad_norm": 0.9241997003555298,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 56750
+    },
+    {
+      "epoch": 373.42105263157896,
+      "grad_norm": 0.7907554507255554,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 56760
+    },
+    {
+      "epoch": 373.4868421052632,
+      "grad_norm": 1.4424355030059814,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 56770
+    },
+    {
+      "epoch": 373.55263157894734,
+      "grad_norm": 0.7469786405563354,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 56780
+    },
+    {
+      "epoch": 373.61842105263156,
+      "grad_norm": 1.0877963304519653,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 56790
+    },
+    {
+      "epoch": 373.6842105263158,
+      "grad_norm": 1.020845890045166,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 56800
+    },
+    {
+      "epoch": 373.75,
+      "grad_norm": 0.8404970765113831,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 56810
+    },
+    {
+      "epoch": 373.8157894736842,
+      "grad_norm": 1.032677412033081,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 56820
+    },
+    {
+      "epoch": 373.88157894736844,
+      "grad_norm": 1.0752617120742798,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 56830
+    },
+    {
+      "epoch": 373.94736842105266,
+      "grad_norm": 0.9850388169288635,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 56840
+    },
+    {
+      "epoch": 374.0131578947368,
+      "grad_norm": 0.8292937874794006,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 56850
+    },
+    {
+      "epoch": 374.07894736842104,
+      "grad_norm": 1.0605348348617554,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 56860
+    },
+    {
+      "epoch": 374.14473684210526,
+      "grad_norm": 1.2353730201721191,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 56870
+    },
+    {
+      "epoch": 374.2105263157895,
+      "grad_norm": 0.9276547431945801,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 56880
+    },
+    {
+      "epoch": 374.2763157894737,
+      "grad_norm": 1.3930708169937134,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 56890
+    },
+    {
+      "epoch": 374.3421052631579,
+      "grad_norm": 1.2086683511734009,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 56900
+    },
+    {
+      "epoch": 374.4078947368421,
+      "grad_norm": 1.2379398345947266,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 56910
+    },
+    {
+      "epoch": 374.4736842105263,
+      "grad_norm": 1.1630572080612183,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 56920
+    },
+    {
+      "epoch": 374.5394736842105,
+      "grad_norm": 1.2085312604904175,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 56930
+    },
+    {
+      "epoch": 374.60526315789474,
+      "grad_norm": 1.1504122018814087,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 56940
+    },
+    {
+      "epoch": 374.67105263157896,
+      "grad_norm": 1.1568752527236938,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 56950
+    },
+    {
+      "epoch": 374.7368421052632,
+      "grad_norm": 1.122700572013855,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 56960
+    },
+    {
+      "epoch": 374.80263157894734,
+      "grad_norm": 0.869761049747467,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 56970
+    },
+    {
+      "epoch": 374.86842105263156,
+      "grad_norm": 1.219969630241394,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 56980
+    },
+    {
+      "epoch": 374.9342105263158,
+      "grad_norm": 1.0810168981552124,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 56990
+    },
+    {
+      "epoch": 375.0,
+      "grad_norm": 0.8487389087677002,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 57000
+    },
+    {
+      "epoch": 375.0657894736842,
+      "grad_norm": 1.3084897994995117,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 57010
+    },
+    {
+      "epoch": 375.13157894736844,
+      "grad_norm": 1.0866003036499023,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 57020
+    },
+    {
+      "epoch": 375.19736842105266,
+      "grad_norm": 1.215729832649231,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 57030
+    },
+    {
+      "epoch": 375.2631578947368,
+      "grad_norm": 1.321714162826538,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 57040
+    },
+    {
+      "epoch": 375.32894736842104,
+      "grad_norm": 1.21835458278656,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 57050
+    },
+    {
+      "epoch": 375.39473684210526,
+      "grad_norm": 1.352271318435669,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 57060
+    },
+    {
+      "epoch": 375.4605263157895,
+      "grad_norm": 1.1468392610549927,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 57070
+    },
+    {
+      "epoch": 375.5263157894737,
+      "grad_norm": 0.9708956480026245,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 57080
+    },
+    {
+      "epoch": 375.5921052631579,
+      "grad_norm": 1.2509098052978516,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 57090
+    },
+    {
+      "epoch": 375.6578947368421,
+      "grad_norm": 0.8894394636154175,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 57100
+    },
+    {
+      "epoch": 375.7236842105263,
+      "grad_norm": 1.4251313209533691,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 57110
+    },
+    {
+      "epoch": 375.7894736842105,
+      "grad_norm": 1.3613892793655396,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 57120
+    },
+    {
+      "epoch": 375.85526315789474,
+      "grad_norm": 1.2195764780044556,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 57130
+    },
+    {
+      "epoch": 375.92105263157896,
+      "grad_norm": 1.2804150581359863,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 57140
+    },
+    {
+      "epoch": 375.9868421052632,
+      "grad_norm": 1.2926859855651855,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 57150
+    },
+    {
+      "epoch": 376.05263157894734,
+      "grad_norm": 1.2950412034988403,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 57160
+    },
+    {
+      "epoch": 376.11842105263156,
+      "grad_norm": 1.122531771659851,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 57170
+    },
+    {
+      "epoch": 376.1842105263158,
+      "grad_norm": 1.0365455150604248,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 57180
+    },
+    {
+      "epoch": 376.25,
+      "grad_norm": 0.9025917649269104,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 57190
+    },
+    {
+      "epoch": 376.3157894736842,
+      "grad_norm": 1.360355257987976,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 57200
+    },
+    {
+      "epoch": 376.38157894736844,
+      "grad_norm": 1.0496774911880493,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 57210
+    },
+    {
+      "epoch": 376.44736842105266,
+      "grad_norm": 0.9033643007278442,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 57220
+    },
+    {
+      "epoch": 376.5131578947368,
+      "grad_norm": 1.0611474514007568,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 57230
+    },
+    {
+      "epoch": 376.57894736842104,
+      "grad_norm": 1.022553563117981,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 57240
+    },
+    {
+      "epoch": 376.64473684210526,
+      "grad_norm": 1.3348984718322754,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 57250
+    },
+    {
+      "epoch": 376.7105263157895,
+      "grad_norm": 1.362504005432129,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 57260
+    },
+    {
+      "epoch": 376.7763157894737,
+      "grad_norm": 1.2323849201202393,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 57270
+    },
+    {
+      "epoch": 376.8421052631579,
+      "grad_norm": 1.3414735794067383,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 57280
+    },
+    {
+      "epoch": 376.9078947368421,
+      "grad_norm": 1.2798577547073364,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 57290
+    },
+    {
+      "epoch": 376.9736842105263,
+      "grad_norm": 0.9005674719810486,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 57300
+    },
+    {
+      "epoch": 377.0394736842105,
+      "grad_norm": 1.0160363912582397,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 57310
+    },
+    {
+      "epoch": 377.10526315789474,
+      "grad_norm": 1.1084673404693604,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 57320
+    },
+    {
+      "epoch": 377.17105263157896,
+      "grad_norm": 1.4065401554107666,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 57330
+    },
+    {
+      "epoch": 377.2368421052632,
+      "grad_norm": 1.1100188493728638,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 57340
+    },
+    {
+      "epoch": 377.30263157894734,
+      "grad_norm": 0.9800202250480652,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 57350
+    },
+    {
+      "epoch": 377.36842105263156,
+      "grad_norm": 1.887536644935608,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 57360
+    },
+    {
+      "epoch": 377.4342105263158,
+      "grad_norm": 1.2453606128692627,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 57370
+    },
+    {
+      "epoch": 377.5,
+      "grad_norm": 1.2225714921951294,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 57380
+    },
+    {
+      "epoch": 377.5657894736842,
+      "grad_norm": 1.2892736196517944,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 57390
+    },
+    {
+      "epoch": 377.63157894736844,
+      "grad_norm": 1.242989420890808,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 57400
+    },
+    {
+      "epoch": 377.69736842105266,
+      "grad_norm": 1.3066517114639282,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 57410
+    },
+    {
+      "epoch": 377.7631578947368,
+      "grad_norm": 1.126461386680603,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 57420
+    },
+    {
+      "epoch": 377.82894736842104,
+      "grad_norm": 1.0878877639770508,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 57430
+    },
+    {
+      "epoch": 377.89473684210526,
+      "grad_norm": 1.2524346113204956,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 57440
+    },
+    {
+      "epoch": 377.9605263157895,
+      "grad_norm": 1.3451015949249268,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 57450
+    },
+    {
+      "epoch": 378.0263157894737,
+      "grad_norm": 1.1526377201080322,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 57460
+    },
+    {
+      "epoch": 378.0921052631579,
+      "grad_norm": 0.9030545353889465,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 57470
+    },
+    {
+      "epoch": 378.1578947368421,
+      "grad_norm": 0.7027451992034912,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 57480
+    },
+    {
+      "epoch": 378.2236842105263,
+      "grad_norm": 0.7617095708847046,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 57490
+    },
+    {
+      "epoch": 378.2894736842105,
+      "grad_norm": 1.2447706460952759,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 57500
+    },
+    {
+      "epoch": 378.35526315789474,
+      "grad_norm": 1.3532993793487549,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 57510
+    },
+    {
+      "epoch": 378.42105263157896,
+      "grad_norm": 1.154651165008545,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 57520
+    },
+    {
+      "epoch": 378.4868421052632,
+      "grad_norm": 1.0880413055419922,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 57530
+    },
+    {
+      "epoch": 378.55263157894734,
+      "grad_norm": 0.9911805987358093,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 57540
+    },
+    {
+      "epoch": 378.61842105263156,
+      "grad_norm": 1.3125237226486206,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 57550
+    },
+    {
+      "epoch": 378.6842105263158,
+      "grad_norm": 1.2095872163772583,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 57560
+    },
+    {
+      "epoch": 378.75,
+      "grad_norm": 1.300068736076355,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 57570
+    },
+    {
+      "epoch": 378.8157894736842,
+      "grad_norm": 0.9355725049972534,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 57580
+    },
+    {
+      "epoch": 378.88157894736844,
+      "grad_norm": 1.0073492527008057,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 57590
+    },
+    {
+      "epoch": 378.94736842105266,
+      "grad_norm": 1.3095647096633911,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 57600
+    },
+    {
+      "epoch": 379.0131578947368,
+      "grad_norm": 1.3785039186477661,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 57610
+    },
+    {
+      "epoch": 379.07894736842104,
+      "grad_norm": 1.2601360082626343,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 57620
+    },
+    {
+      "epoch": 379.14473684210526,
+      "grad_norm": 1.2998868227005005,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 57630
+    },
+    {
+      "epoch": 379.2105263157895,
+      "grad_norm": 1.2391390800476074,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 57640
+    },
+    {
+      "epoch": 379.2763157894737,
+      "grad_norm": 1.320417046546936,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 57650
+    },
+    {
+      "epoch": 379.3421052631579,
+      "grad_norm": 1.2020084857940674,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 57660
+    },
+    {
+      "epoch": 379.4078947368421,
+      "grad_norm": 1.2088592052459717,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 57670
+    },
+    {
+      "epoch": 379.4736842105263,
+      "grad_norm": 0.9938647747039795,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 57680
+    },
+    {
+      "epoch": 379.5394736842105,
+      "grad_norm": 1.057723879814148,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 57690
+    },
+    {
+      "epoch": 379.60526315789474,
+      "grad_norm": 0.9933873414993286,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 57700
+    },
+    {
+      "epoch": 379.67105263157896,
+      "grad_norm": 0.7416632771492004,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 57710
+    },
+    {
+      "epoch": 379.7368421052632,
+      "grad_norm": 1.292233943939209,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 57720
+    },
+    {
+      "epoch": 379.80263157894734,
+      "grad_norm": 1.1518487930297852,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 57730
+    },
+    {
+      "epoch": 379.86842105263156,
+      "grad_norm": 1.1070126295089722,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 57740
+    },
+    {
+      "epoch": 379.9342105263158,
+      "grad_norm": 0.9124812483787537,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 57750
+    },
+    {
+      "epoch": 380.0,
+      "grad_norm": 1.2951626777648926,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 57760
+    },
+    {
+      "epoch": 380.0657894736842,
+      "grad_norm": 1.2988245487213135,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 57770
+    },
+    {
+      "epoch": 380.13157894736844,
+      "grad_norm": 1.5885379314422607,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 57780
+    },
+    {
+      "epoch": 380.19736842105266,
+      "grad_norm": 1.343135118484497,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 57790
+    },
+    {
+      "epoch": 380.2631578947368,
+      "grad_norm": 1.0885659456253052,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 57800
+    },
+    {
+      "epoch": 380.32894736842104,
+      "grad_norm": 1.7573096752166748,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 57810
+    },
+    {
+      "epoch": 380.39473684210526,
+      "grad_norm": 1.4243195056915283,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 57820
+    },
+    {
+      "epoch": 380.4605263157895,
+      "grad_norm": 1.1941052675247192,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 57830
+    },
+    {
+      "epoch": 380.5263157894737,
+      "grad_norm": 1.3269431591033936,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 57840
+    },
+    {
+      "epoch": 380.5921052631579,
+      "grad_norm": 1.0225571393966675,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 57850
+    },
+    {
+      "epoch": 380.6578947368421,
+      "grad_norm": 1.412798523902893,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 57860
+    },
+    {
+      "epoch": 380.7236842105263,
+      "grad_norm": 1.382020354270935,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 57870
+    },
+    {
+      "epoch": 380.7894736842105,
+      "grad_norm": 1.5677051544189453,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 57880
+    },
+    {
+      "epoch": 380.85526315789474,
+      "grad_norm": 1.2945635318756104,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 57890
+    },
+    {
+      "epoch": 380.92105263157896,
+      "grad_norm": 1.3230671882629395,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 57900
+    },
+    {
+      "epoch": 380.9868421052632,
+      "grad_norm": 1.1088383197784424,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 57910
+    },
+    {
+      "epoch": 381.05263157894734,
+      "grad_norm": 1.479811668395996,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 57920
+    },
+    {
+      "epoch": 381.11842105263156,
+      "grad_norm": 1.2625958919525146,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 57930
+    },
+    {
+      "epoch": 381.1842105263158,
+      "grad_norm": 1.2790255546569824,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 57940
+    },
+    {
+      "epoch": 381.25,
+      "grad_norm": 1.589355707168579,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 57950
+    },
+    {
+      "epoch": 381.3157894736842,
+      "grad_norm": 1.3272757530212402,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 57960
+    },
+    {
+      "epoch": 381.38157894736844,
+      "grad_norm": 0.8817294239997864,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 57970
+    },
+    {
+      "epoch": 381.44736842105266,
+      "grad_norm": 1.2385287284851074,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 57980
+    },
+    {
+      "epoch": 381.5131578947368,
+      "grad_norm": 1.324324131011963,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 57990
+    },
+    {
+      "epoch": 381.57894736842104,
+      "grad_norm": 1.4512399435043335,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 58000
+    },
+    {
+      "epoch": 381.64473684210526,
+      "grad_norm": 1.277093529701233,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 58010
+    },
+    {
+      "epoch": 381.7105263157895,
+      "grad_norm": 1.15549635887146,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 58020
+    },
+    {
+      "epoch": 381.7763157894737,
+      "grad_norm": 0.7976158261299133,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 58030
+    },
+    {
+      "epoch": 381.8421052631579,
+      "grad_norm": 1.189330816268921,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 58040
+    },
+    {
+      "epoch": 381.9078947368421,
+      "grad_norm": 1.5687916278839111,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 58050
+    },
+    {
+      "epoch": 381.9736842105263,
+      "grad_norm": 1.33320152759552,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 58060
+    },
+    {
+      "epoch": 382.0394736842105,
+      "grad_norm": 1.0838332176208496,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 58070
+    },
+    {
+      "epoch": 382.10526315789474,
+      "grad_norm": 1.0593252182006836,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 58080
+    },
+    {
+      "epoch": 382.17105263157896,
+      "grad_norm": 1.1587085723876953,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 58090
+    },
+    {
+      "epoch": 382.2368421052632,
+      "grad_norm": 1.2897980213165283,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 58100
+    },
+    {
+      "epoch": 382.30263157894734,
+      "grad_norm": 1.4974095821380615,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 58110
+    },
+    {
+      "epoch": 382.36842105263156,
+      "grad_norm": 1.2067298889160156,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 58120
+    },
+    {
+      "epoch": 382.4342105263158,
+      "grad_norm": 1.151006817817688,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 58130
+    },
+    {
+      "epoch": 382.5,
+      "grad_norm": 0.9617295861244202,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 58140
+    },
+    {
+      "epoch": 382.5657894736842,
+      "grad_norm": 1.2148723602294922,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 58150
+    },
+    {
+      "epoch": 382.63157894736844,
+      "grad_norm": 0.9488323926925659,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 58160
+    },
+    {
+      "epoch": 382.69736842105266,
+      "grad_norm": 1.2400751113891602,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 58170
+    },
+    {
+      "epoch": 382.7631578947368,
+      "grad_norm": 1.123399019241333,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 58180
+    },
+    {
+      "epoch": 382.82894736842104,
+      "grad_norm": 1.2090132236480713,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 58190
+    },
+    {
+      "epoch": 382.89473684210526,
+      "grad_norm": 1.1165274381637573,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 58200
+    },
+    {
+      "epoch": 382.9605263157895,
+      "grad_norm": 1.304680585861206,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 58210
+    },
+    {
+      "epoch": 383.0263157894737,
+      "grad_norm": 1.3738752603530884,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 58220
+    },
+    {
+      "epoch": 383.0921052631579,
+      "grad_norm": 0.8342259526252747,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 58230
+    },
+    {
+      "epoch": 383.1578947368421,
+      "grad_norm": 1.0694853067398071,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 58240
+    },
+    {
+      "epoch": 383.2236842105263,
+      "grad_norm": 1.22238028049469,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 58250
+    },
+    {
+      "epoch": 383.2894736842105,
+      "grad_norm": 0.9267398715019226,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 58260
+    },
+    {
+      "epoch": 383.35526315789474,
+      "grad_norm": 0.9709849953651428,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 58270
+    },
+    {
+      "epoch": 383.42105263157896,
+      "grad_norm": 1.3545488119125366,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 58280
+    },
+    {
+      "epoch": 383.4868421052632,
+      "grad_norm": 1.2838683128356934,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 58290
+    },
+    {
+      "epoch": 383.55263157894734,
+      "grad_norm": 0.764008641242981,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 58300
+    },
+    {
+      "epoch": 383.61842105263156,
+      "grad_norm": 0.821171224117279,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 58310
+    },
+    {
+      "epoch": 383.6842105263158,
+      "grad_norm": 0.8788691759109497,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 58320
+    },
+    {
+      "epoch": 383.75,
+      "grad_norm": 0.9442429542541504,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 58330
+    },
+    {
+      "epoch": 383.8157894736842,
+      "grad_norm": 1.0095479488372803,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 58340
+    },
+    {
+      "epoch": 383.88157894736844,
+      "grad_norm": 1.0548217296600342,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 58350
+    },
+    {
+      "epoch": 383.94736842105266,
+      "grad_norm": 0.9758161902427673,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 58360
+    },
+    {
+      "epoch": 384.0131578947368,
+      "grad_norm": 1.2047322988510132,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 58370
+    },
+    {
+      "epoch": 384.07894736842104,
+      "grad_norm": 0.9442251324653625,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 58380
+    },
+    {
+      "epoch": 384.14473684210526,
+      "grad_norm": 1.4745181798934937,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 58390
+    },
+    {
+      "epoch": 384.2105263157895,
+      "grad_norm": 1.0786917209625244,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 58400
+    },
+    {
+      "epoch": 384.2763157894737,
+      "grad_norm": 0.8698955774307251,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 58410
+    },
+    {
+      "epoch": 384.3421052631579,
+      "grad_norm": 1.3999569416046143,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 58420
+    },
+    {
+      "epoch": 384.4078947368421,
+      "grad_norm": 1.203365445137024,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 58430
+    },
+    {
+      "epoch": 384.4736842105263,
+      "grad_norm": 1.1867121458053589,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 58440
+    },
+    {
+      "epoch": 384.5394736842105,
+      "grad_norm": 1.7259279489517212,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 58450
+    },
+    {
+      "epoch": 384.60526315789474,
+      "grad_norm": 1.2636997699737549,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 58460
+    },
+    {
+      "epoch": 384.67105263157896,
+      "grad_norm": 1.0157842636108398,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 58470
+    },
+    {
+      "epoch": 384.7368421052632,
+      "grad_norm": 1.3222954273223877,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 58480
+    },
+    {
+      "epoch": 384.80263157894734,
+      "grad_norm": 1.118708610534668,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 58490
+    },
+    {
+      "epoch": 384.86842105263156,
+      "grad_norm": 0.8435554504394531,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 58500
+    },
+    {
+      "epoch": 384.9342105263158,
+      "grad_norm": 0.8175836205482483,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 58510
+    },
+    {
+      "epoch": 385.0,
+      "grad_norm": 1.041205883026123,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 58520
+    },
+    {
+      "epoch": 385.0657894736842,
+      "grad_norm": 0.7904459238052368,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 58530
+    },
+    {
+      "epoch": 385.13157894736844,
+      "grad_norm": 1.1305633783340454,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 58540
+    },
+    {
+      "epoch": 385.19736842105266,
+      "grad_norm": 0.7400175333023071,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 58550
+    },
+    {
+      "epoch": 385.2631578947368,
+      "grad_norm": 0.870387077331543,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 58560
+    },
+    {
+      "epoch": 385.32894736842104,
+      "grad_norm": 0.8553933501243591,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 58570
+    },
+    {
+      "epoch": 385.39473684210526,
+      "grad_norm": 0.9809901714324951,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 58580
+    },
+    {
+      "epoch": 385.4605263157895,
+      "grad_norm": 0.8871316313743591,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 58590
+    },
+    {
+      "epoch": 385.5263157894737,
+      "grad_norm": 1.1074786186218262,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 58600
+    },
+    {
+      "epoch": 385.5921052631579,
+      "grad_norm": 0.7304058074951172,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 58610
+    },
+    {
+      "epoch": 385.6578947368421,
+      "grad_norm": 1.168735384941101,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 58620
+    },
+    {
+      "epoch": 385.7236842105263,
+      "grad_norm": 0.8042843341827393,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 58630
+    },
+    {
+      "epoch": 385.7894736842105,
+      "grad_norm": 0.9500011801719666,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 58640
+    },
+    {
+      "epoch": 385.85526315789474,
+      "grad_norm": 0.9175438284873962,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 58650
+    },
+    {
+      "epoch": 385.92105263157896,
+      "grad_norm": 1.302797555923462,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 58660
+    },
+    {
+      "epoch": 385.9868421052632,
+      "grad_norm": 1.2630707025527954,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 58670
+    },
+    {
+      "epoch": 386.05263157894734,
+      "grad_norm": 0.7925125360488892,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 58680
+    },
+    {
+      "epoch": 386.11842105263156,
+      "grad_norm": 0.8696709275245667,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 58690
+    },
+    {
+      "epoch": 386.1842105263158,
+      "grad_norm": 1.196243166923523,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 58700
+    },
+    {
+      "epoch": 386.25,
+      "grad_norm": 1.1323686838150024,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 58710
+    },
+    {
+      "epoch": 386.3157894736842,
+      "grad_norm": 0.9869086742401123,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 58720
+    },
+    {
+      "epoch": 386.38157894736844,
+      "grad_norm": 1.3381530046463013,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 58730
+    },
+    {
+      "epoch": 386.44736842105266,
+      "grad_norm": 1.3545961380004883,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 58740
+    },
+    {
+      "epoch": 386.5131578947368,
+      "grad_norm": 1.1433833837509155,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 58750
+    },
+    {
+      "epoch": 386.57894736842104,
+      "grad_norm": 1.3282660245895386,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 58760
+    },
+    {
+      "epoch": 386.64473684210526,
+      "grad_norm": 1.518915057182312,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 58770
+    },
+    {
+      "epoch": 386.7105263157895,
+      "grad_norm": 1.1334174871444702,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 58780
+    },
+    {
+      "epoch": 386.7763157894737,
+      "grad_norm": 1.2663590908050537,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 58790
+    },
+    {
+      "epoch": 386.8421052631579,
+      "grad_norm": 1.148506999015808,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 58800
+    },
+    {
+      "epoch": 386.9078947368421,
+      "grad_norm": 1.2444261312484741,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 58810
+    },
+    {
+      "epoch": 386.9736842105263,
+      "grad_norm": 1.1038265228271484,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 58820
+    },
+    {
+      "epoch": 387.0394736842105,
+      "grad_norm": 0.9502344131469727,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 58830
+    },
+    {
+      "epoch": 387.10526315789474,
+      "grad_norm": 1.0720306634902954,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 58840
+    },
+    {
+      "epoch": 387.17105263157896,
+      "grad_norm": 1.243300199508667,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 58850
+    },
+    {
+      "epoch": 387.2368421052632,
+      "grad_norm": 0.916761577129364,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 58860
+    },
+    {
+      "epoch": 387.30263157894734,
+      "grad_norm": 1.0160788297653198,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 58870
+    },
+    {
+      "epoch": 387.36842105263156,
+      "grad_norm": 1.1823748350143433,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 58880
+    },
+    {
+      "epoch": 387.4342105263158,
+      "grad_norm": 1.0291345119476318,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 58890
+    },
+    {
+      "epoch": 387.5,
+      "grad_norm": 1.0827511548995972,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 58900
+    },
+    {
+      "epoch": 387.5657894736842,
+      "grad_norm": 1.3324905633926392,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 58910
+    },
+    {
+      "epoch": 387.63157894736844,
+      "grad_norm": 1.159676432609558,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 58920
+    },
+    {
+      "epoch": 387.69736842105266,
+      "grad_norm": 1.6756174564361572,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 58930
+    },
+    {
+      "epoch": 387.7631578947368,
+      "grad_norm": 1.63514244556427,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 58940
+    },
+    {
+      "epoch": 387.82894736842104,
+      "grad_norm": 1.3889161348342896,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 58950
+    },
+    {
+      "epoch": 387.89473684210526,
+      "grad_norm": 1.27745521068573,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 58960
+    },
+    {
+      "epoch": 387.9605263157895,
+      "grad_norm": 1.3350027799606323,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 58970
+    },
+    {
+      "epoch": 388.0263157894737,
+      "grad_norm": 1.5261503458023071,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 58980
+    },
+    {
+      "epoch": 388.0921052631579,
+      "grad_norm": 1.6746344566345215,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 58990
+    },
+    {
+      "epoch": 388.1578947368421,
+      "grad_norm": 1.0707603693008423,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 59000
+    },
+    {
+      "epoch": 388.2236842105263,
+      "grad_norm": 1.0903775691986084,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 59010
+    },
+    {
+      "epoch": 388.2894736842105,
+      "grad_norm": 1.2945852279663086,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 59020
+    },
+    {
+      "epoch": 388.35526315789474,
+      "grad_norm": 0.8267077803611755,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 59030
+    },
+    {
+      "epoch": 388.42105263157896,
+      "grad_norm": 1.4991341829299927,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 59040
+    },
+    {
+      "epoch": 388.4868421052632,
+      "grad_norm": 1.3253583908081055,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 59050
+    },
+    {
+      "epoch": 388.55263157894734,
+      "grad_norm": 0.9886115789413452,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 59060
+    },
+    {
+      "epoch": 388.61842105263156,
+      "grad_norm": 1.1212751865386963,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 59070
+    },
+    {
+      "epoch": 388.6842105263158,
+      "grad_norm": 0.9368122220039368,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 59080
+    },
+    {
+      "epoch": 388.75,
+      "grad_norm": 1.0069820880889893,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 59090
+    },
+    {
+      "epoch": 388.8157894736842,
+      "grad_norm": 1.3330034017562866,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 59100
+    },
+    {
+      "epoch": 388.88157894736844,
+      "grad_norm": 1.5406800508499146,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 59110
+    },
+    {
+      "epoch": 388.94736842105266,
+      "grad_norm": 1.2882165908813477,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 59120
+    },
+    {
+      "epoch": 389.0131578947368,
+      "grad_norm": 1.138337254524231,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 59130
+    },
+    {
+      "epoch": 389.07894736842104,
+      "grad_norm": 1.0221810340881348,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 59140
+    },
+    {
+      "epoch": 389.14473684210526,
+      "grad_norm": 1.281477928161621,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 59150
+    },
+    {
+      "epoch": 389.2105263157895,
+      "grad_norm": 1.1402971744537354,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 59160
+    },
+    {
+      "epoch": 389.2763157894737,
+      "grad_norm": 0.914799153804779,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 59170
+    },
+    {
+      "epoch": 389.3421052631579,
+      "grad_norm": 1.0505526065826416,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 59180
+    },
+    {
+      "epoch": 389.4078947368421,
+      "grad_norm": 1.2654054164886475,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 59190
+    },
+    {
+      "epoch": 389.4736842105263,
+      "grad_norm": 1.0078588724136353,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 59200
+    },
+    {
+      "epoch": 389.5394736842105,
+      "grad_norm": 0.7537126541137695,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 59210
+    },
+    {
+      "epoch": 389.60526315789474,
+      "grad_norm": 1.1669580936431885,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 59220
+    },
+    {
+      "epoch": 389.67105263157896,
+      "grad_norm": 1.2333866357803345,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 59230
+    },
+    {
+      "epoch": 389.7368421052632,
+      "grad_norm": 1.1542506217956543,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 59240
+    },
+    {
+      "epoch": 389.80263157894734,
+      "grad_norm": 1.2886197566986084,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 59250
+    },
+    {
+      "epoch": 389.86842105263156,
+      "grad_norm": 1.0714430809020996,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 59260
+    },
+    {
+      "epoch": 389.9342105263158,
+      "grad_norm": 0.9094032049179077,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 59270
+    },
+    {
+      "epoch": 390.0,
+      "grad_norm": 1.1040136814117432,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 59280
+    },
+    {
+      "epoch": 390.0657894736842,
+      "grad_norm": 0.876976490020752,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 59290
+    },
+    {
+      "epoch": 390.13157894736844,
+      "grad_norm": 0.86655193567276,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 59300
+    },
+    {
+      "epoch": 390.19736842105266,
+      "grad_norm": 1.018251657485962,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 59310
+    },
+    {
+      "epoch": 390.2631578947368,
+      "grad_norm": 1.1418795585632324,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 59320
+    },
+    {
+      "epoch": 390.32894736842104,
+      "grad_norm": 1.039706826210022,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 59330
+    },
+    {
+      "epoch": 390.39473684210526,
+      "grad_norm": 1.0030986070632935,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 59340
+    },
+    {
+      "epoch": 390.4605263157895,
+      "grad_norm": 1.1898283958435059,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 59350
+    },
+    {
+      "epoch": 390.5263157894737,
+      "grad_norm": 0.8979069590568542,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 59360
+    },
+    {
+      "epoch": 390.5921052631579,
+      "grad_norm": 1.1258528232574463,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 59370
+    },
+    {
+      "epoch": 390.6578947368421,
+      "grad_norm": 1.3766101598739624,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 59380
+    },
+    {
+      "epoch": 390.7236842105263,
+      "grad_norm": 1.3407342433929443,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 59390
+    },
+    {
+      "epoch": 390.7894736842105,
+      "grad_norm": 0.9154465198516846,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 59400
+    },
+    {
+      "epoch": 390.85526315789474,
+      "grad_norm": 1.2275203466415405,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 59410
+    },
+    {
+      "epoch": 390.92105263157896,
+      "grad_norm": 1.105592131614685,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 59420
+    },
+    {
+      "epoch": 390.9868421052632,
+      "grad_norm": 1.2495988607406616,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 59430
+    },
+    {
+      "epoch": 391.05263157894734,
+      "grad_norm": 1.3405605554580688,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 59440
+    },
+    {
+      "epoch": 391.11842105263156,
+      "grad_norm": 1.2704845666885376,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 59450
+    },
+    {
+      "epoch": 391.1842105263158,
+      "grad_norm": 1.029582142829895,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 59460
+    },
+    {
+      "epoch": 391.25,
+      "grad_norm": 0.8142033219337463,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 59470
+    },
+    {
+      "epoch": 391.3157894736842,
+      "grad_norm": 1.6257590055465698,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 59480
+    },
+    {
+      "epoch": 391.38157894736844,
+      "grad_norm": 1.2035874128341675,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 59490
+    },
+    {
+      "epoch": 391.44736842105266,
+      "grad_norm": 0.7711625099182129,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 59500
+    },
+    {
+      "epoch": 391.5131578947368,
+      "grad_norm": 1.076262354850769,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 59510
+    },
+    {
+      "epoch": 391.57894736842104,
+      "grad_norm": 1.2418280839920044,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 59520
+    },
+    {
+      "epoch": 391.64473684210526,
+      "grad_norm": 1.1036405563354492,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 59530
+    },
+    {
+      "epoch": 391.7105263157895,
+      "grad_norm": 1.4087510108947754,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 59540
+    },
+    {
+      "epoch": 391.7763157894737,
+      "grad_norm": 0.9523179531097412,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 59550
+    },
+    {
+      "epoch": 391.8421052631579,
+      "grad_norm": 1.2418298721313477,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 59560
+    },
+    {
+      "epoch": 391.9078947368421,
+      "grad_norm": 1.00105881690979,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 59570
+    },
+    {
+      "epoch": 391.9736842105263,
+      "grad_norm": 1.2458897829055786,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 59580
+    },
+    {
+      "epoch": 392.0394736842105,
+      "grad_norm": 1.1352124214172363,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 59590
+    },
+    {
+      "epoch": 392.10526315789474,
+      "grad_norm": 1.072657823562622,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 59600
+    },
+    {
+      "epoch": 392.17105263157896,
+      "grad_norm": 1.4077866077423096,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 59610
+    },
+    {
+      "epoch": 392.2368421052632,
+      "grad_norm": 0.9759162664413452,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 59620
+    },
+    {
+      "epoch": 392.30263157894734,
+      "grad_norm": 1.0127092599868774,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 59630
+    },
+    {
+      "epoch": 392.36842105263156,
+      "grad_norm": 1.2567495107650757,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 59640
+    },
+    {
+      "epoch": 392.4342105263158,
+      "grad_norm": 1.346420168876648,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 59650
+    },
+    {
+      "epoch": 392.5,
+      "grad_norm": 1.125648856163025,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 59660
+    },
+    {
+      "epoch": 392.5657894736842,
+      "grad_norm": 1.004570484161377,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 59670
+    },
+    {
+      "epoch": 392.63157894736844,
+      "grad_norm": 1.5204274654388428,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 59680
+    },
+    {
+      "epoch": 392.69736842105266,
+      "grad_norm": 1.4187020063400269,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 59690
+    },
+    {
+      "epoch": 392.7631578947368,
+      "grad_norm": 1.4313414096832275,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 59700
+    },
+    {
+      "epoch": 392.82894736842104,
+      "grad_norm": 1.398133635520935,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 59710
+    },
+    {
+      "epoch": 392.89473684210526,
+      "grad_norm": 1.279089093208313,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 59720
+    },
+    {
+      "epoch": 392.9605263157895,
+      "grad_norm": 1.511715292930603,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 59730
+    },
+    {
+      "epoch": 393.0263157894737,
+      "grad_norm": 1.0526278018951416,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 59740
+    },
+    {
+      "epoch": 393.0921052631579,
+      "grad_norm": 1.0821075439453125,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 59750
+    },
+    {
+      "epoch": 393.1578947368421,
+      "grad_norm": 1.2545526027679443,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 59760
+    },
+    {
+      "epoch": 393.2236842105263,
+      "grad_norm": 0.8105134963989258,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 59770
+    },
+    {
+      "epoch": 393.2894736842105,
+      "grad_norm": 1.4059394598007202,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 59780
+    },
+    {
+      "epoch": 393.35526315789474,
+      "grad_norm": 1.157401204109192,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 59790
+    },
+    {
+      "epoch": 393.42105263157896,
+      "grad_norm": 0.9264048337936401,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 59800
+    },
+    {
+      "epoch": 393.4868421052632,
+      "grad_norm": 0.9000381827354431,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 59810
+    },
+    {
+      "epoch": 393.55263157894734,
+      "grad_norm": 1.0139583349227905,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 59820
+    },
+    {
+      "epoch": 393.61842105263156,
+      "grad_norm": 0.9514049887657166,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 59830
+    },
+    {
+      "epoch": 393.6842105263158,
+      "grad_norm": 1.2830848693847656,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 59840
+    },
+    {
+      "epoch": 393.75,
+      "grad_norm": 0.9854030013084412,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 59850
+    },
+    {
+      "epoch": 393.8157894736842,
+      "grad_norm": 1.196966290473938,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 59860
+    },
+    {
+      "epoch": 393.88157894736844,
+      "grad_norm": 1.024109125137329,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 59870
+    },
+    {
+      "epoch": 393.94736842105266,
+      "grad_norm": 1.1376116275787354,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 59880
+    },
+    {
+      "epoch": 394.0131578947368,
+      "grad_norm": 1.670544147491455,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 59890
+    },
+    {
+      "epoch": 394.07894736842104,
+      "grad_norm": 1.2976588010787964,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 59900
+    },
+    {
+      "epoch": 394.14473684210526,
+      "grad_norm": 1.2100739479064941,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 59910
+    },
+    {
+      "epoch": 394.2105263157895,
+      "grad_norm": 1.5674223899841309,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 59920
+    },
+    {
+      "epoch": 394.2763157894737,
+      "grad_norm": 1.4684234857559204,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 59930
+    },
+    {
+      "epoch": 394.3421052631579,
+      "grad_norm": 1.1361076831817627,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 59940
+    },
+    {
+      "epoch": 394.4078947368421,
+      "grad_norm": 1.0538053512573242,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 59950
+    },
+    {
+      "epoch": 394.4736842105263,
+      "grad_norm": 1.3147025108337402,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 59960
+    },
+    {
+      "epoch": 394.5394736842105,
+      "grad_norm": 1.4038904905319214,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 59970
+    },
+    {
+      "epoch": 394.60526315789474,
+      "grad_norm": 1.126792073249817,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 59980
+    },
+    {
+      "epoch": 394.67105263157896,
+      "grad_norm": 1.2345644235610962,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 59990
+    },
+    {
+      "epoch": 394.7368421052632,
+      "grad_norm": 1.2304173707962036,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 60000
+    },
+    {
+      "epoch": 394.80263157894734,
+      "grad_norm": 1.4393924474716187,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 60010
+    },
+    {
+      "epoch": 394.86842105263156,
+      "grad_norm": 1.3260493278503418,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 60020
+    },
+    {
+      "epoch": 394.9342105263158,
+      "grad_norm": 1.1898632049560547,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 60030
+    },
+    {
+      "epoch": 395.0,
+      "grad_norm": 1.163697361946106,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 60040
+    },
+    {
+      "epoch": 395.0657894736842,
+      "grad_norm": 1.0178470611572266,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 60050
+    },
+    {
+      "epoch": 395.13157894736844,
+      "grad_norm": 1.5706970691680908,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 60060
+    },
+    {
+      "epoch": 395.19736842105266,
+      "grad_norm": 1.2328516244888306,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 60070
+    },
+    {
+      "epoch": 395.2631578947368,
+      "grad_norm": 1.3404053449630737,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 60080
+    },
+    {
+      "epoch": 395.32894736842104,
+      "grad_norm": 1.3610615730285645,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 60090
+    },
+    {
+      "epoch": 395.39473684210526,
+      "grad_norm": 1.486354947090149,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 60100
+    },
+    {
+      "epoch": 395.4605263157895,
+      "grad_norm": 1.392723560333252,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 60110
+    },
+    {
+      "epoch": 395.5263157894737,
+      "grad_norm": 1.290682077407837,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 60120
+    },
+    {
+      "epoch": 395.5921052631579,
+      "grad_norm": 0.9399741291999817,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 60130
+    },
+    {
+      "epoch": 395.6578947368421,
+      "grad_norm": 1.4426350593566895,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 60140
+    },
+    {
+      "epoch": 395.7236842105263,
+      "grad_norm": 1.2685151100158691,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 60150
+    },
+    {
+      "epoch": 395.7894736842105,
+      "grad_norm": 0.9836694002151489,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 60160
+    },
+    {
+      "epoch": 395.85526315789474,
+      "grad_norm": 1.0677541494369507,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 60170
+    },
+    {
+      "epoch": 395.92105263157896,
+      "grad_norm": 1.1794040203094482,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 60180
+    },
+    {
+      "epoch": 395.9868421052632,
+      "grad_norm": 1.7301592826843262,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 60190
+    },
+    {
+      "epoch": 396.05263157894734,
+      "grad_norm": 1.1919747591018677,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 60200
+    },
+    {
+      "epoch": 396.11842105263156,
+      "grad_norm": 1.3119267225265503,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 60210
+    },
+    {
+      "epoch": 396.1842105263158,
+      "grad_norm": 1.2656959295272827,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 60220
+    },
+    {
+      "epoch": 396.25,
+      "grad_norm": 1.2613000869750977,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 60230
+    },
+    {
+      "epoch": 396.3157894736842,
+      "grad_norm": 1.25479257106781,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 60240
+    },
+    {
+      "epoch": 396.38157894736844,
+      "grad_norm": 1.2844350337982178,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 60250
+    },
+    {
+      "epoch": 396.44736842105266,
+      "grad_norm": 1.0911136865615845,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 60260
+    },
+    {
+      "epoch": 396.5131578947368,
+      "grad_norm": 1.0953329801559448,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 60270
+    },
+    {
+      "epoch": 396.57894736842104,
+      "grad_norm": 1.4672765731811523,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 60280
+    },
+    {
+      "epoch": 396.64473684210526,
+      "grad_norm": 1.0957013368606567,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 60290
+    },
+    {
+      "epoch": 396.7105263157895,
+      "grad_norm": 1.128840684890747,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 60300
+    },
+    {
+      "epoch": 396.7763157894737,
+      "grad_norm": 0.9832814931869507,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 60310
+    },
+    {
+      "epoch": 396.8421052631579,
+      "grad_norm": 1.023591160774231,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 60320
+    },
+    {
+      "epoch": 396.9078947368421,
+      "grad_norm": 0.9705013036727905,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 60330
+    },
+    {
+      "epoch": 396.9736842105263,
+      "grad_norm": 1.1543126106262207,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 60340
+    },
+    {
+      "epoch": 397.0394736842105,
+      "grad_norm": 0.7225308418273926,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 60350
+    },
+    {
+      "epoch": 397.10526315789474,
+      "grad_norm": 1.5401467084884644,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 60360
+    },
+    {
+      "epoch": 397.17105263157896,
+      "grad_norm": 1.3008463382720947,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 60370
+    },
+    {
+      "epoch": 397.2368421052632,
+      "grad_norm": 0.7832499742507935,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 60380
+    },
+    {
+      "epoch": 397.30263157894734,
+      "grad_norm": 0.9722589254379272,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 60390
+    },
+    {
+      "epoch": 397.36842105263156,
+      "grad_norm": 0.9584972858428955,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 60400
+    },
+    {
+      "epoch": 397.4342105263158,
+      "grad_norm": 1.1281046867370605,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 60410
+    },
+    {
+      "epoch": 397.5,
+      "grad_norm": 0.9634712338447571,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 60420
+    },
+    {
+      "epoch": 397.5657894736842,
+      "grad_norm": 0.8092069029808044,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 60430
+    },
+    {
+      "epoch": 397.63157894736844,
+      "grad_norm": 0.941852331161499,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 60440
+    },
+    {
+      "epoch": 397.69736842105266,
+      "grad_norm": 1.1537492275238037,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 60450
+    },
+    {
+      "epoch": 397.7631578947368,
+      "grad_norm": 1.2303005456924438,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 60460
+    },
+    {
+      "epoch": 397.82894736842104,
+      "grad_norm": 0.761372983455658,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 60470
+    },
+    {
+      "epoch": 397.89473684210526,
+      "grad_norm": 0.8683788180351257,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 60480
+    },
+    {
+      "epoch": 397.9605263157895,
+      "grad_norm": 0.7769721150398254,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 60490
+    },
+    {
+      "epoch": 398.0263157894737,
+      "grad_norm": 0.8893076777458191,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 60500
+    },
+    {
+      "epoch": 398.0921052631579,
+      "grad_norm": 1.5258387327194214,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 60510
+    },
+    {
+      "epoch": 398.1578947368421,
+      "grad_norm": 1.3462433815002441,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 60520
+    },
+    {
+      "epoch": 398.2236842105263,
+      "grad_norm": 0.9892566800117493,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 60530
+    },
+    {
+      "epoch": 398.2894736842105,
+      "grad_norm": 1.024162769317627,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 60540
+    },
+    {
+      "epoch": 398.35526315789474,
+      "grad_norm": 0.9343032240867615,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 60550
+    },
+    {
+      "epoch": 398.42105263157896,
+      "grad_norm": 1.1336026191711426,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 60560
+    },
+    {
+      "epoch": 398.4868421052632,
+      "grad_norm": 1.1892248392105103,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 60570
+    },
+    {
+      "epoch": 398.55263157894734,
+      "grad_norm": 1.184756875038147,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 60580
+    },
+    {
+      "epoch": 398.61842105263156,
+      "grad_norm": 1.1197586059570312,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 60590
+    },
+    {
+      "epoch": 398.6842105263158,
+      "grad_norm": 1.4261908531188965,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 60600
+    },
+    {
+      "epoch": 398.75,
+      "grad_norm": 1.3351398706436157,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 60610
+    },
+    {
+      "epoch": 398.8157894736842,
+      "grad_norm": 1.4747203588485718,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 60620
+    },
+    {
+      "epoch": 398.88157894736844,
+      "grad_norm": 1.32119882106781,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 60630
+    },
+    {
+      "epoch": 398.94736842105266,
+      "grad_norm": 1.3960144519805908,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 60640
+    },
+    {
+      "epoch": 399.0131578947368,
+      "grad_norm": 1.1450289487838745,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 60650
+    },
+    {
+      "epoch": 399.07894736842104,
+      "grad_norm": 1.1878561973571777,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 60660
+    },
+    {
+      "epoch": 399.14473684210526,
+      "grad_norm": 1.0264465808868408,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 60670
+    },
+    {
+      "epoch": 399.2105263157895,
+      "grad_norm": 1.1604735851287842,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 60680
+    },
+    {
+      "epoch": 399.2763157894737,
+      "grad_norm": 1.373092532157898,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 60690
+    },
+    {
+      "epoch": 399.3421052631579,
+      "grad_norm": 1.260588526725769,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 60700
+    },
+    {
+      "epoch": 399.4078947368421,
+      "grad_norm": 1.2193946838378906,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 60710
+    },
+    {
+      "epoch": 399.4736842105263,
+      "grad_norm": 1.2826273441314697,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 60720
+    },
+    {
+      "epoch": 399.5394736842105,
+      "grad_norm": 0.9738739132881165,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 60730
+    },
+    {
+      "epoch": 399.60526315789474,
+      "grad_norm": 1.2081356048583984,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 60740
+    },
+    {
+      "epoch": 399.67105263157896,
+      "grad_norm": 1.198371171951294,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 60750
+    },
+    {
+      "epoch": 399.7368421052632,
+      "grad_norm": 1.064581274986267,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 60760
+    },
+    {
+      "epoch": 399.80263157894734,
+      "grad_norm": 0.9279325604438782,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 60770
+    },
+    {
+      "epoch": 399.86842105263156,
+      "grad_norm": 0.7750979661941528,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 60780
+    },
+    {
+      "epoch": 399.9342105263158,
+      "grad_norm": 0.9537990689277649,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 60790
+    },
+    {
+      "epoch": 400.0,
+      "grad_norm": 1.227004885673523,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 60800
+    },
+    {
+      "epoch": 400.0657894736842,
+      "grad_norm": 0.8979701399803162,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 60810
+    },
+    {
+      "epoch": 400.13157894736844,
+      "grad_norm": 1.022169828414917,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 60820
+    },
+    {
+      "epoch": 400.19736842105266,
+      "grad_norm": 1.129315733909607,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 60830
+    },
+    {
+      "epoch": 400.2631578947368,
+      "grad_norm": 1.0634799003601074,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 60840
+    },
+    {
+      "epoch": 400.32894736842104,
+      "grad_norm": 1.2709938287734985,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 60850
+    },
+    {
+      "epoch": 400.39473684210526,
+      "grad_norm": 1.0252529382705688,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 60860
+    },
+    {
+      "epoch": 400.4605263157895,
+      "grad_norm": 1.4631614685058594,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 60870
+    },
+    {
+      "epoch": 400.5263157894737,
+      "grad_norm": 1.0839378833770752,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 60880
+    },
+    {
+      "epoch": 400.5921052631579,
+      "grad_norm": 0.9850987792015076,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 60890
+    },
+    {
+      "epoch": 400.6578947368421,
+      "grad_norm": 1.6288450956344604,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 60900
+    },
+    {
+      "epoch": 400.7236842105263,
+      "grad_norm": 1.1877349615097046,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 60910
+    },
+    {
+      "epoch": 400.7894736842105,
+      "grad_norm": 1.2342051267623901,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 60920
+    },
+    {
+      "epoch": 400.85526315789474,
+      "grad_norm": 1.1505542993545532,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 60930
+    },
+    {
+      "epoch": 400.92105263157896,
+      "grad_norm": 0.6840497851371765,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 60940
+    },
+    {
+      "epoch": 400.9868421052632,
+      "grad_norm": 1.1132681369781494,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 60950
+    },
+    {
+      "epoch": 401.05263157894734,
+      "grad_norm": 0.9815819263458252,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 60960
+    },
+    {
+      "epoch": 401.11842105263156,
+      "grad_norm": 1.063504695892334,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 60970
+    },
+    {
+      "epoch": 401.1842105263158,
+      "grad_norm": 0.8839133977890015,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 60980
+    },
+    {
+      "epoch": 401.25,
+      "grad_norm": 1.3246194124221802,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 60990
+    },
+    {
+      "epoch": 401.3157894736842,
+      "grad_norm": 1.150536060333252,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 61000
+    },
+    {
+      "epoch": 401.38157894736844,
+      "grad_norm": 1.1703293323516846,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 61010
+    },
+    {
+      "epoch": 401.44736842105266,
+      "grad_norm": 1.1966885328292847,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 61020
+    },
+    {
+      "epoch": 401.5131578947368,
+      "grad_norm": 1.2253130674362183,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 61030
+    },
+    {
+      "epoch": 401.57894736842104,
+      "grad_norm": 0.6977288126945496,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 61040
+    },
+    {
+      "epoch": 401.64473684210526,
+      "grad_norm": 1.388411045074463,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 61050
+    },
+    {
+      "epoch": 401.7105263157895,
+      "grad_norm": 1.45159113407135,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 61060
+    },
+    {
+      "epoch": 401.7763157894737,
+      "grad_norm": 1.221388578414917,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 61070
+    },
+    {
+      "epoch": 401.8421052631579,
+      "grad_norm": 1.2289520502090454,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 61080
+    },
+    {
+      "epoch": 401.9078947368421,
+      "grad_norm": 1.2441534996032715,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 61090
+    },
+    {
+      "epoch": 401.9736842105263,
+      "grad_norm": 1.4623695611953735,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 61100
+    },
+    {
+      "epoch": 402.0394736842105,
+      "grad_norm": 1.220936894416809,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 61110
+    },
+    {
+      "epoch": 402.10526315789474,
+      "grad_norm": 1.2815364599227905,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61120
+    },
+    {
+      "epoch": 402.17105263157896,
+      "grad_norm": 1.2547277212142944,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 61130
+    },
+    {
+      "epoch": 402.2368421052632,
+      "grad_norm": 0.9268215894699097,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 61140
+    },
+    {
+      "epoch": 402.30263157894734,
+      "grad_norm": 1.4284296035766602,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 61150
+    },
+    {
+      "epoch": 402.36842105263156,
+      "grad_norm": 1.183083415031433,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 61160
+    },
+    {
+      "epoch": 402.4342105263158,
+      "grad_norm": 1.3317415714263916,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 61170
+    },
+    {
+      "epoch": 402.5,
+      "grad_norm": 0.9961628317832947,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 61180
+    },
+    {
+      "epoch": 402.5657894736842,
+      "grad_norm": 1.011443018913269,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 61190
+    },
+    {
+      "epoch": 402.63157894736844,
+      "grad_norm": 1.214849829673767,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 61200
+    },
+    {
+      "epoch": 402.69736842105266,
+      "grad_norm": 1.2691972255706787,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 61210
+    },
+    {
+      "epoch": 402.7631578947368,
+      "grad_norm": 0.8713511228561401,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 61220
+    },
+    {
+      "epoch": 402.82894736842104,
+      "grad_norm": 1.2095887660980225,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 61230
+    },
+    {
+      "epoch": 402.89473684210526,
+      "grad_norm": 1.3552407026290894,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 61240
+    },
+    {
+      "epoch": 402.9605263157895,
+      "grad_norm": 1.3354055881500244,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61250
+    },
+    {
+      "epoch": 403.0263157894737,
+      "grad_norm": 1.0809153318405151,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 61260
+    },
+    {
+      "epoch": 403.0921052631579,
+      "grad_norm": 0.9687100052833557,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 61270
+    },
+    {
+      "epoch": 403.1578947368421,
+      "grad_norm": 1.0305453538894653,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 61280
+    },
+    {
+      "epoch": 403.2236842105263,
+      "grad_norm": 1.0682860612869263,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 61290
+    },
+    {
+      "epoch": 403.2894736842105,
+      "grad_norm": 1.3021142482757568,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 61300
+    },
+    {
+      "epoch": 403.35526315789474,
+      "grad_norm": 1.2152711153030396,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 61310
+    },
+    {
+      "epoch": 403.42105263157896,
+      "grad_norm": 1.1612188816070557,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 61320
+    },
+    {
+      "epoch": 403.4868421052632,
+      "grad_norm": 1.2532055377960205,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 61330
+    },
+    {
+      "epoch": 403.55263157894734,
+      "grad_norm": 1.164185881614685,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 61340
+    },
+    {
+      "epoch": 403.61842105263156,
+      "grad_norm": 1.5072556734085083,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 61350
+    },
+    {
+      "epoch": 403.6842105263158,
+      "grad_norm": 1.4515098333358765,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61360
+    },
+    {
+      "epoch": 403.75,
+      "grad_norm": 1.3423659801483154,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61370
+    },
+    {
+      "epoch": 403.8157894736842,
+      "grad_norm": 1.5325146913528442,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 61380
+    },
+    {
+      "epoch": 403.88157894736844,
+      "grad_norm": 1.323291301727295,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61390
+    },
+    {
+      "epoch": 403.94736842105266,
+      "grad_norm": 1.3427553176879883,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 61400
+    },
+    {
+      "epoch": 404.0131578947368,
+      "grad_norm": 1.0898405313491821,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 61410
+    },
+    {
+      "epoch": 404.07894736842104,
+      "grad_norm": 1.0834609270095825,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 61420
+    },
+    {
+      "epoch": 404.14473684210526,
+      "grad_norm": 1.5015757083892822,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 61430
+    },
+    {
+      "epoch": 404.2105263157895,
+      "grad_norm": 1.4444079399108887,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61440
+    },
+    {
+      "epoch": 404.2763157894737,
+      "grad_norm": 1.0211026668548584,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61450
+    },
+    {
+      "epoch": 404.3421052631579,
+      "grad_norm": 1.4435980319976807,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 61460
+    },
+    {
+      "epoch": 404.4078947368421,
+      "grad_norm": 0.9525414109230042,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 61470
+    },
+    {
+      "epoch": 404.4736842105263,
+      "grad_norm": 1.289888620376587,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 61480
+    },
+    {
+      "epoch": 404.5394736842105,
+      "grad_norm": 1.652472734451294,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 61490
+    },
+    {
+      "epoch": 404.60526315789474,
+      "grad_norm": 1.2742666006088257,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 61500
+    },
+    {
+      "epoch": 404.67105263157896,
+      "grad_norm": 0.8357386589050293,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 61510
+    },
+    {
+      "epoch": 404.7368421052632,
+      "grad_norm": 0.9393722414970398,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 61520
+    },
+    {
+      "epoch": 404.80263157894734,
+      "grad_norm": 1.255674123764038,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 61530
+    },
+    {
+      "epoch": 404.86842105263156,
+      "grad_norm": 1.0511587858200073,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 61540
+    },
+    {
+      "epoch": 404.9342105263158,
+      "grad_norm": 0.9492988586425781,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 61550
+    },
+    {
+      "epoch": 405.0,
+      "grad_norm": 1.1605418920516968,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 61560
+    },
+    {
+      "epoch": 405.0657894736842,
+      "grad_norm": 1.098800778388977,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 61570
+    },
+    {
+      "epoch": 405.13157894736844,
+      "grad_norm": 1.138261079788208,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 61580
+    },
+    {
+      "epoch": 405.19736842105266,
+      "grad_norm": 1.2927042245864868,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 61590
+    },
+    {
+      "epoch": 405.2631578947368,
+      "grad_norm": 1.1841543912887573,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 61600
+    },
+    {
+      "epoch": 405.32894736842104,
+      "grad_norm": 1.0832710266113281,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 61610
+    },
+    {
+      "epoch": 405.39473684210526,
+      "grad_norm": 1.35568106174469,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 61620
+    },
+    {
+      "epoch": 405.4605263157895,
+      "grad_norm": 1.0459115505218506,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 61630
+    },
+    {
+      "epoch": 405.5263157894737,
+      "grad_norm": 1.2705081701278687,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 61640
+    },
+    {
+      "epoch": 405.5921052631579,
+      "grad_norm": 0.9033043384552002,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 61650
+    },
+    {
+      "epoch": 405.6578947368421,
+      "grad_norm": 1.2756457328796387,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 61660
+    },
+    {
+      "epoch": 405.7236842105263,
+      "grad_norm": 1.5137383937835693,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 61670
+    },
+    {
+      "epoch": 405.7894736842105,
+      "grad_norm": 0.9483539462089539,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 61680
+    },
+    {
+      "epoch": 405.85526315789474,
+      "grad_norm": 1.1545567512512207,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 61690
+    },
+    {
+      "epoch": 405.92105263157896,
+      "grad_norm": 0.8387587666511536,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 61700
+    },
+    {
+      "epoch": 405.9868421052632,
+      "grad_norm": 0.9866654276847839,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61710
+    },
+    {
+      "epoch": 406.05263157894734,
+      "grad_norm": 0.9306283593177795,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 61720
+    },
+    {
+      "epoch": 406.11842105263156,
+      "grad_norm": 1.089195966720581,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61730
+    },
+    {
+      "epoch": 406.1842105263158,
+      "grad_norm": 1.0379259586334229,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 61740
+    },
+    {
+      "epoch": 406.25,
+      "grad_norm": 0.8618190288543701,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 61750
+    },
+    {
+      "epoch": 406.3157894736842,
+      "grad_norm": 1.3050987720489502,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 61760
+    },
+    {
+      "epoch": 406.38157894736844,
+      "grad_norm": 1.1287416219711304,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 61770
+    },
+    {
+      "epoch": 406.44736842105266,
+      "grad_norm": 0.9702133536338806,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 61780
+    },
+    {
+      "epoch": 406.5131578947368,
+      "grad_norm": 0.9845720529556274,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 61790
+    },
+    {
+      "epoch": 406.57894736842104,
+      "grad_norm": 1.2057915925979614,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 61800
+    },
+    {
+      "epoch": 406.64473684210526,
+      "grad_norm": 1.1550894975662231,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 61810
+    },
+    {
+      "epoch": 406.7105263157895,
+      "grad_norm": 1.2562119960784912,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 61820
+    },
+    {
+      "epoch": 406.7763157894737,
+      "grad_norm": 1.1351388692855835,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 61830
+    },
+    {
+      "epoch": 406.8421052631579,
+      "grad_norm": 1.1127445697784424,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 61840
+    },
+    {
+      "epoch": 406.9078947368421,
+      "grad_norm": 1.1058247089385986,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 61850
+    },
+    {
+      "epoch": 406.9736842105263,
+      "grad_norm": 0.9056158661842346,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 61860
+    },
+    {
+      "epoch": 407.0394736842105,
+      "grad_norm": 1.2106934785842896,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 61870
+    },
+    {
+      "epoch": 407.10526315789474,
+      "grad_norm": 1.2813785076141357,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 61880
+    },
+    {
+      "epoch": 407.17105263157896,
+      "grad_norm": 1.3114577531814575,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 61890
+    },
+    {
+      "epoch": 407.2368421052632,
+      "grad_norm": 1.1716445684432983,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 61900
+    },
+    {
+      "epoch": 407.30263157894734,
+      "grad_norm": 1.2091526985168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 61910
+    },
+    {
+      "epoch": 407.36842105263156,
+      "grad_norm": 0.9812834858894348,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 61920
+    },
+    {
+      "epoch": 407.4342105263158,
+      "grad_norm": 0.9171328544616699,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 61930
+    },
+    {
+      "epoch": 407.5,
+      "grad_norm": 0.8399735689163208,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 61940
+    },
+    {
+      "epoch": 407.5657894736842,
+      "grad_norm": 1.0794377326965332,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 61950
+    },
+    {
+      "epoch": 407.63157894736844,
+      "grad_norm": 1.1794744729995728,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 61960
+    },
+    {
+      "epoch": 407.69736842105266,
+      "grad_norm": 1.2339738607406616,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 61970
+    },
+    {
+      "epoch": 407.7631578947368,
+      "grad_norm": 1.1554800271987915,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 61980
+    },
+    {
+      "epoch": 407.82894736842104,
+      "grad_norm": 0.769442081451416,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 61990
+    },
+    {
+      "epoch": 407.89473684210526,
+      "grad_norm": 0.9051490426063538,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 62000
+    },
+    {
+      "epoch": 407.9605263157895,
+      "grad_norm": 0.8999045491218567,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 62010
+    },
+    {
+      "epoch": 408.0263157894737,
+      "grad_norm": 1.5016242265701294,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 62020
+    },
+    {
+      "epoch": 408.0921052631579,
+      "grad_norm": 1.4286948442459106,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 62030
+    },
+    {
+      "epoch": 408.1578947368421,
+      "grad_norm": 1.1384950876235962,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 62040
+    },
+    {
+      "epoch": 408.2236842105263,
+      "grad_norm": 1.6049597263336182,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 62050
+    },
+    {
+      "epoch": 408.2894736842105,
+      "grad_norm": 1.4698026180267334,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 62060
+    },
+    {
+      "epoch": 408.35526315789474,
+      "grad_norm": 0.9157135486602783,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 62070
+    },
+    {
+      "epoch": 408.42105263157896,
+      "grad_norm": 1.380117654800415,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 62080
+    },
+    {
+      "epoch": 408.4868421052632,
+      "grad_norm": 1.374470829963684,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 62090
+    },
+    {
+      "epoch": 408.55263157894734,
+      "grad_norm": 1.056437611579895,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 62100
+    },
+    {
+      "epoch": 408.61842105263156,
+      "grad_norm": 1.3114112615585327,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 62110
+    },
+    {
+      "epoch": 408.6842105263158,
+      "grad_norm": 1.187699794769287,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 62120
+    },
+    {
+      "epoch": 408.75,
+      "grad_norm": 1.2368489503860474,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 62130
+    },
+    {
+      "epoch": 408.8157894736842,
+      "grad_norm": 1.251200556755066,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 62140
+    },
+    {
+      "epoch": 408.88157894736844,
+      "grad_norm": 1.1192817687988281,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 62150
+    },
+    {
+      "epoch": 408.94736842105266,
+      "grad_norm": 1.289333701133728,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 62160
+    },
+    {
+      "epoch": 409.0131578947368,
+      "grad_norm": 1.1822633743286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 62170
+    },
+    {
+      "epoch": 409.07894736842104,
+      "grad_norm": 0.9145843982696533,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 62180
+    },
+    {
+      "epoch": 409.14473684210526,
+      "grad_norm": 1.0487996339797974,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 62190
+    },
+    {
+      "epoch": 409.2105263157895,
+      "grad_norm": 1.1056833267211914,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 62200
+    },
+    {
+      "epoch": 409.2763157894737,
+      "grad_norm": 1.2669976949691772,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 62210
+    },
+    {
+      "epoch": 409.3421052631579,
+      "grad_norm": 0.501373827457428,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 62220
+    },
+    {
+      "epoch": 409.4078947368421,
+      "grad_norm": 1.0608346462249756,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 62230
+    },
+    {
+      "epoch": 409.4736842105263,
+      "grad_norm": 1.1155251264572144,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 62240
+    },
+    {
+      "epoch": 409.5394736842105,
+      "grad_norm": 1.1546201705932617,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 62250
+    },
+    {
+      "epoch": 409.60526315789474,
+      "grad_norm": 1.0794380903244019,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 62260
+    },
+    {
+      "epoch": 409.67105263157896,
+      "grad_norm": 1.337540626525879,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 62270
+    },
+    {
+      "epoch": 409.7368421052632,
+      "grad_norm": 1.1186962127685547,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 62280
+    },
+    {
+      "epoch": 409.80263157894734,
+      "grad_norm": 1.0747431516647339,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 62290
+    },
+    {
+      "epoch": 409.86842105263156,
+      "grad_norm": 1.2456459999084473,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 62300
+    },
+    {
+      "epoch": 409.9342105263158,
+      "grad_norm": 1.2872223854064941,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 62310
+    },
+    {
+      "epoch": 410.0,
+      "grad_norm": 1.3749200105667114,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 62320
+    },
+    {
+      "epoch": 410.0657894736842,
+      "grad_norm": 1.159676432609558,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 62330
+    },
+    {
+      "epoch": 410.13157894736844,
+      "grad_norm": 1.3028908967971802,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 62340
+    },
+    {
+      "epoch": 410.19736842105266,
+      "grad_norm": 0.9679288864135742,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 62350
+    },
+    {
+      "epoch": 410.2631578947368,
+      "grad_norm": 1.2073003053665161,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 62360
+    },
+    {
+      "epoch": 410.32894736842104,
+      "grad_norm": 1.3585573434829712,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 62370
+    },
+    {
+      "epoch": 410.39473684210526,
+      "grad_norm": 1.4543274641036987,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 62380
+    },
+    {
+      "epoch": 410.4605263157895,
+      "grad_norm": 1.1905204057693481,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 62390
+    },
+    {
+      "epoch": 410.5263157894737,
+      "grad_norm": 1.5460284948349,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 62400
+    },
+    {
+      "epoch": 410.5921052631579,
+      "grad_norm": 1.3457225561141968,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 62410
+    },
+    {
+      "epoch": 410.6578947368421,
+      "grad_norm": 1.2957866191864014,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 62420
+    },
+    {
+      "epoch": 410.7236842105263,
+      "grad_norm": 1.269675374031067,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 62430
+    },
+    {
+      "epoch": 410.7894736842105,
+      "grad_norm": 1.0206223726272583,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 62440
+    },
+    {
+      "epoch": 410.85526315789474,
+      "grad_norm": 1.2072843313217163,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 62450
+    },
+    {
+      "epoch": 410.92105263157896,
+      "grad_norm": 0.9825667142868042,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 62460
+    },
+    {
+      "epoch": 410.9868421052632,
+      "grad_norm": 1.3091356754302979,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 62470
+    },
+    {
+      "epoch": 411.05263157894734,
+      "grad_norm": 1.3110969066619873,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 62480
+    },
+    {
+      "epoch": 411.11842105263156,
+      "grad_norm": 1.2493022680282593,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 62490
+    },
+    {
+      "epoch": 411.1842105263158,
+      "grad_norm": 1.1567295789718628,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 62500
+    },
+    {
+      "epoch": 411.25,
+      "grad_norm": 1.2471519708633423,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 62510
+    },
+    {
+      "epoch": 411.3157894736842,
+      "grad_norm": 0.6940414905548096,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 62520
+    },
+    {
+      "epoch": 411.38157894736844,
+      "grad_norm": 0.9060186743736267,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 62530
+    },
+    {
+      "epoch": 411.44736842105266,
+      "grad_norm": 1.0752925872802734,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 62540
+    },
+    {
+      "epoch": 411.5131578947368,
+      "grad_norm": 1.0197933912277222,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 62550
+    },
+    {
+      "epoch": 411.57894736842104,
+      "grad_norm": 0.9554465413093567,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 62560
+    },
+    {
+      "epoch": 411.64473684210526,
+      "grad_norm": 1.5054553747177124,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 62570
+    },
+    {
+      "epoch": 411.7105263157895,
+      "grad_norm": 1.1572545766830444,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 62580
+    },
+    {
+      "epoch": 411.7763157894737,
+      "grad_norm": 1.1669964790344238,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 62590
+    },
+    {
+      "epoch": 411.8421052631579,
+      "grad_norm": 1.5255320072174072,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 62600
+    },
+    {
+      "epoch": 411.9078947368421,
+      "grad_norm": 1.322806715965271,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 62610
+    },
+    {
+      "epoch": 411.9736842105263,
+      "grad_norm": 1.0405142307281494,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 62620
+    },
+    {
+      "epoch": 412.0394736842105,
+      "grad_norm": 1.3252968788146973,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 62630
+    },
+    {
+      "epoch": 412.10526315789474,
+      "grad_norm": 0.9440876841545105,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 62640
+    },
+    {
+      "epoch": 412.17105263157896,
+      "grad_norm": 1.6244385242462158,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 62650
+    },
+    {
+      "epoch": 412.2368421052632,
+      "grad_norm": 0.9535412192344666,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 62660
+    },
+    {
+      "epoch": 412.30263157894734,
+      "grad_norm": 1.2329378128051758,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 62670
+    },
+    {
+      "epoch": 412.36842105263156,
+      "grad_norm": 0.9353330135345459,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 62680
+    },
+    {
+      "epoch": 412.4342105263158,
+      "grad_norm": 1.4033151865005493,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 62690
+    },
+    {
+      "epoch": 412.5,
+      "grad_norm": 1.1142581701278687,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 62700
+    },
+    {
+      "epoch": 412.5657894736842,
+      "grad_norm": 1.250436544418335,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 62710
+    },
+    {
+      "epoch": 412.63157894736844,
+      "grad_norm": 1.180382490158081,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 62720
+    },
+    {
+      "epoch": 412.69736842105266,
+      "grad_norm": 1.2273600101470947,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 62730
+    },
+    {
+      "epoch": 412.7631578947368,
+      "grad_norm": 0.9863313436508179,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 62740
+    },
+    {
+      "epoch": 412.82894736842104,
+      "grad_norm": 1.3861109018325806,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 62750
+    },
+    {
+      "epoch": 412.89473684210526,
+      "grad_norm": 1.238087773323059,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 62760
+    },
+    {
+      "epoch": 412.9605263157895,
+      "grad_norm": 1.0432476997375488,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 62770
+    },
+    {
+      "epoch": 413.0263157894737,
+      "grad_norm": 1.2043246030807495,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 62780
+    },
+    {
+      "epoch": 413.0921052631579,
+      "grad_norm": 1.2075798511505127,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 62790
+    },
+    {
+      "epoch": 413.1578947368421,
+      "grad_norm": 0.8619531393051147,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 62800
+    },
+    {
+      "epoch": 413.2236842105263,
+      "grad_norm": 1.2081847190856934,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 62810
+    },
+    {
+      "epoch": 413.2894736842105,
+      "grad_norm": 1.0683201551437378,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 62820
+    },
+    {
+      "epoch": 413.35526315789474,
+      "grad_norm": 1.2931760549545288,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 62830
+    },
+    {
+      "epoch": 413.42105263157896,
+      "grad_norm": 0.9082301259040833,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 62840
+    },
+    {
+      "epoch": 413.4868421052632,
+      "grad_norm": 1.485244631767273,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 62850
+    },
+    {
+      "epoch": 413.55263157894734,
+      "grad_norm": 1.0334351062774658,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 62860
+    },
+    {
+      "epoch": 413.61842105263156,
+      "grad_norm": 1.2499173879623413,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 62870
+    },
+    {
+      "epoch": 413.6842105263158,
+      "grad_norm": 1.141554832458496,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 62880
+    },
+    {
+      "epoch": 413.75,
+      "grad_norm": 1.2541395425796509,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 62890
+    },
+    {
+      "epoch": 413.8157894736842,
+      "grad_norm": 1.2289867401123047,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 62900
+    },
+    {
+      "epoch": 413.88157894736844,
+      "grad_norm": 1.5980114936828613,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 62910
+    },
+    {
+      "epoch": 413.94736842105266,
+      "grad_norm": 0.9283551573753357,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 62920
+    },
+    {
+      "epoch": 414.0131578947368,
+      "grad_norm": 1.0979995727539062,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 62930
+    },
+    {
+      "epoch": 414.07894736842104,
+      "grad_norm": 0.8771706819534302,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 62940
+    },
+    {
+      "epoch": 414.14473684210526,
+      "grad_norm": 1.0368901491165161,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 62950
+    },
+    {
+      "epoch": 414.2105263157895,
+      "grad_norm": 1.0799942016601562,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 62960
+    },
+    {
+      "epoch": 414.2763157894737,
+      "grad_norm": 1.2244056463241577,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 62970
+    },
+    {
+      "epoch": 414.3421052631579,
+      "grad_norm": 0.9773150086402893,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 62980
+    },
+    {
+      "epoch": 414.4078947368421,
+      "grad_norm": 1.0069860219955444,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 62990
+    },
+    {
+      "epoch": 414.4736842105263,
+      "grad_norm": 0.9400529265403748,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 63000
+    },
+    {
+      "epoch": 414.5394736842105,
+      "grad_norm": 1.3097772598266602,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 63010
+    },
+    {
+      "epoch": 414.60526315789474,
+      "grad_norm": 1.036408543586731,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 63020
+    },
+    {
+      "epoch": 414.67105263157896,
+      "grad_norm": 0.8755251169204712,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 63030
+    },
+    {
+      "epoch": 414.7368421052632,
+      "grad_norm": 0.8445789813995361,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 63040
+    },
+    {
+      "epoch": 414.80263157894734,
+      "grad_norm": 1.4002634286880493,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 63050
+    },
+    {
+      "epoch": 414.86842105263156,
+      "grad_norm": 1.1597133874893188,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 63060
+    },
+    {
+      "epoch": 414.9342105263158,
+      "grad_norm": 1.0446454286575317,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 63070
+    },
+    {
+      "epoch": 415.0,
+      "grad_norm": 1.177336573600769,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 63080
+    },
+    {
+      "epoch": 415.0657894736842,
+      "grad_norm": 1.3958925008773804,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 63090
+    },
+    {
+      "epoch": 415.13157894736844,
+      "grad_norm": 1.2025551795959473,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 63100
+    },
+    {
+      "epoch": 415.19736842105266,
+      "grad_norm": 1.3240025043487549,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 63110
+    },
+    {
+      "epoch": 415.2631578947368,
+      "grad_norm": 1.2941898107528687,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 63120
+    },
+    {
+      "epoch": 415.32894736842104,
+      "grad_norm": 1.2462248802185059,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 63130
+    },
+    {
+      "epoch": 415.39473684210526,
+      "grad_norm": 1.523111343383789,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 63140
+    },
+    {
+      "epoch": 415.4605263157895,
+      "grad_norm": 1.2437251806259155,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 63150
+    },
+    {
+      "epoch": 415.5263157894737,
+      "grad_norm": 1.0009071826934814,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 63160
+    },
+    {
+      "epoch": 415.5921052631579,
+      "grad_norm": 1.008365511894226,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 63170
+    },
+    {
+      "epoch": 415.6578947368421,
+      "grad_norm": 1.136559009552002,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 63180
+    },
+    {
+      "epoch": 415.7236842105263,
+      "grad_norm": 0.6877762675285339,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 63190
+    },
+    {
+      "epoch": 415.7894736842105,
+      "grad_norm": 1.1004842519760132,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 63200
+    },
+    {
+      "epoch": 415.85526315789474,
+      "grad_norm": 0.9160225987434387,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 63210
+    },
+    {
+      "epoch": 415.92105263157896,
+      "grad_norm": 1.2179346084594727,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 63220
+    },
+    {
+      "epoch": 415.9868421052632,
+      "grad_norm": 1.299672245979309,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 63230
+    },
+    {
+      "epoch": 416.05263157894734,
+      "grad_norm": 1.1643931865692139,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 63240
+    },
+    {
+      "epoch": 416.11842105263156,
+      "grad_norm": 1.36052405834198,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 63250
+    },
+    {
+      "epoch": 416.1842105263158,
+      "grad_norm": 1.2271571159362793,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 63260
+    },
+    {
+      "epoch": 416.25,
+      "grad_norm": 1.2492791414260864,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 63270
+    },
+    {
+      "epoch": 416.3157894736842,
+      "grad_norm": 1.2201145887374878,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 63280
+    },
+    {
+      "epoch": 416.38157894736844,
+      "grad_norm": 1.615103006362915,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 63290
+    },
+    {
+      "epoch": 416.44736842105266,
+      "grad_norm": 1.0031245946884155,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 63300
+    },
+    {
+      "epoch": 416.5131578947368,
+      "grad_norm": 1.1543900966644287,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 63310
+    },
+    {
+      "epoch": 416.57894736842104,
+      "grad_norm": 1.0617386102676392,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 63320
+    },
+    {
+      "epoch": 416.64473684210526,
+      "grad_norm": 1.2433397769927979,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 63330
+    },
+    {
+      "epoch": 416.7105263157895,
+      "grad_norm": 1.307660698890686,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 63340
+    },
+    {
+      "epoch": 416.7763157894737,
+      "grad_norm": 1.402388334274292,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 63350
+    },
+    {
+      "epoch": 416.8421052631579,
+      "grad_norm": 1.2867043018341064,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 63360
+    },
+    {
+      "epoch": 416.9078947368421,
+      "grad_norm": 1.1766114234924316,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 63370
+    },
+    {
+      "epoch": 416.9736842105263,
+      "grad_norm": 1.2115635871887207,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 63380
+    },
+    {
+      "epoch": 417.0394736842105,
+      "grad_norm": 1.2058789730072021,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 63390
+    },
+    {
+      "epoch": 417.10526315789474,
+      "grad_norm": 1.194610834121704,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 63400
+    },
+    {
+      "epoch": 417.17105263157896,
+      "grad_norm": 0.9211453199386597,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 63410
+    },
+    {
+      "epoch": 417.2368421052632,
+      "grad_norm": 1.093869924545288,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 63420
+    },
+    {
+      "epoch": 417.30263157894734,
+      "grad_norm": 0.9869666695594788,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 63430
+    },
+    {
+      "epoch": 417.36842105263156,
+      "grad_norm": 1.1537610292434692,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 63440
+    },
+    {
+      "epoch": 417.4342105263158,
+      "grad_norm": 0.9441319108009338,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 63450
+    },
+    {
+      "epoch": 417.5,
+      "grad_norm": 1.3330210447311401,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 63460
+    },
+    {
+      "epoch": 417.5657894736842,
+      "grad_norm": 1.4678821563720703,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 63470
+    },
+    {
+      "epoch": 417.63157894736844,
+      "grad_norm": 1.0894778966903687,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 63480
+    },
+    {
+      "epoch": 417.69736842105266,
+      "grad_norm": 1.3861744403839111,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 63490
+    },
+    {
+      "epoch": 417.7631578947368,
+      "grad_norm": 1.1523667573928833,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 63500
+    },
+    {
+      "epoch": 417.82894736842104,
+      "grad_norm": 1.058275818824768,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 63510
+    },
+    {
+      "epoch": 417.89473684210526,
+      "grad_norm": 1.2407093048095703,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 63520
+    },
+    {
+      "epoch": 417.9605263157895,
+      "grad_norm": 0.861865222454071,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 63530
+    },
+    {
+      "epoch": 418.0263157894737,
+      "grad_norm": 1.3533257246017456,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 63540
+    },
+    {
+      "epoch": 418.0921052631579,
+      "grad_norm": 0.8494672179222107,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 63550
+    },
+    {
+      "epoch": 418.1578947368421,
+      "grad_norm": 0.9071248769760132,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 63560
+    },
+    {
+      "epoch": 418.2236842105263,
+      "grad_norm": 1.4930342435836792,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 63570
+    },
+    {
+      "epoch": 418.2894736842105,
+      "grad_norm": 1.4266362190246582,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 63580
+    },
+    {
+      "epoch": 418.35526315789474,
+      "grad_norm": 1.1286511421203613,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 63590
+    },
+    {
+      "epoch": 418.42105263157896,
+      "grad_norm": 1.0800775289535522,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 63600
+    },
+    {
+      "epoch": 418.4868421052632,
+      "grad_norm": 1.0998226404190063,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 63610
+    },
+    {
+      "epoch": 418.55263157894734,
+      "grad_norm": 1.3095847368240356,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 63620
+    },
+    {
+      "epoch": 418.61842105263156,
+      "grad_norm": 1.0724563598632812,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 63630
+    },
+    {
+      "epoch": 418.6842105263158,
+      "grad_norm": 0.9985296726226807,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 63640
+    },
+    {
+      "epoch": 418.75,
+      "grad_norm": 1.2534633874893188,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 63650
+    },
+    {
+      "epoch": 418.8157894736842,
+      "grad_norm": 1.4504799842834473,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 63660
+    },
+    {
+      "epoch": 418.88157894736844,
+      "grad_norm": 1.180578589439392,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 63670
+    },
+    {
+      "epoch": 418.94736842105266,
+      "grad_norm": 1.6236175298690796,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 63680
+    },
+    {
+      "epoch": 419.0131578947368,
+      "grad_norm": 1.1032344102859497,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 63690
+    },
+    {
+      "epoch": 419.07894736842104,
+      "grad_norm": 1.1350429058074951,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 63700
+    },
+    {
+      "epoch": 419.14473684210526,
+      "grad_norm": 1.1316243410110474,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 63710
+    },
+    {
+      "epoch": 419.2105263157895,
+      "grad_norm": 1.0160430669784546,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 63720
+    },
+    {
+      "epoch": 419.2763157894737,
+      "grad_norm": 0.9599151015281677,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 63730
+    },
+    {
+      "epoch": 419.3421052631579,
+      "grad_norm": 1.0853590965270996,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 63740
+    },
+    {
+      "epoch": 419.4078947368421,
+      "grad_norm": 1.0868717432022095,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 63750
+    },
+    {
+      "epoch": 419.4736842105263,
+      "grad_norm": 0.9881595373153687,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 63760
+    },
+    {
+      "epoch": 419.5394736842105,
+      "grad_norm": 1.1161201000213623,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 63770
+    },
+    {
+      "epoch": 419.60526315789474,
+      "grad_norm": 0.8587058186531067,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 63780
+    },
+    {
+      "epoch": 419.67105263157896,
+      "grad_norm": 0.959867537021637,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 63790
+    },
+    {
+      "epoch": 419.7368421052632,
+      "grad_norm": 1.1373491287231445,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 63800
+    },
+    {
+      "epoch": 419.80263157894734,
+      "grad_norm": 1.2124371528625488,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 63810
+    },
+    {
+      "epoch": 419.86842105263156,
+      "grad_norm": 1.1370086669921875,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 63820
+    },
+    {
+      "epoch": 419.9342105263158,
+      "grad_norm": 1.2051914930343628,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 63830
+    },
+    {
+      "epoch": 420.0,
+      "grad_norm": 1.598705530166626,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 63840
+    },
+    {
+      "epoch": 420.0657894736842,
+      "grad_norm": 1.2667107582092285,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 63850
+    },
+    {
+      "epoch": 420.13157894736844,
+      "grad_norm": 1.4038127660751343,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 63860
+    },
+    {
+      "epoch": 420.19736842105266,
+      "grad_norm": 1.233218789100647,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 63870
+    },
+    {
+      "epoch": 420.2631578947368,
+      "grad_norm": 1.2096288204193115,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 63880
+    },
+    {
+      "epoch": 420.32894736842104,
+      "grad_norm": 1.050657033920288,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 63890
+    },
+    {
+      "epoch": 420.39473684210526,
+      "grad_norm": 1.7145295143127441,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 63900
+    },
+    {
+      "epoch": 420.4605263157895,
+      "grad_norm": 1.58820378780365,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 63910
+    },
+    {
+      "epoch": 420.5263157894737,
+      "grad_norm": 0.9367581009864807,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 63920
+    },
+    {
+      "epoch": 420.5921052631579,
+      "grad_norm": 1.4414033889770508,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 63930
+    },
+    {
+      "epoch": 420.6578947368421,
+      "grad_norm": 1.310313105583191,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 63940
+    },
+    {
+      "epoch": 420.7236842105263,
+      "grad_norm": 0.850493848323822,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 63950
+    },
+    {
+      "epoch": 420.7894736842105,
+      "grad_norm": 1.3179919719696045,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 63960
+    },
+    {
+      "epoch": 420.85526315789474,
+      "grad_norm": 1.3989521265029907,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 63970
+    },
+    {
+      "epoch": 420.92105263157896,
+      "grad_norm": 1.3190523386001587,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 63980
+    },
+    {
+      "epoch": 420.9868421052632,
+      "grad_norm": 0.8655247688293457,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 63990
+    },
+    {
+      "epoch": 421.05263157894734,
+      "grad_norm": 1.5341448783874512,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 64000
+    },
+    {
+      "epoch": 421.11842105263156,
+      "grad_norm": 1.0463502407073975,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 64010
+    },
+    {
+      "epoch": 421.1842105263158,
+      "grad_norm": 0.6681509613990784,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 64020
+    },
+    {
+      "epoch": 421.25,
+      "grad_norm": 1.323490858078003,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 64030
+    },
+    {
+      "epoch": 421.3157894736842,
+      "grad_norm": 0.8754792213439941,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 64040
+    },
+    {
+      "epoch": 421.38157894736844,
+      "grad_norm": 1.2560418844223022,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 64050
+    },
+    {
+      "epoch": 421.44736842105266,
+      "grad_norm": 1.2175922393798828,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 64060
+    },
+    {
+      "epoch": 421.5131578947368,
+      "grad_norm": 1.4118285179138184,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 64070
+    },
+    {
+      "epoch": 421.57894736842104,
+      "grad_norm": 1.3266385793685913,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 64080
+    },
+    {
+      "epoch": 421.64473684210526,
+      "grad_norm": 1.113729476928711,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 64090
+    },
+    {
+      "epoch": 421.7105263157895,
+      "grad_norm": 1.134850025177002,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 64100
+    },
+    {
+      "epoch": 421.7763157894737,
+      "grad_norm": 0.9437716603279114,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 64110
+    },
+    {
+      "epoch": 421.8421052631579,
+      "grad_norm": 1.0176291465759277,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 64120
+    },
+    {
+      "epoch": 421.9078947368421,
+      "grad_norm": 1.2647275924682617,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 64130
+    },
+    {
+      "epoch": 421.9736842105263,
+      "grad_norm": 0.7509058713912964,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 64140
+    },
+    {
+      "epoch": 422.0394736842105,
+      "grad_norm": 0.9064149260520935,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 64150
+    },
+    {
+      "epoch": 422.10526315789474,
+      "grad_norm": 1.20452880859375,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 64160
+    },
+    {
+      "epoch": 422.17105263157896,
+      "grad_norm": 1.04318368434906,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 64170
+    },
+    {
+      "epoch": 422.2368421052632,
+      "grad_norm": 1.1167349815368652,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 64180
+    },
+    {
+      "epoch": 422.30263157894734,
+      "grad_norm": 1.3725717067718506,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 64190
+    },
+    {
+      "epoch": 422.36842105263156,
+      "grad_norm": 1.0042188167572021,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 64200
+    },
+    {
+      "epoch": 422.4342105263158,
+      "grad_norm": 0.822090208530426,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 64210
+    },
+    {
+      "epoch": 422.5,
+      "grad_norm": 1.2058206796646118,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 64220
+    },
+    {
+      "epoch": 422.5657894736842,
+      "grad_norm": 0.9390090703964233,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 64230
+    },
+    {
+      "epoch": 422.63157894736844,
+      "grad_norm": 1.168169379234314,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 64240
+    },
+    {
+      "epoch": 422.69736842105266,
+      "grad_norm": 1.0627769231796265,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 64250
+    },
+    {
+      "epoch": 422.7631578947368,
+      "grad_norm": 0.819271981716156,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 64260
+    },
+    {
+      "epoch": 422.82894736842104,
+      "grad_norm": 1.1444475650787354,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 64270
+    },
+    {
+      "epoch": 422.89473684210526,
+      "grad_norm": 1.187440276145935,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 64280
+    },
+    {
+      "epoch": 422.9605263157895,
+      "grad_norm": 1.1766507625579834,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 64290
+    },
+    {
+      "epoch": 423.0263157894737,
+      "grad_norm": 1.2091718912124634,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 64300
+    },
+    {
+      "epoch": 423.0921052631579,
+      "grad_norm": 0.8333958983421326,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 64310
+    },
+    {
+      "epoch": 423.1578947368421,
+      "grad_norm": 1.0338822603225708,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 64320
+    },
+    {
+      "epoch": 423.2236842105263,
+      "grad_norm": 1.3641862869262695,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 64330
+    },
+    {
+      "epoch": 423.2894736842105,
+      "grad_norm": 1.1286553144454956,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 64340
+    },
+    {
+      "epoch": 423.35526315789474,
+      "grad_norm": 1.0030713081359863,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 64350
+    },
+    {
+      "epoch": 423.42105263157896,
+      "grad_norm": 0.8953898549079895,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 64360
+    },
+    {
+      "epoch": 423.4868421052632,
+      "grad_norm": 0.9812647700309753,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 64370
+    },
+    {
+      "epoch": 423.55263157894734,
+      "grad_norm": 1.1874929666519165,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 64380
+    },
+    {
+      "epoch": 423.61842105263156,
+      "grad_norm": 1.0838314294815063,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 64390
+    },
+    {
+      "epoch": 423.6842105263158,
+      "grad_norm": 1.1809245347976685,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 64400
+    },
+    {
+      "epoch": 423.75,
+      "grad_norm": 1.1488882303237915,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 64410
+    },
+    {
+      "epoch": 423.8157894736842,
+      "grad_norm": 1.5134193897247314,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 64420
+    },
+    {
+      "epoch": 423.88157894736844,
+      "grad_norm": 1.3163988590240479,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 64430
+    },
+    {
+      "epoch": 423.94736842105266,
+      "grad_norm": 1.3154017925262451,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 64440
+    },
+    {
+      "epoch": 424.0131578947368,
+      "grad_norm": 0.980704665184021,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 64450
+    },
+    {
+      "epoch": 424.07894736842104,
+      "grad_norm": 1.4469650983810425,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 64460
+    },
+    {
+      "epoch": 424.14473684210526,
+      "grad_norm": 1.6586270332336426,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 64470
+    },
+    {
+      "epoch": 424.2105263157895,
+      "grad_norm": 1.164552092552185,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 64480
+    },
+    {
+      "epoch": 424.2763157894737,
+      "grad_norm": 1.2287418842315674,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 64490
+    },
+    {
+      "epoch": 424.3421052631579,
+      "grad_norm": 1.2211633920669556,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 64500
+    },
+    {
+      "epoch": 424.4078947368421,
+      "grad_norm": 0.9924276471138,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 64510
+    },
+    {
+      "epoch": 424.4736842105263,
+      "grad_norm": 1.504193663597107,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 64520
+    },
+    {
+      "epoch": 424.5394736842105,
+      "grad_norm": 1.217288613319397,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 64530
+    },
+    {
+      "epoch": 424.60526315789474,
+      "grad_norm": 0.943494439125061,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 64540
+    },
+    {
+      "epoch": 424.67105263157896,
+      "grad_norm": 1.0527056455612183,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 64550
+    },
+    {
+      "epoch": 424.7368421052632,
+      "grad_norm": 0.7593855261802673,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 64560
+    },
+    {
+      "epoch": 424.80263157894734,
+      "grad_norm": 0.9078150391578674,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 64570
+    },
+    {
+      "epoch": 424.86842105263156,
+      "grad_norm": 1.2566509246826172,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 64580
+    },
+    {
+      "epoch": 424.9342105263158,
+      "grad_norm": 0.9278975129127502,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 64590
+    },
+    {
+      "epoch": 425.0,
+      "grad_norm": 1.0791572332382202,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 64600
+    },
+    {
+      "epoch": 425.0657894736842,
+      "grad_norm": 1.1237635612487793,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 64610
+    },
+    {
+      "epoch": 425.13157894736844,
+      "grad_norm": 1.0109001398086548,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 64620
+    },
+    {
+      "epoch": 425.19736842105266,
+      "grad_norm": 1.1927449703216553,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 64630
+    },
+    {
+      "epoch": 425.2631578947368,
+      "grad_norm": 0.9462395310401917,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 64640
+    },
+    {
+      "epoch": 425.32894736842104,
+      "grad_norm": 1.2187925577163696,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 64650
+    },
+    {
+      "epoch": 425.39473684210526,
+      "grad_norm": 0.9968974590301514,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 64660
+    },
+    {
+      "epoch": 425.4605263157895,
+      "grad_norm": 1.0321686267852783,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 64670
+    },
+    {
+      "epoch": 425.5263157894737,
+      "grad_norm": 0.9968619346618652,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 64680
+    },
+    {
+      "epoch": 425.5921052631579,
+      "grad_norm": 1.1771488189697266,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 64690
+    },
+    {
+      "epoch": 425.6578947368421,
+      "grad_norm": 1.3401336669921875,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 64700
+    },
+    {
+      "epoch": 425.7236842105263,
+      "grad_norm": 1.4088808298110962,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 64710
+    },
+    {
+      "epoch": 425.7894736842105,
+      "grad_norm": 0.976675808429718,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 64720
+    },
+    {
+      "epoch": 425.85526315789474,
+      "grad_norm": 1.2826030254364014,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 64730
+    },
+    {
+      "epoch": 425.92105263157896,
+      "grad_norm": 0.7183810472488403,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 64740
+    },
+    {
+      "epoch": 425.9868421052632,
+      "grad_norm": 1.3914217948913574,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 64750
+    },
+    {
+      "epoch": 426.05263157894734,
+      "grad_norm": 1.1009012460708618,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 64760
+    },
+    {
+      "epoch": 426.11842105263156,
+      "grad_norm": 1.2613898515701294,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 64770
+    },
+    {
+      "epoch": 426.1842105263158,
+      "grad_norm": 1.666262149810791,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 64780
+    },
+    {
+      "epoch": 426.25,
+      "grad_norm": 1.3029481172561646,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 64790
+    },
+    {
+      "epoch": 426.3157894736842,
+      "grad_norm": 1.0970615148544312,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 64800
+    },
+    {
+      "epoch": 426.38157894736844,
+      "grad_norm": 0.8303019404411316,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 64810
+    },
+    {
+      "epoch": 426.44736842105266,
+      "grad_norm": 1.0774695873260498,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 64820
+    },
+    {
+      "epoch": 426.5131578947368,
+      "grad_norm": 1.2249752283096313,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 64830
+    },
+    {
+      "epoch": 426.57894736842104,
+      "grad_norm": 1.4472177028656006,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 64840
+    },
+    {
+      "epoch": 426.64473684210526,
+      "grad_norm": 1.1271650791168213,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 64850
+    },
+    {
+      "epoch": 426.7105263157895,
+      "grad_norm": 1.2733358144760132,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 64860
+    },
+    {
+      "epoch": 426.7763157894737,
+      "grad_norm": 0.8488833904266357,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 64870
+    },
+    {
+      "epoch": 426.8421052631579,
+      "grad_norm": 0.9887498021125793,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 64880
+    },
+    {
+      "epoch": 426.9078947368421,
+      "grad_norm": 1.160360336303711,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 64890
+    },
+    {
+      "epoch": 426.9736842105263,
+      "grad_norm": 1.204254388809204,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 64900
+    },
+    {
+      "epoch": 427.0394736842105,
+      "grad_norm": 1.3775442838668823,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 64910
+    },
+    {
+      "epoch": 427.10526315789474,
+      "grad_norm": 1.2718161344528198,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 64920
+    },
+    {
+      "epoch": 427.17105263157896,
+      "grad_norm": 1.3851796388626099,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 64930
+    },
+    {
+      "epoch": 427.2368421052632,
+      "grad_norm": 1.498613715171814,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 64940
+    },
+    {
+      "epoch": 427.30263157894734,
+      "grad_norm": 0.9563484191894531,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 64950
+    },
+    {
+      "epoch": 427.36842105263156,
+      "grad_norm": 1.0629527568817139,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 64960
+    },
+    {
+      "epoch": 427.4342105263158,
+      "grad_norm": 1.2376042604446411,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 64970
+    },
+    {
+      "epoch": 427.5,
+      "grad_norm": 1.0314258337020874,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 64980
+    },
+    {
+      "epoch": 427.5657894736842,
+      "grad_norm": 1.2879270315170288,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 64990
+    },
+    {
+      "epoch": 427.63157894736844,
+      "grad_norm": 0.971474289894104,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 65000
+    },
+    {
+      "epoch": 427.69736842105266,
+      "grad_norm": 0.9196737408638,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 65010
+    },
+    {
+      "epoch": 427.7631578947368,
+      "grad_norm": 0.862593412399292,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 65020
+    },
+    {
+      "epoch": 427.82894736842104,
+      "grad_norm": 1.113861083984375,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 65030
+    },
+    {
+      "epoch": 427.89473684210526,
+      "grad_norm": 1.189150333404541,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 65040
+    },
+    {
+      "epoch": 427.9605263157895,
+      "grad_norm": 1.0264296531677246,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 65050
+    },
+    {
+      "epoch": 428.0263157894737,
+      "grad_norm": 1.2653850317001343,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 65060
+    },
+    {
+      "epoch": 428.0921052631579,
+      "grad_norm": 1.260859489440918,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 65070
+    },
+    {
+      "epoch": 428.1578947368421,
+      "grad_norm": 1.6127079725265503,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 65080
+    },
+    {
+      "epoch": 428.2236842105263,
+      "grad_norm": 1.2209359407424927,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 65090
+    },
+    {
+      "epoch": 428.2894736842105,
+      "grad_norm": 1.2434695959091187,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 65100
+    },
+    {
+      "epoch": 428.35526315789474,
+      "grad_norm": 1.273813009262085,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 65110
+    },
+    {
+      "epoch": 428.42105263157896,
+      "grad_norm": 1.580181360244751,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 65120
+    },
+    {
+      "epoch": 428.4868421052632,
+      "grad_norm": 1.0633710622787476,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 65130
+    },
+    {
+      "epoch": 428.55263157894734,
+      "grad_norm": 1.0825523138046265,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 65140
+    },
+    {
+      "epoch": 428.61842105263156,
+      "grad_norm": 1.143089771270752,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 65150
+    },
+    {
+      "epoch": 428.6842105263158,
+      "grad_norm": 0.7292158603668213,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 65160
+    },
+    {
+      "epoch": 428.75,
+      "grad_norm": 1.3087058067321777,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 65170
+    },
+    {
+      "epoch": 428.8157894736842,
+      "grad_norm": 1.2735743522644043,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 65180
+    },
+    {
+      "epoch": 428.88157894736844,
+      "grad_norm": 1.169930100440979,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 65190
+    },
+    {
+      "epoch": 428.94736842105266,
+      "grad_norm": 1.1776015758514404,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 65200
+    },
+    {
+      "epoch": 429.0131578947368,
+      "grad_norm": 1.1052436828613281,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 65210
+    },
+    {
+      "epoch": 429.07894736842104,
+      "grad_norm": 0.8525586724281311,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 65220
+    },
+    {
+      "epoch": 429.14473684210526,
+      "grad_norm": 0.9319892525672913,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 65230
+    },
+    {
+      "epoch": 429.2105263157895,
+      "grad_norm": 0.9886743426322937,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 65240
+    },
+    {
+      "epoch": 429.2763157894737,
+      "grad_norm": 0.850257158279419,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 65250
+    },
+    {
+      "epoch": 429.3421052631579,
+      "grad_norm": 1.0104154348373413,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 65260
+    },
+    {
+      "epoch": 429.4078947368421,
+      "grad_norm": 0.8095145225524902,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 65270
+    },
+    {
+      "epoch": 429.4736842105263,
+      "grad_norm": 1.1854900121688843,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 65280
+    },
+    {
+      "epoch": 429.5394736842105,
+      "grad_norm": 1.049673080444336,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 65290
+    },
+    {
+      "epoch": 429.60526315789474,
+      "grad_norm": 1.3695430755615234,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 65300
+    },
+    {
+      "epoch": 429.67105263157896,
+      "grad_norm": 1.3096168041229248,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 65310
+    },
+    {
+      "epoch": 429.7368421052632,
+      "grad_norm": 1.11677086353302,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 65320
+    },
+    {
+      "epoch": 429.80263157894734,
+      "grad_norm": 1.0054956674575806,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 65330
+    },
+    {
+      "epoch": 429.86842105263156,
+      "grad_norm": 1.0261496305465698,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 65340
+    },
+    {
+      "epoch": 429.9342105263158,
+      "grad_norm": 0.8702152967453003,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 65350
+    },
+    {
+      "epoch": 430.0,
+      "grad_norm": 1.1188616752624512,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 65360
+    },
+    {
+      "epoch": 430.0657894736842,
+      "grad_norm": 0.9277343153953552,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 65370
+    },
+    {
+      "epoch": 430.13157894736844,
+      "grad_norm": 0.8487063646316528,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 65380
+    },
+    {
+      "epoch": 430.19736842105266,
+      "grad_norm": 1.0749578475952148,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 65390
+    },
+    {
+      "epoch": 430.2631578947368,
+      "grad_norm": 1.4141275882720947,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 65400
+    },
+    {
+      "epoch": 430.32894736842104,
+      "grad_norm": 1.4153060913085938,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 65410
+    },
+    {
+      "epoch": 430.39473684210526,
+      "grad_norm": 1.0824086666107178,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 65420
+    },
+    {
+      "epoch": 430.4605263157895,
+      "grad_norm": 1.1492947340011597,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 65430
+    },
+    {
+      "epoch": 430.5263157894737,
+      "grad_norm": 0.9777079820632935,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 65440
+    },
+    {
+      "epoch": 430.5921052631579,
+      "grad_norm": 1.278460144996643,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 65450
+    },
+    {
+      "epoch": 430.6578947368421,
+      "grad_norm": 1.0033302307128906,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 65460
+    },
+    {
+      "epoch": 430.7236842105263,
+      "grad_norm": 1.1041438579559326,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 65470
+    },
+    {
+      "epoch": 430.7894736842105,
+      "grad_norm": 1.6266603469848633,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 65480
+    },
+    {
+      "epoch": 430.85526315789474,
+      "grad_norm": 1.2963449954986572,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 65490
+    },
+    {
+      "epoch": 430.92105263157896,
+      "grad_norm": 1.226628303527832,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 65500
+    },
+    {
+      "epoch": 430.9868421052632,
+      "grad_norm": 1.3734506368637085,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 65510
+    },
+    {
+      "epoch": 431.05263157894734,
+      "grad_norm": 1.5446348190307617,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 65520
+    },
+    {
+      "epoch": 431.11842105263156,
+      "grad_norm": 1.2585902214050293,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 65530
+    },
+    {
+      "epoch": 431.1842105263158,
+      "grad_norm": 1.1303677558898926,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 65540
+    },
+    {
+      "epoch": 431.25,
+      "grad_norm": 1.005300521850586,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 65550
+    },
+    {
+      "epoch": 431.3157894736842,
+      "grad_norm": 1.325517177581787,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 65560
+    },
+    {
+      "epoch": 431.38157894736844,
+      "grad_norm": 0.9297947883605957,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 65570
+    },
+    {
+      "epoch": 431.44736842105266,
+      "grad_norm": 1.0581676959991455,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 65580
+    },
+    {
+      "epoch": 431.5131578947368,
+      "grad_norm": 0.9958070516586304,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 65590
+    },
+    {
+      "epoch": 431.57894736842104,
+      "grad_norm": 1.3091784715652466,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 65600
+    },
+    {
+      "epoch": 431.64473684210526,
+      "grad_norm": 1.091407060623169,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 65610
+    },
+    {
+      "epoch": 431.7105263157895,
+      "grad_norm": 1.4302153587341309,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 65620
+    },
+    {
+      "epoch": 431.7763157894737,
+      "grad_norm": 1.0654196739196777,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 65630
+    },
+    {
+      "epoch": 431.8421052631579,
+      "grad_norm": 1.413366436958313,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 65640
+    },
+    {
+      "epoch": 431.9078947368421,
+      "grad_norm": 1.09171462059021,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 65650
+    },
+    {
+      "epoch": 431.9736842105263,
+      "grad_norm": 1.5487229824066162,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 65660
+    },
+    {
+      "epoch": 432.0394736842105,
+      "grad_norm": 1.2298110723495483,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 65670
+    },
+    {
+      "epoch": 432.10526315789474,
+      "grad_norm": 1.2410331964492798,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 65680
+    },
+    {
+      "epoch": 432.17105263157896,
+      "grad_norm": 1.3613755702972412,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 65690
+    },
+    {
+      "epoch": 432.2368421052632,
+      "grad_norm": 1.2268283367156982,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 65700
+    },
+    {
+      "epoch": 432.30263157894734,
+      "grad_norm": 1.8424831628799438,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 65710
+    },
+    {
+      "epoch": 432.36842105263156,
+      "grad_norm": 1.5524377822875977,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 65720
+    },
+    {
+      "epoch": 432.4342105263158,
+      "grad_norm": 1.7905634641647339,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 65730
+    },
+    {
+      "epoch": 432.5,
+      "grad_norm": 1.3174190521240234,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 65740
+    },
+    {
+      "epoch": 432.5657894736842,
+      "grad_norm": 1.387406587600708,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 65750
+    },
+    {
+      "epoch": 432.63157894736844,
+      "grad_norm": 1.184760570526123,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 65760
+    },
+    {
+      "epoch": 432.69736842105266,
+      "grad_norm": 1.0794061422348022,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 65770
+    },
+    {
+      "epoch": 432.7631578947368,
+      "grad_norm": 1.1350884437561035,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 65780
+    },
+    {
+      "epoch": 432.82894736842104,
+      "grad_norm": 1.20332670211792,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 65790
+    },
+    {
+      "epoch": 432.89473684210526,
+      "grad_norm": 1.2901190519332886,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 65800
+    },
+    {
+      "epoch": 432.9605263157895,
+      "grad_norm": 0.93093341588974,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 65810
+    },
+    {
+      "epoch": 433.0263157894737,
+      "grad_norm": 1.0623509883880615,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 65820
+    },
+    {
+      "epoch": 433.0921052631579,
+      "grad_norm": 1.2282602787017822,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 65830
+    },
+    {
+      "epoch": 433.1578947368421,
+      "grad_norm": 1.1212533712387085,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 65840
+    },
+    {
+      "epoch": 433.2236842105263,
+      "grad_norm": 1.0316697359085083,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 65850
+    },
+    {
+      "epoch": 433.2894736842105,
+      "grad_norm": 0.9327832460403442,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 65860
+    },
+    {
+      "epoch": 433.35526315789474,
+      "grad_norm": 1.0273808240890503,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 65870
+    },
+    {
+      "epoch": 433.42105263157896,
+      "grad_norm": 0.8748319745063782,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 65880
+    },
+    {
+      "epoch": 433.4868421052632,
+      "grad_norm": 0.8934853076934814,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 65890
+    },
+    {
+      "epoch": 433.55263157894734,
+      "grad_norm": 0.9318240880966187,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 65900
+    },
+    {
+      "epoch": 433.61842105263156,
+      "grad_norm": 1.1659666299819946,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 65910
+    },
+    {
+      "epoch": 433.6842105263158,
+      "grad_norm": 0.9181238412857056,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 65920
+    },
+    {
+      "epoch": 433.75,
+      "grad_norm": 0.6898985505104065,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 65930
+    },
+    {
+      "epoch": 433.8157894736842,
+      "grad_norm": 0.89139324426651,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 65940
+    },
+    {
+      "epoch": 433.88157894736844,
+      "grad_norm": 1.3110442161560059,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 65950
+    },
+    {
+      "epoch": 433.94736842105266,
+      "grad_norm": 1.3319469690322876,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 65960
+    },
+    {
+      "epoch": 434.0131578947368,
+      "grad_norm": 0.9168400168418884,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 65970
+    },
+    {
+      "epoch": 434.07894736842104,
+      "grad_norm": 1.151136875152588,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 65980
+    },
+    {
+      "epoch": 434.14473684210526,
+      "grad_norm": 1.2665919065475464,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 65990
+    },
+    {
+      "epoch": 434.2105263157895,
+      "grad_norm": 1.0252939462661743,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 66000
+    },
+    {
+      "epoch": 434.2763157894737,
+      "grad_norm": 1.1638813018798828,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 66010
+    },
+    {
+      "epoch": 434.3421052631579,
+      "grad_norm": 0.9939305186271667,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 66020
+    },
+    {
+      "epoch": 434.4078947368421,
+      "grad_norm": 1.4579282999038696,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 66030
+    },
+    {
+      "epoch": 434.4736842105263,
+      "grad_norm": 1.1966112852096558,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 66040
+    },
+    {
+      "epoch": 434.5394736842105,
+      "grad_norm": 1.0983270406723022,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 66050
+    },
+    {
+      "epoch": 434.60526315789474,
+      "grad_norm": 1.0120450258255005,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 66060
+    },
+    {
+      "epoch": 434.67105263157896,
+      "grad_norm": 1.0957145690917969,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 66070
+    },
+    {
+      "epoch": 434.7368421052632,
+      "grad_norm": 1.030774474143982,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 66080
+    },
+    {
+      "epoch": 434.80263157894734,
+      "grad_norm": 1.2620954513549805,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 66090
+    },
+    {
+      "epoch": 434.86842105263156,
+      "grad_norm": 0.9031997919082642,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 66100
+    },
+    {
+      "epoch": 434.9342105263158,
+      "grad_norm": 1.1857893466949463,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 66110
+    },
+    {
+      "epoch": 435.0,
+      "grad_norm": 1.083285927772522,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 66120
+    },
+    {
+      "epoch": 435.0657894736842,
+      "grad_norm": 1.357467770576477,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 66130
+    },
+    {
+      "epoch": 435.13157894736844,
+      "grad_norm": 1.4382094144821167,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 66140
+    },
+    {
+      "epoch": 435.19736842105266,
+      "grad_norm": 1.0308399200439453,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 66150
+    },
+    {
+      "epoch": 435.2631578947368,
+      "grad_norm": 1.303141713142395,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 66160
+    },
+    {
+      "epoch": 435.32894736842104,
+      "grad_norm": 1.2150816917419434,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 66170
+    },
+    {
+      "epoch": 435.39473684210526,
+      "grad_norm": 0.9982725977897644,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 66180
+    },
+    {
+      "epoch": 435.4605263157895,
+      "grad_norm": 0.9391024112701416,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 66190
+    },
+    {
+      "epoch": 435.5263157894737,
+      "grad_norm": 0.7449613809585571,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 66200
+    },
+    {
+      "epoch": 435.5921052631579,
+      "grad_norm": 1.597276210784912,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 66210
+    },
+    {
+      "epoch": 435.6578947368421,
+      "grad_norm": 1.5920016765594482,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 66220
+    },
+    {
+      "epoch": 435.7236842105263,
+      "grad_norm": 1.151368260383606,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 66230
+    },
+    {
+      "epoch": 435.7894736842105,
+      "grad_norm": 1.2534699440002441,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 66240
+    },
+    {
+      "epoch": 435.85526315789474,
+      "grad_norm": 1.220913052558899,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 66250
+    },
+    {
+      "epoch": 435.92105263157896,
+      "grad_norm": 1.2110836505889893,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 66260
+    },
+    {
+      "epoch": 435.9868421052632,
+      "grad_norm": 0.7796202898025513,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 66270
+    },
+    {
+      "epoch": 436.05263157894734,
+      "grad_norm": 1.0769124031066895,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 66280
+    },
+    {
+      "epoch": 436.11842105263156,
+      "grad_norm": 0.732404887676239,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 66290
+    },
+    {
+      "epoch": 436.1842105263158,
+      "grad_norm": 0.9934023022651672,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 66300
+    },
+    {
+      "epoch": 436.25,
+      "grad_norm": 1.068806529045105,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 66310
+    },
+    {
+      "epoch": 436.3157894736842,
+      "grad_norm": 0.819196343421936,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 66320
+    },
+    {
+      "epoch": 436.38157894736844,
+      "grad_norm": 1.0213623046875,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 66330
+    },
+    {
+      "epoch": 436.44736842105266,
+      "grad_norm": 1.4873377084732056,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 66340
+    },
+    {
+      "epoch": 436.5131578947368,
+      "grad_norm": 0.9596549272537231,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 66350
+    },
+    {
+      "epoch": 436.57894736842104,
+      "grad_norm": 1.3335829973220825,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 66360
+    },
+    {
+      "epoch": 436.64473684210526,
+      "grad_norm": 1.1521795988082886,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 66370
+    },
+    {
+      "epoch": 436.7105263157895,
+      "grad_norm": 1.2757539749145508,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 66380
+    },
+    {
+      "epoch": 436.7763157894737,
+      "grad_norm": 1.143549919128418,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 66390
+    },
+    {
+      "epoch": 436.8421052631579,
+      "grad_norm": 1.0633903741836548,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 66400
+    },
+    {
+      "epoch": 436.9078947368421,
+      "grad_norm": 0.8937993049621582,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 66410
+    },
+    {
+      "epoch": 436.9736842105263,
+      "grad_norm": 1.0790894031524658,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 66420
+    },
+    {
+      "epoch": 437.0394736842105,
+      "grad_norm": 1.1991922855377197,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 66430
+    },
+    {
+      "epoch": 437.10526315789474,
+      "grad_norm": 1.1077334880828857,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 66440
+    },
+    {
+      "epoch": 437.17105263157896,
+      "grad_norm": 0.9122137427330017,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 66450
+    },
+    {
+      "epoch": 437.2368421052632,
+      "grad_norm": 1.2790870666503906,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 66460
+    },
+    {
+      "epoch": 437.30263157894734,
+      "grad_norm": 1.557098388671875,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 66470
+    },
+    {
+      "epoch": 437.36842105263156,
+      "grad_norm": 1.4872727394104004,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 66480
+    },
+    {
+      "epoch": 437.4342105263158,
+      "grad_norm": 1.1132712364196777,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 66490
+    },
+    {
+      "epoch": 437.5,
+      "grad_norm": 1.3181506395339966,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 66500
+    },
+    {
+      "epoch": 437.5657894736842,
+      "grad_norm": 0.976253867149353,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 66510
+    },
+    {
+      "epoch": 437.63157894736844,
+      "grad_norm": 0.8042446970939636,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 66520
+    },
+    {
+      "epoch": 437.69736842105266,
+      "grad_norm": 0.6995433568954468,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 66530
+    },
+    {
+      "epoch": 437.7631578947368,
+      "grad_norm": 1.2673261165618896,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 66540
+    },
+    {
+      "epoch": 437.82894736842104,
+      "grad_norm": 1.1991665363311768,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 66550
+    },
+    {
+      "epoch": 437.89473684210526,
+      "grad_norm": 1.2135796546936035,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 66560
+    },
+    {
+      "epoch": 437.9605263157895,
+      "grad_norm": 1.4307721853256226,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 66570
+    },
+    {
+      "epoch": 438.0263157894737,
+      "grad_norm": 1.061484694480896,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 66580
+    },
+    {
+      "epoch": 438.0921052631579,
+      "grad_norm": 1.2536671161651611,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 66590
+    },
+    {
+      "epoch": 438.1578947368421,
+      "grad_norm": 1.211685299873352,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 66600
+    },
+    {
+      "epoch": 438.2236842105263,
+      "grad_norm": 0.9517151117324829,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 66610
+    },
+    {
+      "epoch": 438.2894736842105,
+      "grad_norm": 1.3910490274429321,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 66620
+    },
+    {
+      "epoch": 438.35526315789474,
+      "grad_norm": 1.2333219051361084,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 66630
+    },
+    {
+      "epoch": 438.42105263157896,
+      "grad_norm": 1.4747133255004883,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 66640
+    },
+    {
+      "epoch": 438.4868421052632,
+      "grad_norm": 1.4502803087234497,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 66650
+    },
+    {
+      "epoch": 438.55263157894734,
+      "grad_norm": 1.0021837949752808,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 66660
+    },
+    {
+      "epoch": 438.61842105263156,
+      "grad_norm": 1.2110358476638794,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 66670
+    },
+    {
+      "epoch": 438.6842105263158,
+      "grad_norm": 1.2951200008392334,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 66680
+    },
+    {
+      "epoch": 438.75,
+      "grad_norm": 1.4728100299835205,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 66690
+    },
+    {
+      "epoch": 438.8157894736842,
+      "grad_norm": 1.3396544456481934,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 66700
+    },
+    {
+      "epoch": 438.88157894736844,
+      "grad_norm": 0.9344916343688965,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 66710
+    },
+    {
+      "epoch": 438.94736842105266,
+      "grad_norm": 0.8151391744613647,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 66720
+    },
+    {
+      "epoch": 439.0131578947368,
+      "grad_norm": 0.8161630630493164,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 66730
+    },
+    {
+      "epoch": 439.07894736842104,
+      "grad_norm": 1.26533043384552,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 66740
+    },
+    {
+      "epoch": 439.14473684210526,
+      "grad_norm": 1.3989818096160889,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 66750
+    },
+    {
+      "epoch": 439.2105263157895,
+      "grad_norm": 1.301026463508606,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 66760
+    },
+    {
+      "epoch": 439.2763157894737,
+      "grad_norm": 0.9039576649665833,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 66770
+    },
+    {
+      "epoch": 439.3421052631579,
+      "grad_norm": 1.453680396080017,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 66780
+    },
+    {
+      "epoch": 439.4078947368421,
+      "grad_norm": 1.273836612701416,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 66790
+    },
+    {
+      "epoch": 439.4736842105263,
+      "grad_norm": 1.5056235790252686,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 66800
+    },
+    {
+      "epoch": 439.5394736842105,
+      "grad_norm": 1.0324246883392334,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 66810
+    },
+    {
+      "epoch": 439.60526315789474,
+      "grad_norm": 1.2244229316711426,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 66820
+    },
+    {
+      "epoch": 439.67105263157896,
+      "grad_norm": 1.2605669498443604,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 66830
+    },
+    {
+      "epoch": 439.7368421052632,
+      "grad_norm": 1.2557387351989746,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 66840
+    },
+    {
+      "epoch": 439.80263157894734,
+      "grad_norm": 1.641819953918457,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 66850
+    },
+    {
+      "epoch": 439.86842105263156,
+      "grad_norm": 1.0714577436447144,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 66860
+    },
+    {
+      "epoch": 439.9342105263158,
+      "grad_norm": 1.5716043710708618,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 66870
+    },
+    {
+      "epoch": 440.0,
+      "grad_norm": 1.3139293193817139,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 66880
+    },
+    {
+      "epoch": 440.0657894736842,
+      "grad_norm": 1.3056998252868652,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 66890
+    },
+    {
+      "epoch": 440.13157894736844,
+      "grad_norm": 1.1983897686004639,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 66900
+    },
+    {
+      "epoch": 440.19736842105266,
+      "grad_norm": 1.2643394470214844,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 66910
+    },
+    {
+      "epoch": 440.2631578947368,
+      "grad_norm": 1.501151204109192,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 66920
+    },
+    {
+      "epoch": 440.32894736842104,
+      "grad_norm": 1.387885570526123,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 66930
+    },
+    {
+      "epoch": 440.39473684210526,
+      "grad_norm": 1.3784682750701904,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 66940
+    },
+    {
+      "epoch": 440.4605263157895,
+      "grad_norm": 0.9655961394309998,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 66950
+    },
+    {
+      "epoch": 440.5263157894737,
+      "grad_norm": 0.8454356789588928,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 66960
+    },
+    {
+      "epoch": 440.5921052631579,
+      "grad_norm": 1.001245141029358,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 66970
+    },
+    {
+      "epoch": 440.6578947368421,
+      "grad_norm": 1.1278152465820312,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 66980
+    },
+    {
+      "epoch": 440.7236842105263,
+      "grad_norm": 0.9985861778259277,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 66990
+    },
+    {
+      "epoch": 440.7894736842105,
+      "grad_norm": 1.476648211479187,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 67000
+    },
+    {
+      "epoch": 440.85526315789474,
+      "grad_norm": 1.0903433561325073,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 67010
+    },
+    {
+      "epoch": 440.92105263157896,
+      "grad_norm": 1.2222623825073242,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 67020
+    },
+    {
+      "epoch": 440.9868421052632,
+      "grad_norm": 1.298284888267517,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 67030
+    },
+    {
+      "epoch": 441.05263157894734,
+      "grad_norm": 1.4199997186660767,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 67040
+    },
+    {
+      "epoch": 441.11842105263156,
+      "grad_norm": 0.9580684900283813,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 67050
+    },
+    {
+      "epoch": 441.1842105263158,
+      "grad_norm": 1.2852739095687866,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 67060
+    },
+    {
+      "epoch": 441.25,
+      "grad_norm": 0.9127504825592041,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 67070
+    },
+    {
+      "epoch": 441.3157894736842,
+      "grad_norm": 1.0054773092269897,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 67080
+    },
+    {
+      "epoch": 441.38157894736844,
+      "grad_norm": 1.0165013074874878,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 67090
+    },
+    {
+      "epoch": 441.44736842105266,
+      "grad_norm": 1.144066572189331,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 67100
+    },
+    {
+      "epoch": 441.5131578947368,
+      "grad_norm": 0.9844335913658142,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 67110
+    },
+    {
+      "epoch": 441.57894736842104,
+      "grad_norm": 1.0714805126190186,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 67120
+    },
+    {
+      "epoch": 441.64473684210526,
+      "grad_norm": 1.3069833517074585,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 67130
+    },
+    {
+      "epoch": 441.7105263157895,
+      "grad_norm": 1.189154028892517,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 67140
+    },
+    {
+      "epoch": 441.7763157894737,
+      "grad_norm": 1.2548660039901733,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 67150
+    },
+    {
+      "epoch": 441.8421052631579,
+      "grad_norm": 1.6267309188842773,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 67160
+    },
+    {
+      "epoch": 441.9078947368421,
+      "grad_norm": 1.1438740491867065,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 67170
+    },
+    {
+      "epoch": 441.9736842105263,
+      "grad_norm": 1.1230552196502686,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 67180
+    },
+    {
+      "epoch": 442.0394736842105,
+      "grad_norm": 1.0231391191482544,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 67190
+    },
+    {
+      "epoch": 442.10526315789474,
+      "grad_norm": 1.1591103076934814,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 67200
+    },
+    {
+      "epoch": 442.17105263157896,
+      "grad_norm": 1.3125256299972534,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 67210
+    },
+    {
+      "epoch": 442.2368421052632,
+      "grad_norm": 0.9967691898345947,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 67220
+    },
+    {
+      "epoch": 442.30263157894734,
+      "grad_norm": 1.2353827953338623,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 67230
+    },
+    {
+      "epoch": 442.36842105263156,
+      "grad_norm": 0.9676706194877625,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 67240
+    },
+    {
+      "epoch": 442.4342105263158,
+      "grad_norm": 0.89171302318573,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 67250
+    },
+    {
+      "epoch": 442.5,
+      "grad_norm": 0.8268073797225952,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 67260
+    },
+    {
+      "epoch": 442.5657894736842,
+      "grad_norm": 0.8292796015739441,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 67270
+    },
+    {
+      "epoch": 442.63157894736844,
+      "grad_norm": 1.1891852617263794,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 67280
+    },
+    {
+      "epoch": 442.69736842105266,
+      "grad_norm": 1.0386085510253906,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 67290
+    },
+    {
+      "epoch": 442.7631578947368,
+      "grad_norm": 1.4309957027435303,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 67300
+    },
+    {
+      "epoch": 442.82894736842104,
+      "grad_norm": 1.046149730682373,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 67310
+    },
+    {
+      "epoch": 442.89473684210526,
+      "grad_norm": 1.2012232542037964,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 67320
+    },
+    {
+      "epoch": 442.9605263157895,
+      "grad_norm": 0.963539183139801,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 67330
+    },
+    {
+      "epoch": 443.0263157894737,
+      "grad_norm": 0.8693314790725708,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 67340
+    },
+    {
+      "epoch": 443.0921052631579,
+      "grad_norm": 1.503976821899414,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 67350
+    },
+    {
+      "epoch": 443.1578947368421,
+      "grad_norm": 0.8034859895706177,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 67360
+    },
+    {
+      "epoch": 443.2236842105263,
+      "grad_norm": 1.4391164779663086,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 67370
+    },
+    {
+      "epoch": 443.2894736842105,
+      "grad_norm": 1.2415558099746704,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 67380
+    },
+    {
+      "epoch": 443.35526315789474,
+      "grad_norm": 1.0258815288543701,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 67390
+    },
+    {
+      "epoch": 443.42105263157896,
+      "grad_norm": 0.7889366149902344,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 67400
+    },
+    {
+      "epoch": 443.4868421052632,
+      "grad_norm": 0.8310263752937317,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 67410
+    },
+    {
+      "epoch": 443.55263157894734,
+      "grad_norm": 1.102107286453247,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 67420
+    },
+    {
+      "epoch": 443.61842105263156,
+      "grad_norm": 0.891818106174469,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 67430
+    },
+    {
+      "epoch": 443.6842105263158,
+      "grad_norm": 1.2549525499343872,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 67440
+    },
+    {
+      "epoch": 443.75,
+      "grad_norm": 1.475273847579956,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 67450
+    },
+    {
+      "epoch": 443.8157894736842,
+      "grad_norm": 1.5199368000030518,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 67460
+    },
+    {
+      "epoch": 443.88157894736844,
+      "grad_norm": 1.0391242504119873,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 67470
+    },
+    {
+      "epoch": 443.94736842105266,
+      "grad_norm": 1.2882357835769653,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 67480
+    },
+    {
+      "epoch": 444.0131578947368,
+      "grad_norm": 1.4327231645584106,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 67490
+    },
+    {
+      "epoch": 444.07894736842104,
+      "grad_norm": 1.25886869430542,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 67500
+    },
+    {
+      "epoch": 444.14473684210526,
+      "grad_norm": 0.9697027802467346,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 67510
+    },
+    {
+      "epoch": 444.2105263157895,
+      "grad_norm": 1.3872424364089966,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 67520
+    },
+    {
+      "epoch": 444.2763157894737,
+      "grad_norm": 1.1043860912322998,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 67530
+    },
+    {
+      "epoch": 444.3421052631579,
+      "grad_norm": 1.202193021774292,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 67540
+    },
+    {
+      "epoch": 444.4078947368421,
+      "grad_norm": 1.0167324542999268,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 67550
+    },
+    {
+      "epoch": 444.4736842105263,
+      "grad_norm": 0.9748640656471252,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 67560
+    },
+    {
+      "epoch": 444.5394736842105,
+      "grad_norm": 1.1496931314468384,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 67570
+    },
+    {
+      "epoch": 444.60526315789474,
+      "grad_norm": 1.193697452545166,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 67580
+    },
+    {
+      "epoch": 444.67105263157896,
+      "grad_norm": 1.2025853395462036,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 67590
+    },
+    {
+      "epoch": 444.7368421052632,
+      "grad_norm": 0.9726876020431519,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 67600
+    },
+    {
+      "epoch": 444.80263157894734,
+      "grad_norm": 0.8311697840690613,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 67610
+    },
+    {
+      "epoch": 444.86842105263156,
+      "grad_norm": 0.8062876462936401,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 67620
+    },
+    {
+      "epoch": 444.9342105263158,
+      "grad_norm": 0.8064841628074646,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 67630
+    },
+    {
+      "epoch": 445.0,
+      "grad_norm": 0.9301232695579529,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 67640
+    },
+    {
+      "epoch": 445.0657894736842,
+      "grad_norm": 1.1378428936004639,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 67650
+    },
+    {
+      "epoch": 445.13157894736844,
+      "grad_norm": 1.1521905660629272,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 67660
+    },
+    {
+      "epoch": 445.19736842105266,
+      "grad_norm": 1.2799955606460571,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 67670
+    },
+    {
+      "epoch": 445.2631578947368,
+      "grad_norm": 1.008537769317627,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 67680
+    },
+    {
+      "epoch": 445.32894736842104,
+      "grad_norm": 1.2457149028778076,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 67690
+    },
+    {
+      "epoch": 445.39473684210526,
+      "grad_norm": 1.1175915002822876,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 67700
+    },
+    {
+      "epoch": 445.4605263157895,
+      "grad_norm": 1.0198794603347778,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 67710
+    },
+    {
+      "epoch": 445.5263157894737,
+      "grad_norm": 1.1624159812927246,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 67720
+    },
+    {
+      "epoch": 445.5921052631579,
+      "grad_norm": 1.1959896087646484,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 67730
+    },
+    {
+      "epoch": 445.6578947368421,
+      "grad_norm": 1.2008717060089111,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 67740
+    },
+    {
+      "epoch": 445.7236842105263,
+      "grad_norm": 1.321126937866211,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 67750
+    },
+    {
+      "epoch": 445.7894736842105,
+      "grad_norm": 1.1660131216049194,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 67760
+    },
+    {
+      "epoch": 445.85526315789474,
+      "grad_norm": 1.2999013662338257,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 67770
+    },
+    {
+      "epoch": 445.92105263157896,
+      "grad_norm": 0.8440237045288086,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 67780
+    },
+    {
+      "epoch": 445.9868421052632,
+      "grad_norm": 1.0197548866271973,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 67790
+    },
+    {
+      "epoch": 446.05263157894734,
+      "grad_norm": 1.0732282400131226,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 67800
+    },
+    {
+      "epoch": 446.11842105263156,
+      "grad_norm": 1.2190271615982056,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 67810
+    },
+    {
+      "epoch": 446.1842105263158,
+      "grad_norm": 1.098029375076294,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 67820
+    },
+    {
+      "epoch": 446.25,
+      "grad_norm": 1.064057469367981,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 67830
+    },
+    {
+      "epoch": 446.3157894736842,
+      "grad_norm": 1.0440378189086914,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 67840
+    },
+    {
+      "epoch": 446.38157894736844,
+      "grad_norm": 1.2264587879180908,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 67850
+    },
+    {
+      "epoch": 446.44736842105266,
+      "grad_norm": 0.7877911925315857,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 67860
+    },
+    {
+      "epoch": 446.5131578947368,
+      "grad_norm": 1.3286290168762207,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 67870
+    },
+    {
+      "epoch": 446.57894736842104,
+      "grad_norm": 0.9419989585876465,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 67880
+    },
+    {
+      "epoch": 446.64473684210526,
+      "grad_norm": 1.018985629081726,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 67890
+    },
+    {
+      "epoch": 446.7105263157895,
+      "grad_norm": 0.9366230964660645,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 67900
+    },
+    {
+      "epoch": 446.7763157894737,
+      "grad_norm": 0.7390005588531494,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 67910
+    },
+    {
+      "epoch": 446.8421052631579,
+      "grad_norm": 0.7992033362388611,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 67920
+    },
+    {
+      "epoch": 446.9078947368421,
+      "grad_norm": 1.0633325576782227,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 67930
+    },
+    {
+      "epoch": 446.9736842105263,
+      "grad_norm": 1.1128759384155273,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 67940
+    },
+    {
+      "epoch": 447.0394736842105,
+      "grad_norm": 1.405713677406311,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 67950
+    },
+    {
+      "epoch": 447.10526315789474,
+      "grad_norm": 1.0440222024917603,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 67960
+    },
+    {
+      "epoch": 447.17105263157896,
+      "grad_norm": 1.1741487979888916,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 67970
+    },
+    {
+      "epoch": 447.2368421052632,
+      "grad_norm": 1.0820094347000122,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 67980
+    },
+    {
+      "epoch": 447.30263157894734,
+      "grad_norm": 1.0338839292526245,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 67990
+    },
+    {
+      "epoch": 447.36842105263156,
+      "grad_norm": 1.3230096101760864,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 68000
+    },
+    {
+      "epoch": 447.4342105263158,
+      "grad_norm": 0.8464519381523132,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 68010
+    },
+    {
+      "epoch": 447.5,
+      "grad_norm": 1.3465580940246582,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 68020
+    },
+    {
+      "epoch": 447.5657894736842,
+      "grad_norm": 1.2503917217254639,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 68030
+    },
+    {
+      "epoch": 447.63157894736844,
+      "grad_norm": 1.2269294261932373,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 68040
+    },
+    {
+      "epoch": 447.69736842105266,
+      "grad_norm": 0.9101136922836304,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 68050
+    },
+    {
+      "epoch": 447.7631578947368,
+      "grad_norm": 1.1109895706176758,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 68060
+    },
+    {
+      "epoch": 447.82894736842104,
+      "grad_norm": 1.2800877094268799,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 68070
+    },
+    {
+      "epoch": 447.89473684210526,
+      "grad_norm": 1.3578050136566162,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 68080
+    },
+    {
+      "epoch": 447.9605263157895,
+      "grad_norm": 1.1380995512008667,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 68090
+    },
+    {
+      "epoch": 448.0263157894737,
+      "grad_norm": 1.510892629623413,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 68100
+    },
+    {
+      "epoch": 448.0921052631579,
+      "grad_norm": 1.1936112642288208,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 68110
+    },
+    {
+      "epoch": 448.1578947368421,
+      "grad_norm": 1.3569926023483276,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 68120
+    },
+    {
+      "epoch": 448.2236842105263,
+      "grad_norm": 1.0964100360870361,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 68130
+    },
+    {
+      "epoch": 448.2894736842105,
+      "grad_norm": 0.7216523289680481,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 68140
+    },
+    {
+      "epoch": 448.35526315789474,
+      "grad_norm": 1.2212159633636475,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 68150
+    },
+    {
+      "epoch": 448.42105263157896,
+      "grad_norm": 1.0280985832214355,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 68160
+    },
+    {
+      "epoch": 448.4868421052632,
+      "grad_norm": 1.3027311563491821,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 68170
+    },
+    {
+      "epoch": 448.55263157894734,
+      "grad_norm": 1.462243914604187,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 68180
+    },
+    {
+      "epoch": 448.61842105263156,
+      "grad_norm": 1.231683373451233,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 68190
+    },
+    {
+      "epoch": 448.6842105263158,
+      "grad_norm": 1.2947146892547607,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 68200
+    },
+    {
+      "epoch": 448.75,
+      "grad_norm": 1.0188627243041992,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 68210
+    },
+    {
+      "epoch": 448.8157894736842,
+      "grad_norm": 1.1780815124511719,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 68220
+    },
+    {
+      "epoch": 448.88157894736844,
+      "grad_norm": 1.1166077852249146,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 68230
+    },
+    {
+      "epoch": 448.94736842105266,
+      "grad_norm": 0.9867680668830872,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 68240
+    },
+    {
+      "epoch": 449.0131578947368,
+      "grad_norm": 1.014134168624878,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 68250
+    },
+    {
+      "epoch": 449.07894736842104,
+      "grad_norm": 0.8704817891120911,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 68260
+    },
+    {
+      "epoch": 449.14473684210526,
+      "grad_norm": 1.1056900024414062,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 68270
+    },
+    {
+      "epoch": 449.2105263157895,
+      "grad_norm": 1.131632924079895,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 68280
+    },
+    {
+      "epoch": 449.2763157894737,
+      "grad_norm": 1.3262722492218018,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 68290
+    },
+    {
+      "epoch": 449.3421052631579,
+      "grad_norm": 1.328003168106079,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 68300
+    },
+    {
+      "epoch": 449.4078947368421,
+      "grad_norm": 1.1644665002822876,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 68310
+    },
+    {
+      "epoch": 449.4736842105263,
+      "grad_norm": 1.419346809387207,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 68320
+    },
+    {
+      "epoch": 449.5394736842105,
+      "grad_norm": 0.9281734228134155,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 68330
+    },
+    {
+      "epoch": 449.60526315789474,
+      "grad_norm": 1.394044280052185,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 68340
+    },
+    {
+      "epoch": 449.67105263157896,
+      "grad_norm": 1.435802936553955,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 68350
+    },
+    {
+      "epoch": 449.7368421052632,
+      "grad_norm": 1.2658971548080444,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 68360
+    },
+    {
+      "epoch": 449.80263157894734,
+      "grad_norm": 1.4946037530899048,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 68370
+    },
+    {
+      "epoch": 449.86842105263156,
+      "grad_norm": 0.7400012016296387,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 68380
+    },
+    {
+      "epoch": 449.9342105263158,
+      "grad_norm": 1.1774287223815918,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 68390
+    },
+    {
+      "epoch": 450.0,
+      "grad_norm": 1.2723482847213745,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 68400
+    },
+    {
+      "epoch": 450.0657894736842,
+      "grad_norm": 1.0337828397750854,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 68410
+    },
+    {
+      "epoch": 450.13157894736844,
+      "grad_norm": 0.836054265499115,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 68420
+    },
+    {
+      "epoch": 450.19736842105266,
+      "grad_norm": 1.1785285472869873,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 68430
+    },
+    {
+      "epoch": 450.2631578947368,
+      "grad_norm": 1.1145614385604858,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 68440
+    },
+    {
+      "epoch": 450.32894736842104,
+      "grad_norm": 1.3522475957870483,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 68450
+    },
+    {
+      "epoch": 450.39473684210526,
+      "grad_norm": 1.1406996250152588,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 68460
+    },
+    {
+      "epoch": 450.4605263157895,
+      "grad_norm": 1.2381994724273682,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 68470
+    },
+    {
+      "epoch": 450.5263157894737,
+      "grad_norm": 1.4865307807922363,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 68480
+    },
+    {
+      "epoch": 450.5921052631579,
+      "grad_norm": 1.2501206398010254,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 68490
+    },
+    {
+      "epoch": 450.6578947368421,
+      "grad_norm": 1.3360950946807861,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 68500
+    },
+    {
+      "epoch": 450.7236842105263,
+      "grad_norm": 1.5637435913085938,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 68510
+    },
+    {
+      "epoch": 450.7894736842105,
+      "grad_norm": 1.383474349975586,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 68520
+    },
+    {
+      "epoch": 450.85526315789474,
+      "grad_norm": 1.3459497690200806,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 68530
+    },
+    {
+      "epoch": 450.92105263157896,
+      "grad_norm": 0.9514352679252625,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 68540
+    },
+    {
+      "epoch": 450.9868421052632,
+      "grad_norm": 1.1830934286117554,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 68550
+    },
+    {
+      "epoch": 451.05263157894734,
+      "grad_norm": 1.131381630897522,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 68560
+    },
+    {
+      "epoch": 451.11842105263156,
+      "grad_norm": 0.9725364446640015,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 68570
+    },
+    {
+      "epoch": 451.1842105263158,
+      "grad_norm": 1.4071309566497803,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 68580
+    },
+    {
+      "epoch": 451.25,
+      "grad_norm": 1.218841552734375,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 68590
+    },
+    {
+      "epoch": 451.3157894736842,
+      "grad_norm": 1.2471004724502563,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 68600
+    },
+    {
+      "epoch": 451.38157894736844,
+      "grad_norm": 1.2200270891189575,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 68610
+    },
+    {
+      "epoch": 451.44736842105266,
+      "grad_norm": 0.7983037233352661,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 68620
+    },
+    {
+      "epoch": 451.5131578947368,
+      "grad_norm": 1.0153566598892212,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 68630
+    },
+    {
+      "epoch": 451.57894736842104,
+      "grad_norm": 0.8792093396186829,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 68640
+    },
+    {
+      "epoch": 451.64473684210526,
+      "grad_norm": 1.1908783912658691,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 68650
+    },
+    {
+      "epoch": 451.7105263157895,
+      "grad_norm": 1.2272428274154663,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 68660
+    },
+    {
+      "epoch": 451.7763157894737,
+      "grad_norm": 0.9203829765319824,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 68670
+    },
+    {
+      "epoch": 451.8421052631579,
+      "grad_norm": 1.4180694818496704,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 68680
+    },
+    {
+      "epoch": 451.9078947368421,
+      "grad_norm": 0.8548322319984436,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 68690
+    },
+    {
+      "epoch": 451.9736842105263,
+      "grad_norm": 1.259071946144104,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 68700
+    },
+    {
+      "epoch": 452.0394736842105,
+      "grad_norm": 1.4840049743652344,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 68710
+    },
+    {
+      "epoch": 452.10526315789474,
+      "grad_norm": 0.9539170861244202,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 68720
+    },
+    {
+      "epoch": 452.17105263157896,
+      "grad_norm": 1.287971019744873,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 68730
+    },
+    {
+      "epoch": 452.2368421052632,
+      "grad_norm": 1.5346850156784058,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 68740
+    },
+    {
+      "epoch": 452.30263157894734,
+      "grad_norm": 1.4423147439956665,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 68750
+    },
+    {
+      "epoch": 452.36842105263156,
+      "grad_norm": 1.0236411094665527,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 68760
+    },
+    {
+      "epoch": 452.4342105263158,
+      "grad_norm": 0.9541664719581604,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 68770
+    },
+    {
+      "epoch": 452.5,
+      "grad_norm": 0.7958992719650269,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 68780
+    },
+    {
+      "epoch": 452.5657894736842,
+      "grad_norm": 1.2157082557678223,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 68790
+    },
+    {
+      "epoch": 452.63157894736844,
+      "grad_norm": 0.8588937520980835,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 68800
+    },
+    {
+      "epoch": 452.69736842105266,
+      "grad_norm": 1.442825436592102,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 68810
+    },
+    {
+      "epoch": 452.7631578947368,
+      "grad_norm": 1.1870979070663452,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 68820
+    },
+    {
+      "epoch": 452.82894736842104,
+      "grad_norm": 1.28781259059906,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 68830
+    },
+    {
+      "epoch": 452.89473684210526,
+      "grad_norm": 0.9649443030357361,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 68840
+    },
+    {
+      "epoch": 452.9605263157895,
+      "grad_norm": 0.9508198499679565,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 68850
+    },
+    {
+      "epoch": 453.0263157894737,
+      "grad_norm": 1.5486016273498535,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 68860
+    },
+    {
+      "epoch": 453.0921052631579,
+      "grad_norm": 1.217083215713501,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 68870
+    },
+    {
+      "epoch": 453.1578947368421,
+      "grad_norm": 1.2917876243591309,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 68880
+    },
+    {
+      "epoch": 453.2236842105263,
+      "grad_norm": 1.0706918239593506,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 68890
+    },
+    {
+      "epoch": 453.2894736842105,
+      "grad_norm": 1.453224778175354,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 68900
+    },
+    {
+      "epoch": 453.35526315789474,
+      "grad_norm": 1.209774374961853,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 68910
+    },
+    {
+      "epoch": 453.42105263157896,
+      "grad_norm": 0.9678460955619812,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 68920
+    },
+    {
+      "epoch": 453.4868421052632,
+      "grad_norm": 1.2222375869750977,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 68930
+    },
+    {
+      "epoch": 453.55263157894734,
+      "grad_norm": 0.9485657215118408,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 68940
+    },
+    {
+      "epoch": 453.61842105263156,
+      "grad_norm": 1.2582299709320068,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 68950
+    },
+    {
+      "epoch": 453.6842105263158,
+      "grad_norm": 1.2892476320266724,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 68960
+    },
+    {
+      "epoch": 453.75,
+      "grad_norm": 1.3842599391937256,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 68970
+    },
+    {
+      "epoch": 453.8157894736842,
+      "grad_norm": 1.095976710319519,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 68980
+    },
+    {
+      "epoch": 453.88157894736844,
+      "grad_norm": 1.2450053691864014,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 68990
+    },
+    {
+      "epoch": 453.94736842105266,
+      "grad_norm": 0.8790797591209412,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 69000
+    },
+    {
+      "epoch": 454.0131578947368,
+      "grad_norm": 1.2739403247833252,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 69010
+    },
+    {
+      "epoch": 454.07894736842104,
+      "grad_norm": 1.2191137075424194,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 69020
+    },
+    {
+      "epoch": 454.14473684210526,
+      "grad_norm": 1.4736047983169556,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 69030
+    },
+    {
+      "epoch": 454.2105263157895,
+      "grad_norm": 0.9068169593811035,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 69040
+    },
+    {
+      "epoch": 454.2763157894737,
+      "grad_norm": 1.5603197813034058,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 69050
+    },
+    {
+      "epoch": 454.3421052631579,
+      "grad_norm": 1.4251413345336914,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 69060
+    },
+    {
+      "epoch": 454.4078947368421,
+      "grad_norm": 1.295698642730713,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 69070
+    },
+    {
+      "epoch": 454.4736842105263,
+      "grad_norm": 1.536226511001587,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 69080
+    },
+    {
+      "epoch": 454.5394736842105,
+      "grad_norm": 1.6740570068359375,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 69090
+    },
+    {
+      "epoch": 454.60526315789474,
+      "grad_norm": 1.091256856918335,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 69100
+    },
+    {
+      "epoch": 454.67105263157896,
+      "grad_norm": 0.9562748670578003,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 69110
+    },
+    {
+      "epoch": 454.7368421052632,
+      "grad_norm": 1.3585184812545776,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 69120
+    },
+    {
+      "epoch": 454.80263157894734,
+      "grad_norm": 1.2994377613067627,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 69130
+    },
+    {
+      "epoch": 454.86842105263156,
+      "grad_norm": 1.152564525604248,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 69140
+    },
+    {
+      "epoch": 454.9342105263158,
+      "grad_norm": 1.6018366813659668,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 69150
+    },
+    {
+      "epoch": 455.0,
+      "grad_norm": 1.2731026411056519,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 69160
+    },
+    {
+      "epoch": 455.0657894736842,
+      "grad_norm": 1.0202285051345825,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 69170
+    },
+    {
+      "epoch": 455.13157894736844,
+      "grad_norm": 1.0699840784072876,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 69180
+    },
+    {
+      "epoch": 455.19736842105266,
+      "grad_norm": 1.36112380027771,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 69190
+    },
+    {
+      "epoch": 455.2631578947368,
+      "grad_norm": 1.1160879135131836,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 69200
+    },
+    {
+      "epoch": 455.32894736842104,
+      "grad_norm": 1.1175243854522705,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 69210
+    },
+    {
+      "epoch": 455.39473684210526,
+      "grad_norm": 1.1258723735809326,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 69220
+    },
+    {
+      "epoch": 455.4605263157895,
+      "grad_norm": 1.0541523694992065,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 69230
+    },
+    {
+      "epoch": 455.5263157894737,
+      "grad_norm": 0.8870919942855835,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 69240
+    },
+    {
+      "epoch": 455.5921052631579,
+      "grad_norm": 1.1106444597244263,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 69250
+    },
+    {
+      "epoch": 455.6578947368421,
+      "grad_norm": 1.2560456991195679,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 69260
+    },
+    {
+      "epoch": 455.7236842105263,
+      "grad_norm": 0.831874668598175,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 69270
+    },
+    {
+      "epoch": 455.7894736842105,
+      "grad_norm": 1.3205691576004028,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 69280
+    },
+    {
+      "epoch": 455.85526315789474,
+      "grad_norm": 1.1079440116882324,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 69290
+    },
+    {
+      "epoch": 455.92105263157896,
+      "grad_norm": 1.0580300092697144,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 69300
+    },
+    {
+      "epoch": 455.9868421052632,
+      "grad_norm": 1.5250871181488037,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 69310
+    },
+    {
+      "epoch": 456.05263157894734,
+      "grad_norm": 1.4987032413482666,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 69320
+    },
+    {
+      "epoch": 456.11842105263156,
+      "grad_norm": 1.686471939086914,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 69330
+    },
+    {
+      "epoch": 456.1842105263158,
+      "grad_norm": 1.4890836477279663,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 69340
+    },
+    {
+      "epoch": 456.25,
+      "grad_norm": 1.3004556894302368,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 69350
+    },
+    {
+      "epoch": 456.3157894736842,
+      "grad_norm": 1.3867583274841309,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 69360
+    },
+    {
+      "epoch": 456.38157894736844,
+      "grad_norm": 1.4305163621902466,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 69370
+    },
+    {
+      "epoch": 456.44736842105266,
+      "grad_norm": 0.9518430233001709,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 69380
+    },
+    {
+      "epoch": 456.5131578947368,
+      "grad_norm": 1.2672168016433716,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 69390
+    },
+    {
+      "epoch": 456.57894736842104,
+      "grad_norm": 1.313014030456543,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 69400
+    },
+    {
+      "epoch": 456.64473684210526,
+      "grad_norm": 0.7348408699035645,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 69410
+    },
+    {
+      "epoch": 456.7105263157895,
+      "grad_norm": 1.0799474716186523,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 69420
+    },
+    {
+      "epoch": 456.7763157894737,
+      "grad_norm": 0.9113180637359619,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 69430
+    },
+    {
+      "epoch": 456.8421052631579,
+      "grad_norm": 0.8408834934234619,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 69440
+    },
+    {
+      "epoch": 456.9078947368421,
+      "grad_norm": 0.9288105964660645,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 69450
+    },
+    {
+      "epoch": 456.9736842105263,
+      "grad_norm": 1.152346134185791,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 69460
+    },
+    {
+      "epoch": 457.0394736842105,
+      "grad_norm": 1.110255241394043,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 69470
+    },
+    {
+      "epoch": 457.10526315789474,
+      "grad_norm": 0.9234119057655334,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 69480
+    },
+    {
+      "epoch": 457.17105263157896,
+      "grad_norm": 1.131162405014038,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 69490
+    },
+    {
+      "epoch": 457.2368421052632,
+      "grad_norm": 1.2793774604797363,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 69500
+    },
+    {
+      "epoch": 457.30263157894734,
+      "grad_norm": 0.9325578808784485,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 69510
+    },
+    {
+      "epoch": 457.36842105263156,
+      "grad_norm": 0.8188279867172241,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 69520
+    },
+    {
+      "epoch": 457.4342105263158,
+      "grad_norm": 1.163884162902832,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 69530
+    },
+    {
+      "epoch": 457.5,
+      "grad_norm": 0.8260605335235596,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 69540
+    },
+    {
+      "epoch": 457.5657894736842,
+      "grad_norm": 0.9177809357643127,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 69550
+    },
+    {
+      "epoch": 457.63157894736844,
+      "grad_norm": 1.1885771751403809,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 69560
+    },
+    {
+      "epoch": 457.69736842105266,
+      "grad_norm": 0.9311556816101074,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 69570
+    },
+    {
+      "epoch": 457.7631578947368,
+      "grad_norm": 1.2804585695266724,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 69580
+    },
+    {
+      "epoch": 457.82894736842104,
+      "grad_norm": 1.344652771949768,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 69590
+    },
+    {
+      "epoch": 457.89473684210526,
+      "grad_norm": 1.4862884283065796,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 69600
+    },
+    {
+      "epoch": 457.9605263157895,
+      "grad_norm": 1.1936073303222656,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 69610
+    },
+    {
+      "epoch": 458.0263157894737,
+      "grad_norm": 1.193771243095398,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 69620
+    },
+    {
+      "epoch": 458.0921052631579,
+      "grad_norm": 1.1889840364456177,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 69630
+    },
+    {
+      "epoch": 458.1578947368421,
+      "grad_norm": 1.3708604574203491,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 69640
+    },
+    {
+      "epoch": 458.2236842105263,
+      "grad_norm": 0.5873807072639465,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 69650
+    },
+    {
+      "epoch": 458.2894736842105,
+      "grad_norm": 1.2545485496520996,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 69660
+    },
+    {
+      "epoch": 458.35526315789474,
+      "grad_norm": 1.2837920188903809,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 69670
+    },
+    {
+      "epoch": 458.42105263157896,
+      "grad_norm": 1.2512160539627075,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 69680
+    },
+    {
+      "epoch": 458.4868421052632,
+      "grad_norm": 0.9939183592796326,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 69690
+    },
+    {
+      "epoch": 458.55263157894734,
+      "grad_norm": 0.8284690380096436,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 69700
+    },
+    {
+      "epoch": 458.61842105263156,
+      "grad_norm": 1.1884512901306152,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 69710
+    },
+    {
+      "epoch": 458.6842105263158,
+      "grad_norm": 1.069785475730896,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 69720
+    },
+    {
+      "epoch": 458.75,
+      "grad_norm": 1.294502854347229,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 69730
+    },
+    {
+      "epoch": 458.8157894736842,
+      "grad_norm": 1.4734914302825928,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 69740
+    },
+    {
+      "epoch": 458.88157894736844,
+      "grad_norm": 0.9889464974403381,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 69750
+    },
+    {
+      "epoch": 458.94736842105266,
+      "grad_norm": 1.268761396408081,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 69760
+    },
+    {
+      "epoch": 459.0131578947368,
+      "grad_norm": 1.004045009613037,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 69770
+    },
+    {
+      "epoch": 459.07894736842104,
+      "grad_norm": 0.9832544326782227,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 69780
+    },
+    {
+      "epoch": 459.14473684210526,
+      "grad_norm": 1.2626286745071411,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 69790
+    },
+    {
+      "epoch": 459.2105263157895,
+      "grad_norm": 1.5737202167510986,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 69800
+    },
+    {
+      "epoch": 459.2763157894737,
+      "grad_norm": 1.1485472917556763,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 69810
+    },
+    {
+      "epoch": 459.3421052631579,
+      "grad_norm": 0.9378661513328552,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 69820
+    },
+    {
+      "epoch": 459.4078947368421,
+      "grad_norm": 0.9889057874679565,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 69830
+    },
+    {
+      "epoch": 459.4736842105263,
+      "grad_norm": 0.7639946341514587,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 69840
+    },
+    {
+      "epoch": 459.5394736842105,
+      "grad_norm": 1.084297776222229,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 69850
+    },
+    {
+      "epoch": 459.60526315789474,
+      "grad_norm": 1.2830708026885986,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 69860
+    },
+    {
+      "epoch": 459.67105263157896,
+      "grad_norm": 0.9912347793579102,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 69870
+    },
+    {
+      "epoch": 459.7368421052632,
+      "grad_norm": 1.7279759645462036,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 69880
+    },
+    {
+      "epoch": 459.80263157894734,
+      "grad_norm": 1.2045061588287354,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 69890
+    },
+    {
+      "epoch": 459.86842105263156,
+      "grad_norm": 1.4160776138305664,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 69900
+    },
+    {
+      "epoch": 459.9342105263158,
+      "grad_norm": 0.9676499366760254,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 69910
+    },
+    {
+      "epoch": 460.0,
+      "grad_norm": 0.8560894131660461,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 69920
+    },
+    {
+      "epoch": 460.0657894736842,
+      "grad_norm": 1.248684287071228,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 69930
+    },
+    {
+      "epoch": 460.13157894736844,
+      "grad_norm": 0.8064592480659485,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 69940
+    },
+    {
+      "epoch": 460.19736842105266,
+      "grad_norm": 1.0934714078903198,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 69950
+    },
+    {
+      "epoch": 460.2631578947368,
+      "grad_norm": 1.0918971300125122,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 69960
+    },
+    {
+      "epoch": 460.32894736842104,
+      "grad_norm": 1.164820909500122,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 69970
+    },
+    {
+      "epoch": 460.39473684210526,
+      "grad_norm": 0.8863160610198975,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 69980
+    },
+    {
+      "epoch": 460.4605263157895,
+      "grad_norm": 1.3123438358306885,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 69990
+    },
+    {
+      "epoch": 460.5263157894737,
+      "grad_norm": 1.3433982133865356,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 70000
+    },
+    {
+      "epoch": 460.5921052631579,
+      "grad_norm": 1.089854121208191,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 70010
+    },
+    {
+      "epoch": 460.6578947368421,
+      "grad_norm": 1.2574924230575562,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 70020
+    },
+    {
+      "epoch": 460.7236842105263,
+      "grad_norm": 1.087710976600647,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 70030
+    },
+    {
+      "epoch": 460.7894736842105,
+      "grad_norm": 1.1236116886138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 70040
+    },
+    {
+      "epoch": 460.85526315789474,
+      "grad_norm": 1.117353081703186,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 70050
+    },
+    {
+      "epoch": 460.92105263157896,
+      "grad_norm": 1.1128649711608887,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 70060
+    },
+    {
+      "epoch": 460.9868421052632,
+      "grad_norm": 1.0730403661727905,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 70070
+    },
+    {
+      "epoch": 461.05263157894734,
+      "grad_norm": 1.179450273513794,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 70080
+    },
+    {
+      "epoch": 461.11842105263156,
+      "grad_norm": 1.2371575832366943,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 70090
+    },
+    {
+      "epoch": 461.1842105263158,
+      "grad_norm": 1.42441725730896,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 70100
+    },
+    {
+      "epoch": 461.25,
+      "grad_norm": 1.179229497909546,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 70110
+    },
+    {
+      "epoch": 461.3157894736842,
+      "grad_norm": 0.8924153447151184,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 70120
+    },
+    {
+      "epoch": 461.38157894736844,
+      "grad_norm": 1.3798414468765259,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 70130
+    },
+    {
+      "epoch": 461.44736842105266,
+      "grad_norm": 1.3410454988479614,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 70140
+    },
+    {
+      "epoch": 461.5131578947368,
+      "grad_norm": 0.9191851019859314,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 70150
+    },
+    {
+      "epoch": 461.57894736842104,
+      "grad_norm": 1.0787506103515625,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 70160
+    },
+    {
+      "epoch": 461.64473684210526,
+      "grad_norm": 1.022510290145874,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 70170
+    },
+    {
+      "epoch": 461.7105263157895,
+      "grad_norm": 1.0433993339538574,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 70180
+    },
+    {
+      "epoch": 461.7763157894737,
+      "grad_norm": 1.2744972705841064,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 70190
+    },
+    {
+      "epoch": 461.8421052631579,
+      "grad_norm": 1.1916465759277344,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 70200
+    },
+    {
+      "epoch": 461.9078947368421,
+      "grad_norm": 1.2578779458999634,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 70210
+    },
+    {
+      "epoch": 461.9736842105263,
+      "grad_norm": 1.0983797311782837,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 70220
+    },
+    {
+      "epoch": 462.0394736842105,
+      "grad_norm": 1.3907415866851807,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 70230
+    },
+    {
+      "epoch": 462.10526315789474,
+      "grad_norm": 1.0465130805969238,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 70240
+    },
+    {
+      "epoch": 462.17105263157896,
+      "grad_norm": 1.1414015293121338,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 70250
+    },
+    {
+      "epoch": 462.2368421052632,
+      "grad_norm": 1.4109493494033813,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 70260
+    },
+    {
+      "epoch": 462.30263157894734,
+      "grad_norm": 1.1486443281173706,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 70270
+    },
+    {
+      "epoch": 462.36842105263156,
+      "grad_norm": 1.240187168121338,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 70280
+    },
+    {
+      "epoch": 462.4342105263158,
+      "grad_norm": 1.2985851764678955,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 70290
+    },
+    {
+      "epoch": 462.5,
+      "grad_norm": 1.3922537565231323,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 70300
+    },
+    {
+      "epoch": 462.5657894736842,
+      "grad_norm": 1.1055222749710083,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 70310
+    },
+    {
+      "epoch": 462.63157894736844,
+      "grad_norm": 1.1833173036575317,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 70320
+    },
+    {
+      "epoch": 462.69736842105266,
+      "grad_norm": 1.226995587348938,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 70330
+    },
+    {
+      "epoch": 462.7631578947368,
+      "grad_norm": 1.0808686017990112,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 70340
+    },
+    {
+      "epoch": 462.82894736842104,
+      "grad_norm": 1.443953275680542,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 70350
+    },
+    {
+      "epoch": 462.89473684210526,
+      "grad_norm": 1.205458402633667,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 70360
+    },
+    {
+      "epoch": 462.9605263157895,
+      "grad_norm": 1.6020748615264893,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 70370
+    },
+    {
+      "epoch": 463.0263157894737,
+      "grad_norm": 1.2139602899551392,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 70380
+    },
+    {
+      "epoch": 463.0921052631579,
+      "grad_norm": 1.0997470617294312,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 70390
+    },
+    {
+      "epoch": 463.1578947368421,
+      "grad_norm": 0.7857748866081238,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 70400
+    },
+    {
+      "epoch": 463.2236842105263,
+      "grad_norm": 1.3937057256698608,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 70410
+    },
+    {
+      "epoch": 463.2894736842105,
+      "grad_norm": 1.2709908485412598,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 70420
+    },
+    {
+      "epoch": 463.35526315789474,
+      "grad_norm": 1.2485954761505127,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 70430
+    },
+    {
+      "epoch": 463.42105263157896,
+      "grad_norm": 1.241417646408081,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 70440
+    },
+    {
+      "epoch": 463.4868421052632,
+      "grad_norm": 1.4835032224655151,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 70450
+    },
+    {
+      "epoch": 463.55263157894734,
+      "grad_norm": 1.4047082662582397,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 70460
+    },
+    {
+      "epoch": 463.61842105263156,
+      "grad_norm": 1.533784031867981,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 70470
+    },
+    {
+      "epoch": 463.6842105263158,
+      "grad_norm": 1.3087002038955688,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 70480
+    },
+    {
+      "epoch": 463.75,
+      "grad_norm": 1.1302224397659302,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 70490
+    },
+    {
+      "epoch": 463.8157894736842,
+      "grad_norm": 0.9580501317977905,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 70500
+    },
+    {
+      "epoch": 463.88157894736844,
+      "grad_norm": 0.8459773063659668,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 70510
+    },
+    {
+      "epoch": 463.94736842105266,
+      "grad_norm": 1.1055728197097778,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 70520
+    },
+    {
+      "epoch": 464.0131578947368,
+      "grad_norm": 1.2454581260681152,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 70530
+    },
+    {
+      "epoch": 464.07894736842104,
+      "grad_norm": 1.3780239820480347,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 70540
+    },
+    {
+      "epoch": 464.14473684210526,
+      "grad_norm": 1.2484023571014404,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 70550
+    },
+    {
+      "epoch": 464.2105263157895,
+      "grad_norm": 1.0500701665878296,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 70560
+    },
+    {
+      "epoch": 464.2763157894737,
+      "grad_norm": 0.9749228358268738,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 70570
+    },
+    {
+      "epoch": 464.3421052631579,
+      "grad_norm": 1.0054138898849487,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 70580
+    },
+    {
+      "epoch": 464.4078947368421,
+      "grad_norm": 1.2907658815383911,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 70590
+    },
+    {
+      "epoch": 464.4736842105263,
+      "grad_norm": 1.1337512731552124,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 70600
+    },
+    {
+      "epoch": 464.5394736842105,
+      "grad_norm": 1.3894444704055786,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 70610
+    },
+    {
+      "epoch": 464.60526315789474,
+      "grad_norm": 1.5000969171524048,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 70620
+    },
+    {
+      "epoch": 464.67105263157896,
+      "grad_norm": 1.1303669214248657,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 70630
+    },
+    {
+      "epoch": 464.7368421052632,
+      "grad_norm": 0.8212418556213379,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 70640
+    },
+    {
+      "epoch": 464.80263157894734,
+      "grad_norm": 1.0030567646026611,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 70650
+    },
+    {
+      "epoch": 464.86842105263156,
+      "grad_norm": 1.0976468324661255,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 70660
+    },
+    {
+      "epoch": 464.9342105263158,
+      "grad_norm": 1.2463607788085938,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 70670
+    },
+    {
+      "epoch": 465.0,
+      "grad_norm": 0.891297459602356,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 70680
+    },
+    {
+      "epoch": 465.0657894736842,
+      "grad_norm": 1.2342113256454468,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 70690
+    },
+    {
+      "epoch": 465.13157894736844,
+      "grad_norm": 0.8251991868019104,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 70700
+    },
+    {
+      "epoch": 465.19736842105266,
+      "grad_norm": 0.9566839337348938,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 70710
+    },
+    {
+      "epoch": 465.2631578947368,
+      "grad_norm": 0.8764354586601257,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 70720
+    },
+    {
+      "epoch": 465.32894736842104,
+      "grad_norm": 1.0923231840133667,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 70730
+    },
+    {
+      "epoch": 465.39473684210526,
+      "grad_norm": 1.1273107528686523,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 70740
+    },
+    {
+      "epoch": 465.4605263157895,
+      "grad_norm": 1.488062858581543,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 70750
+    },
+    {
+      "epoch": 465.5263157894737,
+      "grad_norm": 1.199041724205017,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 70760
+    },
+    {
+      "epoch": 465.5921052631579,
+      "grad_norm": 1.3205357789993286,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 70770
+    },
+    {
+      "epoch": 465.6578947368421,
+      "grad_norm": 1.4035863876342773,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 70780
+    },
+    {
+      "epoch": 465.7236842105263,
+      "grad_norm": 1.1187028884887695,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 70790
+    },
+    {
+      "epoch": 465.7894736842105,
+      "grad_norm": 1.0568621158599854,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 70800
+    },
+    {
+      "epoch": 465.85526315789474,
+      "grad_norm": 1.3467577695846558,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 70810
+    },
+    {
+      "epoch": 465.92105263157896,
+      "grad_norm": 1.3389008045196533,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 70820
+    },
+    {
+      "epoch": 465.9868421052632,
+      "grad_norm": 1.372363567352295,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 70830
+    },
+    {
+      "epoch": 466.05263157894734,
+      "grad_norm": 1.0468113422393799,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 70840
+    },
+    {
+      "epoch": 466.11842105263156,
+      "grad_norm": 0.9883679747581482,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 70850
+    },
+    {
+      "epoch": 466.1842105263158,
+      "grad_norm": 0.710003674030304,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 70860
+    },
+    {
+      "epoch": 466.25,
+      "grad_norm": 1.3106662034988403,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 70870
+    },
+    {
+      "epoch": 466.3157894736842,
+      "grad_norm": 1.022836685180664,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 70880
+    },
+    {
+      "epoch": 466.38157894736844,
+      "grad_norm": 1.1058964729309082,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 70890
+    },
+    {
+      "epoch": 466.44736842105266,
+      "grad_norm": 0.9566864371299744,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 70900
+    },
+    {
+      "epoch": 466.5131578947368,
+      "grad_norm": 1.0572712421417236,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 70910
+    },
+    {
+      "epoch": 466.57894736842104,
+      "grad_norm": 0.6852365136146545,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 70920
+    },
+    {
+      "epoch": 466.64473684210526,
+      "grad_norm": 0.9085026383399963,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 70930
+    },
+    {
+      "epoch": 466.7105263157895,
+      "grad_norm": 1.1825387477874756,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 70940
+    },
+    {
+      "epoch": 466.7763157894737,
+      "grad_norm": 1.0037176609039307,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 70950
+    },
+    {
+      "epoch": 466.8421052631579,
+      "grad_norm": 1.0735546350479126,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 70960
+    },
+    {
+      "epoch": 466.9078947368421,
+      "grad_norm": 1.3803229331970215,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 70970
+    },
+    {
+      "epoch": 466.9736842105263,
+      "grad_norm": 0.845694363117218,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 70980
+    },
+    {
+      "epoch": 467.0394736842105,
+      "grad_norm": 1.186300277709961,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 70990
+    },
+    {
+      "epoch": 467.10526315789474,
+      "grad_norm": 1.171091079711914,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 71000
+    },
+    {
+      "epoch": 467.17105263157896,
+      "grad_norm": 1.3238587379455566,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 71010
+    },
+    {
+      "epoch": 467.2368421052632,
+      "grad_norm": 1.0442191362380981,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 71020
+    },
+    {
+      "epoch": 467.30263157894734,
+      "grad_norm": 0.9936031103134155,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 71030
+    },
+    {
+      "epoch": 467.36842105263156,
+      "grad_norm": 0.8942242860794067,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 71040
+    },
+    {
+      "epoch": 467.4342105263158,
+      "grad_norm": 1.1458888053894043,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 71050
+    },
+    {
+      "epoch": 467.5,
+      "grad_norm": 1.038323163986206,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 71060
+    },
+    {
+      "epoch": 467.5657894736842,
+      "grad_norm": 1.2702325582504272,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 71070
+    },
+    {
+      "epoch": 467.63157894736844,
+      "grad_norm": 1.037888526916504,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 71080
+    },
+    {
+      "epoch": 467.69736842105266,
+      "grad_norm": 1.5120564699172974,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 71090
+    },
+    {
+      "epoch": 467.7631578947368,
+      "grad_norm": 1.0400851964950562,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 71100
+    },
+    {
+      "epoch": 467.82894736842104,
+      "grad_norm": 1.0576149225234985,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 71110
+    },
+    {
+      "epoch": 467.89473684210526,
+      "grad_norm": 1.1322977542877197,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 71120
+    },
+    {
+      "epoch": 467.9605263157895,
+      "grad_norm": 1.2082958221435547,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 71130
+    },
+    {
+      "epoch": 468.0263157894737,
+      "grad_norm": 0.967847466468811,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 71140
+    },
+    {
+      "epoch": 468.0921052631579,
+      "grad_norm": 1.2820326089859009,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 71150
+    },
+    {
+      "epoch": 468.1578947368421,
+      "grad_norm": 1.3719661235809326,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 71160
+    },
+    {
+      "epoch": 468.2236842105263,
+      "grad_norm": 1.6810905933380127,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 71170
+    },
+    {
+      "epoch": 468.2894736842105,
+      "grad_norm": 0.994966447353363,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 71180
+    },
+    {
+      "epoch": 468.35526315789474,
+      "grad_norm": 1.2583346366882324,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 71190
+    },
+    {
+      "epoch": 468.42105263157896,
+      "grad_norm": 1.1598786115646362,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 71200
+    },
+    {
+      "epoch": 468.4868421052632,
+      "grad_norm": 1.2424436807632446,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 71210
+    },
+    {
+      "epoch": 468.55263157894734,
+      "grad_norm": 1.1447334289550781,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 71220
+    },
+    {
+      "epoch": 468.61842105263156,
+      "grad_norm": 0.9471054077148438,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 71230
+    },
+    {
+      "epoch": 468.6842105263158,
+      "grad_norm": 1.2661805152893066,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 71240
+    },
+    {
+      "epoch": 468.75,
+      "grad_norm": 1.1477277278900146,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 71250
+    },
+    {
+      "epoch": 468.8157894736842,
+      "grad_norm": 1.0798248052597046,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 71260
+    },
+    {
+      "epoch": 468.88157894736844,
+      "grad_norm": 1.0489845275878906,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 71270
+    },
+    {
+      "epoch": 468.94736842105266,
+      "grad_norm": 1.2444311380386353,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 71280
+    },
+    {
+      "epoch": 469.0131578947368,
+      "grad_norm": 1.1508184671401978,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 71290
+    },
+    {
+      "epoch": 469.07894736842104,
+      "grad_norm": 1.0724323987960815,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 71300
+    },
+    {
+      "epoch": 469.14473684210526,
+      "grad_norm": 1.3677020072937012,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 71310
+    },
+    {
+      "epoch": 469.2105263157895,
+      "grad_norm": 0.8909698128700256,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 71320
+    },
+    {
+      "epoch": 469.2763157894737,
+      "grad_norm": 1.2063177824020386,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 71330
+    },
+    {
+      "epoch": 469.3421052631579,
+      "grad_norm": 1.4263769388198853,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 71340
+    },
+    {
+      "epoch": 469.4078947368421,
+      "grad_norm": 1.343396782875061,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 71350
+    },
+    {
+      "epoch": 469.4736842105263,
+      "grad_norm": 1.0640443563461304,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 71360
+    },
+    {
+      "epoch": 469.5394736842105,
+      "grad_norm": 1.131130576133728,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 71370
+    },
+    {
+      "epoch": 469.60526315789474,
+      "grad_norm": 0.8437464833259583,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 71380
+    },
+    {
+      "epoch": 469.67105263157896,
+      "grad_norm": 0.8952385187149048,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 71390
+    },
+    {
+      "epoch": 469.7368421052632,
+      "grad_norm": 0.8962030410766602,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 71400
+    },
+    {
+      "epoch": 469.80263157894734,
+      "grad_norm": 0.953825056552887,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 71410
+    },
+    {
+      "epoch": 469.86842105263156,
+      "grad_norm": 0.8790578246116638,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 71420
+    },
+    {
+      "epoch": 469.9342105263158,
+      "grad_norm": 1.3036227226257324,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 71430
+    },
+    {
+      "epoch": 470.0,
+      "grad_norm": 0.9618934392929077,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 71440
+    },
+    {
+      "epoch": 470.0657894736842,
+      "grad_norm": 0.908938467502594,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 71450
+    },
+    {
+      "epoch": 470.13157894736844,
+      "grad_norm": 1.156169056892395,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 71460
+    },
+    {
+      "epoch": 470.19736842105266,
+      "grad_norm": 0.8988484144210815,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 71470
+    },
+    {
+      "epoch": 470.2631578947368,
+      "grad_norm": 1.0380802154541016,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 71480
+    },
+    {
+      "epoch": 470.32894736842104,
+      "grad_norm": 1.1827173233032227,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 71490
+    },
+    {
+      "epoch": 470.39473684210526,
+      "grad_norm": 0.9451979994773865,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 71500
+    },
+    {
+      "epoch": 470.4605263157895,
+      "grad_norm": 0.8743581771850586,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 71510
+    },
+    {
+      "epoch": 470.5263157894737,
+      "grad_norm": 0.9873955249786377,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 71520
+    },
+    {
+      "epoch": 470.5921052631579,
+      "grad_norm": 0.9436299204826355,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 71530
+    },
+    {
+      "epoch": 470.6578947368421,
+      "grad_norm": 0.8567780256271362,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 71540
+    },
+    {
+      "epoch": 470.7236842105263,
+      "grad_norm": 1.0895605087280273,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 71550
+    },
+    {
+      "epoch": 470.7894736842105,
+      "grad_norm": 1.3269134759902954,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 71560
+    },
+    {
+      "epoch": 470.85526315789474,
+      "grad_norm": 1.2100285291671753,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 71570
+    },
+    {
+      "epoch": 470.92105263157896,
+      "grad_norm": 1.2074404954910278,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 71580
+    },
+    {
+      "epoch": 470.9868421052632,
+      "grad_norm": 1.3929208517074585,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 71590
+    },
+    {
+      "epoch": 471.05263157894734,
+      "grad_norm": 1.3554884195327759,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 71600
+    },
+    {
+      "epoch": 471.11842105263156,
+      "grad_norm": 1.3486567735671997,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 71610
+    },
+    {
+      "epoch": 471.1842105263158,
+      "grad_norm": 0.9685068726539612,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 71620
+    },
+    {
+      "epoch": 471.25,
+      "grad_norm": 1.542588233947754,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 71630
+    },
+    {
+      "epoch": 471.3157894736842,
+      "grad_norm": 1.5774450302124023,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 71640
+    },
+    {
+      "epoch": 471.38157894736844,
+      "grad_norm": 1.1362987756729126,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 71650
+    },
+    {
+      "epoch": 471.44736842105266,
+      "grad_norm": 1.0866764783859253,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 71660
+    },
+    {
+      "epoch": 471.5131578947368,
+      "grad_norm": 1.3600131273269653,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 71670
+    },
+    {
+      "epoch": 471.57894736842104,
+      "grad_norm": 1.3982903957366943,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 71680
+    },
+    {
+      "epoch": 471.64473684210526,
+      "grad_norm": 1.3128663301467896,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 71690
+    },
+    {
+      "epoch": 471.7105263157895,
+      "grad_norm": 1.2559845447540283,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 71700
+    },
+    {
+      "epoch": 471.7763157894737,
+      "grad_norm": 0.9143256545066833,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 71710
+    },
+    {
+      "epoch": 471.8421052631579,
+      "grad_norm": 1.2619221210479736,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 71720
+    },
+    {
+      "epoch": 471.9078947368421,
+      "grad_norm": 1.3303905725479126,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 71730
+    },
+    {
+      "epoch": 471.9736842105263,
+      "grad_norm": 1.1719235181808472,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 71740
+    },
+    {
+      "epoch": 472.0394736842105,
+      "grad_norm": 1.1172962188720703,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 71750
+    },
+    {
+      "epoch": 472.10526315789474,
+      "grad_norm": 1.4545469284057617,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 71760
+    },
+    {
+      "epoch": 472.17105263157896,
+      "grad_norm": 0.806731641292572,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 71770
+    },
+    {
+      "epoch": 472.2368421052632,
+      "grad_norm": 1.0789495706558228,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 71780
+    },
+    {
+      "epoch": 472.30263157894734,
+      "grad_norm": 1.059844732284546,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 71790
+    },
+    {
+      "epoch": 472.36842105263156,
+      "grad_norm": 1.3796194791793823,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 71800
+    },
+    {
+      "epoch": 472.4342105263158,
+      "grad_norm": 1.1321066617965698,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 71810
+    },
+    {
+      "epoch": 472.5,
+      "grad_norm": 0.8206098675727844,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 71820
+    },
+    {
+      "epoch": 472.5657894736842,
+      "grad_norm": 0.9593137502670288,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 71830
+    },
+    {
+      "epoch": 472.63157894736844,
+      "grad_norm": 1.0930147171020508,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 71840
+    },
+    {
+      "epoch": 472.69736842105266,
+      "grad_norm": 1.3240526914596558,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 71850
+    },
+    {
+      "epoch": 472.7631578947368,
+      "grad_norm": 1.3134987354278564,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 71860
+    },
+    {
+      "epoch": 472.82894736842104,
+      "grad_norm": 1.409283995628357,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 71870
+    },
+    {
+      "epoch": 472.89473684210526,
+      "grad_norm": 0.9180440306663513,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 71880
+    },
+    {
+      "epoch": 472.9605263157895,
+      "grad_norm": 0.8743194937705994,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 71890
+    },
+    {
+      "epoch": 473.0263157894737,
+      "grad_norm": 1.0829284191131592,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 71900
+    },
+    {
+      "epoch": 473.0921052631579,
+      "grad_norm": 1.0314230918884277,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 71910
+    },
+    {
+      "epoch": 473.1578947368421,
+      "grad_norm": 1.0930405855178833,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 71920
+    },
+    {
+      "epoch": 473.2236842105263,
+      "grad_norm": 1.387052059173584,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 71930
+    },
+    {
+      "epoch": 473.2894736842105,
+      "grad_norm": 1.1095280647277832,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 71940
+    },
+    {
+      "epoch": 473.35526315789474,
+      "grad_norm": 1.1823383569717407,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 71950
+    },
+    {
+      "epoch": 473.42105263157896,
+      "grad_norm": 1.2319684028625488,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 71960
+    },
+    {
+      "epoch": 473.4868421052632,
+      "grad_norm": 1.3296653032302856,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 71970
+    },
+    {
+      "epoch": 473.55263157894734,
+      "grad_norm": 1.1372100114822388,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 71980
+    },
+    {
+      "epoch": 473.61842105263156,
+      "grad_norm": 1.3765093088150024,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 71990
+    },
+    {
+      "epoch": 473.6842105263158,
+      "grad_norm": 1.0475393533706665,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 72000
+    },
+    {
+      "epoch": 473.75,
+      "grad_norm": 0.8706821203231812,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 72010
+    },
+    {
+      "epoch": 473.8157894736842,
+      "grad_norm": 1.1622743606567383,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 72020
+    },
+    {
+      "epoch": 473.88157894736844,
+      "grad_norm": 1.101678729057312,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 72030
+    },
+    {
+      "epoch": 473.94736842105266,
+      "grad_norm": 1.4966200590133667,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 72040
+    },
+    {
+      "epoch": 474.0131578947368,
+      "grad_norm": 0.8615991473197937,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 72050
+    },
+    {
+      "epoch": 474.07894736842104,
+      "grad_norm": 1.0953624248504639,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 72060
+    },
+    {
+      "epoch": 474.14473684210526,
+      "grad_norm": 1.2191064357757568,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 72070
+    },
+    {
+      "epoch": 474.2105263157895,
+      "grad_norm": 0.9888071417808533,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 72080
+    },
+    {
+      "epoch": 474.2763157894737,
+      "grad_norm": 1.138601541519165,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 72090
+    },
+    {
+      "epoch": 474.3421052631579,
+      "grad_norm": 0.9573279023170471,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 72100
+    },
+    {
+      "epoch": 474.4078947368421,
+      "grad_norm": 1.5391374826431274,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 72110
+    },
+    {
+      "epoch": 474.4736842105263,
+      "grad_norm": 1.4656072854995728,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 72120
+    },
+    {
+      "epoch": 474.5394736842105,
+      "grad_norm": 1.4683994054794312,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 72130
+    },
+    {
+      "epoch": 474.60526315789474,
+      "grad_norm": 1.0790272951126099,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 72140
+    },
+    {
+      "epoch": 474.67105263157896,
+      "grad_norm": 1.186042070388794,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 72150
+    },
+    {
+      "epoch": 474.7368421052632,
+      "grad_norm": 1.169363021850586,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 72160
+    },
+    {
+      "epoch": 474.80263157894734,
+      "grad_norm": 1.1990368366241455,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 72170
+    },
+    {
+      "epoch": 474.86842105263156,
+      "grad_norm": 1.200440764427185,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 72180
+    },
+    {
+      "epoch": 474.9342105263158,
+      "grad_norm": 1.1117382049560547,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 72190
+    },
+    {
+      "epoch": 475.0,
+      "grad_norm": 1.0739638805389404,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 72200
+    },
+    {
+      "epoch": 475.0657894736842,
+      "grad_norm": 1.1158939599990845,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 72210
+    },
+    {
+      "epoch": 475.13157894736844,
+      "grad_norm": 1.6269727945327759,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 72220
+    },
+    {
+      "epoch": 475.19736842105266,
+      "grad_norm": 1.75429368019104,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 72230
+    },
+    {
+      "epoch": 475.2631578947368,
+      "grad_norm": 1.5069760084152222,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 72240
+    },
+    {
+      "epoch": 475.32894736842104,
+      "grad_norm": 1.1883662939071655,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 72250
+    },
+    {
+      "epoch": 475.39473684210526,
+      "grad_norm": 1.3766677379608154,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 72260
+    },
+    {
+      "epoch": 475.4605263157895,
+      "grad_norm": 1.2644368410110474,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 72270
+    },
+    {
+      "epoch": 475.5263157894737,
+      "grad_norm": 1.295596718788147,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 72280
+    },
+    {
+      "epoch": 475.5921052631579,
+      "grad_norm": 1.2578319311141968,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 72290
+    },
+    {
+      "epoch": 475.6578947368421,
+      "grad_norm": 1.1036814451217651,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 72300
+    },
+    {
+      "epoch": 475.7236842105263,
+      "grad_norm": 1.1769661903381348,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 72310
+    },
+    {
+      "epoch": 475.7894736842105,
+      "grad_norm": 0.9555389285087585,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 72320
+    },
+    {
+      "epoch": 475.85526315789474,
+      "grad_norm": 1.3688865900039673,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 72330
+    },
+    {
+      "epoch": 475.92105263157896,
+      "grad_norm": 1.4835799932479858,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 72340
+    },
+    {
+      "epoch": 475.9868421052632,
+      "grad_norm": 1.1539369821548462,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 72350
+    },
+    {
+      "epoch": 476.05263157894734,
+      "grad_norm": 1.3297797441482544,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 72360
+    },
+    {
+      "epoch": 476.11842105263156,
+      "grad_norm": 1.5591670274734497,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 72370
+    },
+    {
+      "epoch": 476.1842105263158,
+      "grad_norm": 1.2825270891189575,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 72380
+    },
+    {
+      "epoch": 476.25,
+      "grad_norm": 0.7293868064880371,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 72390
+    },
+    {
+      "epoch": 476.3157894736842,
+      "grad_norm": 1.2907273769378662,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 72400
+    },
+    {
+      "epoch": 476.38157894736844,
+      "grad_norm": 0.7899346351623535,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 72410
+    },
+    {
+      "epoch": 476.44736842105266,
+      "grad_norm": 1.0589332580566406,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 72420
+    },
+    {
+      "epoch": 476.5131578947368,
+      "grad_norm": 1.1962878704071045,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 72430
+    },
+    {
+      "epoch": 476.57894736842104,
+      "grad_norm": 1.0695892572402954,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 72440
+    },
+    {
+      "epoch": 476.64473684210526,
+      "grad_norm": 0.8697053790092468,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 72450
+    },
+    {
+      "epoch": 476.7105263157895,
+      "grad_norm": 0.8389332890510559,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 72460
+    },
+    {
+      "epoch": 476.7763157894737,
+      "grad_norm": 1.0303555727005005,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 72470
+    },
+    {
+      "epoch": 476.8421052631579,
+      "grad_norm": 0.8726432919502258,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 72480
+    },
+    {
+      "epoch": 476.9078947368421,
+      "grad_norm": 1.1818912029266357,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 72490
+    },
+    {
+      "epoch": 476.9736842105263,
+      "grad_norm": 1.0324101448059082,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 72500
+    },
+    {
+      "epoch": 477.0394736842105,
+      "grad_norm": 1.1190623044967651,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 72510
+    },
+    {
+      "epoch": 477.10526315789474,
+      "grad_norm": 1.3751294612884521,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 72520
+    },
+    {
+      "epoch": 477.17105263157896,
+      "grad_norm": 1.0744208097457886,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 72530
+    },
+    {
+      "epoch": 477.2368421052632,
+      "grad_norm": 1.5082820653915405,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 72540
+    },
+    {
+      "epoch": 477.30263157894734,
+      "grad_norm": 0.6048269867897034,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 72550
+    },
+    {
+      "epoch": 477.36842105263156,
+      "grad_norm": 1.0494825839996338,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 72560
+    },
+    {
+      "epoch": 477.4342105263158,
+      "grad_norm": 1.2498975992202759,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 72570
+    },
+    {
+      "epoch": 477.5,
+      "grad_norm": 0.980216383934021,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 72580
+    },
+    {
+      "epoch": 477.5657894736842,
+      "grad_norm": 1.551719069480896,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 72590
+    },
+    {
+      "epoch": 477.63157894736844,
+      "grad_norm": 0.9513759016990662,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 72600
+    },
+    {
+      "epoch": 477.69736842105266,
+      "grad_norm": 1.2828733921051025,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 72610
+    },
+    {
+      "epoch": 477.7631578947368,
+      "grad_norm": 0.9803738594055176,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 72620
+    },
+    {
+      "epoch": 477.82894736842104,
+      "grad_norm": 1.0510716438293457,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 72630
+    },
+    {
+      "epoch": 477.89473684210526,
+      "grad_norm": 0.7186201214790344,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 72640
+    },
+    {
+      "epoch": 477.9605263157895,
+      "grad_norm": 1.2259303331375122,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 72650
+    },
+    {
+      "epoch": 478.0263157894737,
+      "grad_norm": 0.948920726776123,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 72660
+    },
+    {
+      "epoch": 478.0921052631579,
+      "grad_norm": 1.4578410387039185,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 72670
+    },
+    {
+      "epoch": 478.1578947368421,
+      "grad_norm": 1.2089077234268188,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 72680
+    },
+    {
+      "epoch": 478.2236842105263,
+      "grad_norm": 1.2857468128204346,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 72690
+    },
+    {
+      "epoch": 478.2894736842105,
+      "grad_norm": 0.8931973576545715,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 72700
+    },
+    {
+      "epoch": 478.35526315789474,
+      "grad_norm": 0.7080501317977905,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 72710
+    },
+    {
+      "epoch": 478.42105263157896,
+      "grad_norm": 1.3679475784301758,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 72720
+    },
+    {
+      "epoch": 478.4868421052632,
+      "grad_norm": 1.0242080688476562,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 72730
+    },
+    {
+      "epoch": 478.55263157894734,
+      "grad_norm": 1.1003470420837402,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 72740
+    },
+    {
+      "epoch": 478.61842105263156,
+      "grad_norm": 1.2705953121185303,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 72750
+    },
+    {
+      "epoch": 478.6842105263158,
+      "grad_norm": 0.9441986680030823,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 72760
+    },
+    {
+      "epoch": 478.75,
+      "grad_norm": 1.438714861869812,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 72770
+    },
+    {
+      "epoch": 478.8157894736842,
+      "grad_norm": 1.4461417198181152,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 72780
+    },
+    {
+      "epoch": 478.88157894736844,
+      "grad_norm": 1.4734560251235962,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 72790
+    },
+    {
+      "epoch": 478.94736842105266,
+      "grad_norm": 1.064142107963562,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 72800
+    },
+    {
+      "epoch": 479.0131578947368,
+      "grad_norm": 0.7491738796234131,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 72810
+    },
+    {
+      "epoch": 479.07894736842104,
+      "grad_norm": 1.0148205757141113,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 72820
+    },
+    {
+      "epoch": 479.14473684210526,
+      "grad_norm": 0.9256365299224854,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 72830
+    },
+    {
+      "epoch": 479.2105263157895,
+      "grad_norm": 1.0574623346328735,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 72840
+    },
+    {
+      "epoch": 479.2763157894737,
+      "grad_norm": 1.2569656372070312,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 72850
+    },
+    {
+      "epoch": 479.3421052631579,
+      "grad_norm": 0.737408459186554,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 72860
+    },
+    {
+      "epoch": 479.4078947368421,
+      "grad_norm": 1.0380948781967163,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 72870
+    },
+    {
+      "epoch": 479.4736842105263,
+      "grad_norm": 0.9933061599731445,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 72880
+    },
+    {
+      "epoch": 479.5394736842105,
+      "grad_norm": 1.3550677299499512,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 72890
+    },
+    {
+      "epoch": 479.60526315789474,
+      "grad_norm": 1.2398386001586914,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 72900
+    },
+    {
+      "epoch": 479.67105263157896,
+      "grad_norm": 1.213581919670105,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 72910
+    },
+    {
+      "epoch": 479.7368421052632,
+      "grad_norm": 1.4044569730758667,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 72920
+    },
+    {
+      "epoch": 479.80263157894734,
+      "grad_norm": 0.956494152545929,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 72930
+    },
+    {
+      "epoch": 479.86842105263156,
+      "grad_norm": 0.9288977384567261,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 72940
+    },
+    {
+      "epoch": 479.9342105263158,
+      "grad_norm": 1.0925542116165161,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 72950
+    },
+    {
+      "epoch": 480.0,
+      "grad_norm": 1.3831939697265625,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 72960
+    },
+    {
+      "epoch": 480.0657894736842,
+      "grad_norm": 1.177885890007019,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 72970
+    },
+    {
+      "epoch": 480.13157894736844,
+      "grad_norm": 1.154605507850647,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 72980
+    },
+    {
+      "epoch": 480.19736842105266,
+      "grad_norm": 1.4074510335922241,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 72990
+    },
+    {
+      "epoch": 480.2631578947368,
+      "grad_norm": 1.3587064743041992,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 73000
+    },
+    {
+      "epoch": 480.32894736842104,
+      "grad_norm": 1.0905214548110962,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 73010
+    },
+    {
+      "epoch": 480.39473684210526,
+      "grad_norm": 1.2721132040023804,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 73020
+    },
+    {
+      "epoch": 480.4605263157895,
+      "grad_norm": 1.2455694675445557,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 73030
+    },
+    {
+      "epoch": 480.5263157894737,
+      "grad_norm": 1.1374874114990234,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 73040
+    },
+    {
+      "epoch": 480.5921052631579,
+      "grad_norm": 0.9991102814674377,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 73050
+    },
+    {
+      "epoch": 480.6578947368421,
+      "grad_norm": 0.9660846590995789,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 73060
+    },
+    {
+      "epoch": 480.7236842105263,
+      "grad_norm": 1.3656002283096313,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 73070
+    },
+    {
+      "epoch": 480.7894736842105,
+      "grad_norm": 1.039703607559204,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 73080
+    },
+    {
+      "epoch": 480.85526315789474,
+      "grad_norm": 1.1544702053070068,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 73090
+    },
+    {
+      "epoch": 480.92105263157896,
+      "grad_norm": 1.4217747449874878,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 73100
+    },
+    {
+      "epoch": 480.9868421052632,
+      "grad_norm": 1.363659143447876,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 73110
+    },
+    {
+      "epoch": 481.05263157894734,
+      "grad_norm": 1.2060271501541138,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 73120
+    },
+    {
+      "epoch": 481.11842105263156,
+      "grad_norm": 0.9276877045631409,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 73130
+    },
+    {
+      "epoch": 481.1842105263158,
+      "grad_norm": 1.148481845855713,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 73140
+    },
+    {
+      "epoch": 481.25,
+      "grad_norm": 1.3816297054290771,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 73150
+    },
+    {
+      "epoch": 481.3157894736842,
+      "grad_norm": 1.1494829654693604,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 73160
+    },
+    {
+      "epoch": 481.38157894736844,
+      "grad_norm": 0.8974570035934448,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 73170
+    },
+    {
+      "epoch": 481.44736842105266,
+      "grad_norm": 1.2027701139450073,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 73180
+    },
+    {
+      "epoch": 481.5131578947368,
+      "grad_norm": 1.0080323219299316,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 73190
+    },
+    {
+      "epoch": 481.57894736842104,
+      "grad_norm": 1.2009938955307007,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 73200
+    },
+    {
+      "epoch": 481.64473684210526,
+      "grad_norm": 1.2606686353683472,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 73210
+    },
+    {
+      "epoch": 481.7105263157895,
+      "grad_norm": 0.9282734990119934,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 73220
+    },
+    {
+      "epoch": 481.7763157894737,
+      "grad_norm": 0.6285842657089233,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 73230
+    },
+    {
+      "epoch": 481.8421052631579,
+      "grad_norm": 0.895392894744873,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 73240
+    },
+    {
+      "epoch": 481.9078947368421,
+      "grad_norm": 1.0975970029830933,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 73250
+    },
+    {
+      "epoch": 481.9736842105263,
+      "grad_norm": 1.0048805475234985,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 73260
+    },
+    {
+      "epoch": 482.0394736842105,
+      "grad_norm": 1.4439254999160767,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 73270
+    },
+    {
+      "epoch": 482.10526315789474,
+      "grad_norm": 1.0936566591262817,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 73280
+    },
+    {
+      "epoch": 482.17105263157896,
+      "grad_norm": 1.1018260717391968,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 73290
+    },
+    {
+      "epoch": 482.2368421052632,
+      "grad_norm": 1.48947012424469,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 73300
+    },
+    {
+      "epoch": 482.30263157894734,
+      "grad_norm": 0.8502678871154785,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 73310
+    },
+    {
+      "epoch": 482.36842105263156,
+      "grad_norm": 1.0486079454421997,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 73320
+    },
+    {
+      "epoch": 482.4342105263158,
+      "grad_norm": 1.3065072298049927,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 73330
+    },
+    {
+      "epoch": 482.5,
+      "grad_norm": 1.218176007270813,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 73340
+    },
+    {
+      "epoch": 482.5657894736842,
+      "grad_norm": 1.2767457962036133,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 73350
+    },
+    {
+      "epoch": 482.63157894736844,
+      "grad_norm": 1.322757363319397,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 73360
+    },
+    {
+      "epoch": 482.69736842105266,
+      "grad_norm": 1.248364806175232,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 73370
+    },
+    {
+      "epoch": 482.7631578947368,
+      "grad_norm": 0.9732344746589661,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 73380
+    },
+    {
+      "epoch": 482.82894736842104,
+      "grad_norm": 1.0928685665130615,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 73390
+    },
+    {
+      "epoch": 482.89473684210526,
+      "grad_norm": 1.0037014484405518,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 73400
+    },
+    {
+      "epoch": 482.9605263157895,
+      "grad_norm": 1.0865426063537598,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 73410
+    },
+    {
+      "epoch": 483.0263157894737,
+      "grad_norm": 1.5085911750793457,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 73420
+    },
+    {
+      "epoch": 483.0921052631579,
+      "grad_norm": 1.8017462491989136,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 73430
+    },
+    {
+      "epoch": 483.1578947368421,
+      "grad_norm": 1.266747236251831,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 73440
+    },
+    {
+      "epoch": 483.2236842105263,
+      "grad_norm": 1.4259337186813354,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 73450
+    },
+    {
+      "epoch": 483.2894736842105,
+      "grad_norm": 1.2923461198806763,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 73460
+    },
+    {
+      "epoch": 483.35526315789474,
+      "grad_norm": 1.453090786933899,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 73470
+    },
+    {
+      "epoch": 483.42105263157896,
+      "grad_norm": 1.3885729312896729,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 73480
+    },
+    {
+      "epoch": 483.4868421052632,
+      "grad_norm": 0.8967188000679016,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 73490
+    },
+    {
+      "epoch": 483.55263157894734,
+      "grad_norm": 1.2121909856796265,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 73500
+    },
+    {
+      "epoch": 483.61842105263156,
+      "grad_norm": 1.01747465133667,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 73510
+    },
+    {
+      "epoch": 483.6842105263158,
+      "grad_norm": 1.1471728086471558,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 73520
+    },
+    {
+      "epoch": 483.75,
+      "grad_norm": 1.196329116821289,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 73530
+    },
+    {
+      "epoch": 483.8157894736842,
+      "grad_norm": 1.0889172554016113,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 73540
+    },
+    {
+      "epoch": 483.88157894736844,
+      "grad_norm": 1.1677290201187134,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 73550
+    },
+    {
+      "epoch": 483.94736842105266,
+      "grad_norm": 1.2402957677841187,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 73560
+    },
+    {
+      "epoch": 484.0131578947368,
+      "grad_norm": 1.241837739944458,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 73570
+    },
+    {
+      "epoch": 484.07894736842104,
+      "grad_norm": 1.1311156749725342,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 73580
+    },
+    {
+      "epoch": 484.14473684210526,
+      "grad_norm": 1.0363242626190186,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 73590
+    },
+    {
+      "epoch": 484.2105263157895,
+      "grad_norm": 1.2085379362106323,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 73600
+    },
+    {
+      "epoch": 484.2763157894737,
+      "grad_norm": 0.8469356298446655,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 73610
+    },
+    {
+      "epoch": 484.3421052631579,
+      "grad_norm": 1.0864289999008179,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 73620
+    },
+    {
+      "epoch": 484.4078947368421,
+      "grad_norm": 0.8979195952415466,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 73630
+    },
+    {
+      "epoch": 484.4736842105263,
+      "grad_norm": 1.0335921049118042,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 73640
+    },
+    {
+      "epoch": 484.5394736842105,
+      "grad_norm": 0.8344136476516724,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 73650
+    },
+    {
+      "epoch": 484.60526315789474,
+      "grad_norm": 1.1805672645568848,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 73660
+    },
+    {
+      "epoch": 484.67105263157896,
+      "grad_norm": 0.9891030192375183,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 73670
+    },
+    {
+      "epoch": 484.7368421052632,
+      "grad_norm": 1.2835693359375,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 73680
+    },
+    {
+      "epoch": 484.80263157894734,
+      "grad_norm": 1.1186614036560059,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 73690
+    },
+    {
+      "epoch": 484.86842105263156,
+      "grad_norm": 0.9351660013198853,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 73700
+    },
+    {
+      "epoch": 484.9342105263158,
+      "grad_norm": 1.1274996995925903,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 73710
+    },
+    {
+      "epoch": 485.0,
+      "grad_norm": 1.3561069965362549,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 73720
+    },
+    {
+      "epoch": 485.0657894736842,
+      "grad_norm": 1.0583314895629883,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 73730
+    },
+    {
+      "epoch": 485.13157894736844,
+      "grad_norm": 1.318225383758545,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 73740
+    },
+    {
+      "epoch": 485.19736842105266,
+      "grad_norm": 1.3357988595962524,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 73750
+    },
+    {
+      "epoch": 485.2631578947368,
+      "grad_norm": 0.7873854637145996,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 73760
+    },
+    {
+      "epoch": 485.32894736842104,
+      "grad_norm": 0.9330838322639465,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 73770
+    },
+    {
+      "epoch": 485.39473684210526,
+      "grad_norm": 0.956305205821991,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 73780
+    },
+    {
+      "epoch": 485.4605263157895,
+      "grad_norm": 1.252623200416565,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 73790
+    },
+    {
+      "epoch": 485.5263157894737,
+      "grad_norm": 1.2766034603118896,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 73800
+    },
+    {
+      "epoch": 485.5921052631579,
+      "grad_norm": 1.2142257690429688,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 73810
+    },
+    {
+      "epoch": 485.6578947368421,
+      "grad_norm": 0.972432017326355,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 73820
+    },
+    {
+      "epoch": 485.7236842105263,
+      "grad_norm": 1.1014715433120728,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 73830
+    },
+    {
+      "epoch": 485.7894736842105,
+      "grad_norm": 1.1053544282913208,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 73840
+    },
+    {
+      "epoch": 485.85526315789474,
+      "grad_norm": 1.4604548215866089,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 73850
+    },
+    {
+      "epoch": 485.92105263157896,
+      "grad_norm": 0.9957262277603149,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 73860
+    },
+    {
+      "epoch": 485.9868421052632,
+      "grad_norm": 0.8840813636779785,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 73870
+    },
+    {
+      "epoch": 486.05263157894734,
+      "grad_norm": 1.2641267776489258,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 73880
+    },
+    {
+      "epoch": 486.11842105263156,
+      "grad_norm": 0.9049034714698792,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 73890
+    },
+    {
+      "epoch": 486.1842105263158,
+      "grad_norm": 0.798561155796051,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 73900
+    },
+    {
+      "epoch": 486.25,
+      "grad_norm": 1.3658287525177002,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 73910
+    },
+    {
+      "epoch": 486.3157894736842,
+      "grad_norm": 1.3198788166046143,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 73920
+    },
+    {
+      "epoch": 486.38157894736844,
+      "grad_norm": 1.0078285932540894,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 73930
+    },
+    {
+      "epoch": 486.44736842105266,
+      "grad_norm": 1.053542971611023,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 73940
+    },
+    {
+      "epoch": 486.5131578947368,
+      "grad_norm": 0.8640962243080139,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 73950
+    },
+    {
+      "epoch": 486.57894736842104,
+      "grad_norm": 1.2860034704208374,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 73960
+    },
+    {
+      "epoch": 486.64473684210526,
+      "grad_norm": 1.0493406057357788,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 73970
+    },
+    {
+      "epoch": 486.7105263157895,
+      "grad_norm": 0.9338852763175964,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 73980
+    },
+    {
+      "epoch": 486.7763157894737,
+      "grad_norm": 1.1906099319458008,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 73990
+    },
+    {
+      "epoch": 486.8421052631579,
+      "grad_norm": 1.1527206897735596,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 74000
+    },
+    {
+      "epoch": 486.9078947368421,
+      "grad_norm": 1.1856426000595093,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 74010
+    },
+    {
+      "epoch": 486.9736842105263,
+      "grad_norm": 1.0806999206542969,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 74020
+    },
+    {
+      "epoch": 487.0394736842105,
+      "grad_norm": 0.960417628288269,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 74030
+    },
+    {
+      "epoch": 487.10526315789474,
+      "grad_norm": 0.9834040403366089,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 74040
+    },
+    {
+      "epoch": 487.17105263157896,
+      "grad_norm": 0.9878292679786682,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 74050
+    },
+    {
+      "epoch": 487.2368421052632,
+      "grad_norm": 1.11509370803833,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 74060
+    },
+    {
+      "epoch": 487.30263157894734,
+      "grad_norm": 1.1951452493667603,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 74070
+    },
+    {
+      "epoch": 487.36842105263156,
+      "grad_norm": 0.9179181456565857,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 74080
+    },
+    {
+      "epoch": 487.4342105263158,
+      "grad_norm": 1.1499123573303223,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 74090
+    },
+    {
+      "epoch": 487.5,
+      "grad_norm": 1.035010576248169,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 74100
+    },
+    {
+      "epoch": 487.5657894736842,
+      "grad_norm": 1.0398813486099243,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 74110
+    },
+    {
+      "epoch": 487.63157894736844,
+      "grad_norm": 1.191170573234558,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 74120
+    },
+    {
+      "epoch": 487.69736842105266,
+      "grad_norm": 1.0563814640045166,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 74130
+    },
+    {
+      "epoch": 487.7631578947368,
+      "grad_norm": 1.0885233879089355,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 74140
+    },
+    {
+      "epoch": 487.82894736842104,
+      "grad_norm": 1.0386284589767456,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 74150
+    },
+    {
+      "epoch": 487.89473684210526,
+      "grad_norm": 1.3478586673736572,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 74160
+    },
+    {
+      "epoch": 487.9605263157895,
+      "grad_norm": 0.8383530378341675,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 74170
+    },
+    {
+      "epoch": 488.0263157894737,
+      "grad_norm": 1.2732715606689453,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 74180
+    },
+    {
+      "epoch": 488.0921052631579,
+      "grad_norm": 1.2103279829025269,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 74190
+    },
+    {
+      "epoch": 488.1578947368421,
+      "grad_norm": 1.6459661722183228,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 74200
+    },
+    {
+      "epoch": 488.2236842105263,
+      "grad_norm": 1.0316122770309448,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 74210
+    },
+    {
+      "epoch": 488.2894736842105,
+      "grad_norm": 1.0485596656799316,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 74220
+    },
+    {
+      "epoch": 488.35526315789474,
+      "grad_norm": 1.27117121219635,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 74230
+    },
+    {
+      "epoch": 488.42105263157896,
+      "grad_norm": 1.240209698677063,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 74240
+    },
+    {
+      "epoch": 488.4868421052632,
+      "grad_norm": 0.7314859628677368,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 74250
+    },
+    {
+      "epoch": 488.55263157894734,
+      "grad_norm": 1.4494084119796753,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 74260
+    },
+    {
+      "epoch": 488.61842105263156,
+      "grad_norm": 1.7090212106704712,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 74270
+    },
+    {
+      "epoch": 488.6842105263158,
+      "grad_norm": 1.248016119003296,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 74280
+    },
+    {
+      "epoch": 488.75,
+      "grad_norm": 1.0164270401000977,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 74290
+    },
+    {
+      "epoch": 488.8157894736842,
+      "grad_norm": 0.8657320141792297,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 74300
+    },
+    {
+      "epoch": 488.88157894736844,
+      "grad_norm": 1.136674165725708,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 74310
+    },
+    {
+      "epoch": 488.94736842105266,
+      "grad_norm": 1.108292579650879,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 74320
+    },
+    {
+      "epoch": 489.0131578947368,
+      "grad_norm": 0.756219744682312,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 74330
+    },
+    {
+      "epoch": 489.07894736842104,
+      "grad_norm": 1.0228742361068726,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 74340
+    },
+    {
+      "epoch": 489.14473684210526,
+      "grad_norm": 0.9281026721000671,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 74350
+    },
+    {
+      "epoch": 489.2105263157895,
+      "grad_norm": 0.9329313039779663,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 74360
+    },
+    {
+      "epoch": 489.2763157894737,
+      "grad_norm": 0.8658884167671204,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 74370
+    },
+    {
+      "epoch": 489.3421052631579,
+      "grad_norm": 0.9127900004386902,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 74380
+    },
+    {
+      "epoch": 489.4078947368421,
+      "grad_norm": 1.145900011062622,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 74390
+    },
+    {
+      "epoch": 489.4736842105263,
+      "grad_norm": 1.131519079208374,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 74400
+    },
+    {
+      "epoch": 489.5394736842105,
+      "grad_norm": 1.071823000907898,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 74410
+    },
+    {
+      "epoch": 489.60526315789474,
+      "grad_norm": 0.8739954233169556,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 74420
+    },
+    {
+      "epoch": 489.67105263157896,
+      "grad_norm": 1.0112184286117554,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 74430
+    },
+    {
+      "epoch": 489.7368421052632,
+      "grad_norm": 0.9393345713615417,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 74440
+    },
+    {
+      "epoch": 489.80263157894734,
+      "grad_norm": 1.0487806797027588,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 74450
+    },
+    {
+      "epoch": 489.86842105263156,
+      "grad_norm": 1.1946709156036377,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 74460
+    },
+    {
+      "epoch": 489.9342105263158,
+      "grad_norm": 1.154524803161621,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 74470
+    },
+    {
+      "epoch": 490.0,
+      "grad_norm": 1.5723423957824707,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 74480
+    },
+    {
+      "epoch": 490.0657894736842,
+      "grad_norm": 1.4050586223602295,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 74490
+    },
+    {
+      "epoch": 490.13157894736844,
+      "grad_norm": 1.064405918121338,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 74500
+    },
+    {
+      "epoch": 490.19736842105266,
+      "grad_norm": 1.1484532356262207,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 74510
+    },
+    {
+      "epoch": 490.2631578947368,
+      "grad_norm": 0.8688686490058899,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 74520
+    },
+    {
+      "epoch": 490.32894736842104,
+      "grad_norm": 1.151997447013855,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 74530
+    },
+    {
+      "epoch": 490.39473684210526,
+      "grad_norm": 1.0896145105361938,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 74540
+    },
+    {
+      "epoch": 490.4605263157895,
+      "grad_norm": 1.170395016670227,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 74550
+    },
+    {
+      "epoch": 490.5263157894737,
+      "grad_norm": 1.2518529891967773,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 74560
+    },
+    {
+      "epoch": 490.5921052631579,
+      "grad_norm": 1.385285496711731,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 74570
+    },
+    {
+      "epoch": 490.6578947368421,
+      "grad_norm": 1.2214215993881226,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 74580
+    },
+    {
+      "epoch": 490.7236842105263,
+      "grad_norm": 1.1095465421676636,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 74590
+    },
+    {
+      "epoch": 490.7894736842105,
+      "grad_norm": 1.1326744556427002,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 74600
+    },
+    {
+      "epoch": 490.85526315789474,
+      "grad_norm": 1.0107364654541016,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 74610
+    },
+    {
+      "epoch": 490.92105263157896,
+      "grad_norm": 1.571817398071289,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 74620
+    },
+    {
+      "epoch": 490.9868421052632,
+      "grad_norm": 1.312083125114441,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 74630
+    },
+    {
+      "epoch": 491.05263157894734,
+      "grad_norm": 1.4065839052200317,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 74640
+    },
+    {
+      "epoch": 491.11842105263156,
+      "grad_norm": 1.1010332107543945,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 74650
+    },
+    {
+      "epoch": 491.1842105263158,
+      "grad_norm": 1.1380709409713745,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 74660
+    },
+    {
+      "epoch": 491.25,
+      "grad_norm": 1.1120402812957764,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 74670
+    },
+    {
+      "epoch": 491.3157894736842,
+      "grad_norm": 1.5647122859954834,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 74680
+    },
+    {
+      "epoch": 491.38157894736844,
+      "grad_norm": 1.1048648357391357,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 74690
+    },
+    {
+      "epoch": 491.44736842105266,
+      "grad_norm": 1.0886578559875488,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 74700
+    },
+    {
+      "epoch": 491.5131578947368,
+      "grad_norm": 1.111810564994812,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 74710
+    },
+    {
+      "epoch": 491.57894736842104,
+      "grad_norm": 1.0857470035552979,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 74720
+    },
+    {
+      "epoch": 491.64473684210526,
+      "grad_norm": 1.4645519256591797,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 74730
+    },
+    {
+      "epoch": 491.7105263157895,
+      "grad_norm": 1.5546987056732178,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 74740
+    },
+    {
+      "epoch": 491.7763157894737,
+      "grad_norm": 1.0544028282165527,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 74750
+    },
+    {
+      "epoch": 491.8421052631579,
+      "grad_norm": 1.1025646924972534,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 74760
+    },
+    {
+      "epoch": 491.9078947368421,
+      "grad_norm": 1.238974928855896,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 74770
+    },
+    {
+      "epoch": 491.9736842105263,
+      "grad_norm": 1.3790466785430908,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 74780
+    },
+    {
+      "epoch": 492.0394736842105,
+      "grad_norm": 1.2064578533172607,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 74790
+    },
+    {
+      "epoch": 492.10526315789474,
+      "grad_norm": 0.9122753143310547,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 74800
+    },
+    {
+      "epoch": 492.17105263157896,
+      "grad_norm": 1.4887069463729858,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 74810
+    },
+    {
+      "epoch": 492.2368421052632,
+      "grad_norm": 1.2702773809432983,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 74820
+    },
+    {
+      "epoch": 492.30263157894734,
+      "grad_norm": 1.2110621929168701,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 74830
+    },
+    {
+      "epoch": 492.36842105263156,
+      "grad_norm": 1.53854238986969,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 74840
+    },
+    {
+      "epoch": 492.4342105263158,
+      "grad_norm": 1.8626569509506226,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 74850
+    },
+    {
+      "epoch": 492.5,
+      "grad_norm": 1.666455864906311,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 74860
+    },
+    {
+      "epoch": 492.5657894736842,
+      "grad_norm": 1.684195876121521,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 74870
+    },
+    {
+      "epoch": 492.63157894736844,
+      "grad_norm": 1.1221078634262085,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 74880
+    },
+    {
+      "epoch": 492.69736842105266,
+      "grad_norm": 1.2585291862487793,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 74890
+    },
+    {
+      "epoch": 492.7631578947368,
+      "grad_norm": 1.376713514328003,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 74900
+    },
+    {
+      "epoch": 492.82894736842104,
+      "grad_norm": 0.9732930064201355,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 74910
+    },
+    {
+      "epoch": 492.89473684210526,
+      "grad_norm": 1.3538066148757935,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 74920
+    },
+    {
+      "epoch": 492.9605263157895,
+      "grad_norm": 1.5460445880889893,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 74930
+    },
+    {
+      "epoch": 493.0263157894737,
+      "grad_norm": 1.332470417022705,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 74940
+    },
+    {
+      "epoch": 493.0921052631579,
+      "grad_norm": 1.0886743068695068,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 74950
+    },
+    {
+      "epoch": 493.1578947368421,
+      "grad_norm": 0.8579538464546204,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 74960
+    },
+    {
+      "epoch": 493.2236842105263,
+      "grad_norm": 1.033247470855713,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 74970
+    },
+    {
+      "epoch": 493.2894736842105,
+      "grad_norm": 1.0347689390182495,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 74980
+    },
+    {
+      "epoch": 493.35526315789474,
+      "grad_norm": 1.149769902229309,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 74990
+    },
+    {
+      "epoch": 493.42105263157896,
+      "grad_norm": 0.8017956614494324,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 75000
+    },
+    {
+      "epoch": 493.4868421052632,
+      "grad_norm": 0.933280348777771,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 75010
+    },
+    {
+      "epoch": 493.55263157894734,
+      "grad_norm": 1.4169752597808838,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 75020
+    },
+    {
+      "epoch": 493.61842105263156,
+      "grad_norm": 1.0053260326385498,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 75030
+    },
+    {
+      "epoch": 493.6842105263158,
+      "grad_norm": 1.0985119342803955,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 75040
+    },
+    {
+      "epoch": 493.75,
+      "grad_norm": 0.89968341588974,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 75050
+    },
+    {
+      "epoch": 493.8157894736842,
+      "grad_norm": 1.0092484951019287,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 75060
+    },
+    {
+      "epoch": 493.88157894736844,
+      "grad_norm": 0.8683974742889404,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 75070
+    },
+    {
+      "epoch": 493.94736842105266,
+      "grad_norm": 1.1474254131317139,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 75080
+    },
+    {
+      "epoch": 494.0131578947368,
+      "grad_norm": 0.9479317665100098,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 75090
+    },
+    {
+      "epoch": 494.07894736842104,
+      "grad_norm": 1.194884181022644,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 75100
+    },
+    {
+      "epoch": 494.14473684210526,
+      "grad_norm": 1.3148432970046997,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 75110
+    },
+    {
+      "epoch": 494.2105263157895,
+      "grad_norm": 1.1125714778900146,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 75120
+    },
+    {
+      "epoch": 494.2763157894737,
+      "grad_norm": 1.0506175756454468,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 75130
+    },
+    {
+      "epoch": 494.3421052631579,
+      "grad_norm": 1.3244073390960693,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 75140
+    },
+    {
+      "epoch": 494.4078947368421,
+      "grad_norm": 1.082236647605896,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 75150
+    },
+    {
+      "epoch": 494.4736842105263,
+      "grad_norm": 1.702755093574524,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 75160
+    },
+    {
+      "epoch": 494.5394736842105,
+      "grad_norm": 1.0720016956329346,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 75170
+    },
+    {
+      "epoch": 494.60526315789474,
+      "grad_norm": 0.8905903100967407,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 75180
+    },
+    {
+      "epoch": 494.67105263157896,
+      "grad_norm": 0.9259252548217773,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 75190
+    },
+    {
+      "epoch": 494.7368421052632,
+      "grad_norm": 1.1381369829177856,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 75200
+    },
+    {
+      "epoch": 494.80263157894734,
+      "grad_norm": 1.1923221349716187,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 75210
+    },
+    {
+      "epoch": 494.86842105263156,
+      "grad_norm": 1.3231667280197144,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 75220
+    },
+    {
+      "epoch": 494.9342105263158,
+      "grad_norm": 1.1546229124069214,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 75230
+    },
+    {
+      "epoch": 495.0,
+      "grad_norm": 0.942509114742279,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 75240
+    },
+    {
+      "epoch": 495.0657894736842,
+      "grad_norm": 1.1736472845077515,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 75250
+    },
+    {
+      "epoch": 495.13157894736844,
+      "grad_norm": 1.43409264087677,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 75260
+    },
+    {
+      "epoch": 495.19736842105266,
+      "grad_norm": 1.1117990016937256,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 75270
+    },
+    {
+      "epoch": 495.2631578947368,
+      "grad_norm": 1.196996808052063,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 75280
+    },
+    {
+      "epoch": 495.32894736842104,
+      "grad_norm": 1.3209561109542847,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 75290
+    },
+    {
+      "epoch": 495.39473684210526,
+      "grad_norm": 1.633660078048706,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 75300
+    },
+    {
+      "epoch": 495.4605263157895,
+      "grad_norm": 1.3215038776397705,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 75310
+    },
+    {
+      "epoch": 495.5263157894737,
+      "grad_norm": 1.1978789567947388,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 75320
+    },
+    {
+      "epoch": 495.5921052631579,
+      "grad_norm": 1.508163332939148,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 75330
+    },
+    {
+      "epoch": 495.6578947368421,
+      "grad_norm": 1.5541365146636963,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 75340
+    },
+    {
+      "epoch": 495.7236842105263,
+      "grad_norm": 1.6904091835021973,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 75350
+    },
+    {
+      "epoch": 495.7894736842105,
+      "grad_norm": 1.0997204780578613,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 75360
+    },
+    {
+      "epoch": 495.85526315789474,
+      "grad_norm": 1.1885972023010254,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 75370
+    },
+    {
+      "epoch": 495.92105263157896,
+      "grad_norm": 1.2370365858078003,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 75380
+    },
+    {
+      "epoch": 495.9868421052632,
+      "grad_norm": 1.3986064195632935,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 75390
+    },
+    {
+      "epoch": 496.05263157894734,
+      "grad_norm": 1.3072096109390259,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 75400
+    },
+    {
+      "epoch": 496.11842105263156,
+      "grad_norm": 1.2789565324783325,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 75410
+    },
+    {
+      "epoch": 496.1842105263158,
+      "grad_norm": 1.0061566829681396,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 75420
+    },
+    {
+      "epoch": 496.25,
+      "grad_norm": 0.9549350738525391,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 75430
+    },
+    {
+      "epoch": 496.3157894736842,
+      "grad_norm": 0.9655198454856873,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 75440
+    },
+    {
+      "epoch": 496.38157894736844,
+      "grad_norm": 0.9944902062416077,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 75450
+    },
+    {
+      "epoch": 496.44736842105266,
+      "grad_norm": 1.0632195472717285,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 75460
+    },
+    {
+      "epoch": 496.5131578947368,
+      "grad_norm": 1.0134518146514893,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 75470
+    },
+    {
+      "epoch": 496.57894736842104,
+      "grad_norm": 1.3689377307891846,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 75480
+    },
+    {
+      "epoch": 496.64473684210526,
+      "grad_norm": 1.2085026502609253,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 75490
+    },
+    {
+      "epoch": 496.7105263157895,
+      "grad_norm": 0.7359601259231567,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 75500
+    },
+    {
+      "epoch": 496.7763157894737,
+      "grad_norm": 1.2168084383010864,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 75510
+    },
+    {
+      "epoch": 496.8421052631579,
+      "grad_norm": 0.99058598279953,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 75520
+    },
+    {
+      "epoch": 496.9078947368421,
+      "grad_norm": 0.9866731762886047,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 75530
+    },
+    {
+      "epoch": 496.9736842105263,
+      "grad_norm": 0.7462350726127625,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 75540
+    },
+    {
+      "epoch": 497.0394736842105,
+      "grad_norm": 0.8804934620857239,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 75550
+    },
+    {
+      "epoch": 497.10526315789474,
+      "grad_norm": 0.9820599555969238,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 75560
+    },
+    {
+      "epoch": 497.17105263157896,
+      "grad_norm": 0.8656883239746094,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 75570
+    },
+    {
+      "epoch": 497.2368421052632,
+      "grad_norm": 0.9848073124885559,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 75580
+    },
+    {
+      "epoch": 497.30263157894734,
+      "grad_norm": 0.7762583494186401,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 75590
+    },
+    {
+      "epoch": 497.36842105263156,
+      "grad_norm": 0.9925325512886047,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 75600
+    },
+    {
+      "epoch": 497.4342105263158,
+      "grad_norm": 0.905873715877533,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 75610
+    },
+    {
+      "epoch": 497.5,
+      "grad_norm": 0.8699862957000732,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 75620
+    },
+    {
+      "epoch": 497.5657894736842,
+      "grad_norm": 1.1519784927368164,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 75630
+    },
+    {
+      "epoch": 497.63157894736844,
+      "grad_norm": 1.285287857055664,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 75640
+    },
+    {
+      "epoch": 497.69736842105266,
+      "grad_norm": 0.9361952543258667,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 75650
+    },
+    {
+      "epoch": 497.7631578947368,
+      "grad_norm": 1.113816499710083,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 75660
+    },
+    {
+      "epoch": 497.82894736842104,
+      "grad_norm": 1.1863288879394531,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 75670
+    },
+    {
+      "epoch": 497.89473684210526,
+      "grad_norm": 0.6203667521476746,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 75680
+    },
+    {
+      "epoch": 497.9605263157895,
+      "grad_norm": 1.0878044366836548,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 75690
+    },
+    {
+      "epoch": 498.0263157894737,
+      "grad_norm": 0.5921301245689392,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 75700
+    },
+    {
+      "epoch": 498.0921052631579,
+      "grad_norm": 0.9742683172225952,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 75710
+    },
+    {
+      "epoch": 498.1578947368421,
+      "grad_norm": 1.1154931783676147,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 75720
+    },
+    {
+      "epoch": 498.2236842105263,
+      "grad_norm": 1.0638843774795532,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 75730
+    },
+    {
+      "epoch": 498.2894736842105,
+      "grad_norm": 1.2512115240097046,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 75740
+    },
+    {
+      "epoch": 498.35526315789474,
+      "grad_norm": 1.373955488204956,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 75750
+    },
+    {
+      "epoch": 498.42105263157896,
+      "grad_norm": 1.1924251317977905,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 75760
+    },
+    {
+      "epoch": 498.4868421052632,
+      "grad_norm": 1.0624274015426636,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 75770
+    },
+    {
+      "epoch": 498.55263157894734,
+      "grad_norm": 0.7194404602050781,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 75780
+    },
+    {
+      "epoch": 498.61842105263156,
+      "grad_norm": 1.0660557746887207,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 75790
+    },
+    {
+      "epoch": 498.6842105263158,
+      "grad_norm": 1.237527847290039,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 75800
+    },
+    {
+      "epoch": 498.75,
+      "grad_norm": 1.1295133829116821,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 75810
+    },
+    {
+      "epoch": 498.8157894736842,
+      "grad_norm": 0.8984070420265198,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 75820
+    },
+    {
+      "epoch": 498.88157894736844,
+      "grad_norm": 1.035714864730835,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 75830
+    },
+    {
+      "epoch": 498.94736842105266,
+      "grad_norm": 1.1130855083465576,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 75840
+    },
+    {
+      "epoch": 499.0131578947368,
+      "grad_norm": 1.1051881313323975,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 75850
+    },
+    {
+      "epoch": 499.07894736842104,
+      "grad_norm": 0.8507428765296936,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 75860
+    },
+    {
+      "epoch": 499.14473684210526,
+      "grad_norm": 0.9365002512931824,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 75870
+    },
+    {
+      "epoch": 499.2105263157895,
+      "grad_norm": 1.0836772918701172,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 75880
+    },
+    {
+      "epoch": 499.2763157894737,
+      "grad_norm": 1.0799518823623657,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 75890
+    },
+    {
+      "epoch": 499.3421052631579,
+      "grad_norm": 1.1762809753417969,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 75900
+    },
+    {
+      "epoch": 499.4078947368421,
+      "grad_norm": 0.9116426110267639,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 75910
+    },
+    {
+      "epoch": 499.4736842105263,
+      "grad_norm": 1.1221206188201904,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 75920
+    },
+    {
+      "epoch": 499.5394736842105,
+      "grad_norm": 0.9344303011894226,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 75930
+    },
+    {
+      "epoch": 499.60526315789474,
+      "grad_norm": 1.2901678085327148,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 75940
+    },
+    {
+      "epoch": 499.67105263157896,
+      "grad_norm": 1.0376948118209839,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 75950
+    },
+    {
+      "epoch": 499.7368421052632,
+      "grad_norm": 0.9793575406074524,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 75960
+    },
+    {
+      "epoch": 499.80263157894734,
+      "grad_norm": 1.1645046472549438,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 75970
+    },
+    {
+      "epoch": 499.86842105263156,
+      "grad_norm": 0.8754516243934631,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 75980
+    },
+    {
+      "epoch": 499.9342105263158,
+      "grad_norm": 1.0461459159851074,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 75990
+    },
+    {
+      "epoch": 500.0,
+      "grad_norm": 0.9663628339767456,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 76000
+    },
+    {
+      "epoch": 500.0657894736842,
+      "grad_norm": 1.32105553150177,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 76010
+    },
+    {
+      "epoch": 500.13157894736844,
+      "grad_norm": 1.3787779808044434,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 76020
+    },
+    {
+      "epoch": 500.19736842105266,
+      "grad_norm": 1.2713334560394287,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 76030
+    },
+    {
+      "epoch": 500.2631578947368,
+      "grad_norm": 1.2451242208480835,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 76040
+    },
+    {
+      "epoch": 500.32894736842104,
+      "grad_norm": 1.1290003061294556,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 76050
+    },
+    {
+      "epoch": 500.39473684210526,
+      "grad_norm": 1.2039153575897217,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 76060
+    },
+    {
+      "epoch": 500.4605263157895,
+      "grad_norm": 1.2840030193328857,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 76070
+    },
+    {
+      "epoch": 500.5263157894737,
+      "grad_norm": 1.0132211446762085,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 76080
+    },
+    {
+      "epoch": 500.5921052631579,
+      "grad_norm": 0.8630693554878235,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 76090
+    },
+    {
+      "epoch": 500.6578947368421,
+      "grad_norm": 1.200491189956665,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 76100
+    },
+    {
+      "epoch": 500.7236842105263,
+      "grad_norm": 1.0840117931365967,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 76110
+    },
+    {
+      "epoch": 500.7894736842105,
+      "grad_norm": 1.058173418045044,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 76120
+    },
+    {
+      "epoch": 500.85526315789474,
+      "grad_norm": 0.7691683769226074,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 76130
+    },
+    {
+      "epoch": 500.92105263157896,
+      "grad_norm": 1.3571420907974243,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 76140
+    },
+    {
+      "epoch": 500.9868421052632,
+      "grad_norm": 1.397667646408081,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 76150
+    },
+    {
+      "epoch": 501.05263157894734,
+      "grad_norm": 1.2376160621643066,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 76160
+    },
+    {
+      "epoch": 501.11842105263156,
+      "grad_norm": 1.193163514137268,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 76170
+    },
+    {
+      "epoch": 501.1842105263158,
+      "grad_norm": 1.4034628868103027,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 76180
+    },
+    {
+      "epoch": 501.25,
+      "grad_norm": 1.207105278968811,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 76190
+    },
+    {
+      "epoch": 501.3157894736842,
+      "grad_norm": 0.7556562423706055,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 76200
+    },
+    {
+      "epoch": 501.38157894736844,
+      "grad_norm": 1.2998090982437134,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 76210
+    },
+    {
+      "epoch": 501.44736842105266,
+      "grad_norm": 0.9922952055931091,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 76220
+    },
+    {
+      "epoch": 501.5131578947368,
+      "grad_norm": 1.225348949432373,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 76230
+    },
+    {
+      "epoch": 501.57894736842104,
+      "grad_norm": 1.3138529062271118,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 76240
+    },
+    {
+      "epoch": 501.64473684210526,
+      "grad_norm": 0.9411857724189758,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 76250
+    },
+    {
+      "epoch": 501.7105263157895,
+      "grad_norm": 1.271523356437683,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 76260
+    },
+    {
+      "epoch": 501.7763157894737,
+      "grad_norm": 1.2197598218917847,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 76270
+    },
+    {
+      "epoch": 501.8421052631579,
+      "grad_norm": 1.3674578666687012,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 76280
+    },
+    {
+      "epoch": 501.9078947368421,
+      "grad_norm": 1.3369501829147339,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 76290
+    },
+    {
+      "epoch": 501.9736842105263,
+      "grad_norm": 1.6227846145629883,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 76300
+    },
+    {
+      "epoch": 502.0394736842105,
+      "grad_norm": 1.3651342391967773,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 76310
+    },
+    {
+      "epoch": 502.10526315789474,
+      "grad_norm": 1.1167755126953125,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 76320
+    },
+    {
+      "epoch": 502.17105263157896,
+      "grad_norm": 1.0122556686401367,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 76330
+    },
+    {
+      "epoch": 502.2368421052632,
+      "grad_norm": 1.339971899986267,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 76340
+    },
+    {
+      "epoch": 502.30263157894734,
+      "grad_norm": 1.1776334047317505,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 76350
+    },
+    {
+      "epoch": 502.36842105263156,
+      "grad_norm": 1.1204994916915894,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 76360
+    },
+    {
+      "epoch": 502.4342105263158,
+      "grad_norm": 1.1037061214447021,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 76370
+    },
+    {
+      "epoch": 502.5,
+      "grad_norm": 1.4613289833068848,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 76380
+    },
+    {
+      "epoch": 502.5657894736842,
+      "grad_norm": 1.3934388160705566,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 76390
+    },
+    {
+      "epoch": 502.63157894736844,
+      "grad_norm": 1.470294713973999,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 76400
+    },
+    {
+      "epoch": 502.69736842105266,
+      "grad_norm": 1.2108807563781738,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 76410
+    },
+    {
+      "epoch": 502.7631578947368,
+      "grad_norm": 1.1446878910064697,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 76420
+    },
+    {
+      "epoch": 502.82894736842104,
+      "grad_norm": 1.1863682270050049,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 76430
+    },
+    {
+      "epoch": 502.89473684210526,
+      "grad_norm": 1.2745025157928467,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 76440
+    },
+    {
+      "epoch": 502.9605263157895,
+      "grad_norm": 1.2708007097244263,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 76450
+    },
+    {
+      "epoch": 503.0263157894737,
+      "grad_norm": 1.109881043434143,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 76460
+    },
+    {
+      "epoch": 503.0921052631579,
+      "grad_norm": 0.837130606174469,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 76470
+    },
+    {
+      "epoch": 503.1578947368421,
+      "grad_norm": 1.1610591411590576,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 76480
+    },
+    {
+      "epoch": 503.2236842105263,
+      "grad_norm": 0.9062598943710327,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 76490
+    },
+    {
+      "epoch": 503.2894736842105,
+      "grad_norm": 0.880432665348053,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 76500
+    },
+    {
+      "epoch": 503.35526315789474,
+      "grad_norm": 1.1003761291503906,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 76510
+    },
+    {
+      "epoch": 503.42105263157896,
+      "grad_norm": 1.0578047037124634,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 76520
+    },
+    {
+      "epoch": 503.4868421052632,
+      "grad_norm": 1.265828251838684,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 76530
+    },
+    {
+      "epoch": 503.55263157894734,
+      "grad_norm": 0.7461774945259094,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 76540
+    },
+    {
+      "epoch": 503.61842105263156,
+      "grad_norm": 1.1446738243103027,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 76550
+    },
+    {
+      "epoch": 503.6842105263158,
+      "grad_norm": 1.0880557298660278,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 76560
+    },
+    {
+      "epoch": 503.75,
+      "grad_norm": 1.399858832359314,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 76570
+    },
+    {
+      "epoch": 503.8157894736842,
+      "grad_norm": 1.1367653608322144,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 76580
+    },
+    {
+      "epoch": 503.88157894736844,
+      "grad_norm": 1.018934965133667,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 76590
+    },
+    {
+      "epoch": 503.94736842105266,
+      "grad_norm": 1.066989779472351,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 76600
+    },
+    {
+      "epoch": 504.0131578947368,
+      "grad_norm": 0.9867700338363647,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 76610
+    },
+    {
+      "epoch": 504.07894736842104,
+      "grad_norm": 1.0449360609054565,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 76620
+    },
+    {
+      "epoch": 504.14473684210526,
+      "grad_norm": 0.7353350520133972,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 76630
+    },
+    {
+      "epoch": 504.2105263157895,
+      "grad_norm": 0.8877101540565491,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 76640
+    },
+    {
+      "epoch": 504.2763157894737,
+      "grad_norm": 1.084471583366394,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 76650
+    },
+    {
+      "epoch": 504.3421052631579,
+      "grad_norm": 1.1777323484420776,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 76660
+    },
+    {
+      "epoch": 504.4078947368421,
+      "grad_norm": 1.193827748298645,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 76670
+    },
+    {
+      "epoch": 504.4736842105263,
+      "grad_norm": 1.2224369049072266,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 76680
+    },
+    {
+      "epoch": 504.5394736842105,
+      "grad_norm": 1.1256766319274902,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 76690
+    },
+    {
+      "epoch": 504.60526315789474,
+      "grad_norm": 1.0881433486938477,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 76700
+    },
+    {
+      "epoch": 504.67105263157896,
+      "grad_norm": 0.9593660831451416,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 76710
+    },
+    {
+      "epoch": 504.7368421052632,
+      "grad_norm": 1.1365207433700562,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 76720
+    },
+    {
+      "epoch": 504.80263157894734,
+      "grad_norm": 1.1816807985305786,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 76730
+    },
+    {
+      "epoch": 504.86842105263156,
+      "grad_norm": 1.0027751922607422,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 76740
+    },
+    {
+      "epoch": 504.9342105263158,
+      "grad_norm": 0.7428399920463562,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 76750
+    },
+    {
+      "epoch": 505.0,
+      "grad_norm": 1.1550109386444092,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 76760
+    },
+    {
+      "epoch": 505.0657894736842,
+      "grad_norm": 0.9820820093154907,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 76770
+    },
+    {
+      "epoch": 505.13157894736844,
+      "grad_norm": 1.3091676235198975,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 76780
+    },
+    {
+      "epoch": 505.19736842105266,
+      "grad_norm": 1.0327069759368896,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 76790
+    },
+    {
+      "epoch": 505.2631578947368,
+      "grad_norm": 0.9336044788360596,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 76800
+    },
+    {
+      "epoch": 505.32894736842104,
+      "grad_norm": 1.4007383584976196,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 76810
+    },
+    {
+      "epoch": 505.39473684210526,
+      "grad_norm": 1.2266377210617065,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 76820
+    },
+    {
+      "epoch": 505.4605263157895,
+      "grad_norm": 1.1764065027236938,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 76830
+    },
+    {
+      "epoch": 505.5263157894737,
+      "grad_norm": 1.091961145401001,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 76840
+    },
+    {
+      "epoch": 505.5921052631579,
+      "grad_norm": 0.9492750763893127,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 76850
+    },
+    {
+      "epoch": 505.6578947368421,
+      "grad_norm": 1.088575839996338,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 76860
+    },
+    {
+      "epoch": 505.7236842105263,
+      "grad_norm": 0.8977634906768799,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 76870
+    },
+    {
+      "epoch": 505.7894736842105,
+      "grad_norm": 0.9890562891960144,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 76880
+    },
+    {
+      "epoch": 505.85526315789474,
+      "grad_norm": 0.8245396018028259,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 76890
+    },
+    {
+      "epoch": 505.92105263157896,
+      "grad_norm": 0.810986340045929,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 76900
+    },
+    {
+      "epoch": 505.9868421052632,
+      "grad_norm": 1.3287197351455688,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 76910
+    },
+    {
+      "epoch": 506.05263157894734,
+      "grad_norm": 0.7860926985740662,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 76920
+    },
+    {
+      "epoch": 506.11842105263156,
+      "grad_norm": 1.179496169090271,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 76930
+    },
+    {
+      "epoch": 506.1842105263158,
+      "grad_norm": 1.606441617012024,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 76940
+    },
+    {
+      "epoch": 506.25,
+      "grad_norm": 1.499569058418274,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 76950
+    },
+    {
+      "epoch": 506.3157894736842,
+      "grad_norm": 1.24739670753479,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 76960
+    },
+    {
+      "epoch": 506.38157894736844,
+      "grad_norm": 1.2441381216049194,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 76970
+    },
+    {
+      "epoch": 506.44736842105266,
+      "grad_norm": 0.8672195672988892,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 76980
+    },
+    {
+      "epoch": 506.5131578947368,
+      "grad_norm": 0.9006538391113281,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 76990
+    },
+    {
+      "epoch": 506.57894736842104,
+      "grad_norm": 0.9952192902565002,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 77000
+    },
+    {
+      "epoch": 506.64473684210526,
+      "grad_norm": 0.8276636004447937,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 77010
+    },
+    {
+      "epoch": 506.7105263157895,
+      "grad_norm": 1.2697261571884155,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 77020
+    },
+    {
+      "epoch": 506.7763157894737,
+      "grad_norm": 1.3981074094772339,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 77030
+    },
+    {
+      "epoch": 506.8421052631579,
+      "grad_norm": 1.0681496858596802,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 77040
+    },
+    {
+      "epoch": 506.9078947368421,
+      "grad_norm": 0.8381850719451904,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 77050
+    },
+    {
+      "epoch": 506.9736842105263,
+      "grad_norm": 1.236061692237854,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 77060
+    },
+    {
+      "epoch": 507.0394736842105,
+      "grad_norm": 0.8843864798545837,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 77070
+    },
+    {
+      "epoch": 507.10526315789474,
+      "grad_norm": 1.0024884939193726,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 77080
+    },
+    {
+      "epoch": 507.17105263157896,
+      "grad_norm": 1.0010193586349487,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 77090
+    },
+    {
+      "epoch": 507.2368421052632,
+      "grad_norm": 1.3840446472167969,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 77100
+    },
+    {
+      "epoch": 507.30263157894734,
+      "grad_norm": 0.9459224343299866,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 77110
+    },
+    {
+      "epoch": 507.36842105263156,
+      "grad_norm": 1.2796701192855835,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 77120
+    },
+    {
+      "epoch": 507.4342105263158,
+      "grad_norm": 0.9489369988441467,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 77130
+    },
+    {
+      "epoch": 507.5,
+      "grad_norm": 0.9329638481140137,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 77140
+    },
+    {
+      "epoch": 507.5657894736842,
+      "grad_norm": 1.1059828996658325,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 77150
+    },
+    {
+      "epoch": 507.63157894736844,
+      "grad_norm": 0.9566002488136292,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 77160
+    },
+    {
+      "epoch": 507.69736842105266,
+      "grad_norm": 1.44321608543396,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 77170
+    },
+    {
+      "epoch": 507.7631578947368,
+      "grad_norm": 0.9149665832519531,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 77180
+    },
+    {
+      "epoch": 507.82894736842104,
+      "grad_norm": 1.2634365558624268,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 77190
+    },
+    {
+      "epoch": 507.89473684210526,
+      "grad_norm": 1.0077171325683594,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 77200
+    },
+    {
+      "epoch": 507.9605263157895,
+      "grad_norm": 1.3527032136917114,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 77210
+    },
+    {
+      "epoch": 508.0263157894737,
+      "grad_norm": 1.3523765802383423,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 77220
+    },
+    {
+      "epoch": 508.0921052631579,
+      "grad_norm": 1.2622536420822144,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 77230
+    },
+    {
+      "epoch": 508.1578947368421,
+      "grad_norm": 1.2173060178756714,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 77240
+    },
+    {
+      "epoch": 508.2236842105263,
+      "grad_norm": 1.077372431755066,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 77250
+    },
+    {
+      "epoch": 508.2894736842105,
+      "grad_norm": 1.0975558757781982,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 77260
+    },
+    {
+      "epoch": 508.35526315789474,
+      "grad_norm": 1.3480600118637085,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 77270
+    },
+    {
+      "epoch": 508.42105263157896,
+      "grad_norm": 1.1693116426467896,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 77280
+    },
+    {
+      "epoch": 508.4868421052632,
+      "grad_norm": 1.1305088996887207,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 77290
+    },
+    {
+      "epoch": 508.55263157894734,
+      "grad_norm": 1.2774187326431274,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 77300
+    },
+    {
+      "epoch": 508.61842105263156,
+      "grad_norm": 1.1564953327178955,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 77310
+    },
+    {
+      "epoch": 508.6842105263158,
+      "grad_norm": 0.8752122521400452,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 77320
+    },
+    {
+      "epoch": 508.75,
+      "grad_norm": 1.2619119882583618,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 77330
+    },
+    {
+      "epoch": 508.8157894736842,
+      "grad_norm": 0.9893479943275452,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 77340
+    },
+    {
+      "epoch": 508.88157894736844,
+      "grad_norm": 1.0533159971237183,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 77350
+    },
+    {
+      "epoch": 508.94736842105266,
+      "grad_norm": 1.032997488975525,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 77360
+    },
+    {
+      "epoch": 509.0131578947368,
+      "grad_norm": 1.0692557096481323,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 77370
+    },
+    {
+      "epoch": 509.07894736842104,
+      "grad_norm": 0.8767366409301758,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 77380
+    },
+    {
+      "epoch": 509.14473684210526,
+      "grad_norm": 1.1660325527191162,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 77390
+    },
+    {
+      "epoch": 509.2105263157895,
+      "grad_norm": 0.8173626661300659,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 77400
+    },
+    {
+      "epoch": 509.2763157894737,
+      "grad_norm": 0.7611241936683655,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 77410
+    },
+    {
+      "epoch": 509.3421052631579,
+      "grad_norm": 1.0339752435684204,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 77420
+    },
+    {
+      "epoch": 509.4078947368421,
+      "grad_norm": 1.1537483930587769,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 77430
+    },
+    {
+      "epoch": 509.4736842105263,
+      "grad_norm": 1.1419742107391357,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 77440
+    },
+    {
+      "epoch": 509.5394736842105,
+      "grad_norm": 1.0991747379302979,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 77450
+    },
+    {
+      "epoch": 509.60526315789474,
+      "grad_norm": 1.048770785331726,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 77460
+    },
+    {
+      "epoch": 509.67105263157896,
+      "grad_norm": 1.1474910974502563,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 77470
+    },
+    {
+      "epoch": 509.7368421052632,
+      "grad_norm": 1.519187331199646,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 77480
+    },
+    {
+      "epoch": 509.80263157894734,
+      "grad_norm": 1.7529782056808472,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 77490
+    },
+    {
+      "epoch": 509.86842105263156,
+      "grad_norm": 1.0271803140640259,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 77500
+    },
+    {
+      "epoch": 509.9342105263158,
+      "grad_norm": 1.1275503635406494,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 77510
+    },
+    {
+      "epoch": 510.0,
+      "grad_norm": 1.0205459594726562,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 77520
+    },
+    {
+      "epoch": 510.0657894736842,
+      "grad_norm": 1.088582158088684,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 77530
+    },
+    {
+      "epoch": 510.13157894736844,
+      "grad_norm": 0.9400649070739746,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 77540
+    },
+    {
+      "epoch": 510.19736842105266,
+      "grad_norm": 1.1834319829940796,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 77550
+    },
+    {
+      "epoch": 510.2631578947368,
+      "grad_norm": 1.0135151147842407,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 77560
+    },
+    {
+      "epoch": 510.32894736842104,
+      "grad_norm": 1.5176630020141602,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 77570
+    },
+    {
+      "epoch": 510.39473684210526,
+      "grad_norm": 1.283932089805603,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 77580
+    },
+    {
+      "epoch": 510.4605263157895,
+      "grad_norm": 1.1694949865341187,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 77590
+    },
+    {
+      "epoch": 510.5263157894737,
+      "grad_norm": 0.772484540939331,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 77600
+    },
+    {
+      "epoch": 510.5921052631579,
+      "grad_norm": 1.1480525732040405,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 77610
+    },
+    {
+      "epoch": 510.6578947368421,
+      "grad_norm": 1.1137710809707642,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 77620
+    },
+    {
+      "epoch": 510.7236842105263,
+      "grad_norm": 0.9193239212036133,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 77630
+    },
+    {
+      "epoch": 510.7894736842105,
+      "grad_norm": 0.9958206415176392,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 77640
+    },
+    {
+      "epoch": 510.85526315789474,
+      "grad_norm": 0.8010199069976807,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 77650
+    },
+    {
+      "epoch": 510.92105263157896,
+      "grad_norm": 1.4143106937408447,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 77660
+    },
+    {
+      "epoch": 510.9868421052632,
+      "grad_norm": 0.9478830695152283,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 77670
+    },
+    {
+      "epoch": 511.05263157894734,
+      "grad_norm": 1.2224031686782837,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 77680
+    },
+    {
+      "epoch": 511.11842105263156,
+      "grad_norm": 1.0343413352966309,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 77690
+    },
+    {
+      "epoch": 511.1842105263158,
+      "grad_norm": 1.211520791053772,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 77700
+    },
+    {
+      "epoch": 511.25,
+      "grad_norm": 1.017286777496338,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 77710
+    },
+    {
+      "epoch": 511.3157894736842,
+      "grad_norm": 0.9983707666397095,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 77720
+    },
+    {
+      "epoch": 511.38157894736844,
+      "grad_norm": 1.5086129903793335,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 77730
+    },
+    {
+      "epoch": 511.44736842105266,
+      "grad_norm": 1.1469098329544067,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 77740
+    },
+    {
+      "epoch": 511.5131578947368,
+      "grad_norm": 1.2588064670562744,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 77750
+    },
+    {
+      "epoch": 511.57894736842104,
+      "grad_norm": 0.9466657638549805,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 77760
+    },
+    {
+      "epoch": 511.64473684210526,
+      "grad_norm": 1.3773024082183838,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 77770
+    },
+    {
+      "epoch": 511.7105263157895,
+      "grad_norm": 1.0853861570358276,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 77780
+    },
+    {
+      "epoch": 511.7763157894737,
+      "grad_norm": 1.2546428442001343,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 77790
+    },
+    {
+      "epoch": 511.8421052631579,
+      "grad_norm": 1.0669403076171875,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 77800
+    },
+    {
+      "epoch": 511.9078947368421,
+      "grad_norm": 1.2506325244903564,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 77810
+    },
+    {
+      "epoch": 511.9736842105263,
+      "grad_norm": 1.007456660270691,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 77820
+    },
+    {
+      "epoch": 512.0394736842105,
+      "grad_norm": 1.0313053131103516,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 77830
+    },
+    {
+      "epoch": 512.1052631578947,
+      "grad_norm": 1.1146388053894043,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 77840
+    },
+    {
+      "epoch": 512.171052631579,
+      "grad_norm": 1.1103726625442505,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 77850
+    },
+    {
+      "epoch": 512.2368421052631,
+      "grad_norm": 1.240702748298645,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 77860
+    },
+    {
+      "epoch": 512.3026315789474,
+      "grad_norm": 1.3598905801773071,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 77870
+    },
+    {
+      "epoch": 512.3684210526316,
+      "grad_norm": 1.3365449905395508,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 77880
+    },
+    {
+      "epoch": 512.4342105263158,
+      "grad_norm": 0.8507790565490723,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 77890
+    },
+    {
+      "epoch": 512.5,
+      "grad_norm": 0.9919387698173523,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 77900
+    },
+    {
+      "epoch": 512.5657894736842,
+      "grad_norm": 0.742108941078186,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 77910
+    },
+    {
+      "epoch": 512.6315789473684,
+      "grad_norm": 1.1056491136550903,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 77920
+    },
+    {
+      "epoch": 512.6973684210526,
+      "grad_norm": 1.045460820198059,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 77930
+    },
+    {
+      "epoch": 512.7631578947369,
+      "grad_norm": 1.2098149061203003,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 77940
+    },
+    {
+      "epoch": 512.828947368421,
+      "grad_norm": 1.1252753734588623,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 77950
+    },
+    {
+      "epoch": 512.8947368421053,
+      "grad_norm": 1.4744460582733154,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 77960
+    },
+    {
+      "epoch": 512.9605263157895,
+      "grad_norm": 0.7933367490768433,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 77970
+    },
+    {
+      "epoch": 513.0263157894736,
+      "grad_norm": 1.1913303136825562,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 77980
+    },
+    {
+      "epoch": 513.0921052631579,
+      "grad_norm": 1.1555919647216797,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 77990
+    },
+    {
+      "epoch": 513.1578947368421,
+      "grad_norm": 1.4469823837280273,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 78000
+    },
+    {
+      "epoch": 513.2236842105264,
+      "grad_norm": 1.104987382888794,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 78010
+    },
+    {
+      "epoch": 513.2894736842105,
+      "grad_norm": 1.0023167133331299,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 78020
+    },
+    {
+      "epoch": 513.3552631578947,
+      "grad_norm": 1.0223348140716553,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 78030
+    },
+    {
+      "epoch": 513.421052631579,
+      "grad_norm": 1.0552802085876465,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 78040
+    },
+    {
+      "epoch": 513.4868421052631,
+      "grad_norm": 1.0780552625656128,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 78050
+    },
+    {
+      "epoch": 513.5526315789474,
+      "grad_norm": 1.1177747249603271,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 78060
+    },
+    {
+      "epoch": 513.6184210526316,
+      "grad_norm": 1.3587802648544312,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 78070
+    },
+    {
+      "epoch": 513.6842105263158,
+      "grad_norm": 1.1935032606124878,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 78080
+    },
+    {
+      "epoch": 513.75,
+      "grad_norm": 1.1695393323898315,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 78090
+    },
+    {
+      "epoch": 513.8157894736842,
+      "grad_norm": 0.7090479135513306,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 78100
+    },
+    {
+      "epoch": 513.8815789473684,
+      "grad_norm": 1.5053452253341675,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 78110
+    },
+    {
+      "epoch": 513.9473684210526,
+      "grad_norm": 1.4671512842178345,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 78120
+    },
+    {
+      "epoch": 514.0131578947369,
+      "grad_norm": 1.2938599586486816,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 78130
+    },
+    {
+      "epoch": 514.078947368421,
+      "grad_norm": 1.09611177444458,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 78140
+    },
+    {
+      "epoch": 514.1447368421053,
+      "grad_norm": 1.0053191184997559,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 78150
+    },
+    {
+      "epoch": 514.2105263157895,
+      "grad_norm": 0.8893487453460693,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 78160
+    },
+    {
+      "epoch": 514.2763157894736,
+      "grad_norm": 1.13539457321167,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 78170
+    },
+    {
+      "epoch": 514.3421052631579,
+      "grad_norm": 1.3186606168746948,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 78180
+    },
+    {
+      "epoch": 514.4078947368421,
+      "grad_norm": 1.296586036682129,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 78190
+    },
+    {
+      "epoch": 514.4736842105264,
+      "grad_norm": 1.1459144353866577,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 78200
+    },
+    {
+      "epoch": 514.5394736842105,
+      "grad_norm": 0.9058517813682556,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 78210
+    },
+    {
+      "epoch": 514.6052631578947,
+      "grad_norm": 1.4812514781951904,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 78220
+    },
+    {
+      "epoch": 514.671052631579,
+      "grad_norm": 1.5883773565292358,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 78230
+    },
+    {
+      "epoch": 514.7368421052631,
+      "grad_norm": 1.2800959348678589,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 78240
+    },
+    {
+      "epoch": 514.8026315789474,
+      "grad_norm": 0.9361305236816406,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 78250
+    },
+    {
+      "epoch": 514.8684210526316,
+      "grad_norm": 1.3019955158233643,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 78260
+    },
+    {
+      "epoch": 514.9342105263158,
+      "grad_norm": 1.1924444437026978,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 78270
+    },
+    {
+      "epoch": 515.0,
+      "grad_norm": 1.103654384613037,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 78280
+    },
+    {
+      "epoch": 515.0657894736842,
+      "grad_norm": 1.3044533729553223,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 78290
+    },
+    {
+      "epoch": 515.1315789473684,
+      "grad_norm": 0.844568133354187,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 78300
+    },
+    {
+      "epoch": 515.1973684210526,
+      "grad_norm": 1.071467399597168,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 78310
+    },
+    {
+      "epoch": 515.2631578947369,
+      "grad_norm": 0.9115777015686035,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 78320
+    },
+    {
+      "epoch": 515.328947368421,
+      "grad_norm": 0.8897896409034729,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 78330
+    },
+    {
+      "epoch": 515.3947368421053,
+      "grad_norm": 1.273638367652893,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 78340
+    },
+    {
+      "epoch": 515.4605263157895,
+      "grad_norm": 0.7495373487472534,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 78350
+    },
+    {
+      "epoch": 515.5263157894736,
+      "grad_norm": 1.1438860893249512,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 78360
+    },
+    {
+      "epoch": 515.5921052631579,
+      "grad_norm": 0.8679733276367188,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 78370
+    },
+    {
+      "epoch": 515.6578947368421,
+      "grad_norm": 1.0994764566421509,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 78380
+    },
+    {
+      "epoch": 515.7236842105264,
+      "grad_norm": 1.1603001356124878,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 78390
+    },
+    {
+      "epoch": 515.7894736842105,
+      "grad_norm": 1.049238920211792,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 78400
+    },
+    {
+      "epoch": 515.8552631578947,
+      "grad_norm": 1.0663156509399414,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 78410
+    },
+    {
+      "epoch": 515.921052631579,
+      "grad_norm": 0.7960723638534546,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 78420
+    },
+    {
+      "epoch": 515.9868421052631,
+      "grad_norm": 0.9979690909385681,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 78430
+    },
+    {
+      "epoch": 516.0526315789474,
+      "grad_norm": 0.8607260584831238,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 78440
+    },
+    {
+      "epoch": 516.1184210526316,
+      "grad_norm": 1.2933063507080078,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 78450
+    },
+    {
+      "epoch": 516.1842105263158,
+      "grad_norm": 0.9800442457199097,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 78460
+    },
+    {
+      "epoch": 516.25,
+      "grad_norm": 1.1290419101715088,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 78470
+    },
+    {
+      "epoch": 516.3157894736842,
+      "grad_norm": 1.2492495775222778,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 78480
+    },
+    {
+      "epoch": 516.3815789473684,
+      "grad_norm": 1.2778388261795044,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 78490
+    },
+    {
+      "epoch": 516.4473684210526,
+      "grad_norm": 1.31378173828125,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 78500
+    },
+    {
+      "epoch": 516.5131578947369,
+      "grad_norm": 0.9775269627571106,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 78510
+    },
+    {
+      "epoch": 516.578947368421,
+      "grad_norm": 0.8219498991966248,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 78520
+    },
+    {
+      "epoch": 516.6447368421053,
+      "grad_norm": 0.7330734729766846,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 78530
+    },
+    {
+      "epoch": 516.7105263157895,
+      "grad_norm": 0.9409795999526978,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 78540
+    },
+    {
+      "epoch": 516.7763157894736,
+      "grad_norm": 1.4014899730682373,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 78550
+    },
+    {
+      "epoch": 516.8421052631579,
+      "grad_norm": 1.3369898796081543,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 78560
+    },
+    {
+      "epoch": 516.9078947368421,
+      "grad_norm": 0.7518098950386047,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 78570
+    },
+    {
+      "epoch": 516.9736842105264,
+      "grad_norm": 0.977200448513031,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 78580
+    },
+    {
+      "epoch": 517.0394736842105,
+      "grad_norm": 0.9853875041007996,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 78590
+    },
+    {
+      "epoch": 517.1052631578947,
+      "grad_norm": 0.9010801911354065,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 78600
+    },
+    {
+      "epoch": 517.171052631579,
+      "grad_norm": 0.7902107238769531,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 78610
+    },
+    {
+      "epoch": 517.2368421052631,
+      "grad_norm": 0.7945990562438965,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 78620
+    },
+    {
+      "epoch": 517.3026315789474,
+      "grad_norm": 0.8374760746955872,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 78630
+    },
+    {
+      "epoch": 517.3684210526316,
+      "grad_norm": 0.8473724722862244,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 78640
+    },
+    {
+      "epoch": 517.4342105263158,
+      "grad_norm": 1.1389491558074951,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 78650
+    },
+    {
+      "epoch": 517.5,
+      "grad_norm": 1.1223610639572144,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 78660
+    },
+    {
+      "epoch": 517.5657894736842,
+      "grad_norm": 0.9881941080093384,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 78670
+    },
+    {
+      "epoch": 517.6315789473684,
+      "grad_norm": 1.021945834159851,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 78680
+    },
+    {
+      "epoch": 517.6973684210526,
+      "grad_norm": 0.9731574058532715,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 78690
+    },
+    {
+      "epoch": 517.7631578947369,
+      "grad_norm": 1.2203140258789062,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 78700
+    },
+    {
+      "epoch": 517.828947368421,
+      "grad_norm": 1.5560880899429321,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 78710
+    },
+    {
+      "epoch": 517.8947368421053,
+      "grad_norm": 1.2174243927001953,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 78720
+    },
+    {
+      "epoch": 517.9605263157895,
+      "grad_norm": 1.308231234550476,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 78730
+    },
+    {
+      "epoch": 518.0263157894736,
+      "grad_norm": 1.0401585102081299,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 78740
+    },
+    {
+      "epoch": 518.0921052631579,
+      "grad_norm": 1.1766376495361328,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 78750
+    },
+    {
+      "epoch": 518.1578947368421,
+      "grad_norm": 1.2370878458023071,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 78760
+    },
+    {
+      "epoch": 518.2236842105264,
+      "grad_norm": 1.2282683849334717,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 78770
+    },
+    {
+      "epoch": 518.2894736842105,
+      "grad_norm": 1.6565943956375122,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 78780
+    },
+    {
+      "epoch": 518.3552631578947,
+      "grad_norm": 1.6417547464370728,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 78790
+    },
+    {
+      "epoch": 518.421052631579,
+      "grad_norm": 1.1247665882110596,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 78800
+    },
+    {
+      "epoch": 518.4868421052631,
+      "grad_norm": 1.1730952262878418,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 78810
+    },
+    {
+      "epoch": 518.5526315789474,
+      "grad_norm": 1.2181925773620605,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 78820
+    },
+    {
+      "epoch": 518.6184210526316,
+      "grad_norm": 1.0916987657546997,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 78830
+    },
+    {
+      "epoch": 518.6842105263158,
+      "grad_norm": 0.9300577640533447,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 78840
+    },
+    {
+      "epoch": 518.75,
+      "grad_norm": 1.0200120210647583,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 78850
+    },
+    {
+      "epoch": 518.8157894736842,
+      "grad_norm": 1.1790283918380737,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 78860
+    },
+    {
+      "epoch": 518.8815789473684,
+      "grad_norm": 1.4343522787094116,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 78870
+    },
+    {
+      "epoch": 518.9473684210526,
+      "grad_norm": 1.2748308181762695,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 78880
+    },
+    {
+      "epoch": 519.0131578947369,
+      "grad_norm": 1.3263063430786133,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 78890
+    },
+    {
+      "epoch": 519.078947368421,
+      "grad_norm": 0.8143279552459717,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 78900
+    },
+    {
+      "epoch": 519.1447368421053,
+      "grad_norm": 0.8361022472381592,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 78910
+    },
+    {
+      "epoch": 519.2105263157895,
+      "grad_norm": 1.0913989543914795,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 78920
+    },
+    {
+      "epoch": 519.2763157894736,
+      "grad_norm": 1.3421107530593872,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 78930
+    },
+    {
+      "epoch": 519.3421052631579,
+      "grad_norm": 1.0078271627426147,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 78940
+    },
+    {
+      "epoch": 519.4078947368421,
+      "grad_norm": 1.1363328695297241,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 78950
+    },
+    {
+      "epoch": 519.4736842105264,
+      "grad_norm": 1.1798573732376099,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 78960
+    },
+    {
+      "epoch": 519.5394736842105,
+      "grad_norm": 1.3588061332702637,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 78970
+    },
+    {
+      "epoch": 519.6052631578947,
+      "grad_norm": 0.9301666021347046,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 78980
+    },
+    {
+      "epoch": 519.671052631579,
+      "grad_norm": 1.2737902402877808,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 78990
+    },
+    {
+      "epoch": 519.7368421052631,
+      "grad_norm": 1.1327887773513794,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 79000
+    },
+    {
+      "epoch": 519.8026315789474,
+      "grad_norm": 1.0071492195129395,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 79010
+    },
+    {
+      "epoch": 519.8684210526316,
+      "grad_norm": 1.227921724319458,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 79020
+    },
+    {
+      "epoch": 519.9342105263158,
+      "grad_norm": 1.2975742816925049,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 79030
+    },
+    {
+      "epoch": 520.0,
+      "grad_norm": 1.392256259918213,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 79040
+    },
+    {
+      "epoch": 520.0657894736842,
+      "grad_norm": 1.469244360923767,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 79050
+    },
+    {
+      "epoch": 520.1315789473684,
+      "grad_norm": 0.9325305223464966,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 79060
+    },
+    {
+      "epoch": 520.1973684210526,
+      "grad_norm": 1.2577964067459106,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 79070
+    },
+    {
+      "epoch": 520.2631578947369,
+      "grad_norm": 1.347522497177124,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 79080
+    },
+    {
+      "epoch": 520.328947368421,
+      "grad_norm": 1.0849223136901855,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 79090
+    },
+    {
+      "epoch": 520.3947368421053,
+      "grad_norm": 1.3118022680282593,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 79100
+    },
+    {
+      "epoch": 520.4605263157895,
+      "grad_norm": 1.0926787853240967,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 79110
+    },
+    {
+      "epoch": 520.5263157894736,
+      "grad_norm": 1.3140438795089722,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 79120
+    },
+    {
+      "epoch": 520.5921052631579,
+      "grad_norm": 1.4527376890182495,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 79130
+    },
+    {
+      "epoch": 520.6578947368421,
+      "grad_norm": 0.9648956656455994,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 79140
+    },
+    {
+      "epoch": 520.7236842105264,
+      "grad_norm": 0.9755305051803589,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 79150
+    },
+    {
+      "epoch": 520.7894736842105,
+      "grad_norm": 1.085988163948059,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 79160
+    },
+    {
+      "epoch": 520.8552631578947,
+      "grad_norm": 0.982824444770813,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 79170
+    },
+    {
+      "epoch": 520.921052631579,
+      "grad_norm": 1.0134260654449463,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 79180
+    },
+    {
+      "epoch": 520.9868421052631,
+      "grad_norm": 1.0143835544586182,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 79190
+    },
+    {
+      "epoch": 521.0526315789474,
+      "grad_norm": 1.0507410764694214,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 79200
+    },
+    {
+      "epoch": 521.1184210526316,
+      "grad_norm": 0.9346352815628052,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 79210
+    },
+    {
+      "epoch": 521.1842105263158,
+      "grad_norm": 0.6871551871299744,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 79220
+    },
+    {
+      "epoch": 521.25,
+      "grad_norm": 1.1676675081253052,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 79230
+    },
+    {
+      "epoch": 521.3157894736842,
+      "grad_norm": 0.6791279911994934,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 79240
+    },
+    {
+      "epoch": 521.3815789473684,
+      "grad_norm": 0.735756516456604,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 79250
+    },
+    {
+      "epoch": 521.4473684210526,
+      "grad_norm": 1.068332314491272,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 79260
+    },
+    {
+      "epoch": 521.5131578947369,
+      "grad_norm": 1.0991226434707642,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 79270
+    },
+    {
+      "epoch": 521.578947368421,
+      "grad_norm": 0.9471065402030945,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 79280
+    },
+    {
+      "epoch": 521.6447368421053,
+      "grad_norm": 1.1460009813308716,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 79290
+    },
+    {
+      "epoch": 521.7105263157895,
+      "grad_norm": 1.2541179656982422,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 79300
+    },
+    {
+      "epoch": 521.7763157894736,
+      "grad_norm": 1.3320528268814087,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 79310
+    },
+    {
+      "epoch": 521.8421052631579,
+      "grad_norm": 1.2252389192581177,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 79320
+    },
+    {
+      "epoch": 521.9078947368421,
+      "grad_norm": 1.0492901802062988,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 79330
+    },
+    {
+      "epoch": 521.9736842105264,
+      "grad_norm": 0.755221962928772,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 79340
+    },
+    {
+      "epoch": 522.0394736842105,
+      "grad_norm": 0.9267183542251587,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 79350
+    },
+    {
+      "epoch": 522.1052631578947,
+      "grad_norm": 1.2305091619491577,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 79360
+    },
+    {
+      "epoch": 522.171052631579,
+      "grad_norm": 1.3195940256118774,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 79370
+    },
+    {
+      "epoch": 522.2368421052631,
+      "grad_norm": 1.2802159786224365,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 79380
+    },
+    {
+      "epoch": 522.3026315789474,
+      "grad_norm": 1.2666503190994263,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 79390
+    },
+    {
+      "epoch": 522.3684210526316,
+      "grad_norm": 1.2319374084472656,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 79400
+    },
+    {
+      "epoch": 522.4342105263158,
+      "grad_norm": 1.361430287361145,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 79410
+    },
+    {
+      "epoch": 522.5,
+      "grad_norm": 1.0288817882537842,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 79420
+    },
+    {
+      "epoch": 522.5657894736842,
+      "grad_norm": 1.1962461471557617,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 79430
+    },
+    {
+      "epoch": 522.6315789473684,
+      "grad_norm": 1.5438764095306396,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 79440
+    },
+    {
+      "epoch": 522.6973684210526,
+      "grad_norm": 1.5173245668411255,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 79450
+    },
+    {
+      "epoch": 522.7631578947369,
+      "grad_norm": 1.4040913581848145,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 79460
+    },
+    {
+      "epoch": 522.828947368421,
+      "grad_norm": 1.4810621738433838,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 79470
+    },
+    {
+      "epoch": 522.8947368421053,
+      "grad_norm": 1.279215931892395,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 79480
+    },
+    {
+      "epoch": 522.9605263157895,
+      "grad_norm": 1.1824666261672974,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 79490
+    },
+    {
+      "epoch": 523.0263157894736,
+      "grad_norm": 1.2264102697372437,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 79500
+    },
+    {
+      "epoch": 523.0921052631579,
+      "grad_norm": 1.2089831829071045,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 79510
+    },
+    {
+      "epoch": 523.1578947368421,
+      "grad_norm": 1.6189210414886475,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 79520
+    },
+    {
+      "epoch": 523.2236842105264,
+      "grad_norm": 1.5252612829208374,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 79530
+    },
+    {
+      "epoch": 523.2894736842105,
+      "grad_norm": 1.3406212329864502,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 79540
+    },
+    {
+      "epoch": 523.3552631578947,
+      "grad_norm": 1.1579642295837402,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 79550
+    },
+    {
+      "epoch": 523.421052631579,
+      "grad_norm": 1.012661337852478,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 79560
+    },
+    {
+      "epoch": 523.4868421052631,
+      "grad_norm": 0.8289533257484436,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 79570
+    },
+    {
+      "epoch": 523.5526315789474,
+      "grad_norm": 1.1507660150527954,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 79580
+    },
+    {
+      "epoch": 523.6184210526316,
+      "grad_norm": 1.1765552759170532,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 79590
+    },
+    {
+      "epoch": 523.6842105263158,
+      "grad_norm": 1.2260336875915527,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 79600
+    },
+    {
+      "epoch": 523.75,
+      "grad_norm": 1.341565728187561,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 79610
+    },
+    {
+      "epoch": 523.8157894736842,
+      "grad_norm": 0.9649227261543274,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 79620
+    },
+    {
+      "epoch": 523.8815789473684,
+      "grad_norm": 1.075209617614746,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 79630
+    },
+    {
+      "epoch": 523.9473684210526,
+      "grad_norm": 1.4141005277633667,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 79640
+    },
+    {
+      "epoch": 524.0131578947369,
+      "grad_norm": 1.0523087978363037,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 79650
+    },
+    {
+      "epoch": 524.078947368421,
+      "grad_norm": 1.1604161262512207,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 79660
+    },
+    {
+      "epoch": 524.1447368421053,
+      "grad_norm": 0.8030272722244263,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 79670
+    },
+    {
+      "epoch": 524.2105263157895,
+      "grad_norm": 0.9838864207267761,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 79680
+    },
+    {
+      "epoch": 524.2763157894736,
+      "grad_norm": 1.2279644012451172,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 79690
+    },
+    {
+      "epoch": 524.3421052631579,
+      "grad_norm": 1.4876803159713745,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 79700
+    },
+    {
+      "epoch": 524.4078947368421,
+      "grad_norm": 1.1998071670532227,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 79710
+    },
+    {
+      "epoch": 524.4736842105264,
+      "grad_norm": 1.1491254568099976,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 79720
+    },
+    {
+      "epoch": 524.5394736842105,
+      "grad_norm": 1.4121824502944946,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 79730
+    },
+    {
+      "epoch": 524.6052631578947,
+      "grad_norm": 0.9082379341125488,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 79740
+    },
+    {
+      "epoch": 524.671052631579,
+      "grad_norm": 1.1881701946258545,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 79750
+    },
+    {
+      "epoch": 524.7368421052631,
+      "grad_norm": 1.3639180660247803,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 79760
+    },
+    {
+      "epoch": 524.8026315789474,
+      "grad_norm": 1.2215701341629028,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 79770
+    },
+    {
+      "epoch": 524.8684210526316,
+      "grad_norm": 0.7113543748855591,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 79780
+    },
+    {
+      "epoch": 524.9342105263158,
+      "grad_norm": 1.2361555099487305,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 79790
+    },
+    {
+      "epoch": 525.0,
+      "grad_norm": 1.0537495613098145,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 79800
+    },
+    {
+      "epoch": 525.0657894736842,
+      "grad_norm": 1.0916693210601807,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 79810
+    },
+    {
+      "epoch": 525.1315789473684,
+      "grad_norm": 1.286116600036621,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 79820
+    },
+    {
+      "epoch": 525.1973684210526,
+      "grad_norm": 1.1319963932037354,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 79830
+    },
+    {
+      "epoch": 525.2631578947369,
+      "grad_norm": 1.149547815322876,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 79840
+    },
+    {
+      "epoch": 525.328947368421,
+      "grad_norm": 1.2941960096359253,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 79850
+    },
+    {
+      "epoch": 525.3947368421053,
+      "grad_norm": 0.9617957472801208,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 79860
+    },
+    {
+      "epoch": 525.4605263157895,
+      "grad_norm": 1.2239959239959717,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 79870
+    },
+    {
+      "epoch": 525.5263157894736,
+      "grad_norm": 1.0392863750457764,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 79880
+    },
+    {
+      "epoch": 525.5921052631579,
+      "grad_norm": 1.2805525064468384,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 79890
+    },
+    {
+      "epoch": 525.6578947368421,
+      "grad_norm": 1.201557993888855,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 79900
+    },
+    {
+      "epoch": 525.7236842105264,
+      "grad_norm": 1.165624737739563,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 79910
+    },
+    {
+      "epoch": 525.7894736842105,
+      "grad_norm": 1.2472856044769287,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 79920
+    },
+    {
+      "epoch": 525.8552631578947,
+      "grad_norm": 0.8875550627708435,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 79930
+    },
+    {
+      "epoch": 525.921052631579,
+      "grad_norm": 1.0350874662399292,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 79940
+    },
+    {
+      "epoch": 525.9868421052631,
+      "grad_norm": 1.2986034154891968,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 79950
+    },
+    {
+      "epoch": 526.0526315789474,
+      "grad_norm": 1.1310527324676514,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 79960
+    },
+    {
+      "epoch": 526.1184210526316,
+      "grad_norm": 1.1110562086105347,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 79970
+    },
+    {
+      "epoch": 526.1842105263158,
+      "grad_norm": 1.1927745342254639,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 79980
+    },
+    {
+      "epoch": 526.25,
+      "grad_norm": 1.0965501070022583,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 79990
+    },
+    {
+      "epoch": 526.3157894736842,
+      "grad_norm": 1.4071152210235596,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 80000
+    },
+    {
+      "epoch": 526.3815789473684,
+      "grad_norm": 1.0651636123657227,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 80010
+    },
+    {
+      "epoch": 526.4473684210526,
+      "grad_norm": 1.3853951692581177,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 80020
+    },
+    {
+      "epoch": 526.5131578947369,
+      "grad_norm": 1.1486934423446655,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 80030
+    },
+    {
+      "epoch": 526.578947368421,
+      "grad_norm": 1.088996171951294,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 80040
+    },
+    {
+      "epoch": 526.6447368421053,
+      "grad_norm": 1.4724292755126953,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 80050
+    },
+    {
+      "epoch": 526.7105263157895,
+      "grad_norm": 1.1345316171646118,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 80060
+    },
+    {
+      "epoch": 526.7763157894736,
+      "grad_norm": 1.065085768699646,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 80070
+    },
+    {
+      "epoch": 526.8421052631579,
+      "grad_norm": 1.199918270111084,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 80080
+    },
+    {
+      "epoch": 526.9078947368421,
+      "grad_norm": 1.019289255142212,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 80090
+    },
+    {
+      "epoch": 526.9736842105264,
+      "grad_norm": 1.2872799634933472,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 80100
+    },
+    {
+      "epoch": 527.0394736842105,
+      "grad_norm": 1.3231974840164185,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 80110
+    },
+    {
+      "epoch": 527.1052631578947,
+      "grad_norm": 1.0737744569778442,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 80120
+    },
+    {
+      "epoch": 527.171052631579,
+      "grad_norm": 0.7930030822753906,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 80130
+    },
+    {
+      "epoch": 527.2368421052631,
+      "grad_norm": 1.4024803638458252,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 80140
+    },
+    {
+      "epoch": 527.3026315789474,
+      "grad_norm": 0.9458417892456055,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 80150
+    },
+    {
+      "epoch": 527.3684210526316,
+      "grad_norm": 1.3698056936264038,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 80160
+    },
+    {
+      "epoch": 527.4342105263158,
+      "grad_norm": 0.9708862900733948,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 80170
+    },
+    {
+      "epoch": 527.5,
+      "grad_norm": 0.9804370403289795,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 80180
+    },
+    {
+      "epoch": 527.5657894736842,
+      "grad_norm": 1.2980481386184692,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 80190
+    },
+    {
+      "epoch": 527.6315789473684,
+      "grad_norm": 1.3205853700637817,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 80200
+    },
+    {
+      "epoch": 527.6973684210526,
+      "grad_norm": 0.9288889765739441,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 80210
+    },
+    {
+      "epoch": 527.7631578947369,
+      "grad_norm": 1.5022870302200317,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 80220
+    },
+    {
+      "epoch": 527.828947368421,
+      "grad_norm": 1.310860276222229,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 80230
+    },
+    {
+      "epoch": 527.8947368421053,
+      "grad_norm": 1.0890276432037354,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 80240
+    },
+    {
+      "epoch": 527.9605263157895,
+      "grad_norm": 1.25932776927948,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 80250
+    },
+    {
+      "epoch": 528.0263157894736,
+      "grad_norm": 1.037662386894226,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 80260
+    },
+    {
+      "epoch": 528.0921052631579,
+      "grad_norm": 1.1433483362197876,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 80270
+    },
+    {
+      "epoch": 528.1578947368421,
+      "grad_norm": 1.212409257888794,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 80280
+    },
+    {
+      "epoch": 528.2236842105264,
+      "grad_norm": 1.226341962814331,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 80290
+    },
+    {
+      "epoch": 528.2894736842105,
+      "grad_norm": 1.4430090188980103,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 80300
+    },
+    {
+      "epoch": 528.3552631578947,
+      "grad_norm": 1.1568644046783447,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 80310
+    },
+    {
+      "epoch": 528.421052631579,
+      "grad_norm": 1.183970332145691,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 80320
+    },
+    {
+      "epoch": 528.4868421052631,
+      "grad_norm": 1.3099122047424316,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 80330
+    },
+    {
+      "epoch": 528.5526315789474,
+      "grad_norm": 1.2902318239212036,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 80340
+    },
+    {
+      "epoch": 528.6184210526316,
+      "grad_norm": 1.0034809112548828,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 80350
+    },
+    {
+      "epoch": 528.6842105263158,
+      "grad_norm": 0.9680561423301697,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 80360
+    },
+    {
+      "epoch": 528.75,
+      "grad_norm": 1.1589789390563965,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 80370
+    },
+    {
+      "epoch": 528.8157894736842,
+      "grad_norm": 1.2295933961868286,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 80380
+    },
+    {
+      "epoch": 528.8815789473684,
+      "grad_norm": 1.5712645053863525,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 80390
+    },
+    {
+      "epoch": 528.9473684210526,
+      "grad_norm": 1.4577996730804443,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 80400
+    },
+    {
+      "epoch": 529.0131578947369,
+      "grad_norm": 1.1000990867614746,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 80410
+    },
+    {
+      "epoch": 529.078947368421,
+      "grad_norm": 1.2241917848587036,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 80420
+    },
+    {
+      "epoch": 529.1447368421053,
+      "grad_norm": 1.0974723100662231,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 80430
+    },
+    {
+      "epoch": 529.2105263157895,
+      "grad_norm": 0.9757032990455627,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 80440
+    },
+    {
+      "epoch": 529.2763157894736,
+      "grad_norm": 1.317093849182129,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 80450
+    },
+    {
+      "epoch": 529.3421052631579,
+      "grad_norm": 1.1558804512023926,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 80460
+    },
+    {
+      "epoch": 529.4078947368421,
+      "grad_norm": 1.0101091861724854,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 80470
+    },
+    {
+      "epoch": 529.4736842105264,
+      "grad_norm": 1.2322200536727905,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 80480
+    },
+    {
+      "epoch": 529.5394736842105,
+      "grad_norm": 1.0501878261566162,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 80490
+    },
+    {
+      "epoch": 529.6052631578947,
+      "grad_norm": 1.1091357469558716,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 80500
+    },
+    {
+      "epoch": 529.671052631579,
+      "grad_norm": 1.2675368785858154,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 80510
+    },
+    {
+      "epoch": 529.7368421052631,
+      "grad_norm": 1.0387545824050903,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 80520
+    },
+    {
+      "epoch": 529.8026315789474,
+      "grad_norm": 1.0804754495620728,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 80530
+    },
+    {
+      "epoch": 529.8684210526316,
+      "grad_norm": 1.0754399299621582,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 80540
+    },
+    {
+      "epoch": 529.9342105263158,
+      "grad_norm": 1.0641555786132812,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 80550
+    },
+    {
+      "epoch": 530.0,
+      "grad_norm": 1.3832042217254639,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 80560
+    },
+    {
+      "epoch": 530.0657894736842,
+      "grad_norm": 1.0079296827316284,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 80570
+    },
+    {
+      "epoch": 530.1315789473684,
+      "grad_norm": 0.7935236692428589,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 80580
+    },
+    {
+      "epoch": 530.1973684210526,
+      "grad_norm": 1.2828476428985596,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 80590
+    },
+    {
+      "epoch": 530.2631578947369,
+      "grad_norm": 1.0356460809707642,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 80600
+    },
+    {
+      "epoch": 530.328947368421,
+      "grad_norm": 1.1222877502441406,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 80610
+    },
+    {
+      "epoch": 530.3947368421053,
+      "grad_norm": 0.8804997801780701,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 80620
+    },
+    {
+      "epoch": 530.4605263157895,
+      "grad_norm": 1.159664273262024,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 80630
+    },
+    {
+      "epoch": 530.5263157894736,
+      "grad_norm": 1.4239946603775024,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 80640
+    },
+    {
+      "epoch": 530.5921052631579,
+      "grad_norm": 1.7034857273101807,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 80650
+    },
+    {
+      "epoch": 530.6578947368421,
+      "grad_norm": 1.6954020261764526,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 80660
+    },
+    {
+      "epoch": 530.7236842105264,
+      "grad_norm": 1.291788935661316,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 80670
+    },
+    {
+      "epoch": 530.7894736842105,
+      "grad_norm": 2.2857964038848877,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 80680
+    },
+    {
+      "epoch": 530.8552631578947,
+      "grad_norm": 1.9390625953674316,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 80690
+    },
+    {
+      "epoch": 530.921052631579,
+      "grad_norm": 1.31684410572052,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 80700
+    },
+    {
+      "epoch": 530.9868421052631,
+      "grad_norm": 1.6871960163116455,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 80710
+    },
+    {
+      "epoch": 531.0526315789474,
+      "grad_norm": 1.471971869468689,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 80720
+    },
+    {
+      "epoch": 531.1184210526316,
+      "grad_norm": 1.5256435871124268,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 80730
+    },
+    {
+      "epoch": 531.1842105263158,
+      "grad_norm": 1.4224085807800293,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 80740
+    },
+    {
+      "epoch": 531.25,
+      "grad_norm": 1.179168701171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 80750
+    },
+    {
+      "epoch": 531.3157894736842,
+      "grad_norm": 1.2524338960647583,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 80760
+    },
+    {
+      "epoch": 531.3815789473684,
+      "grad_norm": 1.2265777587890625,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 80770
+    },
+    {
+      "epoch": 531.4473684210526,
+      "grad_norm": 1.4335110187530518,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 80780
+    },
+    {
+      "epoch": 531.5131578947369,
+      "grad_norm": 1.5284316539764404,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 80790
+    },
+    {
+      "epoch": 531.578947368421,
+      "grad_norm": 1.42241370677948,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 80800
+    },
+    {
+      "epoch": 531.6447368421053,
+      "grad_norm": 1.530569314956665,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 80810
+    },
+    {
+      "epoch": 531.7105263157895,
+      "grad_norm": 0.7496344447135925,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 80820
+    },
+    {
+      "epoch": 531.7763157894736,
+      "grad_norm": 1.7025457620620728,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 80830
+    },
+    {
+      "epoch": 531.8421052631579,
+      "grad_norm": 1.1982378959655762,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 80840
+    },
+    {
+      "epoch": 531.9078947368421,
+      "grad_norm": 1.0408469438552856,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 80850
+    },
+    {
+      "epoch": 531.9736842105264,
+      "grad_norm": 1.2916014194488525,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 80860
+    },
+    {
+      "epoch": 532.0394736842105,
+      "grad_norm": 1.271699070930481,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 80870
+    },
+    {
+      "epoch": 532.1052631578947,
+      "grad_norm": 1.313772439956665,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 80880
+    },
+    {
+      "epoch": 532.171052631579,
+      "grad_norm": 1.413476824760437,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 80890
+    },
+    {
+      "epoch": 532.2368421052631,
+      "grad_norm": 1.160772442817688,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 80900
+    },
+    {
+      "epoch": 532.3026315789474,
+      "grad_norm": 1.0700751543045044,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 80910
+    },
+    {
+      "epoch": 532.3684210526316,
+      "grad_norm": 1.1918394565582275,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 80920
+    },
+    {
+      "epoch": 532.4342105263158,
+      "grad_norm": 0.9265491366386414,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 80930
+    },
+    {
+      "epoch": 532.5,
+      "grad_norm": 0.8995620012283325,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 80940
+    },
+    {
+      "epoch": 532.5657894736842,
+      "grad_norm": 1.0701876878738403,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 80950
+    },
+    {
+      "epoch": 532.6315789473684,
+      "grad_norm": 1.1642286777496338,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 80960
+    },
+    {
+      "epoch": 532.6973684210526,
+      "grad_norm": 1.027518630027771,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 80970
+    },
+    {
+      "epoch": 532.7631578947369,
+      "grad_norm": 1.2189431190490723,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 80980
+    },
+    {
+      "epoch": 532.828947368421,
+      "grad_norm": 1.1508433818817139,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 80990
+    },
+    {
+      "epoch": 532.8947368421053,
+      "grad_norm": 1.2702674865722656,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 81000
+    },
+    {
+      "epoch": 532.9605263157895,
+      "grad_norm": 1.1186507940292358,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 81010
+    },
+    {
+      "epoch": 533.0263157894736,
+      "grad_norm": 1.3020246028900146,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 81020
+    },
+    {
+      "epoch": 533.0921052631579,
+      "grad_norm": 1.0293231010437012,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 81030
+    },
+    {
+      "epoch": 533.1578947368421,
+      "grad_norm": 1.2916498184204102,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 81040
+    },
+    {
+      "epoch": 533.2236842105264,
+      "grad_norm": 1.2180614471435547,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 81050
+    },
+    {
+      "epoch": 533.2894736842105,
+      "grad_norm": 0.9033607840538025,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 81060
+    },
+    {
+      "epoch": 533.3552631578947,
+      "grad_norm": 1.1329870223999023,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 81070
+    },
+    {
+      "epoch": 533.421052631579,
+      "grad_norm": 0.9711558818817139,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 81080
+    },
+    {
+      "epoch": 533.4868421052631,
+      "grad_norm": 1.2064048051834106,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 81090
+    },
+    {
+      "epoch": 533.5526315789474,
+      "grad_norm": 1.200500726699829,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 81100
+    },
+    {
+      "epoch": 533.6184210526316,
+      "grad_norm": 1.3031874895095825,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 81110
+    },
+    {
+      "epoch": 533.6842105263158,
+      "grad_norm": 1.249746322631836,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 81120
+    },
+    {
+      "epoch": 533.75,
+      "grad_norm": 1.4198533296585083,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 81130
+    },
+    {
+      "epoch": 533.8157894736842,
+      "grad_norm": 1.0344791412353516,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 81140
+    },
+    {
+      "epoch": 533.8815789473684,
+      "grad_norm": 1.2430800199508667,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 81150
+    },
+    {
+      "epoch": 533.9473684210526,
+      "grad_norm": 1.0078041553497314,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 81160
+    },
+    {
+      "epoch": 534.0131578947369,
+      "grad_norm": 1.1461764574050903,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 81170
+    },
+    {
+      "epoch": 534.078947368421,
+      "grad_norm": 1.1230483055114746,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 81180
+    },
+    {
+      "epoch": 534.1447368421053,
+      "grad_norm": 1.142557144165039,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 81190
+    },
+    {
+      "epoch": 534.2105263157895,
+      "grad_norm": 1.1560312509536743,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 81200
+    },
+    {
+      "epoch": 534.2763157894736,
+      "grad_norm": 1.1191767454147339,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 81210
+    },
+    {
+      "epoch": 534.3421052631579,
+      "grad_norm": 0.7681576013565063,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 81220
+    },
+    {
+      "epoch": 534.4078947368421,
+      "grad_norm": 0.7272207736968994,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 81230
+    },
+    {
+      "epoch": 534.4736842105264,
+      "grad_norm": 0.9689534306526184,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 81240
+    },
+    {
+      "epoch": 534.5394736842105,
+      "grad_norm": 1.1911404132843018,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 81250
+    },
+    {
+      "epoch": 534.6052631578947,
+      "grad_norm": 1.2596523761749268,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 81260
+    },
+    {
+      "epoch": 534.671052631579,
+      "grad_norm": 1.0386167764663696,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 81270
+    },
+    {
+      "epoch": 534.7368421052631,
+      "grad_norm": 0.7270429730415344,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 81280
+    },
+    {
+      "epoch": 534.8026315789474,
+      "grad_norm": 0.9695709347724915,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 81290
+    },
+    {
+      "epoch": 534.8684210526316,
+      "grad_norm": 1.0142713785171509,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 81300
+    },
+    {
+      "epoch": 534.9342105263158,
+      "grad_norm": 1.2047271728515625,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 81310
+    },
+    {
+      "epoch": 535.0,
+      "grad_norm": 1.0015687942504883,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 81320
+    },
+    {
+      "epoch": 535.0657894736842,
+      "grad_norm": 1.3443199396133423,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 81330
+    },
+    {
+      "epoch": 535.1315789473684,
+      "grad_norm": 1.1419992446899414,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 81340
+    },
+    {
+      "epoch": 535.1973684210526,
+      "grad_norm": 1.3758596181869507,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 81350
+    },
+    {
+      "epoch": 535.2631578947369,
+      "grad_norm": 1.384379267692566,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 81360
+    },
+    {
+      "epoch": 535.328947368421,
+      "grad_norm": 0.993719220161438,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 81370
+    },
+    {
+      "epoch": 535.3947368421053,
+      "grad_norm": 0.9903963804244995,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 81380
+    },
+    {
+      "epoch": 535.4605263157895,
+      "grad_norm": 1.840928077697754,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 81390
+    },
+    {
+      "epoch": 535.5263157894736,
+      "grad_norm": 1.8315232992172241,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 81400
+    },
+    {
+      "epoch": 535.5921052631579,
+      "grad_norm": 1.0046427249908447,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 81410
+    },
+    {
+      "epoch": 535.6578947368421,
+      "grad_norm": 1.2056866884231567,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 81420
+    },
+    {
+      "epoch": 535.7236842105264,
+      "grad_norm": 1.0682623386383057,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 81430
+    },
+    {
+      "epoch": 535.7894736842105,
+      "grad_norm": 1.022611141204834,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 81440
+    },
+    {
+      "epoch": 535.8552631578947,
+      "grad_norm": 0.8441836833953857,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 81450
+    },
+    {
+      "epoch": 535.921052631579,
+      "grad_norm": 0.9319405555725098,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 81460
+    },
+    {
+      "epoch": 535.9868421052631,
+      "grad_norm": 1.3196568489074707,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 81470
+    },
+    {
+      "epoch": 536.0526315789474,
+      "grad_norm": 0.9573938846588135,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 81480
+    },
+    {
+      "epoch": 536.1184210526316,
+      "grad_norm": 1.1913952827453613,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 81490
+    },
+    {
+      "epoch": 536.1842105263158,
+      "grad_norm": 1.4894800186157227,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 81500
+    },
+    {
+      "epoch": 536.25,
+      "grad_norm": 1.2386541366577148,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 81510
+    },
+    {
+      "epoch": 536.3157894736842,
+      "grad_norm": 1.4592559337615967,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 81520
+    },
+    {
+      "epoch": 536.3815789473684,
+      "grad_norm": 1.2014906406402588,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 81530
+    },
+    {
+      "epoch": 536.4473684210526,
+      "grad_norm": 0.9843008518218994,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 81540
+    },
+    {
+      "epoch": 536.5131578947369,
+      "grad_norm": 1.1141507625579834,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 81550
+    },
+    {
+      "epoch": 536.578947368421,
+      "grad_norm": 1.0911942720413208,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 81560
+    },
+    {
+      "epoch": 536.6447368421053,
+      "grad_norm": 0.9761918783187866,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 81570
+    },
+    {
+      "epoch": 536.7105263157895,
+      "grad_norm": 1.1536732912063599,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 81580
+    },
+    {
+      "epoch": 536.7763157894736,
+      "grad_norm": 0.9874551296234131,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 81590
+    },
+    {
+      "epoch": 536.8421052631579,
+      "grad_norm": 1.2264351844787598,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 81600
+    },
+    {
+      "epoch": 536.9078947368421,
+      "grad_norm": 1.0342381000518799,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 81610
+    },
+    {
+      "epoch": 536.9736842105264,
+      "grad_norm": 0.8672528266906738,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 81620
+    },
+    {
+      "epoch": 537.0394736842105,
+      "grad_norm": 1.4549974203109741,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 81630
+    },
+    {
+      "epoch": 537.1052631578947,
+      "grad_norm": 1.4086586236953735,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 81640
+    },
+    {
+      "epoch": 537.171052631579,
+      "grad_norm": 0.8391483426094055,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 81650
+    },
+    {
+      "epoch": 537.2368421052631,
+      "grad_norm": 1.2846829891204834,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 81660
+    },
+    {
+      "epoch": 537.3026315789474,
+      "grad_norm": 0.9655391573905945,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 81670
+    },
+    {
+      "epoch": 537.3684210526316,
+      "grad_norm": 1.599820613861084,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 81680
+    },
+    {
+      "epoch": 537.4342105263158,
+      "grad_norm": 0.971474826335907,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 81690
+    },
+    {
+      "epoch": 537.5,
+      "grad_norm": 0.8270283937454224,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 81700
+    },
+    {
+      "epoch": 537.5657894736842,
+      "grad_norm": 1.2286924123764038,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 81710
+    },
+    {
+      "epoch": 537.6315789473684,
+      "grad_norm": 0.9310139417648315,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 81720
+    },
+    {
+      "epoch": 537.6973684210526,
+      "grad_norm": 0.9931386113166809,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 81730
+    },
+    {
+      "epoch": 537.7631578947369,
+      "grad_norm": 1.1506545543670654,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 81740
+    },
+    {
+      "epoch": 537.828947368421,
+      "grad_norm": 0.7511075735092163,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 81750
+    },
+    {
+      "epoch": 537.8947368421053,
+      "grad_norm": 1.1868315935134888,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 81760
+    },
+    {
+      "epoch": 537.9605263157895,
+      "grad_norm": 1.2605093717575073,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 81770
+    },
+    {
+      "epoch": 538.0263157894736,
+      "grad_norm": 0.8703752756118774,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 81780
+    },
+    {
+      "epoch": 538.0921052631579,
+      "grad_norm": 0.8889430165290833,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 81790
+    },
+    {
+      "epoch": 538.1578947368421,
+      "grad_norm": 1.4555449485778809,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 81800
+    },
+    {
+      "epoch": 538.2236842105264,
+      "grad_norm": 0.8129600882530212,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 81810
+    },
+    {
+      "epoch": 538.2894736842105,
+      "grad_norm": 0.6337363123893738,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 81820
+    },
+    {
+      "epoch": 538.3552631578947,
+      "grad_norm": 1.1285789012908936,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 81830
+    },
+    {
+      "epoch": 538.421052631579,
+      "grad_norm": 1.1603407859802246,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 81840
+    },
+    {
+      "epoch": 538.4868421052631,
+      "grad_norm": 0.8500170707702637,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 81850
+    },
+    {
+      "epoch": 538.5526315789474,
+      "grad_norm": 0.9496303200721741,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 81860
+    },
+    {
+      "epoch": 538.6184210526316,
+      "grad_norm": 1.0320367813110352,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 81870
+    },
+    {
+      "epoch": 538.6842105263158,
+      "grad_norm": 0.8967204689979553,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 81880
+    },
+    {
+      "epoch": 538.75,
+      "grad_norm": 1.439362645149231,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 81890
+    },
+    {
+      "epoch": 538.8157894736842,
+      "grad_norm": 1.2981364727020264,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 81900
+    },
+    {
+      "epoch": 538.8815789473684,
+      "grad_norm": 0.7729794979095459,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 81910
+    },
+    {
+      "epoch": 538.9473684210526,
+      "grad_norm": 0.9322214722633362,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 81920
+    },
+    {
+      "epoch": 539.0131578947369,
+      "grad_norm": 1.2167713642120361,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 81930
+    },
+    {
+      "epoch": 539.078947368421,
+      "grad_norm": 0.9902080297470093,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 81940
+    },
+    {
+      "epoch": 539.1447368421053,
+      "grad_norm": 1.3800474405288696,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 81950
+    },
+    {
+      "epoch": 539.2105263157895,
+      "grad_norm": 1.2342554330825806,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 81960
+    },
+    {
+      "epoch": 539.2763157894736,
+      "grad_norm": 1.1256905794143677,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 81970
+    },
+    {
+      "epoch": 539.3421052631579,
+      "grad_norm": 0.8514827489852905,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 81980
+    },
+    {
+      "epoch": 539.4078947368421,
+      "grad_norm": 1.0054851770401,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 81990
+    },
+    {
+      "epoch": 539.4736842105264,
+      "grad_norm": 1.3685269355773926,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 82000
+    },
+    {
+      "epoch": 539.5394736842105,
+      "grad_norm": 0.8945045471191406,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 82010
+    },
+    {
+      "epoch": 539.6052631578947,
+      "grad_norm": 0.7146797776222229,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 82020
+    },
+    {
+      "epoch": 539.671052631579,
+      "grad_norm": 1.4394199848175049,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 82030
+    },
+    {
+      "epoch": 539.7368421052631,
+      "grad_norm": 1.3503402471542358,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 82040
+    },
+    {
+      "epoch": 539.8026315789474,
+      "grad_norm": 0.9955316185951233,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 82050
+    },
+    {
+      "epoch": 539.8684210526316,
+      "grad_norm": 1.2270481586456299,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 82060
+    },
+    {
+      "epoch": 539.9342105263158,
+      "grad_norm": 0.8091347217559814,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 82070
+    },
+    {
+      "epoch": 540.0,
+      "grad_norm": 0.9882093071937561,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 82080
+    },
+    {
+      "epoch": 540.0657894736842,
+      "grad_norm": 1.3003027439117432,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 82090
+    },
+    {
+      "epoch": 540.1315789473684,
+      "grad_norm": 1.113666296005249,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 82100
+    },
+    {
+      "epoch": 540.1973684210526,
+      "grad_norm": 1.5681219100952148,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 82110
+    },
+    {
+      "epoch": 540.2631578947369,
+      "grad_norm": 0.7766351103782654,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 82120
+    },
+    {
+      "epoch": 540.328947368421,
+      "grad_norm": 1.0728567838668823,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 82130
+    },
+    {
+      "epoch": 540.3947368421053,
+      "grad_norm": 0.9811382293701172,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 82140
+    },
+    {
+      "epoch": 540.4605263157895,
+      "grad_norm": 0.8526586890220642,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 82150
+    },
+    {
+      "epoch": 540.5263157894736,
+      "grad_norm": 0.9959100484848022,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 82160
+    },
+    {
+      "epoch": 540.5921052631579,
+      "grad_norm": 1.1363943815231323,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 82170
+    },
+    {
+      "epoch": 540.6578947368421,
+      "grad_norm": 1.0341233015060425,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 82180
+    },
+    {
+      "epoch": 540.7236842105264,
+      "grad_norm": 1.2639089822769165,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 82190
+    },
+    {
+      "epoch": 540.7894736842105,
+      "grad_norm": 1.197579026222229,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 82200
+    },
+    {
+      "epoch": 540.8552631578947,
+      "grad_norm": 1.078098177909851,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 82210
+    },
+    {
+      "epoch": 540.921052631579,
+      "grad_norm": 1.026748538017273,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 82220
+    },
+    {
+      "epoch": 540.9868421052631,
+      "grad_norm": 1.1622309684753418,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 82230
+    },
+    {
+      "epoch": 541.0526315789474,
+      "grad_norm": 0.9618615508079529,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 82240
+    },
+    {
+      "epoch": 541.1184210526316,
+      "grad_norm": 1.2612966299057007,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 82250
+    },
+    {
+      "epoch": 541.1842105263158,
+      "grad_norm": 1.0327303409576416,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 82260
+    },
+    {
+      "epoch": 541.25,
+      "grad_norm": 1.160564661026001,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 82270
+    },
+    {
+      "epoch": 541.3157894736842,
+      "grad_norm": 1.1718395948410034,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 82280
+    },
+    {
+      "epoch": 541.3815789473684,
+      "grad_norm": 0.9807857275009155,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 82290
+    },
+    {
+      "epoch": 541.4473684210526,
+      "grad_norm": 1.2698783874511719,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 82300
+    },
+    {
+      "epoch": 541.5131578947369,
+      "grad_norm": 1.2093521356582642,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 82310
+    },
+    {
+      "epoch": 541.578947368421,
+      "grad_norm": 1.0178953409194946,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 82320
+    },
+    {
+      "epoch": 541.6447368421053,
+      "grad_norm": 0.9447916150093079,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 82330
+    },
+    {
+      "epoch": 541.7105263157895,
+      "grad_norm": 1.0798600912094116,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 82340
+    },
+    {
+      "epoch": 541.7763157894736,
+      "grad_norm": 1.1212877035140991,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 82350
+    },
+    {
+      "epoch": 541.8421052631579,
+      "grad_norm": 1.1646664142608643,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 82360
+    },
+    {
+      "epoch": 541.9078947368421,
+      "grad_norm": 1.073519229888916,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 82370
+    },
+    {
+      "epoch": 541.9736842105264,
+      "grad_norm": 0.8553018569946289,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 82380
+    },
+    {
+      "epoch": 542.0394736842105,
+      "grad_norm": 1.2172927856445312,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 82390
+    },
+    {
+      "epoch": 542.1052631578947,
+      "grad_norm": 1.0330595970153809,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 82400
+    },
+    {
+      "epoch": 542.171052631579,
+      "grad_norm": 0.9683526158332825,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 82410
+    },
+    {
+      "epoch": 542.2368421052631,
+      "grad_norm": 1.1833120584487915,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 82420
+    },
+    {
+      "epoch": 542.3026315789474,
+      "grad_norm": 1.3812700510025024,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 82430
+    },
+    {
+      "epoch": 542.3684210526316,
+      "grad_norm": 1.6291792392730713,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 82440
+    },
+    {
+      "epoch": 542.4342105263158,
+      "grad_norm": 1.2887747287750244,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 82450
+    },
+    {
+      "epoch": 542.5,
+      "grad_norm": 1.5787415504455566,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 82460
+    },
+    {
+      "epoch": 542.5657894736842,
+      "grad_norm": 1.3304678201675415,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 82470
+    },
+    {
+      "epoch": 542.6315789473684,
+      "grad_norm": 1.3938226699829102,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 82480
+    },
+    {
+      "epoch": 542.6973684210526,
+      "grad_norm": 1.4595518112182617,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 82490
+    },
+    {
+      "epoch": 542.7631578947369,
+      "grad_norm": 1.0828487873077393,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 82500
+    },
+    {
+      "epoch": 542.828947368421,
+      "grad_norm": 1.4781620502471924,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 82510
+    },
+    {
+      "epoch": 542.8947368421053,
+      "grad_norm": 1.2808284759521484,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 82520
+    },
+    {
+      "epoch": 542.9605263157895,
+      "grad_norm": 1.0565862655639648,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 82530
+    },
+    {
+      "epoch": 543.0263157894736,
+      "grad_norm": 1.2909079790115356,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 82540
+    },
+    {
+      "epoch": 543.0921052631579,
+      "grad_norm": 1.1060192584991455,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 82550
+    },
+    {
+      "epoch": 543.1578947368421,
+      "grad_norm": 0.9638274908065796,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 82560
+    },
+    {
+      "epoch": 543.2236842105264,
+      "grad_norm": 1.082889437675476,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 82570
+    },
+    {
+      "epoch": 543.2894736842105,
+      "grad_norm": 0.6525837182998657,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 82580
+    },
+    {
+      "epoch": 543.3552631578947,
+      "grad_norm": 1.2893084287643433,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 82590
+    },
+    {
+      "epoch": 543.421052631579,
+      "grad_norm": 1.0150879621505737,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 82600
+    },
+    {
+      "epoch": 543.4868421052631,
+      "grad_norm": 0.5977818965911865,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 82610
+    },
+    {
+      "epoch": 543.5526315789474,
+      "grad_norm": 1.1314287185668945,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 82620
+    },
+    {
+      "epoch": 543.6184210526316,
+      "grad_norm": 1.184816837310791,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 82630
+    },
+    {
+      "epoch": 543.6842105263158,
+      "grad_norm": 0.9463902711868286,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 82640
+    },
+    {
+      "epoch": 543.75,
+      "grad_norm": 1.0592985153198242,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 82650
+    },
+    {
+      "epoch": 543.8157894736842,
+      "grad_norm": 1.3953194618225098,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 82660
+    },
+    {
+      "epoch": 543.8815789473684,
+      "grad_norm": 1.4499620199203491,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 82670
+    },
+    {
+      "epoch": 543.9473684210526,
+      "grad_norm": 1.2184102535247803,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 82680
+    },
+    {
+      "epoch": 544.0131578947369,
+      "grad_norm": 0.9526715874671936,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 82690
+    },
+    {
+      "epoch": 544.078947368421,
+      "grad_norm": 1.111823320388794,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 82700
+    },
+    {
+      "epoch": 544.1447368421053,
+      "grad_norm": 1.0925346612930298,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 82710
+    },
+    {
+      "epoch": 544.2105263157895,
+      "grad_norm": 1.0318348407745361,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 82720
+    },
+    {
+      "epoch": 544.2763157894736,
+      "grad_norm": 1.3430664539337158,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 82730
+    },
+    {
+      "epoch": 544.3421052631579,
+      "grad_norm": 0.8877522945404053,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 82740
+    },
+    {
+      "epoch": 544.4078947368421,
+      "grad_norm": 1.2313371896743774,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 82750
+    },
+    {
+      "epoch": 544.4736842105264,
+      "grad_norm": 1.1085177659988403,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 82760
+    },
+    {
+      "epoch": 544.5394736842105,
+      "grad_norm": 1.0655664205551147,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 82770
+    },
+    {
+      "epoch": 544.6052631578947,
+      "grad_norm": 1.3107541799545288,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 82780
+    },
+    {
+      "epoch": 544.671052631579,
+      "grad_norm": 1.2717570066452026,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 82790
+    },
+    {
+      "epoch": 544.7368421052631,
+      "grad_norm": 1.365206003189087,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 82800
+    },
+    {
+      "epoch": 544.8026315789474,
+      "grad_norm": 1.2015581130981445,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 82810
+    },
+    {
+      "epoch": 544.8684210526316,
+      "grad_norm": 0.8265142440795898,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 82820
+    },
+    {
+      "epoch": 544.9342105263158,
+      "grad_norm": 1.278316617012024,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 82830
+    },
+    {
+      "epoch": 545.0,
+      "grad_norm": 0.8285160064697266,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 82840
+    },
+    {
+      "epoch": 545.0657894736842,
+      "grad_norm": 1.131121277809143,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 82850
+    },
+    {
+      "epoch": 545.1315789473684,
+      "grad_norm": 1.1002274751663208,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 82860
+    },
+    {
+      "epoch": 545.1973684210526,
+      "grad_norm": 1.3011295795440674,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 82870
+    },
+    {
+      "epoch": 545.2631578947369,
+      "grad_norm": 0.9043525457382202,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 82880
+    },
+    {
+      "epoch": 545.328947368421,
+      "grad_norm": 1.1586050987243652,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 82890
+    },
+    {
+      "epoch": 545.3947368421053,
+      "grad_norm": 1.1788783073425293,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 82900
+    },
+    {
+      "epoch": 545.4605263157895,
+      "grad_norm": 1.1490991115570068,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 82910
+    },
+    {
+      "epoch": 545.5263157894736,
+      "grad_norm": 0.9973250031471252,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 82920
+    },
+    {
+      "epoch": 545.5921052631579,
+      "grad_norm": 0.9318479299545288,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 82930
+    },
+    {
+      "epoch": 545.6578947368421,
+      "grad_norm": 0.8547037839889526,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 82940
+    },
+    {
+      "epoch": 545.7236842105264,
+      "grad_norm": 1.1414095163345337,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 82950
+    },
+    {
+      "epoch": 545.7894736842105,
+      "grad_norm": 1.1906640529632568,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 82960
+    },
+    {
+      "epoch": 545.8552631578947,
+      "grad_norm": 0.7530783414840698,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 82970
+    },
+    {
+      "epoch": 545.921052631579,
+      "grad_norm": 1.2797421216964722,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 82980
+    },
+    {
+      "epoch": 545.9868421052631,
+      "grad_norm": 0.995582640171051,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 82990
+    },
+    {
+      "epoch": 546.0526315789474,
+      "grad_norm": 1.1054130792617798,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 83000
+    },
+    {
+      "epoch": 546.1184210526316,
+      "grad_norm": 1.2854208946228027,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 83010
+    },
+    {
+      "epoch": 546.1842105263158,
+      "grad_norm": 1.1343199014663696,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 83020
+    },
+    {
+      "epoch": 546.25,
+      "grad_norm": 0.953602135181427,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 83030
+    },
+    {
+      "epoch": 546.3157894736842,
+      "grad_norm": 1.223362922668457,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 83040
+    },
+    {
+      "epoch": 546.3815789473684,
+      "grad_norm": 1.4338486194610596,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 83050
+    },
+    {
+      "epoch": 546.4473684210526,
+      "grad_norm": 1.1713676452636719,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 83060
+    },
+    {
+      "epoch": 546.5131578947369,
+      "grad_norm": 1.1168043613433838,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 83070
+    },
+    {
+      "epoch": 546.578947368421,
+      "grad_norm": 1.1231666803359985,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 83080
+    },
+    {
+      "epoch": 546.6447368421053,
+      "grad_norm": 1.1550521850585938,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 83090
+    },
+    {
+      "epoch": 546.7105263157895,
+      "grad_norm": 1.1135015487670898,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 83100
+    },
+    {
+      "epoch": 546.7763157894736,
+      "grad_norm": 1.2623625993728638,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 83110
+    },
+    {
+      "epoch": 546.8421052631579,
+      "grad_norm": 0.7757622003555298,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 83120
+    },
+    {
+      "epoch": 546.9078947368421,
+      "grad_norm": 0.9508805274963379,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 83130
+    },
+    {
+      "epoch": 546.9736842105264,
+      "grad_norm": 0.7631132006645203,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 83140
+    },
+    {
+      "epoch": 547.0394736842105,
+      "grad_norm": 0.9571039080619812,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 83150
+    },
+    {
+      "epoch": 547.1052631578947,
+      "grad_norm": 0.9324824213981628,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 83160
+    },
+    {
+      "epoch": 547.171052631579,
+      "grad_norm": 1.1061806678771973,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 83170
+    },
+    {
+      "epoch": 547.2368421052631,
+      "grad_norm": 0.9221788048744202,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 83180
+    },
+    {
+      "epoch": 547.3026315789474,
+      "grad_norm": 1.1765472888946533,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 83190
+    },
+    {
+      "epoch": 547.3684210526316,
+      "grad_norm": 1.0740410089492798,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 83200
+    },
+    {
+      "epoch": 547.4342105263158,
+      "grad_norm": 1.1589864492416382,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 83210
+    },
+    {
+      "epoch": 547.5,
+      "grad_norm": 1.1935651302337646,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 83220
+    },
+    {
+      "epoch": 547.5657894736842,
+      "grad_norm": 0.8413220047950745,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 83230
+    },
+    {
+      "epoch": 547.6315789473684,
+      "grad_norm": 0.8780201077461243,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 83240
+    },
+    {
+      "epoch": 547.6973684210526,
+      "grad_norm": 1.1799852848052979,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 83250
+    },
+    {
+      "epoch": 547.7631578947369,
+      "grad_norm": 0.9087440371513367,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 83260
+    },
+    {
+      "epoch": 547.828947368421,
+      "grad_norm": 1.065704584121704,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 83270
+    },
+    {
+      "epoch": 547.8947368421053,
+      "grad_norm": 1.0333939790725708,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 83280
+    },
+    {
+      "epoch": 547.9605263157895,
+      "grad_norm": 0.951385498046875,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 83290
+    },
+    {
+      "epoch": 548.0263157894736,
+      "grad_norm": 0.7980932593345642,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 83300
+    },
+    {
+      "epoch": 548.0921052631579,
+      "grad_norm": 1.1457808017730713,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 83310
+    },
+    {
+      "epoch": 548.1578947368421,
+      "grad_norm": 0.9355278015136719,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 83320
+    },
+    {
+      "epoch": 548.2236842105264,
+      "grad_norm": 0.7948923707008362,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 83330
+    },
+    {
+      "epoch": 548.2894736842105,
+      "grad_norm": 0.9692309498786926,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 83340
+    },
+    {
+      "epoch": 548.3552631578947,
+      "grad_norm": 1.2351458072662354,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 83350
+    },
+    {
+      "epoch": 548.421052631579,
+      "grad_norm": 0.8993010520935059,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 83360
+    },
+    {
+      "epoch": 548.4868421052631,
+      "grad_norm": 1.3239619731903076,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 83370
+    },
+    {
+      "epoch": 548.5526315789474,
+      "grad_norm": 1.2244657278060913,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 83380
+    },
+    {
+      "epoch": 548.6184210526316,
+      "grad_norm": 1.208211064338684,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 83390
+    },
+    {
+      "epoch": 548.6842105263158,
+      "grad_norm": 1.225913405418396,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 83400
+    },
+    {
+      "epoch": 548.75,
+      "grad_norm": 1.019867181777954,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 83410
+    },
+    {
+      "epoch": 548.8157894736842,
+      "grad_norm": 1.138984203338623,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 83420
+    },
+    {
+      "epoch": 548.8815789473684,
+      "grad_norm": 1.0787227153778076,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 83430
+    },
+    {
+      "epoch": 548.9473684210526,
+      "grad_norm": 1.3269912004470825,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 83440
+    },
+    {
+      "epoch": 549.0131578947369,
+      "grad_norm": 1.1982839107513428,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 83450
+    },
+    {
+      "epoch": 549.078947368421,
+      "grad_norm": 1.4429033994674683,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 83460
+    },
+    {
+      "epoch": 549.1447368421053,
+      "grad_norm": 1.1085598468780518,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 83470
+    },
+    {
+      "epoch": 549.2105263157895,
+      "grad_norm": 1.2021725177764893,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 83480
+    },
+    {
+      "epoch": 549.2763157894736,
+      "grad_norm": 0.8998311161994934,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 83490
+    },
+    {
+      "epoch": 549.3421052631579,
+      "grad_norm": 1.5632169246673584,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 83500
+    },
+    {
+      "epoch": 549.4078947368421,
+      "grad_norm": 1.549469232559204,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 83510
+    },
+    {
+      "epoch": 549.4736842105264,
+      "grad_norm": 1.4514491558074951,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 83520
+    },
+    {
+      "epoch": 549.5394736842105,
+      "grad_norm": 1.0409729480743408,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 83530
+    },
+    {
+      "epoch": 549.6052631578947,
+      "grad_norm": 1.5041582584381104,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 83540
+    },
+    {
+      "epoch": 549.671052631579,
+      "grad_norm": 0.9774188995361328,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 83550
+    },
+    {
+      "epoch": 549.7368421052631,
+      "grad_norm": 1.5498707294464111,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 83560
+    },
+    {
+      "epoch": 549.8026315789474,
+      "grad_norm": 1.5098849534988403,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 83570
+    },
+    {
+      "epoch": 549.8684210526316,
+      "grad_norm": 1.2632389068603516,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 83580
+    },
+    {
+      "epoch": 549.9342105263158,
+      "grad_norm": 1.1932131052017212,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 83590
+    },
+    {
+      "epoch": 550.0,
+      "grad_norm": 1.4617688655853271,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 83600
+    },
+    {
+      "epoch": 550.0657894736842,
+      "grad_norm": 1.0467870235443115,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 83610
+    },
+    {
+      "epoch": 550.1315789473684,
+      "grad_norm": 1.3570033311843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 83620
+    },
+    {
+      "epoch": 550.1973684210526,
+      "grad_norm": 1.4261776208877563,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 83630
+    },
+    {
+      "epoch": 550.2631578947369,
+      "grad_norm": 1.2804388999938965,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 83640
+    },
+    {
+      "epoch": 550.328947368421,
+      "grad_norm": 1.119381070137024,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 83650
+    },
+    {
+      "epoch": 550.3947368421053,
+      "grad_norm": 1.0963352918624878,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 83660
+    },
+    {
+      "epoch": 550.4605263157895,
+      "grad_norm": 0.940981924533844,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 83670
+    },
+    {
+      "epoch": 550.5263157894736,
+      "grad_norm": 1.4243806600570679,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 83680
+    },
+    {
+      "epoch": 550.5921052631579,
+      "grad_norm": 1.049736738204956,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 83690
+    },
+    {
+      "epoch": 550.6578947368421,
+      "grad_norm": 1.2986576557159424,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 83700
+    },
+    {
+      "epoch": 550.7236842105264,
+      "grad_norm": 0.9634553790092468,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 83710
+    },
+    {
+      "epoch": 550.7894736842105,
+      "grad_norm": 1.209815263748169,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 83720
+    },
+    {
+      "epoch": 550.8552631578947,
+      "grad_norm": 1.3946318626403809,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 83730
+    },
+    {
+      "epoch": 550.921052631579,
+      "grad_norm": 1.1185719966888428,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 83740
+    },
+    {
+      "epoch": 550.9868421052631,
+      "grad_norm": 0.8527510762214661,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 83750
+    },
+    {
+      "epoch": 551.0526315789474,
+      "grad_norm": 1.235743761062622,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 83760
+    },
+    {
+      "epoch": 551.1184210526316,
+      "grad_norm": 1.2538394927978516,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 83770
+    },
+    {
+      "epoch": 551.1842105263158,
+      "grad_norm": 1.089035987854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 83780
+    },
+    {
+      "epoch": 551.25,
+      "grad_norm": 1.1372734308242798,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 83790
+    },
+    {
+      "epoch": 551.3157894736842,
+      "grad_norm": 1.026092290878296,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 83800
+    },
+    {
+      "epoch": 551.3815789473684,
+      "grad_norm": 1.2157622575759888,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 83810
+    },
+    {
+      "epoch": 551.4473684210526,
+      "grad_norm": 1.1005877256393433,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 83820
+    },
+    {
+      "epoch": 551.5131578947369,
+      "grad_norm": 1.059844970703125,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 83830
+    },
+    {
+      "epoch": 551.578947368421,
+      "grad_norm": 1.1124544143676758,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 83840
+    },
+    {
+      "epoch": 551.6447368421053,
+      "grad_norm": 1.0764391422271729,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 83850
+    },
+    {
+      "epoch": 551.7105263157895,
+      "grad_norm": 1.031297206878662,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 83860
+    },
+    {
+      "epoch": 551.7763157894736,
+      "grad_norm": 1.022562026977539,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 83870
+    },
+    {
+      "epoch": 551.8421052631579,
+      "grad_norm": 1.1150662899017334,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 83880
+    },
+    {
+      "epoch": 551.9078947368421,
+      "grad_norm": 1.0998282432556152,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 83890
+    },
+    {
+      "epoch": 551.9736842105264,
+      "grad_norm": 1.265975832939148,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 83900
+    },
+    {
+      "epoch": 552.0394736842105,
+      "grad_norm": 1.0744695663452148,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 83910
+    },
+    {
+      "epoch": 552.1052631578947,
+      "grad_norm": 1.1602157354354858,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 83920
+    },
+    {
+      "epoch": 552.171052631579,
+      "grad_norm": 1.2619503736495972,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 83930
+    },
+    {
+      "epoch": 552.2368421052631,
+      "grad_norm": 1.3881173133850098,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 83940
+    },
+    {
+      "epoch": 552.3026315789474,
+      "grad_norm": 1.3225895166397095,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 83950
+    },
+    {
+      "epoch": 552.3684210526316,
+      "grad_norm": 1.4558875560760498,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 83960
+    },
+    {
+      "epoch": 552.4342105263158,
+      "grad_norm": 0.8859315514564514,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 83970
+    },
+    {
+      "epoch": 552.5,
+      "grad_norm": 0.9375401735305786,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 83980
+    },
+    {
+      "epoch": 552.5657894736842,
+      "grad_norm": 1.0013401508331299,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 83990
+    },
+    {
+      "epoch": 552.6315789473684,
+      "grad_norm": 0.8222454190254211,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 84000
+    },
+    {
+      "epoch": 552.6973684210526,
+      "grad_norm": 1.0796682834625244,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 84010
+    },
+    {
+      "epoch": 552.7631578947369,
+      "grad_norm": 1.3614517450332642,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 84020
+    },
+    {
+      "epoch": 552.828947368421,
+      "grad_norm": 1.4528874158859253,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 84030
+    },
+    {
+      "epoch": 552.8947368421053,
+      "grad_norm": 1.2762951850891113,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 84040
+    },
+    {
+      "epoch": 552.9605263157895,
+      "grad_norm": 1.0010347366333008,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 84050
+    },
+    {
+      "epoch": 553.0263157894736,
+      "grad_norm": 1.4103955030441284,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 84060
+    },
+    {
+      "epoch": 553.0921052631579,
+      "grad_norm": 1.2792900800704956,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 84070
+    },
+    {
+      "epoch": 553.1578947368421,
+      "grad_norm": 1.1283965110778809,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 84080
+    },
+    {
+      "epoch": 553.2236842105264,
+      "grad_norm": 1.2378798723220825,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 84090
+    },
+    {
+      "epoch": 553.2894736842105,
+      "grad_norm": 1.0513310432434082,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 84100
+    },
+    {
+      "epoch": 553.3552631578947,
+      "grad_norm": 1.0277812480926514,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 84110
+    },
+    {
+      "epoch": 553.421052631579,
+      "grad_norm": 1.211808681488037,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 84120
+    },
+    {
+      "epoch": 553.4868421052631,
+      "grad_norm": 1.0594956874847412,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 84130
+    },
+    {
+      "epoch": 553.5526315789474,
+      "grad_norm": 1.1370322704315186,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 84140
+    },
+    {
+      "epoch": 553.6184210526316,
+      "grad_norm": 0.897812008857727,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 84150
+    },
+    {
+      "epoch": 553.6842105263158,
+      "grad_norm": 0.9729959964752197,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 84160
+    },
+    {
+      "epoch": 553.75,
+      "grad_norm": 1.1582231521606445,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 84170
+    },
+    {
+      "epoch": 553.8157894736842,
+      "grad_norm": 1.115011215209961,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 84180
+    },
+    {
+      "epoch": 553.8815789473684,
+      "grad_norm": 1.1389967203140259,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 84190
+    },
+    {
+      "epoch": 553.9473684210526,
+      "grad_norm": 1.1534051895141602,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 84200
+    },
+    {
+      "epoch": 554.0131578947369,
+      "grad_norm": 1.1648868322372437,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 84210
+    },
+    {
+      "epoch": 554.078947368421,
+      "grad_norm": 0.9899840354919434,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 84220
+    },
+    {
+      "epoch": 554.1447368421053,
+      "grad_norm": 1.097878336906433,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 84230
+    },
+    {
+      "epoch": 554.2105263157895,
+      "grad_norm": 1.1488139629364014,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 84240
+    },
+    {
+      "epoch": 554.2763157894736,
+      "grad_norm": 1.2230936288833618,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 84250
+    },
+    {
+      "epoch": 554.3421052631579,
+      "grad_norm": 1.2412559986114502,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 84260
+    },
+    {
+      "epoch": 554.4078947368421,
+      "grad_norm": 1.044669270515442,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 84270
+    },
+    {
+      "epoch": 554.4736842105264,
+      "grad_norm": 0.9399918913841248,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 84280
+    },
+    {
+      "epoch": 554.5394736842105,
+      "grad_norm": 0.8007739782333374,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 84290
+    },
+    {
+      "epoch": 554.6052631578947,
+      "grad_norm": 0.68852698802948,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 84300
+    },
+    {
+      "epoch": 554.671052631579,
+      "grad_norm": 1.3844374418258667,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 84310
+    },
+    {
+      "epoch": 554.7368421052631,
+      "grad_norm": 0.897927463054657,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 84320
+    },
+    {
+      "epoch": 554.8026315789474,
+      "grad_norm": 1.1353065967559814,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 84330
+    },
+    {
+      "epoch": 554.8684210526316,
+      "grad_norm": 1.1590604782104492,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 84340
+    },
+    {
+      "epoch": 554.9342105263158,
+      "grad_norm": 1.2227908372879028,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 84350
+    },
+    {
+      "epoch": 555.0,
+      "grad_norm": 1.2319128513336182,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 84360
+    },
+    {
+      "epoch": 555.0657894736842,
+      "grad_norm": 1.0367213487625122,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 84370
+    },
+    {
+      "epoch": 555.1315789473684,
+      "grad_norm": 1.069209098815918,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 84380
+    },
+    {
+      "epoch": 555.1973684210526,
+      "grad_norm": 1.0245308876037598,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 84390
+    },
+    {
+      "epoch": 555.2631578947369,
+      "grad_norm": 1.2172003984451294,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 84400
+    },
+    {
+      "epoch": 555.328947368421,
+      "grad_norm": 1.0955407619476318,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 84410
+    },
+    {
+      "epoch": 555.3947368421053,
+      "grad_norm": 1.0502513647079468,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 84420
+    },
+    {
+      "epoch": 555.4605263157895,
+      "grad_norm": 1.2674477100372314,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 84430
+    },
+    {
+      "epoch": 555.5263157894736,
+      "grad_norm": 1.4851372241973877,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 84440
+    },
+    {
+      "epoch": 555.5921052631579,
+      "grad_norm": 1.4355062246322632,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 84450
+    },
+    {
+      "epoch": 555.6578947368421,
+      "grad_norm": 1.2910557985305786,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 84460
+    },
+    {
+      "epoch": 555.7236842105264,
+      "grad_norm": 0.9084872603416443,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 84470
+    },
+    {
+      "epoch": 555.7894736842105,
+      "grad_norm": 1.1866637468338013,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 84480
+    },
+    {
+      "epoch": 555.8552631578947,
+      "grad_norm": 1.1410084962844849,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 84490
+    },
+    {
+      "epoch": 555.921052631579,
+      "grad_norm": 1.256248116493225,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 84500
+    },
+    {
+      "epoch": 555.9868421052631,
+      "grad_norm": 1.1614428758621216,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 84510
+    },
+    {
+      "epoch": 556.0526315789474,
+      "grad_norm": 1.1941181421279907,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 84520
+    },
+    {
+      "epoch": 556.1184210526316,
+      "grad_norm": 1.1115273237228394,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 84530
+    },
+    {
+      "epoch": 556.1842105263158,
+      "grad_norm": 0.8539004325866699,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 84540
+    },
+    {
+      "epoch": 556.25,
+      "grad_norm": 1.1488637924194336,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 84550
+    },
+    {
+      "epoch": 556.3157894736842,
+      "grad_norm": 1.1279138326644897,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 84560
+    },
+    {
+      "epoch": 556.3815789473684,
+      "grad_norm": 1.3774553537368774,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 84570
+    },
+    {
+      "epoch": 556.4473684210526,
+      "grad_norm": 1.3250354528427124,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 84580
+    },
+    {
+      "epoch": 556.5131578947369,
+      "grad_norm": 1.1280083656311035,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 84590
+    },
+    {
+      "epoch": 556.578947368421,
+      "grad_norm": 0.9719551205635071,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 84600
+    },
+    {
+      "epoch": 556.6447368421053,
+      "grad_norm": 0.8734564781188965,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 84610
+    },
+    {
+      "epoch": 556.7105263157895,
+      "grad_norm": 1.252884864807129,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 84620
+    },
+    {
+      "epoch": 556.7763157894736,
+      "grad_norm": 1.2169554233551025,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 84630
+    },
+    {
+      "epoch": 556.8421052631579,
+      "grad_norm": 1.3537654876708984,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 84640
+    },
+    {
+      "epoch": 556.9078947368421,
+      "grad_norm": 1.1690727472305298,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 84650
+    },
+    {
+      "epoch": 556.9736842105264,
+      "grad_norm": 1.4341109991073608,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 84660
+    },
+    {
+      "epoch": 557.0394736842105,
+      "grad_norm": 1.1728025674819946,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 84670
+    },
+    {
+      "epoch": 557.1052631578947,
+      "grad_norm": 1.1351969242095947,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 84680
+    },
+    {
+      "epoch": 557.171052631579,
+      "grad_norm": 1.049938440322876,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 84690
+    },
+    {
+      "epoch": 557.2368421052631,
+      "grad_norm": 1.0241278409957886,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 84700
+    },
+    {
+      "epoch": 557.3026315789474,
+      "grad_norm": 1.357133150100708,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 84710
+    },
+    {
+      "epoch": 557.3684210526316,
+      "grad_norm": 1.2477604150772095,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 84720
+    },
+    {
+      "epoch": 557.4342105263158,
+      "grad_norm": 1.1618016958236694,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 84730
+    },
+    {
+      "epoch": 557.5,
+      "grad_norm": 1.4906044006347656,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 84740
+    },
+    {
+      "epoch": 557.5657894736842,
+      "grad_norm": 1.6575461626052856,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 84750
+    },
+    {
+      "epoch": 557.6315789473684,
+      "grad_norm": 1.5979561805725098,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 84760
+    },
+    {
+      "epoch": 557.6973684210526,
+      "grad_norm": 1.49028480052948,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 84770
+    },
+    {
+      "epoch": 557.7631578947369,
+      "grad_norm": 1.476001501083374,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 84780
+    },
+    {
+      "epoch": 557.828947368421,
+      "grad_norm": 1.1101016998291016,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 84790
+    },
+    {
+      "epoch": 557.8947368421053,
+      "grad_norm": 1.2816787958145142,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 84800
+    },
+    {
+      "epoch": 557.9605263157895,
+      "grad_norm": 1.4827154874801636,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 84810
+    },
+    {
+      "epoch": 558.0263157894736,
+      "grad_norm": 1.2179068326950073,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 84820
+    },
+    {
+      "epoch": 558.0921052631579,
+      "grad_norm": 1.0725390911102295,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 84830
+    },
+    {
+      "epoch": 558.1578947368421,
+      "grad_norm": 1.1772629022598267,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 84840
+    },
+    {
+      "epoch": 558.2236842105264,
+      "grad_norm": 1.1219836473464966,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 84850
+    },
+    {
+      "epoch": 558.2894736842105,
+      "grad_norm": 1.1598541736602783,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 84860
+    },
+    {
+      "epoch": 558.3552631578947,
+      "grad_norm": 1.2729027271270752,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 84870
+    },
+    {
+      "epoch": 558.421052631579,
+      "grad_norm": 1.107276201248169,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 84880
+    },
+    {
+      "epoch": 558.4868421052631,
+      "grad_norm": 0.9644288420677185,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 84890
+    },
+    {
+      "epoch": 558.5526315789474,
+      "grad_norm": 0.8253970742225647,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 84900
+    },
+    {
+      "epoch": 558.6184210526316,
+      "grad_norm": 0.98391193151474,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 84910
+    },
+    {
+      "epoch": 558.6842105263158,
+      "grad_norm": 1.0832492113113403,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 84920
+    },
+    {
+      "epoch": 558.75,
+      "grad_norm": 0.9063038229942322,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 84930
+    },
+    {
+      "epoch": 558.8157894736842,
+      "grad_norm": 1.0386995077133179,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 84940
+    },
+    {
+      "epoch": 558.8815789473684,
+      "grad_norm": 0.7958237528800964,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 84950
+    },
+    {
+      "epoch": 558.9473684210526,
+      "grad_norm": 0.860396683216095,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 84960
+    },
+    {
+      "epoch": 559.0131578947369,
+      "grad_norm": 0.9728363752365112,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 84970
+    },
+    {
+      "epoch": 559.078947368421,
+      "grad_norm": 1.1759837865829468,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 84980
+    },
+    {
+      "epoch": 559.1447368421053,
+      "grad_norm": 1.0870031118392944,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 84990
+    },
+    {
+      "epoch": 559.2105263157895,
+      "grad_norm": 1.1532487869262695,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 85000
+    },
+    {
+      "epoch": 559.2763157894736,
+      "grad_norm": 1.2575640678405762,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 85010
+    },
+    {
+      "epoch": 559.3421052631579,
+      "grad_norm": 1.3786592483520508,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 85020
+    },
+    {
+      "epoch": 559.4078947368421,
+      "grad_norm": 0.8900845050811768,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 85030
+    },
+    {
+      "epoch": 559.4736842105264,
+      "grad_norm": 1.2417876720428467,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 85040
+    },
+    {
+      "epoch": 559.5394736842105,
+      "grad_norm": 1.2613035440444946,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 85050
+    },
+    {
+      "epoch": 559.6052631578947,
+      "grad_norm": 0.8748085498809814,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 85060
+    },
+    {
+      "epoch": 559.671052631579,
+      "grad_norm": 1.059360384941101,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 85070
+    },
+    {
+      "epoch": 559.7368421052631,
+      "grad_norm": 0.8595458269119263,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 85080
+    },
+    {
+      "epoch": 559.8026315789474,
+      "grad_norm": 1.0253055095672607,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 85090
+    },
+    {
+      "epoch": 559.8684210526316,
+      "grad_norm": 1.1457302570343018,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 85100
+    },
+    {
+      "epoch": 559.9342105263158,
+      "grad_norm": 1.5385023355484009,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 85110
+    },
+    {
+      "epoch": 560.0,
+      "grad_norm": 1.041575312614441,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 85120
+    },
+    {
+      "epoch": 560.0657894736842,
+      "grad_norm": 0.9644649624824524,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 85130
+    },
+    {
+      "epoch": 560.1315789473684,
+      "grad_norm": 1.3442351818084717,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 85140
+    },
+    {
+      "epoch": 560.1973684210526,
+      "grad_norm": 1.4101388454437256,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 85150
+    },
+    {
+      "epoch": 560.2631578947369,
+      "grad_norm": 1.3312458992004395,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 85160
+    },
+    {
+      "epoch": 560.328947368421,
+      "grad_norm": 1.2617542743682861,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 85170
+    },
+    {
+      "epoch": 560.3947368421053,
+      "grad_norm": 0.9706904888153076,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 85180
+    },
+    {
+      "epoch": 560.4605263157895,
+      "grad_norm": 1.0391241312026978,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 85190
+    },
+    {
+      "epoch": 560.5263157894736,
+      "grad_norm": 0.7928785085678101,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 85200
+    },
+    {
+      "epoch": 560.5921052631579,
+      "grad_norm": 0.8561776280403137,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 85210
+    },
+    {
+      "epoch": 560.6578947368421,
+      "grad_norm": 1.097593903541565,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 85220
+    },
+    {
+      "epoch": 560.7236842105264,
+      "grad_norm": 0.9008827805519104,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 85230
+    },
+    {
+      "epoch": 560.7894736842105,
+      "grad_norm": 1.3910067081451416,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 85240
+    },
+    {
+      "epoch": 560.8552631578947,
+      "grad_norm": 1.2383750677108765,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 85250
+    },
+    {
+      "epoch": 560.921052631579,
+      "grad_norm": 1.2001198530197144,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 85260
+    },
+    {
+      "epoch": 560.9868421052631,
+      "grad_norm": 1.0926671028137207,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 85270
+    },
+    {
+      "epoch": 561.0526315789474,
+      "grad_norm": 1.1885424852371216,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 85280
+    },
+    {
+      "epoch": 561.1184210526316,
+      "grad_norm": 1.3044886589050293,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 85290
+    },
+    {
+      "epoch": 561.1842105263158,
+      "grad_norm": 1.361910104751587,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 85300
+    },
+    {
+      "epoch": 561.25,
+      "grad_norm": 1.2132842540740967,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 85310
+    },
+    {
+      "epoch": 561.3157894736842,
+      "grad_norm": 1.2285622358322144,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 85320
+    },
+    {
+      "epoch": 561.3815789473684,
+      "grad_norm": 1.0628094673156738,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 85330
+    },
+    {
+      "epoch": 561.4473684210526,
+      "grad_norm": 1.0431807041168213,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 85340
+    },
+    {
+      "epoch": 561.5131578947369,
+      "grad_norm": 0.8674909472465515,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 85350
+    },
+    {
+      "epoch": 561.578947368421,
+      "grad_norm": 0.8657084107398987,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 85360
+    },
+    {
+      "epoch": 561.6447368421053,
+      "grad_norm": 1.0542387962341309,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 85370
+    },
+    {
+      "epoch": 561.7105263157895,
+      "grad_norm": 1.1342475414276123,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 85380
+    },
+    {
+      "epoch": 561.7763157894736,
+      "grad_norm": 1.0846221446990967,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 85390
+    },
+    {
+      "epoch": 561.8421052631579,
+      "grad_norm": 1.0843048095703125,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 85400
+    },
+    {
+      "epoch": 561.9078947368421,
+      "grad_norm": 1.266693115234375,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 85410
+    },
+    {
+      "epoch": 561.9736842105264,
+      "grad_norm": 1.3362053632736206,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 85420
+    },
+    {
+      "epoch": 562.0394736842105,
+      "grad_norm": 1.1382287740707397,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 85430
+    },
+    {
+      "epoch": 562.1052631578947,
+      "grad_norm": 1.0653843879699707,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 85440
+    },
+    {
+      "epoch": 562.171052631579,
+      "grad_norm": 0.9367539882659912,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 85450
+    },
+    {
+      "epoch": 562.2368421052631,
+      "grad_norm": 0.8476194143295288,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 85460
+    },
+    {
+      "epoch": 562.3026315789474,
+      "grad_norm": 1.0958266258239746,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 85470
+    },
+    {
+      "epoch": 562.3684210526316,
+      "grad_norm": 1.2579114437103271,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 85480
+    },
+    {
+      "epoch": 562.4342105263158,
+      "grad_norm": 1.1456224918365479,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 85490
+    },
+    {
+      "epoch": 562.5,
+      "grad_norm": 1.2678587436676025,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 85500
+    },
+    {
+      "epoch": 562.5657894736842,
+      "grad_norm": 1.0048190355300903,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 85510
+    },
+    {
+      "epoch": 562.6315789473684,
+      "grad_norm": 0.7200008630752563,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 85520
+    },
+    {
+      "epoch": 562.6973684210526,
+      "grad_norm": 1.4088075160980225,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 85530
+    },
+    {
+      "epoch": 562.7631578947369,
+      "grad_norm": 1.2613880634307861,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 85540
+    },
+    {
+      "epoch": 562.828947368421,
+      "grad_norm": 1.0894513130187988,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 85550
+    },
+    {
+      "epoch": 562.8947368421053,
+      "grad_norm": 1.0057373046875,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 85560
+    },
+    {
+      "epoch": 562.9605263157895,
+      "grad_norm": 1.1327048540115356,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 85570
+    },
+    {
+      "epoch": 563.0263157894736,
+      "grad_norm": 1.3295555114746094,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 85580
+    },
+    {
+      "epoch": 563.0921052631579,
+      "grad_norm": 1.098968505859375,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 85590
+    },
+    {
+      "epoch": 563.1578947368421,
+      "grad_norm": 1.1151443719863892,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 85600
+    },
+    {
+      "epoch": 563.2236842105264,
+      "grad_norm": 1.261618971824646,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 85610
+    },
+    {
+      "epoch": 563.2894736842105,
+      "grad_norm": 0.6959012150764465,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 85620
+    },
+    {
+      "epoch": 563.3552631578947,
+      "grad_norm": 0.9894868731498718,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 85630
+    },
+    {
+      "epoch": 563.421052631579,
+      "grad_norm": 0.8784093260765076,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 85640
+    },
+    {
+      "epoch": 563.4868421052631,
+      "grad_norm": 0.5873738527297974,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 85650
+    },
+    {
+      "epoch": 563.5526315789474,
+      "grad_norm": 0.7651575803756714,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 85660
+    },
+    {
+      "epoch": 563.6184210526316,
+      "grad_norm": 0.6802495718002319,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 85670
+    },
+    {
+      "epoch": 563.6842105263158,
+      "grad_norm": 0.948125422000885,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 85680
+    },
+    {
+      "epoch": 563.75,
+      "grad_norm": 1.1198630332946777,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 85690
+    },
+    {
+      "epoch": 563.8157894736842,
+      "grad_norm": 1.2532132863998413,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 85700
+    },
+    {
+      "epoch": 563.8815789473684,
+      "grad_norm": 0.9288182854652405,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 85710
+    },
+    {
+      "epoch": 563.9473684210526,
+      "grad_norm": 1.1498973369598389,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 85720
+    },
+    {
+      "epoch": 564.0131578947369,
+      "grad_norm": 0.9289548397064209,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 85730
+    },
+    {
+      "epoch": 564.078947368421,
+      "grad_norm": 0.6125806570053101,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 85740
+    },
+    {
+      "epoch": 564.1447368421053,
+      "grad_norm": 0.762289822101593,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 85750
+    },
+    {
+      "epoch": 564.2105263157895,
+      "grad_norm": 1.2654229402542114,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 85760
+    },
+    {
+      "epoch": 564.2763157894736,
+      "grad_norm": 1.2169716358184814,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 85770
+    },
+    {
+      "epoch": 564.3421052631579,
+      "grad_norm": 1.0589436292648315,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 85780
+    },
+    {
+      "epoch": 564.4078947368421,
+      "grad_norm": 0.9136533141136169,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 85790
+    },
+    {
+      "epoch": 564.4736842105264,
+      "grad_norm": 0.831851601600647,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 85800
+    },
+    {
+      "epoch": 564.5394736842105,
+      "grad_norm": 0.9365316033363342,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 85810
+    },
+    {
+      "epoch": 564.6052631578947,
+      "grad_norm": 1.4530662298202515,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 85820
+    },
+    {
+      "epoch": 564.671052631579,
+      "grad_norm": 1.1239123344421387,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 85830
+    },
+    {
+      "epoch": 564.7368421052631,
+      "grad_norm": 1.2323276996612549,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 85840
+    },
+    {
+      "epoch": 564.8026315789474,
+      "grad_norm": 0.9077661633491516,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 85850
+    },
+    {
+      "epoch": 564.8684210526316,
+      "grad_norm": 1.2481967210769653,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 85860
+    },
+    {
+      "epoch": 564.9342105263158,
+      "grad_norm": 1.2315006256103516,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 85870
+    },
+    {
+      "epoch": 565.0,
+      "grad_norm": 1.2639683485031128,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 85880
+    },
+    {
+      "epoch": 565.0657894736842,
+      "grad_norm": 1.2742642164230347,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 85890
+    },
+    {
+      "epoch": 565.1315789473684,
+      "grad_norm": 1.1369564533233643,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 85900
+    },
+    {
+      "epoch": 565.1973684210526,
+      "grad_norm": 1.4705597162246704,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 85910
+    },
+    {
+      "epoch": 565.2631578947369,
+      "grad_norm": 1.326097846031189,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 85920
+    },
+    {
+      "epoch": 565.328947368421,
+      "grad_norm": 0.9751961827278137,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 85930
+    },
+    {
+      "epoch": 565.3947368421053,
+      "grad_norm": 1.3746706247329712,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 85940
+    },
+    {
+      "epoch": 565.4605263157895,
+      "grad_norm": 1.2249608039855957,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 85950
+    },
+    {
+      "epoch": 565.5263157894736,
+      "grad_norm": 1.0640735626220703,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 85960
+    },
+    {
+      "epoch": 565.5921052631579,
+      "grad_norm": 1.0786769390106201,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 85970
+    },
+    {
+      "epoch": 565.6578947368421,
+      "grad_norm": 1.5025138854980469,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 85980
+    },
+    {
+      "epoch": 565.7236842105264,
+      "grad_norm": 1.311964511871338,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 85990
+    },
+    {
+      "epoch": 565.7894736842105,
+      "grad_norm": 1.1438058614730835,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 86000
+    },
+    {
+      "epoch": 565.8552631578947,
+      "grad_norm": 0.921307384967804,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 86010
+    },
+    {
+      "epoch": 565.921052631579,
+      "grad_norm": 1.0109132528305054,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 86020
+    },
+    {
+      "epoch": 565.9868421052631,
+      "grad_norm": 1.3036121129989624,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 86030
+    },
+    {
+      "epoch": 566.0526315789474,
+      "grad_norm": 1.1101711988449097,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 86040
+    },
+    {
+      "epoch": 566.1184210526316,
+      "grad_norm": 1.3747954368591309,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 86050
+    },
+    {
+      "epoch": 566.1842105263158,
+      "grad_norm": 1.133420705795288,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 86060
+    },
+    {
+      "epoch": 566.25,
+      "grad_norm": 1.0541547536849976,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 86070
+    },
+    {
+      "epoch": 566.3157894736842,
+      "grad_norm": 1.0774303674697876,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 86080
+    },
+    {
+      "epoch": 566.3815789473684,
+      "grad_norm": 1.3107317686080933,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 86090
+    },
+    {
+      "epoch": 566.4473684210526,
+      "grad_norm": 0.8362441658973694,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 86100
+    },
+    {
+      "epoch": 566.5131578947369,
+      "grad_norm": 1.1291522979736328,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 86110
+    },
+    {
+      "epoch": 566.578947368421,
+      "grad_norm": 1.062232494354248,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 86120
+    },
+    {
+      "epoch": 566.6447368421053,
+      "grad_norm": 1.1691906452178955,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 86130
+    },
+    {
+      "epoch": 566.7105263157895,
+      "grad_norm": 1.1058992147445679,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 86140
+    },
+    {
+      "epoch": 566.7763157894736,
+      "grad_norm": 0.6713123321533203,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 86150
+    },
+    {
+      "epoch": 566.8421052631579,
+      "grad_norm": 0.7876428365707397,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 86160
+    },
+    {
+      "epoch": 566.9078947368421,
+      "grad_norm": 1.0768256187438965,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 86170
+    },
+    {
+      "epoch": 566.9736842105264,
+      "grad_norm": 1.6893032789230347,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 86180
+    },
+    {
+      "epoch": 567.0394736842105,
+      "grad_norm": 0.7960553765296936,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 86190
+    },
+    {
+      "epoch": 567.1052631578947,
+      "grad_norm": 0.6956373453140259,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 86200
+    },
+    {
+      "epoch": 567.171052631579,
+      "grad_norm": 1.1536741256713867,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 86210
+    },
+    {
+      "epoch": 567.2368421052631,
+      "grad_norm": 0.8399643301963806,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 86220
+    },
+    {
+      "epoch": 567.3026315789474,
+      "grad_norm": 0.8098105192184448,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 86230
+    },
+    {
+      "epoch": 567.3684210526316,
+      "grad_norm": 0.8427768349647522,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 86240
+    },
+    {
+      "epoch": 567.4342105263158,
+      "grad_norm": 0.9141896963119507,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 86250
+    },
+    {
+      "epoch": 567.5,
+      "grad_norm": 1.6003632545471191,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 86260
+    },
+    {
+      "epoch": 567.5657894736842,
+      "grad_norm": 1.1154911518096924,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 86270
+    },
+    {
+      "epoch": 567.6315789473684,
+      "grad_norm": 1.1013606786727905,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 86280
+    },
+    {
+      "epoch": 567.6973684210526,
+      "grad_norm": 1.0994194746017456,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 86290
+    },
+    {
+      "epoch": 567.7631578947369,
+      "grad_norm": 0.9053112864494324,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 86300
+    },
+    {
+      "epoch": 567.828947368421,
+      "grad_norm": 1.4429203271865845,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 86310
+    },
+    {
+      "epoch": 567.8947368421053,
+      "grad_norm": 1.0973800420761108,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 86320
+    },
+    {
+      "epoch": 567.9605263157895,
+      "grad_norm": 1.2074874639511108,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 86330
+    },
+    {
+      "epoch": 568.0263157894736,
+      "grad_norm": 1.1671373844146729,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 86340
+    },
+    {
+      "epoch": 568.0921052631579,
+      "grad_norm": 1.0695538520812988,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 86350
+    },
+    {
+      "epoch": 568.1578947368421,
+      "grad_norm": 1.335986852645874,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 86360
+    },
+    {
+      "epoch": 568.2236842105264,
+      "grad_norm": 1.4699814319610596,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 86370
+    },
+    {
+      "epoch": 568.2894736842105,
+      "grad_norm": 1.071112871170044,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 86380
+    },
+    {
+      "epoch": 568.3552631578947,
+      "grad_norm": 1.1350902318954468,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 86390
+    },
+    {
+      "epoch": 568.421052631579,
+      "grad_norm": 1.3547534942626953,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 86400
+    },
+    {
+      "epoch": 568.4868421052631,
+      "grad_norm": 1.5289572477340698,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 86410
+    },
+    {
+      "epoch": 568.5526315789474,
+      "grad_norm": 1.0102062225341797,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 86420
+    },
+    {
+      "epoch": 568.6184210526316,
+      "grad_norm": 0.8501527309417725,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 86430
+    },
+    {
+      "epoch": 568.6842105263158,
+      "grad_norm": 0.9667913317680359,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 86440
+    },
+    {
+      "epoch": 568.75,
+      "grad_norm": 1.1724212169647217,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 86450
+    },
+    {
+      "epoch": 568.8157894736842,
+      "grad_norm": 1.0859756469726562,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 86460
+    },
+    {
+      "epoch": 568.8815789473684,
+      "grad_norm": 1.2196053266525269,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 86470
+    },
+    {
+      "epoch": 568.9473684210526,
+      "grad_norm": 0.9872120022773743,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 86480
+    },
+    {
+      "epoch": 569.0131578947369,
+      "grad_norm": 1.1090824604034424,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 86490
+    },
+    {
+      "epoch": 569.078947368421,
+      "grad_norm": 1.0156619548797607,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 86500
+    },
+    {
+      "epoch": 569.1447368421053,
+      "grad_norm": 0.6294759511947632,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 86510
+    },
+    {
+      "epoch": 569.2105263157895,
+      "grad_norm": 0.8278417587280273,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 86520
+    },
+    {
+      "epoch": 569.2763157894736,
+      "grad_norm": 0.8335527777671814,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 86530
+    },
+    {
+      "epoch": 569.3421052631579,
+      "grad_norm": 0.728941798210144,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 86540
+    },
+    {
+      "epoch": 569.4078947368421,
+      "grad_norm": 1.1835122108459473,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 86550
+    },
+    {
+      "epoch": 569.4736842105264,
+      "grad_norm": 1.3987102508544922,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 86560
+    },
+    {
+      "epoch": 569.5394736842105,
+      "grad_norm": 1.288018822669983,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 86570
+    },
+    {
+      "epoch": 569.6052631578947,
+      "grad_norm": 1.1396912336349487,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 86580
+    },
+    {
+      "epoch": 569.671052631579,
+      "grad_norm": 0.8793830871582031,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 86590
+    },
+    {
+      "epoch": 569.7368421052631,
+      "grad_norm": 0.8946493268013,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 86600
+    },
+    {
+      "epoch": 569.8026315789474,
+      "grad_norm": 0.8746724724769592,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 86610
+    },
+    {
+      "epoch": 569.8684210526316,
+      "grad_norm": 0.7413333654403687,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 86620
+    },
+    {
+      "epoch": 569.9342105263158,
+      "grad_norm": 1.0053696632385254,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 86630
+    },
+    {
+      "epoch": 570.0,
+      "grad_norm": 1.109494924545288,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 86640
+    },
+    {
+      "epoch": 570.0657894736842,
+      "grad_norm": 1.0057399272918701,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 86650
+    },
+    {
+      "epoch": 570.1315789473684,
+      "grad_norm": 0.9789668321609497,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 86660
+    },
+    {
+      "epoch": 570.1973684210526,
+      "grad_norm": 1.0310949087142944,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 86670
+    },
+    {
+      "epoch": 570.2631578947369,
+      "grad_norm": 0.9389269351959229,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 86680
+    },
+    {
+      "epoch": 570.328947368421,
+      "grad_norm": 0.8357144594192505,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 86690
+    },
+    {
+      "epoch": 570.3947368421053,
+      "grad_norm": 0.850374698638916,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 86700
+    },
+    {
+      "epoch": 570.4605263157895,
+      "grad_norm": 0.8948403596878052,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 86710
+    },
+    {
+      "epoch": 570.5263157894736,
+      "grad_norm": 1.1358330249786377,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 86720
+    },
+    {
+      "epoch": 570.5921052631579,
+      "grad_norm": 1.393713355064392,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 86730
+    },
+    {
+      "epoch": 570.6578947368421,
+      "grad_norm": 1.0343717336654663,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 86740
+    },
+    {
+      "epoch": 570.7236842105264,
+      "grad_norm": 0.8011754751205444,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 86750
+    },
+    {
+      "epoch": 570.7894736842105,
+      "grad_norm": 1.0685820579528809,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 86760
+    },
+    {
+      "epoch": 570.8552631578947,
+      "grad_norm": 1.2427990436553955,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 86770
+    },
+    {
+      "epoch": 570.921052631579,
+      "grad_norm": 1.2266359329223633,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 86780
+    },
+    {
+      "epoch": 570.9868421052631,
+      "grad_norm": 1.432370662689209,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 86790
+    },
+    {
+      "epoch": 571.0526315789474,
+      "grad_norm": 1.2090051174163818,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 86800
+    },
+    {
+      "epoch": 571.1184210526316,
+      "grad_norm": 0.8846063017845154,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 86810
+    },
+    {
+      "epoch": 571.1842105263158,
+      "grad_norm": 1.1082240343093872,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 86820
+    },
+    {
+      "epoch": 571.25,
+      "grad_norm": 1.0534160137176514,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 86830
+    },
+    {
+      "epoch": 571.3157894736842,
+      "grad_norm": 1.1436423063278198,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 86840
+    },
+    {
+      "epoch": 571.3815789473684,
+      "grad_norm": 0.8122831583023071,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 86850
+    },
+    {
+      "epoch": 571.4473684210526,
+      "grad_norm": 1.0820256471633911,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 86860
+    },
+    {
+      "epoch": 571.5131578947369,
+      "grad_norm": 0.848965585231781,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 86870
+    },
+    {
+      "epoch": 571.578947368421,
+      "grad_norm": 1.2625044584274292,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 86880
+    },
+    {
+      "epoch": 571.6447368421053,
+      "grad_norm": 1.3473550081253052,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 86890
+    },
+    {
+      "epoch": 571.7105263157895,
+      "grad_norm": 1.336229920387268,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 86900
+    },
+    {
+      "epoch": 571.7763157894736,
+      "grad_norm": 0.9691115617752075,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 86910
+    },
+    {
+      "epoch": 571.8421052631579,
+      "grad_norm": 1.2057015895843506,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 86920
+    },
+    {
+      "epoch": 571.9078947368421,
+      "grad_norm": 1.0640344619750977,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 86930
+    },
+    {
+      "epoch": 571.9736842105264,
+      "grad_norm": 1.1055980920791626,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 86940
+    },
+    {
+      "epoch": 572.0394736842105,
+      "grad_norm": 0.6833211779594421,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 86950
+    },
+    {
+      "epoch": 572.1052631578947,
+      "grad_norm": 1.1310468912124634,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 86960
+    },
+    {
+      "epoch": 572.171052631579,
+      "grad_norm": 1.280174970626831,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 86970
+    },
+    {
+      "epoch": 572.2368421052631,
+      "grad_norm": 1.0595585107803345,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 86980
+    },
+    {
+      "epoch": 572.3026315789474,
+      "grad_norm": 1.1463285684585571,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 86990
+    },
+    {
+      "epoch": 572.3684210526316,
+      "grad_norm": 1.0951709747314453,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 87000
+    },
+    {
+      "epoch": 572.4342105263158,
+      "grad_norm": 1.1616261005401611,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 87010
+    },
+    {
+      "epoch": 572.5,
+      "grad_norm": 1.0127562284469604,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 87020
+    },
+    {
+      "epoch": 572.5657894736842,
+      "grad_norm": 1.3694489002227783,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 87030
+    },
+    {
+      "epoch": 572.6315789473684,
+      "grad_norm": 1.5818332433700562,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 87040
+    },
+    {
+      "epoch": 572.6973684210526,
+      "grad_norm": 1.6021660566329956,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 87050
+    },
+    {
+      "epoch": 572.7631578947369,
+      "grad_norm": 1.4695814847946167,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 87060
+    },
+    {
+      "epoch": 572.828947368421,
+      "grad_norm": 1.1281092166900635,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 87070
+    },
+    {
+      "epoch": 572.8947368421053,
+      "grad_norm": 1.168941617012024,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 87080
+    },
+    {
+      "epoch": 572.9605263157895,
+      "grad_norm": 1.1322778463363647,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 87090
+    },
+    {
+      "epoch": 573.0263157894736,
+      "grad_norm": 1.4014652967453003,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 87100
+    },
+    {
+      "epoch": 573.0921052631579,
+      "grad_norm": 1.2505130767822266,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 87110
+    },
+    {
+      "epoch": 573.1578947368421,
+      "grad_norm": 1.029102087020874,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 87120
+    },
+    {
+      "epoch": 573.2236842105264,
+      "grad_norm": 1.2195322513580322,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 87130
+    },
+    {
+      "epoch": 573.2894736842105,
+      "grad_norm": 1.3658270835876465,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 87140
+    },
+    {
+      "epoch": 573.3552631578947,
+      "grad_norm": 1.0435492992401123,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 87150
+    },
+    {
+      "epoch": 573.421052631579,
+      "grad_norm": 1.1018818616867065,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 87160
+    },
+    {
+      "epoch": 573.4868421052631,
+      "grad_norm": 0.9364510774612427,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 87170
+    },
+    {
+      "epoch": 573.5526315789474,
+      "grad_norm": 1.258358359336853,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 87180
+    },
+    {
+      "epoch": 573.6184210526316,
+      "grad_norm": 1.2375657558441162,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 87190
+    },
+    {
+      "epoch": 573.6842105263158,
+      "grad_norm": 1.2008768320083618,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 87200
+    },
+    {
+      "epoch": 573.75,
+      "grad_norm": 0.818011999130249,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 87210
+    },
+    {
+      "epoch": 573.8157894736842,
+      "grad_norm": 1.2211856842041016,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 87220
+    },
+    {
+      "epoch": 573.8815789473684,
+      "grad_norm": 1.0101572275161743,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 87230
+    },
+    {
+      "epoch": 573.9473684210526,
+      "grad_norm": 1.2342889308929443,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 87240
+    },
+    {
+      "epoch": 574.0131578947369,
+      "grad_norm": 0.9465478658676147,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 87250
+    },
+    {
+      "epoch": 574.078947368421,
+      "grad_norm": 0.9019796848297119,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 87260
+    },
+    {
+      "epoch": 574.1447368421053,
+      "grad_norm": 1.0364303588867188,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 87270
+    },
+    {
+      "epoch": 574.2105263157895,
+      "grad_norm": 1.4639605283737183,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 87280
+    },
+    {
+      "epoch": 574.2763157894736,
+      "grad_norm": 1.2540547847747803,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 87290
+    },
+    {
+      "epoch": 574.3421052631579,
+      "grad_norm": 1.2065496444702148,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 87300
+    },
+    {
+      "epoch": 574.4078947368421,
+      "grad_norm": 1.3892549276351929,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 87310
+    },
+    {
+      "epoch": 574.4736842105264,
+      "grad_norm": 1.1445990800857544,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 87320
+    },
+    {
+      "epoch": 574.5394736842105,
+      "grad_norm": 1.227105975151062,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 87330
+    },
+    {
+      "epoch": 574.6052631578947,
+      "grad_norm": 1.262370228767395,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 87340
+    },
+    {
+      "epoch": 574.671052631579,
+      "grad_norm": 1.0574274063110352,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 87350
+    },
+    {
+      "epoch": 574.7368421052631,
+      "grad_norm": 0.9377652406692505,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 87360
+    },
+    {
+      "epoch": 574.8026315789474,
+      "grad_norm": 0.9124358296394348,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 87370
+    },
+    {
+      "epoch": 574.8684210526316,
+      "grad_norm": 1.0424607992172241,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 87380
+    },
+    {
+      "epoch": 574.9342105263158,
+      "grad_norm": 1.2752419710159302,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 87390
+    },
+    {
+      "epoch": 575.0,
+      "grad_norm": 1.1712169647216797,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 87400
+    },
+    {
+      "epoch": 575.0657894736842,
+      "grad_norm": 1.0214799642562866,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 87410
+    },
+    {
+      "epoch": 575.1315789473684,
+      "grad_norm": 0.9856082797050476,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 87420
+    },
+    {
+      "epoch": 575.1973684210526,
+      "grad_norm": 1.4361339807510376,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 87430
+    },
+    {
+      "epoch": 575.2631578947369,
+      "grad_norm": 1.2261438369750977,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 87440
+    },
+    {
+      "epoch": 575.328947368421,
+      "grad_norm": 1.267825722694397,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 87450
+    },
+    {
+      "epoch": 575.3947368421053,
+      "grad_norm": 1.1174856424331665,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 87460
+    },
+    {
+      "epoch": 575.4605263157895,
+      "grad_norm": 1.3079813718795776,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 87470
+    },
+    {
+      "epoch": 575.5263157894736,
+      "grad_norm": 1.29449462890625,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 87480
+    },
+    {
+      "epoch": 575.5921052631579,
+      "grad_norm": 1.4480912685394287,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 87490
+    },
+    {
+      "epoch": 575.6578947368421,
+      "grad_norm": 1.2274054288864136,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 87500
+    },
+    {
+      "epoch": 575.7236842105264,
+      "grad_norm": 1.114770770072937,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 87510
+    },
+    {
+      "epoch": 575.7894736842105,
+      "grad_norm": 1.2212456464767456,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 87520
+    },
+    {
+      "epoch": 575.8552631578947,
+      "grad_norm": 0.8942835927009583,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 87530
+    },
+    {
+      "epoch": 575.921052631579,
+      "grad_norm": 1.293633222579956,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 87540
+    },
+    {
+      "epoch": 575.9868421052631,
+      "grad_norm": 1.2598819732666016,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 87550
+    },
+    {
+      "epoch": 576.0526315789474,
+      "grad_norm": 1.32999849319458,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 87560
+    },
+    {
+      "epoch": 576.1184210526316,
+      "grad_norm": 1.2802047729492188,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 87570
+    },
+    {
+      "epoch": 576.1842105263158,
+      "grad_norm": 1.7709383964538574,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 87580
+    },
+    {
+      "epoch": 576.25,
+      "grad_norm": 1.3085800409317017,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 87590
+    },
+    {
+      "epoch": 576.3157894736842,
+      "grad_norm": 1.2775263786315918,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 87600
+    },
+    {
+      "epoch": 576.3815789473684,
+      "grad_norm": 1.5602463483810425,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 87610
+    },
+    {
+      "epoch": 576.4473684210526,
+      "grad_norm": 1.5633126497268677,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 87620
+    },
+    {
+      "epoch": 576.5131578947369,
+      "grad_norm": 1.0983535051345825,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 87630
+    },
+    {
+      "epoch": 576.578947368421,
+      "grad_norm": 1.0312081575393677,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 87640
+    },
+    {
+      "epoch": 576.6447368421053,
+      "grad_norm": 1.3446412086486816,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 87650
+    },
+    {
+      "epoch": 576.7105263157895,
+      "grad_norm": 1.1297847032546997,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 87660
+    },
+    {
+      "epoch": 576.7763157894736,
+      "grad_norm": 1.5967837572097778,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 87670
+    },
+    {
+      "epoch": 576.8421052631579,
+      "grad_norm": 1.1126961708068848,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 87680
+    },
+    {
+      "epoch": 576.9078947368421,
+      "grad_norm": 1.0181384086608887,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 87690
+    },
+    {
+      "epoch": 576.9736842105264,
+      "grad_norm": 0.9504018425941467,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 87700
+    },
+    {
+      "epoch": 577.0394736842105,
+      "grad_norm": 1.2466793060302734,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 87710
+    },
+    {
+      "epoch": 577.1052631578947,
+      "grad_norm": 0.9084358215332031,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 87720
+    },
+    {
+      "epoch": 577.171052631579,
+      "grad_norm": 1.0614901781082153,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 87730
+    },
+    {
+      "epoch": 577.2368421052631,
+      "grad_norm": 1.036545991897583,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 87740
+    },
+    {
+      "epoch": 577.3026315789474,
+      "grad_norm": 0.8779444098472595,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 87750
+    },
+    {
+      "epoch": 577.3684210526316,
+      "grad_norm": 0.7535855174064636,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 87760
+    },
+    {
+      "epoch": 577.4342105263158,
+      "grad_norm": 1.1626191139221191,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 87770
+    },
+    {
+      "epoch": 577.5,
+      "grad_norm": 1.0589649677276611,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 87780
+    },
+    {
+      "epoch": 577.5657894736842,
+      "grad_norm": 1.277850866317749,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 87790
+    },
+    {
+      "epoch": 577.6315789473684,
+      "grad_norm": 0.9819774031639099,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 87800
+    },
+    {
+      "epoch": 577.6973684210526,
+      "grad_norm": 0.9045615792274475,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 87810
+    },
+    {
+      "epoch": 577.7631578947369,
+      "grad_norm": 1.1192935705184937,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 87820
+    },
+    {
+      "epoch": 577.828947368421,
+      "grad_norm": 1.2856429815292358,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 87830
+    },
+    {
+      "epoch": 577.8947368421053,
+      "grad_norm": 1.3235225677490234,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 87840
+    },
+    {
+      "epoch": 577.9605263157895,
+      "grad_norm": 0.8897786140441895,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 87850
+    },
+    {
+      "epoch": 578.0263157894736,
+      "grad_norm": 0.9557298421859741,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 87860
+    },
+    {
+      "epoch": 578.0921052631579,
+      "grad_norm": 1.0050173997879028,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 87870
+    },
+    {
+      "epoch": 578.1578947368421,
+      "grad_norm": 0.8767371773719788,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 87880
+    },
+    {
+      "epoch": 578.2236842105264,
+      "grad_norm": 1.2130120992660522,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 87890
+    },
+    {
+      "epoch": 578.2894736842105,
+      "grad_norm": 1.0449823141098022,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 87900
+    },
+    {
+      "epoch": 578.3552631578947,
+      "grad_norm": 0.840207040309906,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 87910
+    },
+    {
+      "epoch": 578.421052631579,
+      "grad_norm": 0.847531795501709,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 87920
+    },
+    {
+      "epoch": 578.4868421052631,
+      "grad_norm": 1.5373563766479492,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 87930
+    },
+    {
+      "epoch": 578.5526315789474,
+      "grad_norm": 1.208318829536438,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 87940
+    },
+    {
+      "epoch": 578.6184210526316,
+      "grad_norm": 1.1403499841690063,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 87950
+    },
+    {
+      "epoch": 578.6842105263158,
+      "grad_norm": 1.375062108039856,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 87960
+    },
+    {
+      "epoch": 578.75,
+      "grad_norm": 0.9031561017036438,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 87970
+    },
+    {
+      "epoch": 578.8157894736842,
+      "grad_norm": 1.3210184574127197,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 87980
+    },
+    {
+      "epoch": 578.8815789473684,
+      "grad_norm": 1.013850450515747,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 87990
+    },
+    {
+      "epoch": 578.9473684210526,
+      "grad_norm": 1.104063630104065,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 88000
+    },
+    {
+      "epoch": 579.0131578947369,
+      "grad_norm": 0.8655098080635071,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 88010
+    },
+    {
+      "epoch": 579.078947368421,
+      "grad_norm": 1.320491075515747,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 88020
+    },
+    {
+      "epoch": 579.1447368421053,
+      "grad_norm": 1.221070408821106,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 88030
+    },
+    {
+      "epoch": 579.2105263157895,
+      "grad_norm": 1.0820084810256958,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 88040
+    },
+    {
+      "epoch": 579.2763157894736,
+      "grad_norm": 1.2061529159545898,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 88050
+    },
+    {
+      "epoch": 579.3421052631579,
+      "grad_norm": 0.9127936959266663,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 88060
+    },
+    {
+      "epoch": 579.4078947368421,
+      "grad_norm": 1.4120718240737915,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 88070
+    },
+    {
+      "epoch": 579.4736842105264,
+      "grad_norm": 1.0153974294662476,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 88080
+    },
+    {
+      "epoch": 579.5394736842105,
+      "grad_norm": 1.3774847984313965,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 88090
+    },
+    {
+      "epoch": 579.6052631578947,
+      "grad_norm": 1.1112881898880005,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 88100
+    },
+    {
+      "epoch": 579.671052631579,
+      "grad_norm": 1.1673177480697632,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 88110
+    },
+    {
+      "epoch": 579.7368421052631,
+      "grad_norm": 1.160556674003601,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 88120
+    },
+    {
+      "epoch": 579.8026315789474,
+      "grad_norm": 1.118118405342102,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 88130
+    },
+    {
+      "epoch": 579.8684210526316,
+      "grad_norm": 0.7637477517127991,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 88140
+    },
+    {
+      "epoch": 579.9342105263158,
+      "grad_norm": 1.3707367181777954,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 88150
+    },
+    {
+      "epoch": 580.0,
+      "grad_norm": 1.009476900100708,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 88160
+    },
+    {
+      "epoch": 580.0657894736842,
+      "grad_norm": 1.3438063859939575,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 88170
+    },
+    {
+      "epoch": 580.1315789473684,
+      "grad_norm": 1.126423716545105,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 88180
+    },
+    {
+      "epoch": 580.1973684210526,
+      "grad_norm": 1.1746346950531006,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 88190
+    },
+    {
+      "epoch": 580.2631578947369,
+      "grad_norm": 1.0661026239395142,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 88200
+    },
+    {
+      "epoch": 580.328947368421,
+      "grad_norm": 0.9947375059127808,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 88210
+    },
+    {
+      "epoch": 580.3947368421053,
+      "grad_norm": 1.178560733795166,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 88220
+    },
+    {
+      "epoch": 580.4605263157895,
+      "grad_norm": 1.3273268938064575,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 88230
+    },
+    {
+      "epoch": 580.5263157894736,
+      "grad_norm": 1.040791392326355,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 88240
+    },
+    {
+      "epoch": 580.5921052631579,
+      "grad_norm": 1.4010847806930542,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 88250
+    },
+    {
+      "epoch": 580.6578947368421,
+      "grad_norm": 1.2355036735534668,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 88260
+    },
+    {
+      "epoch": 580.7236842105264,
+      "grad_norm": 0.9420546889305115,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 88270
+    },
+    {
+      "epoch": 580.7894736842105,
+      "grad_norm": 1.1292775869369507,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 88280
+    },
+    {
+      "epoch": 580.8552631578947,
+      "grad_norm": 1.118950366973877,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 88290
+    },
+    {
+      "epoch": 580.921052631579,
+      "grad_norm": 1.3669413328170776,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 88300
+    },
+    {
+      "epoch": 580.9868421052631,
+      "grad_norm": 1.1475030183792114,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 88310
+    },
+    {
+      "epoch": 581.0526315789474,
+      "grad_norm": 0.9832250475883484,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 88320
+    },
+    {
+      "epoch": 581.1184210526316,
+      "grad_norm": 1.0179389715194702,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 88330
+    },
+    {
+      "epoch": 581.1842105263158,
+      "grad_norm": 1.026995301246643,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 88340
+    },
+    {
+      "epoch": 581.25,
+      "grad_norm": 1.0577194690704346,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 88350
+    },
+    {
+      "epoch": 581.3157894736842,
+      "grad_norm": 0.8815690875053406,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 88360
+    },
+    {
+      "epoch": 581.3815789473684,
+      "grad_norm": 0.9633040428161621,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 88370
+    },
+    {
+      "epoch": 581.4473684210526,
+      "grad_norm": 1.289242148399353,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 88380
+    },
+    {
+      "epoch": 581.5131578947369,
+      "grad_norm": 0.8594449758529663,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 88390
+    },
+    {
+      "epoch": 581.578947368421,
+      "grad_norm": 0.8207187652587891,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 88400
+    },
+    {
+      "epoch": 581.6447368421053,
+      "grad_norm": 1.0943405628204346,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 88410
+    },
+    {
+      "epoch": 581.7105263157895,
+      "grad_norm": 0.834723711013794,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 88420
+    },
+    {
+      "epoch": 581.7763157894736,
+      "grad_norm": 1.02744722366333,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 88430
+    },
+    {
+      "epoch": 581.8421052631579,
+      "grad_norm": 1.1640406847000122,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 88440
+    },
+    {
+      "epoch": 581.9078947368421,
+      "grad_norm": 1.0775976181030273,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 88450
+    },
+    {
+      "epoch": 581.9736842105264,
+      "grad_norm": 1.163527250289917,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 88460
+    },
+    {
+      "epoch": 582.0394736842105,
+      "grad_norm": 0.8929876089096069,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 88470
+    },
+    {
+      "epoch": 582.1052631578947,
+      "grad_norm": 1.3309000730514526,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 88480
+    },
+    {
+      "epoch": 582.171052631579,
+      "grad_norm": 1.0838176012039185,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 88490
+    },
+    {
+      "epoch": 582.2368421052631,
+      "grad_norm": 1.5269906520843506,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 88500
+    },
+    {
+      "epoch": 582.3026315789474,
+      "grad_norm": 0.8997194170951843,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 88510
+    },
+    {
+      "epoch": 582.3684210526316,
+      "grad_norm": 1.0836443901062012,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 88520
+    },
+    {
+      "epoch": 582.4342105263158,
+      "grad_norm": 0.7760714888572693,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 88530
+    },
+    {
+      "epoch": 582.5,
+      "grad_norm": 1.0541595220565796,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 88540
+    },
+    {
+      "epoch": 582.5657894736842,
+      "grad_norm": 1.0740338563919067,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 88550
+    },
+    {
+      "epoch": 582.6315789473684,
+      "grad_norm": 1.234104871749878,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 88560
+    },
+    {
+      "epoch": 582.6973684210526,
+      "grad_norm": 1.1293065547943115,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 88570
+    },
+    {
+      "epoch": 582.7631578947369,
+      "grad_norm": 0.9713568091392517,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 88580
+    },
+    {
+      "epoch": 582.828947368421,
+      "grad_norm": 0.9887568950653076,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 88590
+    },
+    {
+      "epoch": 582.8947368421053,
+      "grad_norm": 1.0662295818328857,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 88600
+    },
+    {
+      "epoch": 582.9605263157895,
+      "grad_norm": 1.09245765209198,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 88610
+    },
+    {
+      "epoch": 583.0263157894736,
+      "grad_norm": 1.12832510471344,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 88620
+    },
+    {
+      "epoch": 583.0921052631579,
+      "grad_norm": 1.6330440044403076,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 88630
+    },
+    {
+      "epoch": 583.1578947368421,
+      "grad_norm": 1.4116557836532593,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 88640
+    },
+    {
+      "epoch": 583.2236842105264,
+      "grad_norm": 1.313237190246582,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 88650
+    },
+    {
+      "epoch": 583.2894736842105,
+      "grad_norm": 1.496812105178833,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 88660
+    },
+    {
+      "epoch": 583.3552631578947,
+      "grad_norm": 1.3389391899108887,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 88670
+    },
+    {
+      "epoch": 583.421052631579,
+      "grad_norm": 0.9950214624404907,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 88680
+    },
+    {
+      "epoch": 583.4868421052631,
+      "grad_norm": 1.0509663820266724,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 88690
+    },
+    {
+      "epoch": 583.5526315789474,
+      "grad_norm": 1.2093925476074219,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 88700
+    },
+    {
+      "epoch": 583.6184210526316,
+      "grad_norm": 1.2887691259384155,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 88710
+    },
+    {
+      "epoch": 583.6842105263158,
+      "grad_norm": 1.2735077142715454,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 88720
+    },
+    {
+      "epoch": 583.75,
+      "grad_norm": 1.4364982843399048,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 88730
+    },
+    {
+      "epoch": 583.8157894736842,
+      "grad_norm": 1.1002883911132812,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 88740
+    },
+    {
+      "epoch": 583.8815789473684,
+      "grad_norm": 0.9703975915908813,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 88750
+    },
+    {
+      "epoch": 583.9473684210526,
+      "grad_norm": 0.8523468971252441,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 88760
+    },
+    {
+      "epoch": 584.0131578947369,
+      "grad_norm": 1.2249200344085693,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 88770
+    },
+    {
+      "epoch": 584.078947368421,
+      "grad_norm": 1.7450804710388184,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 88780
+    },
+    {
+      "epoch": 584.1447368421053,
+      "grad_norm": 1.176170825958252,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 88790
+    },
+    {
+      "epoch": 584.2105263157895,
+      "grad_norm": 0.7471736073493958,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 88800
+    },
+    {
+      "epoch": 584.2763157894736,
+      "grad_norm": 1.824398159980774,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 88810
+    },
+    {
+      "epoch": 584.3421052631579,
+      "grad_norm": 1.3423678874969482,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 88820
+    },
+    {
+      "epoch": 584.4078947368421,
+      "grad_norm": 1.2872895002365112,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 88830
+    },
+    {
+      "epoch": 584.4736842105264,
+      "grad_norm": 1.6336323022842407,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 88840
+    },
+    {
+      "epoch": 584.5394736842105,
+      "grad_norm": 1.0463390350341797,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 88850
+    },
+    {
+      "epoch": 584.6052631578947,
+      "grad_norm": 1.3081940412521362,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 88860
+    },
+    {
+      "epoch": 584.671052631579,
+      "grad_norm": 1.1893011331558228,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 88870
+    },
+    {
+      "epoch": 584.7368421052631,
+      "grad_norm": 1.2873516082763672,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 88880
+    },
+    {
+      "epoch": 584.8026315789474,
+      "grad_norm": 0.9724097847938538,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 88890
+    },
+    {
+      "epoch": 584.8684210526316,
+      "grad_norm": 0.8140742778778076,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 88900
+    },
+    {
+      "epoch": 584.9342105263158,
+      "grad_norm": 1.4383323192596436,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 88910
+    },
+    {
+      "epoch": 585.0,
+      "grad_norm": 0.8941511511802673,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 88920
+    },
+    {
+      "epoch": 585.0657894736842,
+      "grad_norm": 1.3522826433181763,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 88930
+    },
+    {
+      "epoch": 585.1315789473684,
+      "grad_norm": 1.0675338506698608,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 88940
+    },
+    {
+      "epoch": 585.1973684210526,
+      "grad_norm": 1.4727058410644531,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 88950
+    },
+    {
+      "epoch": 585.2631578947369,
+      "grad_norm": 1.2497466802597046,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 88960
+    },
+    {
+      "epoch": 585.328947368421,
+      "grad_norm": 0.8785649538040161,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 88970
+    },
+    {
+      "epoch": 585.3947368421053,
+      "grad_norm": 1.2095212936401367,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 88980
+    },
+    {
+      "epoch": 585.4605263157895,
+      "grad_norm": 1.3274887800216675,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 88990
+    },
+    {
+      "epoch": 585.5263157894736,
+      "grad_norm": 1.0932297706604004,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 89000
+    },
+    {
+      "epoch": 585.5921052631579,
+      "grad_norm": 1.1494570970535278,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 89010
+    },
+    {
+      "epoch": 585.6578947368421,
+      "grad_norm": 1.5100291967391968,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 89020
+    },
+    {
+      "epoch": 585.7236842105264,
+      "grad_norm": 1.3086682558059692,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 89030
+    },
+    {
+      "epoch": 585.7894736842105,
+      "grad_norm": 1.1666781902313232,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 89040
+    },
+    {
+      "epoch": 585.8552631578947,
+      "grad_norm": 0.9592698216438293,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 89050
+    },
+    {
+      "epoch": 585.921052631579,
+      "grad_norm": 0.6694148182868958,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 89060
+    },
+    {
+      "epoch": 585.9868421052631,
+      "grad_norm": 1.0263710021972656,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 89070
+    },
+    {
+      "epoch": 586.0526315789474,
+      "grad_norm": 0.6993970274925232,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 89080
+    },
+    {
+      "epoch": 586.1184210526316,
+      "grad_norm": 0.8670745491981506,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 89090
+    },
+    {
+      "epoch": 586.1842105263158,
+      "grad_norm": 0.7880876064300537,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 89100
+    },
+    {
+      "epoch": 586.25,
+      "grad_norm": 0.9657894968986511,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 89110
+    },
+    {
+      "epoch": 586.3157894736842,
+      "grad_norm": 0.7474163174629211,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 89120
+    },
+    {
+      "epoch": 586.3815789473684,
+      "grad_norm": 1.1458895206451416,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 89130
+    },
+    {
+      "epoch": 586.4473684210526,
+      "grad_norm": 1.0431797504425049,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 89140
+    },
+    {
+      "epoch": 586.5131578947369,
+      "grad_norm": 0.9211791157722473,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 89150
+    },
+    {
+      "epoch": 586.578947368421,
+      "grad_norm": 1.1954262256622314,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 89160
+    },
+    {
+      "epoch": 586.6447368421053,
+      "grad_norm": 1.164695143699646,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 89170
+    },
+    {
+      "epoch": 586.7105263157895,
+      "grad_norm": 1.2543039321899414,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 89180
+    },
+    {
+      "epoch": 586.7763157894736,
+      "grad_norm": 1.077324390411377,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 89190
+    },
+    {
+      "epoch": 586.8421052631579,
+      "grad_norm": 0.7734859585762024,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 89200
+    },
+    {
+      "epoch": 586.9078947368421,
+      "grad_norm": 1.3108208179473877,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 89210
+    },
+    {
+      "epoch": 586.9736842105264,
+      "grad_norm": 1.516038179397583,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 89220
+    },
+    {
+      "epoch": 587.0394736842105,
+      "grad_norm": 1.5509519577026367,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 89230
+    },
+    {
+      "epoch": 587.1052631578947,
+      "grad_norm": 1.3086330890655518,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 89240
+    },
+    {
+      "epoch": 587.171052631579,
+      "grad_norm": 1.3632980585098267,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 89250
+    },
+    {
+      "epoch": 587.2368421052631,
+      "grad_norm": 1.1639214754104614,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 89260
+    },
+    {
+      "epoch": 587.3026315789474,
+      "grad_norm": 1.0251145362854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 89270
+    },
+    {
+      "epoch": 587.3684210526316,
+      "grad_norm": 1.2207335233688354,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 89280
+    },
+    {
+      "epoch": 587.4342105263158,
+      "grad_norm": 1.3697402477264404,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 89290
+    },
+    {
+      "epoch": 587.5,
+      "grad_norm": 1.4069492816925049,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 89300
+    },
+    {
+      "epoch": 587.5657894736842,
+      "grad_norm": 1.0171831846237183,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 89310
+    },
+    {
+      "epoch": 587.6315789473684,
+      "grad_norm": 1.1951297521591187,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 89320
+    },
+    {
+      "epoch": 587.6973684210526,
+      "grad_norm": 0.9013453125953674,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 89330
+    },
+    {
+      "epoch": 587.7631578947369,
+      "grad_norm": 1.1464810371398926,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 89340
+    },
+    {
+      "epoch": 587.828947368421,
+      "grad_norm": 1.0271261930465698,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 89350
+    },
+    {
+      "epoch": 587.8947368421053,
+      "grad_norm": 1.106770396232605,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 89360
+    },
+    {
+      "epoch": 587.9605263157895,
+      "grad_norm": 0.9565857648849487,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 89370
+    },
+    {
+      "epoch": 588.0263157894736,
+      "grad_norm": 1.2642189264297485,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 89380
+    },
+    {
+      "epoch": 588.0921052631579,
+      "grad_norm": 1.1177488565444946,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 89390
+    },
+    {
+      "epoch": 588.1578947368421,
+      "grad_norm": 0.8865790367126465,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 89400
+    },
+    {
+      "epoch": 588.2236842105264,
+      "grad_norm": 1.007317066192627,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 89410
+    },
+    {
+      "epoch": 588.2894736842105,
+      "grad_norm": 1.041730523109436,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 89420
+    },
+    {
+      "epoch": 588.3552631578947,
+      "grad_norm": 1.389570713043213,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 89430
+    },
+    {
+      "epoch": 588.421052631579,
+      "grad_norm": 1.1564185619354248,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 89440
+    },
+    {
+      "epoch": 588.4868421052631,
+      "grad_norm": 1.2922906875610352,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 89450
+    },
+    {
+      "epoch": 588.5526315789474,
+      "grad_norm": 1.2938793897628784,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 89460
+    },
+    {
+      "epoch": 588.6184210526316,
+      "grad_norm": 0.9030840992927551,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 89470
+    },
+    {
+      "epoch": 588.6842105263158,
+      "grad_norm": 1.131744384765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 89480
+    },
+    {
+      "epoch": 588.75,
+      "grad_norm": 1.0331127643585205,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 89490
+    },
+    {
+      "epoch": 588.8157894736842,
+      "grad_norm": 0.9357874989509583,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 89500
+    },
+    {
+      "epoch": 588.8815789473684,
+      "grad_norm": 1.4031165838241577,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 89510
+    },
+    {
+      "epoch": 588.9473684210526,
+      "grad_norm": 1.0815694332122803,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 89520
+    },
+    {
+      "epoch": 589.0131578947369,
+      "grad_norm": 1.360446810722351,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 89530
+    },
+    {
+      "epoch": 589.078947368421,
+      "grad_norm": 1.0539952516555786,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 89540
+    },
+    {
+      "epoch": 589.1447368421053,
+      "grad_norm": 0.9572492837905884,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 89550
+    },
+    {
+      "epoch": 589.2105263157895,
+      "grad_norm": 1.1082371473312378,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 89560
+    },
+    {
+      "epoch": 589.2763157894736,
+      "grad_norm": 1.2691287994384766,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 89570
+    },
+    {
+      "epoch": 589.3421052631579,
+      "grad_norm": 1.6313940286636353,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 89580
+    },
+    {
+      "epoch": 589.4078947368421,
+      "grad_norm": 0.9120645523071289,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 89590
+    },
+    {
+      "epoch": 589.4736842105264,
+      "grad_norm": 1.050746202468872,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 89600
+    },
+    {
+      "epoch": 589.5394736842105,
+      "grad_norm": 1.3637466430664062,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 89610
+    },
+    {
+      "epoch": 589.6052631578947,
+      "grad_norm": 1.2993441820144653,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 89620
+    },
+    {
+      "epoch": 589.671052631579,
+      "grad_norm": 1.2083455324172974,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 89630
+    },
+    {
+      "epoch": 589.7368421052631,
+      "grad_norm": 0.9543986320495605,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 89640
+    },
+    {
+      "epoch": 589.8026315789474,
+      "grad_norm": 1.0169148445129395,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 89650
+    },
+    {
+      "epoch": 589.8684210526316,
+      "grad_norm": 0.9722794890403748,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 89660
+    },
+    {
+      "epoch": 589.9342105263158,
+      "grad_norm": 1.2600023746490479,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 89670
+    },
+    {
+      "epoch": 590.0,
+      "grad_norm": 1.084512710571289,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 89680
+    },
+    {
+      "epoch": 590.0657894736842,
+      "grad_norm": 0.9002571105957031,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 89690
+    },
+    {
+      "epoch": 590.1315789473684,
+      "grad_norm": 1.2132498025894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 89700
+    },
+    {
+      "epoch": 590.1973684210526,
+      "grad_norm": 1.4505194425582886,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 89710
+    },
+    {
+      "epoch": 590.2631578947369,
+      "grad_norm": 1.2071669101715088,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 89720
+    },
+    {
+      "epoch": 590.328947368421,
+      "grad_norm": 1.1249711513519287,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 89730
+    },
+    {
+      "epoch": 590.3947368421053,
+      "grad_norm": 0.9531528353691101,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 89740
+    },
+    {
+      "epoch": 590.4605263157895,
+      "grad_norm": 1.5327131748199463,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 89750
+    },
+    {
+      "epoch": 590.5263157894736,
+      "grad_norm": 1.2193597555160522,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 89760
+    },
+    {
+      "epoch": 590.5921052631579,
+      "grad_norm": 1.114341378211975,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 89770
+    },
+    {
+      "epoch": 590.6578947368421,
+      "grad_norm": 0.7572506070137024,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 89780
+    },
+    {
+      "epoch": 590.7236842105264,
+      "grad_norm": 1.0355122089385986,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 89790
+    },
+    {
+      "epoch": 590.7894736842105,
+      "grad_norm": 1.0882655382156372,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 89800
+    },
+    {
+      "epoch": 590.8552631578947,
+      "grad_norm": 1.2581169605255127,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 89810
+    },
+    {
+      "epoch": 590.921052631579,
+      "grad_norm": 1.334328055381775,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 89820
+    },
+    {
+      "epoch": 590.9868421052631,
+      "grad_norm": 1.0767223834991455,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 89830
+    },
+    {
+      "epoch": 591.0526315789474,
+      "grad_norm": 1.1012510061264038,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 89840
+    },
+    {
+      "epoch": 591.1184210526316,
+      "grad_norm": 0.7526869773864746,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 89850
+    },
+    {
+      "epoch": 591.1842105263158,
+      "grad_norm": 1.0117813348770142,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 89860
+    },
+    {
+      "epoch": 591.25,
+      "grad_norm": 1.2263928651809692,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 89870
+    },
+    {
+      "epoch": 591.3157894736842,
+      "grad_norm": 0.8752433657646179,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 89880
+    },
+    {
+      "epoch": 591.3815789473684,
+      "grad_norm": 1.4674887657165527,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 89890
+    },
+    {
+      "epoch": 591.4473684210526,
+      "grad_norm": 1.1503700017929077,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 89900
+    },
+    {
+      "epoch": 591.5131578947369,
+      "grad_norm": 1.218532681465149,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 89910
+    },
+    {
+      "epoch": 591.578947368421,
+      "grad_norm": 1.5845961570739746,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 89920
+    },
+    {
+      "epoch": 591.6447368421053,
+      "grad_norm": 1.585045576095581,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 89930
+    },
+    {
+      "epoch": 591.7105263157895,
+      "grad_norm": 1.2569578886032104,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 89940
+    },
+    {
+      "epoch": 591.7763157894736,
+      "grad_norm": 1.1375163793563843,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 89950
+    },
+    {
+      "epoch": 591.8421052631579,
+      "grad_norm": 0.8566263914108276,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 89960
+    },
+    {
+      "epoch": 591.9078947368421,
+      "grad_norm": 1.2207688093185425,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 89970
+    },
+    {
+      "epoch": 591.9736842105264,
+      "grad_norm": 1.099183440208435,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 89980
+    },
+    {
+      "epoch": 592.0394736842105,
+      "grad_norm": 1.3385899066925049,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 89990
+    },
+    {
+      "epoch": 592.1052631578947,
+      "grad_norm": 1.2732150554656982,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 90000
+    },
+    {
+      "epoch": 592.171052631579,
+      "grad_norm": 1.053571105003357,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 90010
+    },
+    {
+      "epoch": 592.2368421052631,
+      "grad_norm": 0.9966652393341064,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 90020
+    },
+    {
+      "epoch": 592.3026315789474,
+      "grad_norm": 1.0630309581756592,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 90030
+    },
+    {
+      "epoch": 592.3684210526316,
+      "grad_norm": 1.1324174404144287,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 90040
+    },
+    {
+      "epoch": 592.4342105263158,
+      "grad_norm": 1.2332628965377808,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 90050
+    },
+    {
+      "epoch": 592.5,
+      "grad_norm": 1.151996374130249,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 90060
+    },
+    {
+      "epoch": 592.5657894736842,
+      "grad_norm": 0.7464222311973572,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 90070
+    },
+    {
+      "epoch": 592.6315789473684,
+      "grad_norm": 0.8995870351791382,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 90080
+    },
+    {
+      "epoch": 592.6973684210526,
+      "grad_norm": 1.467040777206421,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 90090
+    },
+    {
+      "epoch": 592.7631578947369,
+      "grad_norm": 0.7274995446205139,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 90100
+    },
+    {
+      "epoch": 592.828947368421,
+      "grad_norm": 1.107151985168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 90110
+    },
+    {
+      "epoch": 592.8947368421053,
+      "grad_norm": 1.1053555011749268,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 90120
+    },
+    {
+      "epoch": 592.9605263157895,
+      "grad_norm": 0.884556770324707,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 90130
+    },
+    {
+      "epoch": 593.0263157894736,
+      "grad_norm": 1.09501051902771,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 90140
+    },
+    {
+      "epoch": 593.0921052631579,
+      "grad_norm": 1.0981241464614868,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 90150
+    },
+    {
+      "epoch": 593.1578947368421,
+      "grad_norm": 0.5903021097183228,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 90160
+    },
+    {
+      "epoch": 593.2236842105264,
+      "grad_norm": 0.7272729873657227,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 90170
+    },
+    {
+      "epoch": 593.2894736842105,
+      "grad_norm": 1.2663187980651855,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 90180
+    },
+    {
+      "epoch": 593.3552631578947,
+      "grad_norm": 1.210874080657959,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 90190
+    },
+    {
+      "epoch": 593.421052631579,
+      "grad_norm": 1.0026808977127075,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 90200
+    },
+    {
+      "epoch": 593.4868421052631,
+      "grad_norm": 1.1828995943069458,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 90210
+    },
+    {
+      "epoch": 593.5526315789474,
+      "grad_norm": 1.2319470643997192,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 90220
+    },
+    {
+      "epoch": 593.6184210526316,
+      "grad_norm": 1.5883439779281616,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 90230
+    },
+    {
+      "epoch": 593.6842105263158,
+      "grad_norm": 1.1823478937149048,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 90240
+    },
+    {
+      "epoch": 593.75,
+      "grad_norm": 1.1434694528579712,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 90250
+    },
+    {
+      "epoch": 593.8157894736842,
+      "grad_norm": 0.8349093794822693,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 90260
+    },
+    {
+      "epoch": 593.8815789473684,
+      "grad_norm": 1.1716779470443726,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 90270
+    },
+    {
+      "epoch": 593.9473684210526,
+      "grad_norm": 0.9748353362083435,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 90280
+    },
+    {
+      "epoch": 594.0131578947369,
+      "grad_norm": 1.1021101474761963,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 90290
+    },
+    {
+      "epoch": 594.078947368421,
+      "grad_norm": 1.007872223854065,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 90300
+    },
+    {
+      "epoch": 594.1447368421053,
+      "grad_norm": 0.8488720059394836,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 90310
+    },
+    {
+      "epoch": 594.2105263157895,
+      "grad_norm": 0.9892115592956543,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 90320
+    },
+    {
+      "epoch": 594.2763157894736,
+      "grad_norm": 1.2533457279205322,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 90330
+    },
+    {
+      "epoch": 594.3421052631579,
+      "grad_norm": 1.1002429723739624,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 90340
+    },
+    {
+      "epoch": 594.4078947368421,
+      "grad_norm": 1.0091798305511475,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 90350
+    },
+    {
+      "epoch": 594.4736842105264,
+      "grad_norm": 1.074102520942688,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 90360
+    },
+    {
+      "epoch": 594.5394736842105,
+      "grad_norm": 0.9539269804954529,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 90370
+    },
+    {
+      "epoch": 594.6052631578947,
+      "grad_norm": 1.0919673442840576,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 90380
+    },
+    {
+      "epoch": 594.671052631579,
+      "grad_norm": 1.65645170211792,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 90390
+    },
+    {
+      "epoch": 594.7368421052631,
+      "grad_norm": 1.5577173233032227,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 90400
+    },
+    {
+      "epoch": 594.8026315789474,
+      "grad_norm": 0.9961990714073181,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 90410
+    },
+    {
+      "epoch": 594.8684210526316,
+      "grad_norm": 1.1721851825714111,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 90420
+    },
+    {
+      "epoch": 594.9342105263158,
+      "grad_norm": 1.203532099723816,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 90430
+    },
+    {
+      "epoch": 595.0,
+      "grad_norm": 1.3049278259277344,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 90440
+    },
+    {
+      "epoch": 595.0657894736842,
+      "grad_norm": 1.4618241786956787,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 90450
+    },
+    {
+      "epoch": 595.1315789473684,
+      "grad_norm": 1.3204786777496338,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 90460
+    },
+    {
+      "epoch": 595.1973684210526,
+      "grad_norm": 1.2331825494766235,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 90470
+    },
+    {
+      "epoch": 595.2631578947369,
+      "grad_norm": 1.2492101192474365,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 90480
+    },
+    {
+      "epoch": 595.328947368421,
+      "grad_norm": 1.1810252666473389,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 90490
+    },
+    {
+      "epoch": 595.3947368421053,
+      "grad_norm": 1.0970319509506226,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 90500
+    },
+    {
+      "epoch": 595.4605263157895,
+      "grad_norm": 1.2086048126220703,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 90510
+    },
+    {
+      "epoch": 595.5263157894736,
+      "grad_norm": 0.9637195467948914,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 90520
+    },
+    {
+      "epoch": 595.5921052631579,
+      "grad_norm": 1.4612727165222168,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 90530
+    },
+    {
+      "epoch": 595.6578947368421,
+      "grad_norm": 1.3866544961929321,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 90540
+    },
+    {
+      "epoch": 595.7236842105264,
+      "grad_norm": 1.3224684000015259,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 90550
+    },
+    {
+      "epoch": 595.7894736842105,
+      "grad_norm": 1.1373405456542969,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 90560
+    },
+    {
+      "epoch": 595.8552631578947,
+      "grad_norm": 0.909879207611084,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 90570
+    },
+    {
+      "epoch": 595.921052631579,
+      "grad_norm": 0.965241551399231,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 90580
+    },
+    {
+      "epoch": 595.9868421052631,
+      "grad_norm": 1.0295147895812988,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 90590
+    },
+    {
+      "epoch": 596.0526315789474,
+      "grad_norm": 1.1842830181121826,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 90600
+    },
+    {
+      "epoch": 596.1184210526316,
+      "grad_norm": 1.119468331336975,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 90610
+    },
+    {
+      "epoch": 596.1842105263158,
+      "grad_norm": 1.0015408992767334,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 90620
+    },
+    {
+      "epoch": 596.25,
+      "grad_norm": 1.0252635478973389,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 90630
+    },
+    {
+      "epoch": 596.3157894736842,
+      "grad_norm": 1.0679528713226318,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 90640
+    },
+    {
+      "epoch": 596.3815789473684,
+      "grad_norm": 1.1197189092636108,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 90650
+    },
+    {
+      "epoch": 596.4473684210526,
+      "grad_norm": 1.1711959838867188,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 90660
+    },
+    {
+      "epoch": 596.5131578947369,
+      "grad_norm": 0.8267062306404114,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 90670
+    },
+    {
+      "epoch": 596.578947368421,
+      "grad_norm": 0.9089040756225586,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 90680
+    },
+    {
+      "epoch": 596.6447368421053,
+      "grad_norm": 1.1336896419525146,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 90690
+    },
+    {
+      "epoch": 596.7105263157895,
+      "grad_norm": 1.5351234674453735,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 90700
+    },
+    {
+      "epoch": 596.7763157894736,
+      "grad_norm": 1.4574313163757324,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 90710
+    },
+    {
+      "epoch": 596.8421052631579,
+      "grad_norm": 1.3572088479995728,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 90720
+    },
+    {
+      "epoch": 596.9078947368421,
+      "grad_norm": 1.556077480316162,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 90730
+    },
+    {
+      "epoch": 596.9736842105264,
+      "grad_norm": 1.4981848001480103,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 90740
+    },
+    {
+      "epoch": 597.0394736842105,
+      "grad_norm": 1.3176113367080688,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 90750
+    },
+    {
+      "epoch": 597.1052631578947,
+      "grad_norm": 1.1030333042144775,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 90760
+    },
+    {
+      "epoch": 597.171052631579,
+      "grad_norm": 1.3109503984451294,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 90770
+    },
+    {
+      "epoch": 597.2368421052631,
+      "grad_norm": 1.4163578748703003,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 90780
+    },
+    {
+      "epoch": 597.3026315789474,
+      "grad_norm": 1.3910415172576904,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 90790
+    },
+    {
+      "epoch": 597.3684210526316,
+      "grad_norm": 1.3340750932693481,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 90800
+    },
+    {
+      "epoch": 597.4342105263158,
+      "grad_norm": 0.9749845862388611,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 90810
+    },
+    {
+      "epoch": 597.5,
+      "grad_norm": 0.9810044169425964,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 90820
+    },
+    {
+      "epoch": 597.5657894736842,
+      "grad_norm": 1.392261266708374,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 90830
+    },
+    {
+      "epoch": 597.6315789473684,
+      "grad_norm": 1.2727516889572144,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 90840
+    },
+    {
+      "epoch": 597.6973684210526,
+      "grad_norm": 1.2531763315200806,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 90850
+    },
+    {
+      "epoch": 597.7631578947369,
+      "grad_norm": 1.1090060472488403,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 90860
+    },
+    {
+      "epoch": 597.828947368421,
+      "grad_norm": 1.283146619796753,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 90870
+    },
+    {
+      "epoch": 597.8947368421053,
+      "grad_norm": 0.8939560055732727,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 90880
+    },
+    {
+      "epoch": 597.9605263157895,
+      "grad_norm": 0.7372931838035583,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 90890
+    },
+    {
+      "epoch": 598.0263157894736,
+      "grad_norm": 0.992813229560852,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 90900
+    },
+    {
+      "epoch": 598.0921052631579,
+      "grad_norm": 1.0757131576538086,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 90910
+    },
+    {
+      "epoch": 598.1578947368421,
+      "grad_norm": 1.3756095170974731,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 90920
+    },
+    {
+      "epoch": 598.2236842105264,
+      "grad_norm": 1.1609660387039185,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 90930
+    },
+    {
+      "epoch": 598.2894736842105,
+      "grad_norm": 1.197822093963623,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 90940
+    },
+    {
+      "epoch": 598.3552631578947,
+      "grad_norm": 1.0572071075439453,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 90950
+    },
+    {
+      "epoch": 598.421052631579,
+      "grad_norm": 1.026586890220642,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 90960
+    },
+    {
+      "epoch": 598.4868421052631,
+      "grad_norm": 1.0258917808532715,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 90970
+    },
+    {
+      "epoch": 598.5526315789474,
+      "grad_norm": 1.1755049228668213,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 90980
+    },
+    {
+      "epoch": 598.6184210526316,
+      "grad_norm": 1.2228240966796875,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 90990
+    },
+    {
+      "epoch": 598.6842105263158,
+      "grad_norm": 1.14155912399292,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 91000
+    },
+    {
+      "epoch": 598.75,
+      "grad_norm": 1.355281114578247,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 91010
+    },
+    {
+      "epoch": 598.8157894736842,
+      "grad_norm": 0.7199444770812988,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 91020
+    },
+    {
+      "epoch": 598.8815789473684,
+      "grad_norm": 0.8066603541374207,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 91030
+    },
+    {
+      "epoch": 598.9473684210526,
+      "grad_norm": 0.8587542772293091,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 91040
+    },
+    {
+      "epoch": 599.0131578947369,
+      "grad_norm": 0.8503077030181885,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 91050
+    },
+    {
+      "epoch": 599.078947368421,
+      "grad_norm": 0.5482606887817383,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 91060
+    },
+    {
+      "epoch": 599.1447368421053,
+      "grad_norm": 1.3587243556976318,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 91070
+    },
+    {
+      "epoch": 599.2105263157895,
+      "grad_norm": 1.3960087299346924,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 91080
+    },
+    {
+      "epoch": 599.2763157894736,
+      "grad_norm": 1.0723680257797241,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 91090
+    },
+    {
+      "epoch": 599.3421052631579,
+      "grad_norm": 0.911795437335968,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 91100
+    },
+    {
+      "epoch": 599.4078947368421,
+      "grad_norm": 1.3153270483016968,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 91110
+    },
+    {
+      "epoch": 599.4736842105264,
+      "grad_norm": 1.3088375329971313,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 91120
+    },
+    {
+      "epoch": 599.5394736842105,
+      "grad_norm": 1.0322259664535522,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 91130
+    },
+    {
+      "epoch": 599.6052631578947,
+      "grad_norm": 0.8440555334091187,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 91140
+    },
+    {
+      "epoch": 599.671052631579,
+      "grad_norm": 1.4241567850112915,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 91150
+    },
+    {
+      "epoch": 599.7368421052631,
+      "grad_norm": 1.298012375831604,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 91160
+    },
+    {
+      "epoch": 599.8026315789474,
+      "grad_norm": 2.0893771648406982,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 91170
+    },
+    {
+      "epoch": 599.8684210526316,
+      "grad_norm": 1.2253968715667725,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 91180
+    },
+    {
+      "epoch": 599.9342105263158,
+      "grad_norm": 1.2711297273635864,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 91190
+    },
+    {
+      "epoch": 600.0,
+      "grad_norm": 0.9054577350616455,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 91200
+    },
+    {
+      "epoch": 600.0657894736842,
+      "grad_norm": 0.9020711779594421,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 91210
+    },
+    {
+      "epoch": 600.1315789473684,
+      "grad_norm": 1.009748935699463,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 91220
+    },
+    {
+      "epoch": 600.1973684210526,
+      "grad_norm": 1.466187834739685,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 91230
+    },
+    {
+      "epoch": 600.2631578947369,
+      "grad_norm": 1.3021607398986816,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 91240
+    },
+    {
+      "epoch": 600.328947368421,
+      "grad_norm": 1.0254582166671753,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 91250
+    },
+    {
+      "epoch": 600.3947368421053,
+      "grad_norm": 0.8893698453903198,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 91260
+    },
+    {
+      "epoch": 600.4605263157895,
+      "grad_norm": 1.190147042274475,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 91270
+    },
+    {
+      "epoch": 600.5263157894736,
+      "grad_norm": 1.1104352474212646,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 91280
+    },
+    {
+      "epoch": 600.5921052631579,
+      "grad_norm": 1.2664473056793213,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 91290
+    },
+    {
+      "epoch": 600.6578947368421,
+      "grad_norm": 1.136460781097412,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 91300
+    },
+    {
+      "epoch": 600.7236842105264,
+      "grad_norm": 1.4673341512680054,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 91310
+    },
+    {
+      "epoch": 600.7894736842105,
+      "grad_norm": 1.533826231956482,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 91320
+    },
+    {
+      "epoch": 600.8552631578947,
+      "grad_norm": 1.178153157234192,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 91330
+    },
+    {
+      "epoch": 600.921052631579,
+      "grad_norm": 1.365976095199585,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 91340
+    },
+    {
+      "epoch": 600.9868421052631,
+      "grad_norm": 1.1566143035888672,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 91350
+    },
+    {
+      "epoch": 601.0526315789474,
+      "grad_norm": 1.1433640718460083,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 91360
+    },
+    {
+      "epoch": 601.1184210526316,
+      "grad_norm": 0.9018469452857971,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 91370
+    },
+    {
+      "epoch": 601.1842105263158,
+      "grad_norm": 1.1393685340881348,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 91380
+    },
+    {
+      "epoch": 601.25,
+      "grad_norm": 1.1455858945846558,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 91390
+    },
+    {
+      "epoch": 601.3157894736842,
+      "grad_norm": 1.4033070802688599,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 91400
+    },
+    {
+      "epoch": 601.3815789473684,
+      "grad_norm": 1.4642856121063232,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 91410
+    },
+    {
+      "epoch": 601.4473684210526,
+      "grad_norm": 1.0535764694213867,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 91420
+    },
+    {
+      "epoch": 601.5131578947369,
+      "grad_norm": 1.0785901546478271,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 91430
+    },
+    {
+      "epoch": 601.578947368421,
+      "grad_norm": 1.2098617553710938,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 91440
+    },
+    {
+      "epoch": 601.6447368421053,
+      "grad_norm": 1.4227787256240845,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 91450
+    },
+    {
+      "epoch": 601.7105263157895,
+      "grad_norm": 1.245746374130249,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 91460
+    },
+    {
+      "epoch": 601.7763157894736,
+      "grad_norm": 1.1183278560638428,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 91470
+    },
+    {
+      "epoch": 601.8421052631579,
+      "grad_norm": 1.2257438898086548,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 91480
+    },
+    {
+      "epoch": 601.9078947368421,
+      "grad_norm": 0.8656227588653564,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 91490
+    },
+    {
+      "epoch": 601.9736842105264,
+      "grad_norm": 0.9605566263198853,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 91500
+    },
+    {
+      "epoch": 602.0394736842105,
+      "grad_norm": 1.148573637008667,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 91510
+    },
+    {
+      "epoch": 602.1052631578947,
+      "grad_norm": 1.1873161792755127,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 91520
+    },
+    {
+      "epoch": 602.171052631579,
+      "grad_norm": 1.2313774824142456,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 91530
+    },
+    {
+      "epoch": 602.2368421052631,
+      "grad_norm": 0.932478666305542,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 91540
+    },
+    {
+      "epoch": 602.3026315789474,
+      "grad_norm": 1.0965076684951782,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 91550
+    },
+    {
+      "epoch": 602.3684210526316,
+      "grad_norm": 1.0548715591430664,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 91560
+    },
+    {
+      "epoch": 602.4342105263158,
+      "grad_norm": 0.6713628172874451,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 91570
+    },
+    {
+      "epoch": 602.5,
+      "grad_norm": 1.1099224090576172,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 91580
+    },
+    {
+      "epoch": 602.5657894736842,
+      "grad_norm": 1.04884672164917,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 91590
+    },
+    {
+      "epoch": 602.6315789473684,
+      "grad_norm": 0.7628430128097534,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 91600
+    },
+    {
+      "epoch": 602.6973684210526,
+      "grad_norm": 0.8330525755882263,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 91610
+    },
+    {
+      "epoch": 602.7631578947369,
+      "grad_norm": 1.0476446151733398,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 91620
+    },
+    {
+      "epoch": 602.828947368421,
+      "grad_norm": 0.9488280415534973,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 91630
+    },
+    {
+      "epoch": 602.8947368421053,
+      "grad_norm": 1.0100157260894775,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 91640
+    },
+    {
+      "epoch": 602.9605263157895,
+      "grad_norm": 1.3274532556533813,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 91650
+    },
+    {
+      "epoch": 603.0263157894736,
+      "grad_norm": 1.0656237602233887,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 91660
+    },
+    {
+      "epoch": 603.0921052631579,
+      "grad_norm": 1.239275336265564,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 91670
+    },
+    {
+      "epoch": 603.1578947368421,
+      "grad_norm": 1.4684178829193115,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 91680
+    },
+    {
+      "epoch": 603.2236842105264,
+      "grad_norm": 1.0184123516082764,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 91690
+    },
+    {
+      "epoch": 603.2894736842105,
+      "grad_norm": 0.9157920479774475,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 91700
+    },
+    {
+      "epoch": 603.3552631578947,
+      "grad_norm": 1.0523053407669067,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 91710
+    },
+    {
+      "epoch": 603.421052631579,
+      "grad_norm": 1.4203752279281616,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 91720
+    },
+    {
+      "epoch": 603.4868421052631,
+      "grad_norm": 1.1770908832550049,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 91730
+    },
+    {
+      "epoch": 603.5526315789474,
+      "grad_norm": 1.181809425354004,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 91740
+    },
+    {
+      "epoch": 603.6184210526316,
+      "grad_norm": 1.3628486394882202,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 91750
+    },
+    {
+      "epoch": 603.6842105263158,
+      "grad_norm": 1.0041279792785645,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 91760
+    },
+    {
+      "epoch": 603.75,
+      "grad_norm": 1.1971017122268677,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 91770
+    },
+    {
+      "epoch": 603.8157894736842,
+      "grad_norm": 0.9669057726860046,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 91780
+    },
+    {
+      "epoch": 603.8815789473684,
+      "grad_norm": 1.1059627532958984,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 91790
+    },
+    {
+      "epoch": 603.9473684210526,
+      "grad_norm": 1.0041719675064087,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 91800
+    },
+    {
+      "epoch": 604.0131578947369,
+      "grad_norm": 0.8442330360412598,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 91810
+    },
+    {
+      "epoch": 604.078947368421,
+      "grad_norm": 0.9047984480857849,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 91820
+    },
+    {
+      "epoch": 604.1447368421053,
+      "grad_norm": 1.000754714012146,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 91830
+    },
+    {
+      "epoch": 604.2105263157895,
+      "grad_norm": 1.2270958423614502,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 91840
+    },
+    {
+      "epoch": 604.2763157894736,
+      "grad_norm": 0.8679501414299011,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 91850
+    },
+    {
+      "epoch": 604.3421052631579,
+      "grad_norm": 0.7244850397109985,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 91860
+    },
+    {
+      "epoch": 604.4078947368421,
+      "grad_norm": 0.9245374202728271,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 91870
+    },
+    {
+      "epoch": 604.4736842105264,
+      "grad_norm": 0.977651834487915,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 91880
+    },
+    {
+      "epoch": 604.5394736842105,
+      "grad_norm": 0.9714836478233337,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 91890
+    },
+    {
+      "epoch": 604.6052631578947,
+      "grad_norm": 1.11795973777771,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 91900
+    },
+    {
+      "epoch": 604.671052631579,
+      "grad_norm": 1.1345611810684204,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 91910
+    },
+    {
+      "epoch": 604.7368421052631,
+      "grad_norm": 1.1849992275238037,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 91920
+    },
+    {
+      "epoch": 604.8026315789474,
+      "grad_norm": 1.2668788433074951,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 91930
+    },
+    {
+      "epoch": 604.8684210526316,
+      "grad_norm": 1.0477993488311768,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 91940
+    },
+    {
+      "epoch": 604.9342105263158,
+      "grad_norm": 1.0195016860961914,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 91950
+    },
+    {
+      "epoch": 605.0,
+      "grad_norm": 0.9753181338310242,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 91960
+    },
+    {
+      "epoch": 605.0657894736842,
+      "grad_norm": 0.8940220475196838,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 91970
+    },
+    {
+      "epoch": 605.1315789473684,
+      "grad_norm": 0.779227077960968,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 91980
+    },
+    {
+      "epoch": 605.1973684210526,
+      "grad_norm": 1.1637389659881592,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 91990
+    },
+    {
+      "epoch": 605.2631578947369,
+      "grad_norm": 1.1492772102355957,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 92000
+    },
+    {
+      "epoch": 605.328947368421,
+      "grad_norm": 1.265363097190857,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 92010
+    },
+    {
+      "epoch": 605.3947368421053,
+      "grad_norm": 1.1617438793182373,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 92020
+    },
+    {
+      "epoch": 605.4605263157895,
+      "grad_norm": 1.0030673742294312,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 92030
+    },
+    {
+      "epoch": 605.5263157894736,
+      "grad_norm": 1.7177019119262695,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 92040
+    },
+    {
+      "epoch": 605.5921052631579,
+      "grad_norm": 1.1032168865203857,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 92050
+    },
+    {
+      "epoch": 605.6578947368421,
+      "grad_norm": 1.0121941566467285,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 92060
+    },
+    {
+      "epoch": 605.7236842105264,
+      "grad_norm": 1.0975619554519653,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 92070
+    },
+    {
+      "epoch": 605.7894736842105,
+      "grad_norm": 0.9619658589363098,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 92080
+    },
+    {
+      "epoch": 605.8552631578947,
+      "grad_norm": 0.8327913284301758,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 92090
+    },
+    {
+      "epoch": 605.921052631579,
+      "grad_norm": 1.230811357498169,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 92100
+    },
+    {
+      "epoch": 605.9868421052631,
+      "grad_norm": 1.1441431045532227,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 92110
+    },
+    {
+      "epoch": 606.0526315789474,
+      "grad_norm": 0.8227072954177856,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 92120
+    },
+    {
+      "epoch": 606.1184210526316,
+      "grad_norm": 0.9248983860015869,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 92130
+    },
+    {
+      "epoch": 606.1842105263158,
+      "grad_norm": 1.613922119140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 92140
+    },
+    {
+      "epoch": 606.25,
+      "grad_norm": 1.2741233110427856,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 92150
+    },
+    {
+      "epoch": 606.3157894736842,
+      "grad_norm": 1.0718578100204468,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 92160
+    },
+    {
+      "epoch": 606.3815789473684,
+      "grad_norm": 1.7179508209228516,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 92170
+    },
+    {
+      "epoch": 606.4473684210526,
+      "grad_norm": 1.6176236867904663,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 92180
+    },
+    {
+      "epoch": 606.5131578947369,
+      "grad_norm": 1.308016300201416,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 92190
+    },
+    {
+      "epoch": 606.578947368421,
+      "grad_norm": 1.2682429552078247,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 92200
+    },
+    {
+      "epoch": 606.6447368421053,
+      "grad_norm": 1.2215003967285156,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 92210
+    },
+    {
+      "epoch": 606.7105263157895,
+      "grad_norm": 1.3663808107376099,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 92220
+    },
+    {
+      "epoch": 606.7763157894736,
+      "grad_norm": 0.8408277630805969,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 92230
+    },
+    {
+      "epoch": 606.8421052631579,
+      "grad_norm": 0.7498778104782104,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 92240
+    },
+    {
+      "epoch": 606.9078947368421,
+      "grad_norm": 1.0909346342086792,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 92250
+    },
+    {
+      "epoch": 606.9736842105264,
+      "grad_norm": 0.9541800618171692,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 92260
+    },
+    {
+      "epoch": 607.0394736842105,
+      "grad_norm": 1.0058972835540771,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 92270
+    },
+    {
+      "epoch": 607.1052631578947,
+      "grad_norm": 0.4746664762496948,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 92280
+    },
+    {
+      "epoch": 607.171052631579,
+      "grad_norm": 0.893031656742096,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 92290
+    },
+    {
+      "epoch": 607.2368421052631,
+      "grad_norm": 1.1034398078918457,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 92300
+    },
+    {
+      "epoch": 607.3026315789474,
+      "grad_norm": 1.0481274127960205,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 92310
+    },
+    {
+      "epoch": 607.3684210526316,
+      "grad_norm": 1.0464096069335938,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 92320
+    },
+    {
+      "epoch": 607.4342105263158,
+      "grad_norm": 1.2462482452392578,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 92330
+    },
+    {
+      "epoch": 607.5,
+      "grad_norm": 1.3764554262161255,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 92340
+    },
+    {
+      "epoch": 607.5657894736842,
+      "grad_norm": 1.0368061065673828,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 92350
+    },
+    {
+      "epoch": 607.6315789473684,
+      "grad_norm": 0.9703469276428223,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 92360
+    },
+    {
+      "epoch": 607.6973684210526,
+      "grad_norm": 1.2898932695388794,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 92370
+    },
+    {
+      "epoch": 607.7631578947369,
+      "grad_norm": 0.900486946105957,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 92380
+    },
+    {
+      "epoch": 607.828947368421,
+      "grad_norm": 1.2100414037704468,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 92390
+    },
+    {
+      "epoch": 607.8947368421053,
+      "grad_norm": 1.1729727983474731,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 92400
+    },
+    {
+      "epoch": 607.9605263157895,
+      "grad_norm": 1.127228856086731,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 92410
+    },
+    {
+      "epoch": 608.0263157894736,
+      "grad_norm": 0.8651881217956543,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 92420
+    },
+    {
+      "epoch": 608.0921052631579,
+      "grad_norm": 0.8712637424468994,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 92430
+    },
+    {
+      "epoch": 608.1578947368421,
+      "grad_norm": 1.1293796300888062,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 92440
+    },
+    {
+      "epoch": 608.2236842105264,
+      "grad_norm": 1.2297700643539429,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 92450
+    },
+    {
+      "epoch": 608.2894736842105,
+      "grad_norm": 1.038426399230957,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 92460
+    },
+    {
+      "epoch": 608.3552631578947,
+      "grad_norm": 1.3792036771774292,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 92470
+    },
+    {
+      "epoch": 608.421052631579,
+      "grad_norm": 1.0525060892105103,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 92480
+    },
+    {
+      "epoch": 608.4868421052631,
+      "grad_norm": 1.0307869911193848,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 92490
+    },
+    {
+      "epoch": 608.5526315789474,
+      "grad_norm": 0.7510896325111389,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 92500
+    },
+    {
+      "epoch": 608.6184210526316,
+      "grad_norm": 0.973735511302948,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 92510
+    },
+    {
+      "epoch": 608.6842105263158,
+      "grad_norm": 1.3305103778839111,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 92520
+    },
+    {
+      "epoch": 608.75,
+      "grad_norm": 0.8219487071037292,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 92530
+    },
+    {
+      "epoch": 608.8157894736842,
+      "grad_norm": 0.836087703704834,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 92540
+    },
+    {
+      "epoch": 608.8815789473684,
+      "grad_norm": 1.397853970527649,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 92550
+    },
+    {
+      "epoch": 608.9473684210526,
+      "grad_norm": 1.1062546968460083,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 92560
+    },
+    {
+      "epoch": 609.0131578947369,
+      "grad_norm": 0.8973669409751892,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 92570
+    },
+    {
+      "epoch": 609.078947368421,
+      "grad_norm": 1.047875165939331,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 92580
+    },
+    {
+      "epoch": 609.1447368421053,
+      "grad_norm": 1.2423298358917236,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 92590
+    },
+    {
+      "epoch": 609.2105263157895,
+      "grad_norm": 0.6068233251571655,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 92600
+    },
+    {
+      "epoch": 609.2763157894736,
+      "grad_norm": 1.0823287963867188,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 92610
+    },
+    {
+      "epoch": 609.3421052631579,
+      "grad_norm": 0.9071270227432251,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 92620
+    },
+    {
+      "epoch": 609.4078947368421,
+      "grad_norm": 0.6887641549110413,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 92630
+    },
+    {
+      "epoch": 609.4736842105264,
+      "grad_norm": 1.3606575727462769,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 92640
+    },
+    {
+      "epoch": 609.5394736842105,
+      "grad_norm": 0.9844822287559509,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 92650
+    },
+    {
+      "epoch": 609.6052631578947,
+      "grad_norm": 1.0416951179504395,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 92660
+    },
+    {
+      "epoch": 609.671052631579,
+      "grad_norm": 1.3930866718292236,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 92670
+    },
+    {
+      "epoch": 609.7368421052631,
+      "grad_norm": 1.41727614402771,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 92680
+    },
+    {
+      "epoch": 609.8026315789474,
+      "grad_norm": 1.0249476432800293,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 92690
+    },
+    {
+      "epoch": 609.8684210526316,
+      "grad_norm": 1.2941230535507202,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 92700
+    },
+    {
+      "epoch": 609.9342105263158,
+      "grad_norm": 1.1592159271240234,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 92710
+    },
+    {
+      "epoch": 610.0,
+      "grad_norm": 1.3490889072418213,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 92720
+    },
+    {
+      "epoch": 610.0657894736842,
+      "grad_norm": 1.1179836988449097,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 92730
+    },
+    {
+      "epoch": 610.1315789473684,
+      "grad_norm": 1.387890100479126,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 92740
+    },
+    {
+      "epoch": 610.1973684210526,
+      "grad_norm": 1.3158421516418457,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 92750
+    },
+    {
+      "epoch": 610.2631578947369,
+      "grad_norm": 1.1624884605407715,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 92760
+    },
+    {
+      "epoch": 610.328947368421,
+      "grad_norm": 0.9471017718315125,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 92770
+    },
+    {
+      "epoch": 610.3947368421053,
+      "grad_norm": 1.1989469528198242,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 92780
+    },
+    {
+      "epoch": 610.4605263157895,
+      "grad_norm": 1.1440564393997192,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 92790
+    },
+    {
+      "epoch": 610.5263157894736,
+      "grad_norm": 1.0892585515975952,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 92800
+    },
+    {
+      "epoch": 610.5921052631579,
+      "grad_norm": 1.0231949090957642,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 92810
+    },
+    {
+      "epoch": 610.6578947368421,
+      "grad_norm": 1.5775789022445679,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 92820
+    },
+    {
+      "epoch": 610.7236842105264,
+      "grad_norm": 1.2640076875686646,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 92830
+    },
+    {
+      "epoch": 610.7894736842105,
+      "grad_norm": 1.052940011024475,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 92840
+    },
+    {
+      "epoch": 610.8552631578947,
+      "grad_norm": 1.000012993812561,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 92850
+    },
+    {
+      "epoch": 610.921052631579,
+      "grad_norm": 1.0184035301208496,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 92860
+    },
+    {
+      "epoch": 610.9868421052631,
+      "grad_norm": 1.1735719442367554,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 92870
+    },
+    {
+      "epoch": 611.0526315789474,
+      "grad_norm": 1.5585322380065918,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 92880
+    },
+    {
+      "epoch": 611.1184210526316,
+      "grad_norm": 1.2368836402893066,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 92890
+    },
+    {
+      "epoch": 611.1842105263158,
+      "grad_norm": 0.9293583035469055,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 92900
+    },
+    {
+      "epoch": 611.25,
+      "grad_norm": 1.3896769285202026,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 92910
+    },
+    {
+      "epoch": 611.3157894736842,
+      "grad_norm": 1.0448602437973022,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 92920
+    },
+    {
+      "epoch": 611.3815789473684,
+      "grad_norm": 1.4385607242584229,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 92930
+    },
+    {
+      "epoch": 611.4473684210526,
+      "grad_norm": 1.271680474281311,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 92940
+    },
+    {
+      "epoch": 611.5131578947369,
+      "grad_norm": 1.3433717489242554,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 92950
+    },
+    {
+      "epoch": 611.578947368421,
+      "grad_norm": 1.157219409942627,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 92960
+    },
+    {
+      "epoch": 611.6447368421053,
+      "grad_norm": 1.4146461486816406,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 92970
+    },
+    {
+      "epoch": 611.7105263157895,
+      "grad_norm": 1.070226788520813,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 92980
+    },
+    {
+      "epoch": 611.7763157894736,
+      "grad_norm": 1.0087385177612305,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 92990
+    },
+    {
+      "epoch": 611.8421052631579,
+      "grad_norm": 1.1002082824707031,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 93000
+    },
+    {
+      "epoch": 611.9078947368421,
+      "grad_norm": 1.0607959032058716,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 93010
+    },
+    {
+      "epoch": 611.9736842105264,
+      "grad_norm": 0.7643837928771973,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 93020
+    },
+    {
+      "epoch": 612.0394736842105,
+      "grad_norm": 1.0315144062042236,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 93030
+    },
+    {
+      "epoch": 612.1052631578947,
+      "grad_norm": 1.1846758127212524,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 93040
+    },
+    {
+      "epoch": 612.171052631579,
+      "grad_norm": 1.1276209354400635,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 93050
+    },
+    {
+      "epoch": 612.2368421052631,
+      "grad_norm": 1.1060857772827148,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 93060
+    },
+    {
+      "epoch": 612.3026315789474,
+      "grad_norm": 0.8714224696159363,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 93070
+    },
+    {
+      "epoch": 612.3684210526316,
+      "grad_norm": 1.2322101593017578,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 93080
+    },
+    {
+      "epoch": 612.4342105263158,
+      "grad_norm": 1.2291126251220703,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 93090
+    },
+    {
+      "epoch": 612.5,
+      "grad_norm": 1.4525691270828247,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 93100
+    },
+    {
+      "epoch": 612.5657894736842,
+      "grad_norm": 0.7913857698440552,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 93110
+    },
+    {
+      "epoch": 612.6315789473684,
+      "grad_norm": 1.1557925939559937,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 93120
+    },
+    {
+      "epoch": 612.6973684210526,
+      "grad_norm": 0.9045695066452026,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 93130
+    },
+    {
+      "epoch": 612.7631578947369,
+      "grad_norm": 1.2888610363006592,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 93140
+    },
+    {
+      "epoch": 612.828947368421,
+      "grad_norm": 1.0652800798416138,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 93150
+    },
+    {
+      "epoch": 612.8947368421053,
+      "grad_norm": 1.246463656425476,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 93160
+    },
+    {
+      "epoch": 612.9605263157895,
+      "grad_norm": 0.9784409403800964,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 93170
+    },
+    {
+      "epoch": 613.0263157894736,
+      "grad_norm": 1.0943495035171509,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 93180
+    },
+    {
+      "epoch": 613.0921052631579,
+      "grad_norm": 1.1042827367782593,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 93190
+    },
+    {
+      "epoch": 613.1578947368421,
+      "grad_norm": 1.0616756677627563,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 93200
+    },
+    {
+      "epoch": 613.2236842105264,
+      "grad_norm": 1.2859280109405518,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 93210
+    },
+    {
+      "epoch": 613.2894736842105,
+      "grad_norm": 1.3570408821105957,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 93220
+    },
+    {
+      "epoch": 613.3552631578947,
+      "grad_norm": 0.9719215035438538,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 93230
+    },
+    {
+      "epoch": 613.421052631579,
+      "grad_norm": 1.1417118310928345,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 93240
+    },
+    {
+      "epoch": 613.4868421052631,
+      "grad_norm": 1.616381287574768,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 93250
+    },
+    {
+      "epoch": 613.5526315789474,
+      "grad_norm": 1.248619794845581,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 93260
+    },
+    {
+      "epoch": 613.6184210526316,
+      "grad_norm": 1.056206226348877,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 93270
+    },
+    {
+      "epoch": 613.6842105263158,
+      "grad_norm": 0.8425273299217224,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 93280
+    },
+    {
+      "epoch": 613.75,
+      "grad_norm": 0.8737837076187134,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 93290
+    },
+    {
+      "epoch": 613.8157894736842,
+      "grad_norm": 0.8249803185462952,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 93300
+    },
+    {
+      "epoch": 613.8815789473684,
+      "grad_norm": 1.2743202447891235,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 93310
+    },
+    {
+      "epoch": 613.9473684210526,
+      "grad_norm": 0.8276104927062988,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 93320
+    },
+    {
+      "epoch": 614.0131578947369,
+      "grad_norm": 0.7873020768165588,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 93330
+    },
+    {
+      "epoch": 614.078947368421,
+      "grad_norm": 0.9627916812896729,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 93340
+    },
+    {
+      "epoch": 614.1447368421053,
+      "grad_norm": 1.3203983306884766,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 93350
+    },
+    {
+      "epoch": 614.2105263157895,
+      "grad_norm": 0.8902283906936646,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 93360
+    },
+    {
+      "epoch": 614.2763157894736,
+      "grad_norm": 1.3011919260025024,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 93370
+    },
+    {
+      "epoch": 614.3421052631579,
+      "grad_norm": 1.173582673072815,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 93380
+    },
+    {
+      "epoch": 614.4078947368421,
+      "grad_norm": 0.8310754299163818,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 93390
+    },
+    {
+      "epoch": 614.4736842105264,
+      "grad_norm": 1.0314079523086548,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 93400
+    },
+    {
+      "epoch": 614.5394736842105,
+      "grad_norm": 0.9039306640625,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 93410
+    },
+    {
+      "epoch": 614.6052631578947,
+      "grad_norm": 0.8027017116546631,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 93420
+    },
+    {
+      "epoch": 614.671052631579,
+      "grad_norm": 0.8218366503715515,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 93430
+    },
+    {
+      "epoch": 614.7368421052631,
+      "grad_norm": 0.8938267827033997,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 93440
+    },
+    {
+      "epoch": 614.8026315789474,
+      "grad_norm": 0.6793553829193115,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 93450
+    },
+    {
+      "epoch": 614.8684210526316,
+      "grad_norm": 1.1240465641021729,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 93460
+    },
+    {
+      "epoch": 614.9342105263158,
+      "grad_norm": 1.2319480180740356,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 93470
+    },
+    {
+      "epoch": 615.0,
+      "grad_norm": 1.1686429977416992,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 93480
+    },
+    {
+      "epoch": 615.0657894736842,
+      "grad_norm": 1.3799974918365479,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 93490
+    },
+    {
+      "epoch": 615.1315789473684,
+      "grad_norm": 0.9599565863609314,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 93500
+    },
+    {
+      "epoch": 615.1973684210526,
+      "grad_norm": 0.8217589855194092,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 93510
+    },
+    {
+      "epoch": 615.2631578947369,
+      "grad_norm": 1.090279221534729,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 93520
+    },
+    {
+      "epoch": 615.328947368421,
+      "grad_norm": 0.7684426307678223,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 93530
+    },
+    {
+      "epoch": 615.3947368421053,
+      "grad_norm": 1.3158767223358154,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 93540
+    },
+    {
+      "epoch": 615.4605263157895,
+      "grad_norm": 1.1429448127746582,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 93550
+    },
+    {
+      "epoch": 615.5263157894736,
+      "grad_norm": 1.2864338159561157,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 93560
+    },
+    {
+      "epoch": 615.5921052631579,
+      "grad_norm": 1.0945682525634766,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 93570
+    },
+    {
+      "epoch": 615.6578947368421,
+      "grad_norm": 0.966946542263031,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 93580
+    },
+    {
+      "epoch": 615.7236842105264,
+      "grad_norm": 0.7336428761482239,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 93590
+    },
+    {
+      "epoch": 615.7894736842105,
+      "grad_norm": 0.6933264136314392,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 93600
+    },
+    {
+      "epoch": 615.8552631578947,
+      "grad_norm": 1.118443489074707,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 93610
+    },
+    {
+      "epoch": 615.921052631579,
+      "grad_norm": 1.2565184831619263,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 93620
+    },
+    {
+      "epoch": 615.9868421052631,
+      "grad_norm": 1.2820919752120972,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 93630
+    },
+    {
+      "epoch": 616.0526315789474,
+      "grad_norm": 1.0800106525421143,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 93640
+    },
+    {
+      "epoch": 616.1184210526316,
+      "grad_norm": 1.172425627708435,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 93650
+    },
+    {
+      "epoch": 616.1842105263158,
+      "grad_norm": 1.2287852764129639,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 93660
+    },
+    {
+      "epoch": 616.25,
+      "grad_norm": 1.2166169881820679,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 93670
+    },
+    {
+      "epoch": 616.3157894736842,
+      "grad_norm": 0.8677031397819519,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 93680
+    },
+    {
+      "epoch": 616.3815789473684,
+      "grad_norm": 1.3982559442520142,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 93690
+    },
+    {
+      "epoch": 616.4473684210526,
+      "grad_norm": 1.3162497282028198,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 93700
+    },
+    {
+      "epoch": 616.5131578947369,
+      "grad_norm": 1.3209203481674194,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 93710
+    },
+    {
+      "epoch": 616.578947368421,
+      "grad_norm": 1.223204493522644,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 93720
+    },
+    {
+      "epoch": 616.6447368421053,
+      "grad_norm": 1.7229893207550049,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 93730
+    },
+    {
+      "epoch": 616.7105263157895,
+      "grad_norm": 1.5067758560180664,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 93740
+    },
+    {
+      "epoch": 616.7763157894736,
+      "grad_norm": 1.7268178462982178,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 93750
+    },
+    {
+      "epoch": 616.8421052631579,
+      "grad_norm": 1.096575140953064,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 93760
+    },
+    {
+      "epoch": 616.9078947368421,
+      "grad_norm": 1.145380973815918,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 93770
+    },
+    {
+      "epoch": 616.9736842105264,
+      "grad_norm": 1.3325169086456299,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 93780
+    },
+    {
+      "epoch": 617.0394736842105,
+      "grad_norm": 1.6675764322280884,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 93790
+    },
+    {
+      "epoch": 617.1052631578947,
+      "grad_norm": 1.2384144067764282,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 93800
+    },
+    {
+      "epoch": 617.171052631579,
+      "grad_norm": 1.222592830657959,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 93810
+    },
+    {
+      "epoch": 617.2368421052631,
+      "grad_norm": 1.2035316228866577,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 93820
+    },
+    {
+      "epoch": 617.3026315789474,
+      "grad_norm": 1.2100892066955566,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 93830
+    },
+    {
+      "epoch": 617.3684210526316,
+      "grad_norm": 1.365777850151062,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 93840
+    },
+    {
+      "epoch": 617.4342105263158,
+      "grad_norm": 1.5900193452835083,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 93850
+    },
+    {
+      "epoch": 617.5,
+      "grad_norm": 1.3592753410339355,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 93860
+    },
+    {
+      "epoch": 617.5657894736842,
+      "grad_norm": 1.417954683303833,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 93870
+    },
+    {
+      "epoch": 617.6315789473684,
+      "grad_norm": 1.517095923423767,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 93880
+    },
+    {
+      "epoch": 617.6973684210526,
+      "grad_norm": 1.3372291326522827,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 93890
+    },
+    {
+      "epoch": 617.7631578947369,
+      "grad_norm": 1.4415357112884521,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 93900
+    },
+    {
+      "epoch": 617.828947368421,
+      "grad_norm": 0.986171543598175,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 93910
+    },
+    {
+      "epoch": 617.8947368421053,
+      "grad_norm": 1.1450518369674683,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 93920
+    },
+    {
+      "epoch": 617.9605263157895,
+      "grad_norm": 1.439809799194336,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 93930
+    },
+    {
+      "epoch": 618.0263157894736,
+      "grad_norm": 1.4576247930526733,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 93940
+    },
+    {
+      "epoch": 618.0921052631579,
+      "grad_norm": 1.307482361793518,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 93950
+    },
+    {
+      "epoch": 618.1578947368421,
+      "grad_norm": 1.5274423360824585,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 93960
+    },
+    {
+      "epoch": 618.2236842105264,
+      "grad_norm": 1.0210192203521729,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 93970
+    },
+    {
+      "epoch": 618.2894736842105,
+      "grad_norm": 1.1291130781173706,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 93980
+    },
+    {
+      "epoch": 618.3552631578947,
+      "grad_norm": 0.7608104944229126,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 93990
+    },
+    {
+      "epoch": 618.421052631579,
+      "grad_norm": 0.9358824491500854,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 94000
+    },
+    {
+      "epoch": 618.4868421052631,
+      "grad_norm": 0.9126713871955872,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 94010
+    },
+    {
+      "epoch": 618.5526315789474,
+      "grad_norm": 0.9021777510643005,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 94020
+    },
+    {
+      "epoch": 618.6184210526316,
+      "grad_norm": 0.9748514890670776,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 94030
+    },
+    {
+      "epoch": 618.6842105263158,
+      "grad_norm": 0.7533281445503235,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 94040
+    },
+    {
+      "epoch": 618.75,
+      "grad_norm": 0.7163051962852478,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 94050
+    },
+    {
+      "epoch": 618.8157894736842,
+      "grad_norm": 1.0862754583358765,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 94060
+    },
+    {
+      "epoch": 618.8815789473684,
+      "grad_norm": 0.8490597009658813,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 94070
+    },
+    {
+      "epoch": 618.9473684210526,
+      "grad_norm": 1.2494447231292725,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 94080
+    },
+    {
+      "epoch": 619.0131578947369,
+      "grad_norm": 1.0552870035171509,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 94090
+    },
+    {
+      "epoch": 619.078947368421,
+      "grad_norm": 0.8766233921051025,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 94100
+    },
+    {
+      "epoch": 619.1447368421053,
+      "grad_norm": 1.181583285331726,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 94110
+    },
+    {
+      "epoch": 619.2105263157895,
+      "grad_norm": 1.25386381149292,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 94120
+    },
+    {
+      "epoch": 619.2763157894736,
+      "grad_norm": 1.0907665491104126,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 94130
+    },
+    {
+      "epoch": 619.3421052631579,
+      "grad_norm": 1.258827567100525,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 94140
+    },
+    {
+      "epoch": 619.4078947368421,
+      "grad_norm": 1.2584177255630493,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 94150
+    },
+    {
+      "epoch": 619.4736842105264,
+      "grad_norm": 1.2424556016921997,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 94160
+    },
+    {
+      "epoch": 619.5394736842105,
+      "grad_norm": 1.395603060722351,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 94170
+    },
+    {
+      "epoch": 619.6052631578947,
+      "grad_norm": 0.9532821774482727,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 94180
+    },
+    {
+      "epoch": 619.671052631579,
+      "grad_norm": 1.263140082359314,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 94190
+    },
+    {
+      "epoch": 619.7368421052631,
+      "grad_norm": 0.8259878754615784,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 94200
+    },
+    {
+      "epoch": 619.8026315789474,
+      "grad_norm": 1.4339922666549683,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 94210
+    },
+    {
+      "epoch": 619.8684210526316,
+      "grad_norm": 0.9723814725875854,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 94220
+    },
+    {
+      "epoch": 619.9342105263158,
+      "grad_norm": 0.9586044549942017,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 94230
+    },
+    {
+      "epoch": 620.0,
+      "grad_norm": 0.9816604852676392,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 94240
+    },
+    {
+      "epoch": 620.0657894736842,
+      "grad_norm": 0.8294615149497986,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 94250
+    },
+    {
+      "epoch": 620.1315789473684,
+      "grad_norm": 0.8866937160491943,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 94260
+    },
+    {
+      "epoch": 620.1973684210526,
+      "grad_norm": 0.9527875781059265,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 94270
+    },
+    {
+      "epoch": 620.2631578947369,
+      "grad_norm": 1.1666209697723389,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 94280
+    },
+    {
+      "epoch": 620.328947368421,
+      "grad_norm": 0.9513602256774902,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 94290
+    },
+    {
+      "epoch": 620.3947368421053,
+      "grad_norm": 0.7834760546684265,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 94300
+    },
+    {
+      "epoch": 620.4605263157895,
+      "grad_norm": 0.9751595258712769,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 94310
+    },
+    {
+      "epoch": 620.5263157894736,
+      "grad_norm": 1.2259643077850342,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 94320
+    },
+    {
+      "epoch": 620.5921052631579,
+      "grad_norm": 0.9753211736679077,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 94330
+    },
+    {
+      "epoch": 620.6578947368421,
+      "grad_norm": 1.5966567993164062,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 94340
+    },
+    {
+      "epoch": 620.7236842105264,
+      "grad_norm": 1.2838525772094727,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 94350
+    },
+    {
+      "epoch": 620.7894736842105,
+      "grad_norm": 1.416764497756958,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 94360
+    },
+    {
+      "epoch": 620.8552631578947,
+      "grad_norm": 1.317114233970642,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 94370
+    },
+    {
+      "epoch": 620.921052631579,
+      "grad_norm": 1.6000310182571411,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 94380
+    },
+    {
+      "epoch": 620.9868421052631,
+      "grad_norm": 1.296604037284851,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 94390
+    },
+    {
+      "epoch": 621.0526315789474,
+      "grad_norm": 1.1899832487106323,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 94400
+    },
+    {
+      "epoch": 621.1184210526316,
+      "grad_norm": 1.3651827573776245,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 94410
+    },
+    {
+      "epoch": 621.1842105263158,
+      "grad_norm": 1.230488657951355,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 94420
+    },
+    {
+      "epoch": 621.25,
+      "grad_norm": 1.2421315908432007,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 94430
+    },
+    {
+      "epoch": 621.3157894736842,
+      "grad_norm": 1.5289520025253296,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 94440
+    },
+    {
+      "epoch": 621.3815789473684,
+      "grad_norm": 1.2667958736419678,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 94450
+    },
+    {
+      "epoch": 621.4473684210526,
+      "grad_norm": 1.2462661266326904,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 94460
+    },
+    {
+      "epoch": 621.5131578947369,
+      "grad_norm": 1.1925386190414429,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 94470
+    },
+    {
+      "epoch": 621.578947368421,
+      "grad_norm": 0.7232915759086609,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 94480
+    },
+    {
+      "epoch": 621.6447368421053,
+      "grad_norm": 0.9784261584281921,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 94490
+    },
+    {
+      "epoch": 621.7105263157895,
+      "grad_norm": 0.8584251999855042,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 94500
+    },
+    {
+      "epoch": 621.7763157894736,
+      "grad_norm": 1.0461781024932861,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 94510
+    },
+    {
+      "epoch": 621.8421052631579,
+      "grad_norm": 0.9463827013969421,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 94520
+    },
+    {
+      "epoch": 621.9078947368421,
+      "grad_norm": 0.9517367482185364,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 94530
+    },
+    {
+      "epoch": 621.9736842105264,
+      "grad_norm": 1.1361017227172852,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 94540
+    },
+    {
+      "epoch": 622.0394736842105,
+      "grad_norm": 1.1935226917266846,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 94550
+    },
+    {
+      "epoch": 622.1052631578947,
+      "grad_norm": 1.004470705986023,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 94560
+    },
+    {
+      "epoch": 622.171052631579,
+      "grad_norm": 1.2536550760269165,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 94570
+    },
+    {
+      "epoch": 622.2368421052631,
+      "grad_norm": 1.262121319770813,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 94580
+    },
+    {
+      "epoch": 622.3026315789474,
+      "grad_norm": 1.241873860359192,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 94590
+    },
+    {
+      "epoch": 622.3684210526316,
+      "grad_norm": 0.9200069308280945,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 94600
+    },
+    {
+      "epoch": 622.4342105263158,
+      "grad_norm": 0.8964002132415771,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 94610
+    },
+    {
+      "epoch": 622.5,
+      "grad_norm": 1.2735228538513184,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 94620
+    },
+    {
+      "epoch": 622.5657894736842,
+      "grad_norm": 1.320853590965271,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 94630
+    },
+    {
+      "epoch": 622.6315789473684,
+      "grad_norm": 1.3654546737670898,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 94640
+    },
+    {
+      "epoch": 622.6973684210526,
+      "grad_norm": 0.8524348139762878,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 94650
+    },
+    {
+      "epoch": 622.7631578947369,
+      "grad_norm": 1.1246964931488037,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 94660
+    },
+    {
+      "epoch": 622.828947368421,
+      "grad_norm": 1.0840219259262085,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 94670
+    },
+    {
+      "epoch": 622.8947368421053,
+      "grad_norm": 0.9175322651863098,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 94680
+    },
+    {
+      "epoch": 622.9605263157895,
+      "grad_norm": 0.6897749900817871,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 94690
+    },
+    {
+      "epoch": 623.0263157894736,
+      "grad_norm": 1.313354730606079,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 94700
+    },
+    {
+      "epoch": 623.0921052631579,
+      "grad_norm": 0.9614347815513611,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 94710
+    },
+    {
+      "epoch": 623.1578947368421,
+      "grad_norm": 1.1975575685501099,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 94720
+    },
+    {
+      "epoch": 623.2236842105264,
+      "grad_norm": 1.0742084980010986,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 94730
+    },
+    {
+      "epoch": 623.2894736842105,
+      "grad_norm": 0.9996278882026672,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 94740
+    },
+    {
+      "epoch": 623.3552631578947,
+      "grad_norm": 1.3011069297790527,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 94750
+    },
+    {
+      "epoch": 623.421052631579,
+      "grad_norm": 0.7885226011276245,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 94760
+    },
+    {
+      "epoch": 623.4868421052631,
+      "grad_norm": 0.9483335018157959,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 94770
+    },
+    {
+      "epoch": 623.5526315789474,
+      "grad_norm": 1.2516369819641113,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 94780
+    },
+    {
+      "epoch": 623.6184210526316,
+      "grad_norm": 1.041446566581726,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 94790
+    },
+    {
+      "epoch": 623.6842105263158,
+      "grad_norm": 1.1910645961761475,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 94800
+    },
+    {
+      "epoch": 623.75,
+      "grad_norm": 1.1067719459533691,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 94810
+    },
+    {
+      "epoch": 623.8157894736842,
+      "grad_norm": 1.0154963731765747,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 94820
+    },
+    {
+      "epoch": 623.8815789473684,
+      "grad_norm": 1.08232843875885,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 94830
+    },
+    {
+      "epoch": 623.9473684210526,
+      "grad_norm": 1.1486544609069824,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 94840
+    },
+    {
+      "epoch": 624.0131578947369,
+      "grad_norm": 1.1820085048675537,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 94850
+    },
+    {
+      "epoch": 624.078947368421,
+      "grad_norm": 1.095815658569336,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 94860
+    },
+    {
+      "epoch": 624.1447368421053,
+      "grad_norm": 0.9863499999046326,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 94870
+    },
+    {
+      "epoch": 624.2105263157895,
+      "grad_norm": 1.2064094543457031,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 94880
+    },
+    {
+      "epoch": 624.2763157894736,
+      "grad_norm": 0.7593415379524231,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 94890
+    },
+    {
+      "epoch": 624.3421052631579,
+      "grad_norm": 0.7509251832962036,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 94900
+    },
+    {
+      "epoch": 624.4078947368421,
+      "grad_norm": 0.8254883289337158,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 94910
+    },
+    {
+      "epoch": 624.4736842105264,
+      "grad_norm": 1.2074943780899048,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 94920
+    },
+    {
+      "epoch": 624.5394736842105,
+      "grad_norm": 1.1988282203674316,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 94930
+    },
+    {
+      "epoch": 624.6052631578947,
+      "grad_norm": 1.4289294481277466,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 94940
+    },
+    {
+      "epoch": 624.671052631579,
+      "grad_norm": 1.4623303413391113,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 94950
+    },
+    {
+      "epoch": 624.7368421052631,
+      "grad_norm": 1.0225207805633545,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 94960
+    },
+    {
+      "epoch": 624.8026315789474,
+      "grad_norm": 1.1724567413330078,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 94970
+    },
+    {
+      "epoch": 624.8684210526316,
+      "grad_norm": 1.0636459589004517,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 94980
+    },
+    {
+      "epoch": 624.9342105263158,
+      "grad_norm": 1.551671028137207,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 94990
+    },
+    {
+      "epoch": 625.0,
+      "grad_norm": 1.9914339780807495,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 95000
+    },
+    {
+      "epoch": 625.0657894736842,
+      "grad_norm": 1.3119971752166748,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 95010
+    },
+    {
+      "epoch": 625.1315789473684,
+      "grad_norm": 1.2091944217681885,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 95020
+    },
+    {
+      "epoch": 625.1973684210526,
+      "grad_norm": 1.333477258682251,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 95030
+    },
+    {
+      "epoch": 625.2631578947369,
+      "grad_norm": 1.5082767009735107,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 95040
+    },
+    {
+      "epoch": 625.328947368421,
+      "grad_norm": 1.2716140747070312,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 95050
+    },
+    {
+      "epoch": 625.3947368421053,
+      "grad_norm": 1.2596821784973145,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 95060
+    },
+    {
+      "epoch": 625.4605263157895,
+      "grad_norm": 1.198418140411377,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 95070
+    },
+    {
+      "epoch": 625.5263157894736,
+      "grad_norm": 1.1842753887176514,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 95080
+    },
+    {
+      "epoch": 625.5921052631579,
+      "grad_norm": 1.162773847579956,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 95090
+    },
+    {
+      "epoch": 625.6578947368421,
+      "grad_norm": 0.8941752910614014,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 95100
+    },
+    {
+      "epoch": 625.7236842105264,
+      "grad_norm": 0.9903761744499207,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 95110
+    },
+    {
+      "epoch": 625.7894736842105,
+      "grad_norm": 1.0713928937911987,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 95120
+    },
+    {
+      "epoch": 625.8552631578947,
+      "grad_norm": 1.1026970148086548,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 95130
+    },
+    {
+      "epoch": 625.921052631579,
+      "grad_norm": 1.091262936592102,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 95140
+    },
+    {
+      "epoch": 625.9868421052631,
+      "grad_norm": 1.0287009477615356,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 95150
+    },
+    {
+      "epoch": 626.0526315789474,
+      "grad_norm": 0.9312505125999451,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 95160
+    },
+    {
+      "epoch": 626.1184210526316,
+      "grad_norm": 1.1395642757415771,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 95170
+    },
+    {
+      "epoch": 626.1842105263158,
+      "grad_norm": 1.1278079748153687,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 95180
+    },
+    {
+      "epoch": 626.25,
+      "grad_norm": 1.112074613571167,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 95190
+    },
+    {
+      "epoch": 626.3157894736842,
+      "grad_norm": 1.41941499710083,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 95200
+    },
+    {
+      "epoch": 626.3815789473684,
+      "grad_norm": 1.5267459154129028,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 95210
+    },
+    {
+      "epoch": 626.4473684210526,
+      "grad_norm": 1.2732293605804443,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 95220
+    },
+    {
+      "epoch": 626.5131578947369,
+      "grad_norm": 1.3131200075149536,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 95230
+    },
+    {
+      "epoch": 626.578947368421,
+      "grad_norm": 1.115815281867981,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 95240
+    },
+    {
+      "epoch": 626.6447368421053,
+      "grad_norm": 1.6979179382324219,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 95250
+    },
+    {
+      "epoch": 626.7105263157895,
+      "grad_norm": 1.280093789100647,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 95260
+    },
+    {
+      "epoch": 626.7763157894736,
+      "grad_norm": 1.349902629852295,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 95270
+    },
+    {
+      "epoch": 626.8421052631579,
+      "grad_norm": 1.4500685930252075,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 95280
+    },
+    {
+      "epoch": 626.9078947368421,
+      "grad_norm": 1.2372456789016724,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 95290
+    },
+    {
+      "epoch": 626.9736842105264,
+      "grad_norm": 1.7116293907165527,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 95300
+    },
+    {
+      "epoch": 627.0394736842105,
+      "grad_norm": 1.3504589796066284,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 95310
+    },
+    {
+      "epoch": 627.1052631578947,
+      "grad_norm": 1.3713061809539795,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 95320
+    },
+    {
+      "epoch": 627.171052631579,
+      "grad_norm": 1.5668679475784302,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 95330
+    },
+    {
+      "epoch": 627.2368421052631,
+      "grad_norm": 1.5698161125183105,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 95340
+    },
+    {
+      "epoch": 627.3026315789474,
+      "grad_norm": 1.002941370010376,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 95350
+    },
+    {
+      "epoch": 627.3684210526316,
+      "grad_norm": 1.125559687614441,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 95360
+    },
+    {
+      "epoch": 627.4342105263158,
+      "grad_norm": 0.9617181420326233,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 95370
+    },
+    {
+      "epoch": 627.5,
+      "grad_norm": 0.9981349110603333,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 95380
+    },
+    {
+      "epoch": 627.5657894736842,
+      "grad_norm": 1.1463162899017334,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 95390
+    },
+    {
+      "epoch": 627.6315789473684,
+      "grad_norm": 1.363034725189209,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 95400
+    },
+    {
+      "epoch": 627.6973684210526,
+      "grad_norm": 1.0812790393829346,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 95410
+    },
+    {
+      "epoch": 627.7631578947369,
+      "grad_norm": 0.9944791793823242,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 95420
+    },
+    {
+      "epoch": 627.828947368421,
+      "grad_norm": 1.1499660015106201,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 95430
+    },
+    {
+      "epoch": 627.8947368421053,
+      "grad_norm": 1.1649742126464844,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 95440
+    },
+    {
+      "epoch": 627.9605263157895,
+      "grad_norm": 0.9862028360366821,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 95450
+    },
+    {
+      "epoch": 628.0263157894736,
+      "grad_norm": 1.0337918996810913,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 95460
+    },
+    {
+      "epoch": 628.0921052631579,
+      "grad_norm": 1.1768397092819214,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 95470
+    },
+    {
+      "epoch": 628.1578947368421,
+      "grad_norm": 1.1246834993362427,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 95480
+    },
+    {
+      "epoch": 628.2236842105264,
+      "grad_norm": 0.6552825570106506,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 95490
+    },
+    {
+      "epoch": 628.2894736842105,
+      "grad_norm": 0.9042338728904724,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 95500
+    },
+    {
+      "epoch": 628.3552631578947,
+      "grad_norm": 1.1680855751037598,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 95510
+    },
+    {
+      "epoch": 628.421052631579,
+      "grad_norm": 1.1862854957580566,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 95520
+    },
+    {
+      "epoch": 628.4868421052631,
+      "grad_norm": 0.9259408712387085,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 95530
+    },
+    {
+      "epoch": 628.5526315789474,
+      "grad_norm": 1.0121794939041138,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 95540
+    },
+    {
+      "epoch": 628.6184210526316,
+      "grad_norm": 1.021821141242981,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 95550
+    },
+    {
+      "epoch": 628.6842105263158,
+      "grad_norm": 1.0887248516082764,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 95560
+    },
+    {
+      "epoch": 628.75,
+      "grad_norm": 1.1664204597473145,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 95570
+    },
+    {
+      "epoch": 628.8157894736842,
+      "grad_norm": 0.9954600930213928,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 95580
+    },
+    {
+      "epoch": 628.8815789473684,
+      "grad_norm": 1.174320101737976,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 95590
+    },
+    {
+      "epoch": 628.9473684210526,
+      "grad_norm": 1.5383800268173218,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 95600
+    },
+    {
+      "epoch": 629.0131578947369,
+      "grad_norm": 1.5063879489898682,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 95610
+    },
+    {
+      "epoch": 629.078947368421,
+      "grad_norm": 1.2808583974838257,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 95620
+    },
+    {
+      "epoch": 629.1447368421053,
+      "grad_norm": 1.3587077856063843,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 95630
+    },
+    {
+      "epoch": 629.2105263157895,
+      "grad_norm": 1.1514275074005127,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 95640
+    },
+    {
+      "epoch": 629.2763157894736,
+      "grad_norm": 1.3568569421768188,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 95650
+    },
+    {
+      "epoch": 629.3421052631579,
+      "grad_norm": 1.1111990213394165,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 95660
+    },
+    {
+      "epoch": 629.4078947368421,
+      "grad_norm": 1.160679578781128,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 95670
+    },
+    {
+      "epoch": 629.4736842105264,
+      "grad_norm": 0.875325083732605,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 95680
+    },
+    {
+      "epoch": 629.5394736842105,
+      "grad_norm": 0.676986575126648,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 95690
+    },
+    {
+      "epoch": 629.6052631578947,
+      "grad_norm": 1.0533658266067505,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 95700
+    },
+    {
+      "epoch": 629.671052631579,
+      "grad_norm": 1.126704454421997,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 95710
+    },
+    {
+      "epoch": 629.7368421052631,
+      "grad_norm": 0.875869631767273,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 95720
+    },
+    {
+      "epoch": 629.8026315789474,
+      "grad_norm": 0.843349277973175,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 95730
+    },
+    {
+      "epoch": 629.8684210526316,
+      "grad_norm": 1.0324947834014893,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 95740
+    },
+    {
+      "epoch": 629.9342105263158,
+      "grad_norm": 1.0827348232269287,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 95750
+    },
+    {
+      "epoch": 630.0,
+      "grad_norm": 0.8079311847686768,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 95760
+    },
+    {
+      "epoch": 630.0657894736842,
+      "grad_norm": 1.0577661991119385,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 95770
+    },
+    {
+      "epoch": 630.1315789473684,
+      "grad_norm": 1.0803674459457397,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 95780
+    },
+    {
+      "epoch": 630.1973684210526,
+      "grad_norm": 1.1568626165390015,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 95790
+    },
+    {
+      "epoch": 630.2631578947369,
+      "grad_norm": 0.7139347195625305,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 95800
+    },
+    {
+      "epoch": 630.328947368421,
+      "grad_norm": 1.54314386844635,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 95810
+    },
+    {
+      "epoch": 630.3947368421053,
+      "grad_norm": 0.8538906574249268,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 95820
+    },
+    {
+      "epoch": 630.4605263157895,
+      "grad_norm": 1.0766382217407227,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 95830
+    },
+    {
+      "epoch": 630.5263157894736,
+      "grad_norm": 0.9496257901191711,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 95840
+    },
+    {
+      "epoch": 630.5921052631579,
+      "grad_norm": 1.2234865427017212,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 95850
+    },
+    {
+      "epoch": 630.6578947368421,
+      "grad_norm": 0.8329818248748779,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 95860
+    },
+    {
+      "epoch": 630.7236842105264,
+      "grad_norm": 1.100710153579712,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 95870
+    },
+    {
+      "epoch": 630.7894736842105,
+      "grad_norm": 0.9329621195793152,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 95880
+    },
+    {
+      "epoch": 630.8552631578947,
+      "grad_norm": 1.2335596084594727,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 95890
+    },
+    {
+      "epoch": 630.921052631579,
+      "grad_norm": 1.0950738191604614,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 95900
+    },
+    {
+      "epoch": 630.9868421052631,
+      "grad_norm": 0.9743636846542358,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 95910
+    },
+    {
+      "epoch": 631.0526315789474,
+      "grad_norm": 0.9001186490058899,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 95920
+    },
+    {
+      "epoch": 631.1184210526316,
+      "grad_norm": 1.3222036361694336,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 95930
+    },
+    {
+      "epoch": 631.1842105263158,
+      "grad_norm": 1.2145307064056396,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 95940
+    },
+    {
+      "epoch": 631.25,
+      "grad_norm": 1.1518114805221558,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 95950
+    },
+    {
+      "epoch": 631.3157894736842,
+      "grad_norm": 1.0182396173477173,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 95960
+    },
+    {
+      "epoch": 631.3815789473684,
+      "grad_norm": 0.998622477054596,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 95970
+    },
+    {
+      "epoch": 631.4473684210526,
+      "grad_norm": 0.9129418730735779,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 95980
+    },
+    {
+      "epoch": 631.5131578947369,
+      "grad_norm": 0.8914371728897095,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 95990
+    },
+    {
+      "epoch": 631.578947368421,
+      "grad_norm": 1.324182152748108,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 96000
+    },
+    {
+      "epoch": 631.6447368421053,
+      "grad_norm": 1.3271923065185547,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 96010
+    },
+    {
+      "epoch": 631.7105263157895,
+      "grad_norm": 1.2438828945159912,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 96020
+    },
+    {
+      "epoch": 631.7763157894736,
+      "grad_norm": 0.808535635471344,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 96030
+    },
+    {
+      "epoch": 631.8421052631579,
+      "grad_norm": 1.3767280578613281,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 96040
+    },
+    {
+      "epoch": 631.9078947368421,
+      "grad_norm": 1.3440818786621094,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 96050
+    },
+    {
+      "epoch": 631.9736842105264,
+      "grad_norm": 1.1138218641281128,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 96060
+    },
+    {
+      "epoch": 632.0394736842105,
+      "grad_norm": 0.868145227432251,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 96070
+    },
+    {
+      "epoch": 632.1052631578947,
+      "grad_norm": 1.3606065511703491,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 96080
+    },
+    {
+      "epoch": 632.171052631579,
+      "grad_norm": 1.3854315280914307,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 96090
+    },
+    {
+      "epoch": 632.2368421052631,
+      "grad_norm": 1.2756513357162476,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 96100
+    },
+    {
+      "epoch": 632.3026315789474,
+      "grad_norm": 1.1794759035110474,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 96110
+    },
+    {
+      "epoch": 632.3684210526316,
+      "grad_norm": 1.2816321849822998,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 96120
+    },
+    {
+      "epoch": 632.4342105263158,
+      "grad_norm": 1.0967599153518677,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 96130
+    },
+    {
+      "epoch": 632.5,
+      "grad_norm": 1.3079113960266113,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 96140
+    },
+    {
+      "epoch": 632.5657894736842,
+      "grad_norm": 1.0151442289352417,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 96150
+    },
+    {
+      "epoch": 632.6315789473684,
+      "grad_norm": 2.014241933822632,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 96160
+    },
+    {
+      "epoch": 632.6973684210526,
+      "grad_norm": 1.0446006059646606,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 96170
+    },
+    {
+      "epoch": 632.7631578947369,
+      "grad_norm": 1.2573105096817017,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 96180
+    },
+    {
+      "epoch": 632.828947368421,
+      "grad_norm": 1.1705162525177002,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 96190
+    },
+    {
+      "epoch": 632.8947368421053,
+      "grad_norm": 1.3962786197662354,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 96200
+    },
+    {
+      "epoch": 632.9605263157895,
+      "grad_norm": 1.2889972925186157,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 96210
+    },
+    {
+      "epoch": 633.0263157894736,
+      "grad_norm": 1.30446457862854,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 96220
+    },
+    {
+      "epoch": 633.0921052631579,
+      "grad_norm": 1.0984090566635132,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 96230
+    },
+    {
+      "epoch": 633.1578947368421,
+      "grad_norm": 1.0214024782180786,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 96240
+    },
+    {
+      "epoch": 633.2236842105264,
+      "grad_norm": 1.3582532405853271,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 96250
+    },
+    {
+      "epoch": 633.2894736842105,
+      "grad_norm": 1.1658719778060913,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 96260
+    },
+    {
+      "epoch": 633.3552631578947,
+      "grad_norm": 1.1683430671691895,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 96270
+    },
+    {
+      "epoch": 633.421052631579,
+      "grad_norm": 1.399514079093933,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 96280
+    },
+    {
+      "epoch": 633.4868421052631,
+      "grad_norm": 1.0703054666519165,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 96290
+    },
+    {
+      "epoch": 633.5526315789474,
+      "grad_norm": 1.1495949029922485,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 96300
+    },
+    {
+      "epoch": 633.6184210526316,
+      "grad_norm": 1.1547905206680298,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 96310
+    },
+    {
+      "epoch": 633.6842105263158,
+      "grad_norm": 1.3389787673950195,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 96320
+    },
+    {
+      "epoch": 633.75,
+      "grad_norm": 0.622372031211853,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 96330
+    },
+    {
+      "epoch": 633.8157894736842,
+      "grad_norm": 0.7797669768333435,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 96340
+    },
+    {
+      "epoch": 633.8815789473684,
+      "grad_norm": 1.225666880607605,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 96350
+    },
+    {
+      "epoch": 633.9473684210526,
+      "grad_norm": 1.2764816284179688,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 96360
+    },
+    {
+      "epoch": 634.0131578947369,
+      "grad_norm": 1.3104716539382935,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 96370
+    },
+    {
+      "epoch": 634.078947368421,
+      "grad_norm": 1.122225046157837,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 96380
+    },
+    {
+      "epoch": 634.1447368421053,
+      "grad_norm": 1.331106424331665,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 96390
+    },
+    {
+      "epoch": 634.2105263157895,
+      "grad_norm": 1.1977465152740479,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 96400
+    },
+    {
+      "epoch": 634.2763157894736,
+      "grad_norm": 1.4676520824432373,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 96410
+    },
+    {
+      "epoch": 634.3421052631579,
+      "grad_norm": 1.1884233951568604,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 96420
+    },
+    {
+      "epoch": 634.4078947368421,
+      "grad_norm": 1.448885440826416,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 96430
+    },
+    {
+      "epoch": 634.4736842105264,
+      "grad_norm": 1.1831235885620117,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 96440
+    },
+    {
+      "epoch": 634.5394736842105,
+      "grad_norm": 1.1921072006225586,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 96450
+    },
+    {
+      "epoch": 634.6052631578947,
+      "grad_norm": 0.9832690954208374,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 96460
+    },
+    {
+      "epoch": 634.671052631579,
+      "grad_norm": 1.3597534894943237,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 96470
+    },
+    {
+      "epoch": 634.7368421052631,
+      "grad_norm": 1.2656805515289307,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 96480
+    },
+    {
+      "epoch": 634.8026315789474,
+      "grad_norm": 0.9572046995162964,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 96490
+    },
+    {
+      "epoch": 634.8684210526316,
+      "grad_norm": 1.0890827178955078,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 96500
+    },
+    {
+      "epoch": 634.9342105263158,
+      "grad_norm": 1.2112818956375122,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 96510
+    },
+    {
+      "epoch": 635.0,
+      "grad_norm": 1.3493852615356445,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 96520
+    },
+    {
+      "epoch": 635.0657894736842,
+      "grad_norm": 0.8873164057731628,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 96530
+    },
+    {
+      "epoch": 635.1315789473684,
+      "grad_norm": 1.3361443281173706,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 96540
+    },
+    {
+      "epoch": 635.1973684210526,
+      "grad_norm": 1.2855825424194336,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 96550
+    },
+    {
+      "epoch": 635.2631578947369,
+      "grad_norm": 1.0987671613693237,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 96560
+    },
+    {
+      "epoch": 635.328947368421,
+      "grad_norm": 0.7303910255432129,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 96570
+    },
+    {
+      "epoch": 635.3947368421053,
+      "grad_norm": 1.2419792413711548,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 96580
+    },
+    {
+      "epoch": 635.4605263157895,
+      "grad_norm": 1.311336874961853,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 96590
+    },
+    {
+      "epoch": 635.5263157894736,
+      "grad_norm": 1.0877469778060913,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 96600
+    },
+    {
+      "epoch": 635.5921052631579,
+      "grad_norm": 1.0030274391174316,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 96610
+    },
+    {
+      "epoch": 635.6578947368421,
+      "grad_norm": 1.3998702764511108,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 96620
+    },
+    {
+      "epoch": 635.7236842105264,
+      "grad_norm": 0.7409038543701172,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 96630
+    },
+    {
+      "epoch": 635.7894736842105,
+      "grad_norm": 0.9985183477401733,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 96640
+    },
+    {
+      "epoch": 635.8552631578947,
+      "grad_norm": 1.0499337911605835,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 96650
+    },
+    {
+      "epoch": 635.921052631579,
+      "grad_norm": 0.9747737050056458,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 96660
+    },
+    {
+      "epoch": 635.9868421052631,
+      "grad_norm": 1.1201547384262085,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 96670
+    },
+    {
+      "epoch": 636.0526315789474,
+      "grad_norm": 1.0508447885513306,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 96680
+    },
+    {
+      "epoch": 636.1184210526316,
+      "grad_norm": 0.8244642019271851,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 96690
+    },
+    {
+      "epoch": 636.1842105263158,
+      "grad_norm": 1.1868029832839966,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 96700
+    },
+    {
+      "epoch": 636.25,
+      "grad_norm": 1.2624951601028442,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 96710
+    },
+    {
+      "epoch": 636.3157894736842,
+      "grad_norm": 1.0682224035263062,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 96720
+    },
+    {
+      "epoch": 636.3815789473684,
+      "grad_norm": 1.52485990524292,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 96730
+    },
+    {
+      "epoch": 636.4473684210526,
+      "grad_norm": 1.0763731002807617,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 96740
+    },
+    {
+      "epoch": 636.5131578947369,
+      "grad_norm": 1.3827406167984009,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 96750
+    },
+    {
+      "epoch": 636.578947368421,
+      "grad_norm": 1.2517787218093872,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 96760
+    },
+    {
+      "epoch": 636.6447368421053,
+      "grad_norm": 1.0848556756973267,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 96770
+    },
+    {
+      "epoch": 636.7105263157895,
+      "grad_norm": 0.9772914052009583,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 96780
+    },
+    {
+      "epoch": 636.7763157894736,
+      "grad_norm": 1.1397275924682617,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 96790
+    },
+    {
+      "epoch": 636.8421052631579,
+      "grad_norm": 0.940963089466095,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 96800
+    },
+    {
+      "epoch": 636.9078947368421,
+      "grad_norm": 0.8980696797370911,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 96810
+    },
+    {
+      "epoch": 636.9736842105264,
+      "grad_norm": 0.9884552359580994,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 96820
+    },
+    {
+      "epoch": 637.0394736842105,
+      "grad_norm": 1.1750383377075195,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 96830
+    },
+    {
+      "epoch": 637.1052631578947,
+      "grad_norm": 1.1034425497055054,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 96840
+    },
+    {
+      "epoch": 637.171052631579,
+      "grad_norm": 0.7074925899505615,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 96850
+    },
+    {
+      "epoch": 637.2368421052631,
+      "grad_norm": 0.977379322052002,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 96860
+    },
+    {
+      "epoch": 637.3026315789474,
+      "grad_norm": 0.942234456539154,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 96870
+    },
+    {
+      "epoch": 637.3684210526316,
+      "grad_norm": 1.336562156677246,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 96880
+    },
+    {
+      "epoch": 637.4342105263158,
+      "grad_norm": 1.1461960077285767,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 96890
+    },
+    {
+      "epoch": 637.5,
+      "grad_norm": 1.023284912109375,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 96900
+    },
+    {
+      "epoch": 637.5657894736842,
+      "grad_norm": 1.1981929540634155,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 96910
+    },
+    {
+      "epoch": 637.6315789473684,
+      "grad_norm": 1.3196550607681274,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 96920
+    },
+    {
+      "epoch": 637.6973684210526,
+      "grad_norm": 1.042585015296936,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 96930
+    },
+    {
+      "epoch": 637.7631578947369,
+      "grad_norm": 0.9002341628074646,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 96940
+    },
+    {
+      "epoch": 637.828947368421,
+      "grad_norm": 0.8911952376365662,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 96950
+    },
+    {
+      "epoch": 637.8947368421053,
+      "grad_norm": 1.2505193948745728,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 96960
+    },
+    {
+      "epoch": 637.9605263157895,
+      "grad_norm": 1.0482158660888672,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 96970
+    },
+    {
+      "epoch": 638.0263157894736,
+      "grad_norm": 1.1949135065078735,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 96980
+    },
+    {
+      "epoch": 638.0921052631579,
+      "grad_norm": 1.1491917371749878,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 96990
+    },
+    {
+      "epoch": 638.1578947368421,
+      "grad_norm": 1.0446337461471558,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 97000
+    },
+    {
+      "epoch": 638.2236842105264,
+      "grad_norm": 1.1364582777023315,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 97010
+    },
+    {
+      "epoch": 638.2894736842105,
+      "grad_norm": 0.856816291809082,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 97020
+    },
+    {
+      "epoch": 638.3552631578947,
+      "grad_norm": 0.8455696702003479,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 97030
+    },
+    {
+      "epoch": 638.421052631579,
+      "grad_norm": 1.37261164188385,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 97040
+    },
+    {
+      "epoch": 638.4868421052631,
+      "grad_norm": 1.065354585647583,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 97050
+    },
+    {
+      "epoch": 638.5526315789474,
+      "grad_norm": 1.0042915344238281,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 97060
+    },
+    {
+      "epoch": 638.6184210526316,
+      "grad_norm": 0.9727343320846558,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 97070
+    },
+    {
+      "epoch": 638.6842105263158,
+      "grad_norm": 0.724888026714325,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 97080
+    },
+    {
+      "epoch": 638.75,
+      "grad_norm": 1.3290162086486816,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 97090
+    },
+    {
+      "epoch": 638.8157894736842,
+      "grad_norm": 1.107700228691101,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 97100
+    },
+    {
+      "epoch": 638.8815789473684,
+      "grad_norm": 1.076773762702942,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 97110
+    },
+    {
+      "epoch": 638.9473684210526,
+      "grad_norm": 0.8548887968063354,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 97120
+    },
+    {
+      "epoch": 639.0131578947369,
+      "grad_norm": 1.2806280851364136,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 97130
+    },
+    {
+      "epoch": 639.078947368421,
+      "grad_norm": 0.99432772397995,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 97140
+    },
+    {
+      "epoch": 639.1447368421053,
+      "grad_norm": 1.1284815073013306,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 97150
+    },
+    {
+      "epoch": 639.2105263157895,
+      "grad_norm": 1.4430863857269287,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 97160
+    },
+    {
+      "epoch": 639.2763157894736,
+      "grad_norm": 0.9166154265403748,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 97170
+    },
+    {
+      "epoch": 639.3421052631579,
+      "grad_norm": 1.2260555028915405,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 97180
+    },
+    {
+      "epoch": 639.4078947368421,
+      "grad_norm": 1.1222833395004272,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 97190
+    },
+    {
+      "epoch": 639.4736842105264,
+      "grad_norm": 1.1462934017181396,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 97200
+    },
+    {
+      "epoch": 639.5394736842105,
+      "grad_norm": 1.0073168277740479,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 97210
+    },
+    {
+      "epoch": 639.6052631578947,
+      "grad_norm": 1.0227516889572144,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 97220
+    },
+    {
+      "epoch": 639.671052631579,
+      "grad_norm": 0.9892854690551758,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 97230
+    },
+    {
+      "epoch": 639.7368421052631,
+      "grad_norm": 0.918364942073822,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 97240
+    },
+    {
+      "epoch": 639.8026315789474,
+      "grad_norm": 0.8302596211433411,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 97250
+    },
+    {
+      "epoch": 639.8684210526316,
+      "grad_norm": 1.3149603605270386,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 97260
+    },
+    {
+      "epoch": 639.9342105263158,
+      "grad_norm": 1.0401725769042969,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 97270
+    },
+    {
+      "epoch": 640.0,
+      "grad_norm": 0.8277792930603027,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 97280
+    },
+    {
+      "epoch": 640.0657894736842,
+      "grad_norm": 0.8482030034065247,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 97290
+    },
+    {
+      "epoch": 640.1315789473684,
+      "grad_norm": 1.0147836208343506,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 97300
+    },
+    {
+      "epoch": 640.1973684210526,
+      "grad_norm": 1.1419466733932495,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 97310
+    },
+    {
+      "epoch": 640.2631578947369,
+      "grad_norm": 0.7621695399284363,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 97320
+    },
+    {
+      "epoch": 640.328947368421,
+      "grad_norm": 1.1174577474594116,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 97330
+    },
+    {
+      "epoch": 640.3947368421053,
+      "grad_norm": 0.8022199273109436,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 97340
+    },
+    {
+      "epoch": 640.4605263157895,
+      "grad_norm": 0.8935840129852295,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 97350
+    },
+    {
+      "epoch": 640.5263157894736,
+      "grad_norm": 1.3368089199066162,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 97360
+    },
+    {
+      "epoch": 640.5921052631579,
+      "grad_norm": 1.0535829067230225,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 97370
+    },
+    {
+      "epoch": 640.6578947368421,
+      "grad_norm": 1.1254475116729736,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 97380
+    },
+    {
+      "epoch": 640.7236842105264,
+      "grad_norm": 1.1946035623550415,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 97390
+    },
+    {
+      "epoch": 640.7894736842105,
+      "grad_norm": 1.2451772689819336,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 97400
+    },
+    {
+      "epoch": 640.8552631578947,
+      "grad_norm": 0.8775976896286011,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 97410
+    },
+    {
+      "epoch": 640.921052631579,
+      "grad_norm": 1.0031647682189941,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 97420
+    },
+    {
+      "epoch": 640.9868421052631,
+      "grad_norm": 1.1627060174942017,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 97430
+    },
+    {
+      "epoch": 641.0526315789474,
+      "grad_norm": 0.9640074372291565,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 97440
+    },
+    {
+      "epoch": 641.1184210526316,
+      "grad_norm": 1.0985318422317505,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 97450
+    },
+    {
+      "epoch": 641.1842105263158,
+      "grad_norm": 0.9738800525665283,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 97460
+    },
+    {
+      "epoch": 641.25,
+      "grad_norm": 0.9755899906158447,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 97470
+    },
+    {
+      "epoch": 641.3157894736842,
+      "grad_norm": 1.0237241983413696,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 97480
+    },
+    {
+      "epoch": 641.3815789473684,
+      "grad_norm": 0.8942602872848511,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 97490
+    },
+    {
+      "epoch": 641.4473684210526,
+      "grad_norm": 0.7703356742858887,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 97500
+    },
+    {
+      "epoch": 641.5131578947369,
+      "grad_norm": 1.2847163677215576,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 97510
+    },
+    {
+      "epoch": 641.578947368421,
+      "grad_norm": 0.8879604339599609,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 97520
+    },
+    {
+      "epoch": 641.6447368421053,
+      "grad_norm": 1.2082008123397827,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 97530
+    },
+    {
+      "epoch": 641.7105263157895,
+      "grad_norm": 1.3666983842849731,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 97540
+    },
+    {
+      "epoch": 641.7763157894736,
+      "grad_norm": 1.454520583152771,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 97550
+    },
+    {
+      "epoch": 641.8421052631579,
+      "grad_norm": 1.0853079557418823,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 97560
+    },
+    {
+      "epoch": 641.9078947368421,
+      "grad_norm": 0.8811721205711365,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 97570
+    },
+    {
+      "epoch": 641.9736842105264,
+      "grad_norm": 1.1882359981536865,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 97580
+    },
+    {
+      "epoch": 642.0394736842105,
+      "grad_norm": 1.194444179534912,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 97590
+    },
+    {
+      "epoch": 642.1052631578947,
+      "grad_norm": 1.390655517578125,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 97600
+    },
+    {
+      "epoch": 642.171052631579,
+      "grad_norm": 0.7747687697410583,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 97610
+    },
+    {
+      "epoch": 642.2368421052631,
+      "grad_norm": 0.9693976640701294,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 97620
+    },
+    {
+      "epoch": 642.3026315789474,
+      "grad_norm": 1.255947470664978,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 97630
+    },
+    {
+      "epoch": 642.3684210526316,
+      "grad_norm": 1.0807551145553589,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 97640
+    },
+    {
+      "epoch": 642.4342105263158,
+      "grad_norm": 1.077979326248169,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 97650
+    },
+    {
+      "epoch": 642.5,
+      "grad_norm": 0.8917891979217529,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 97660
+    },
+    {
+      "epoch": 642.5657894736842,
+      "grad_norm": 1.0527293682098389,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 97670
+    },
+    {
+      "epoch": 642.6315789473684,
+      "grad_norm": 0.8535796403884888,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 97680
+    },
+    {
+      "epoch": 642.6973684210526,
+      "grad_norm": 0.9337286949157715,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 97690
+    },
+    {
+      "epoch": 642.7631578947369,
+      "grad_norm": 1.0351840257644653,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 97700
+    },
+    {
+      "epoch": 642.828947368421,
+      "grad_norm": 0.8320999145507812,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 97710
+    },
+    {
+      "epoch": 642.8947368421053,
+      "grad_norm": 1.2720904350280762,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 97720
+    },
+    {
+      "epoch": 642.9605263157895,
+      "grad_norm": 1.19514799118042,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 97730
+    },
+    {
+      "epoch": 643.0263157894736,
+      "grad_norm": 1.1567554473876953,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 97740
+    },
+    {
+      "epoch": 643.0921052631579,
+      "grad_norm": 0.992160975933075,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 97750
+    },
+    {
+      "epoch": 643.1578947368421,
+      "grad_norm": 0.9545310735702515,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 97760
+    },
+    {
+      "epoch": 643.2236842105264,
+      "grad_norm": 0.5216129422187805,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 97770
+    },
+    {
+      "epoch": 643.2894736842105,
+      "grad_norm": 1.1363471746444702,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 97780
+    },
+    {
+      "epoch": 643.3552631578947,
+      "grad_norm": 1.0840483903884888,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 97790
+    },
+    {
+      "epoch": 643.421052631579,
+      "grad_norm": 0.9298526048660278,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 97800
+    },
+    {
+      "epoch": 643.4868421052631,
+      "grad_norm": 0.9495653510093689,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 97810
+    },
+    {
+      "epoch": 643.5526315789474,
+      "grad_norm": 0.9317263960838318,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 97820
+    },
+    {
+      "epoch": 643.6184210526316,
+      "grad_norm": 1.0475777387619019,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 97830
+    },
+    {
+      "epoch": 643.6842105263158,
+      "grad_norm": 1.4607027769088745,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 97840
+    },
+    {
+      "epoch": 643.75,
+      "grad_norm": 0.8541473150253296,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 97850
+    },
+    {
+      "epoch": 643.8157894736842,
+      "grad_norm": 0.600893497467041,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 97860
+    },
+    {
+      "epoch": 643.8815789473684,
+      "grad_norm": 0.9018236994743347,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 97870
+    },
+    {
+      "epoch": 643.9473684210526,
+      "grad_norm": 0.7661747932434082,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 97880
+    },
+    {
+      "epoch": 644.0131578947369,
+      "grad_norm": 0.8002960085868835,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 97890
+    },
+    {
+      "epoch": 644.078947368421,
+      "grad_norm": 0.8436306715011597,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 97900
+    },
+    {
+      "epoch": 644.1447368421053,
+      "grad_norm": 1.3472542762756348,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 97910
+    },
+    {
+      "epoch": 644.2105263157895,
+      "grad_norm": 0.8664248585700989,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 97920
+    },
+    {
+      "epoch": 644.2763157894736,
+      "grad_norm": 0.9880051612854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 97930
+    },
+    {
+      "epoch": 644.3421052631579,
+      "grad_norm": 1.1191985607147217,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 97940
+    },
+    {
+      "epoch": 644.4078947368421,
+      "grad_norm": 1.2010605335235596,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 97950
+    },
+    {
+      "epoch": 644.4736842105264,
+      "grad_norm": 1.250929594039917,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 97960
+    },
+    {
+      "epoch": 644.5394736842105,
+      "grad_norm": 1.0687397718429565,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 97970
+    },
+    {
+      "epoch": 644.6052631578947,
+      "grad_norm": 0.9296792149543762,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 97980
+    },
+    {
+      "epoch": 644.671052631579,
+      "grad_norm": 1.3585340976715088,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 97990
+    },
+    {
+      "epoch": 644.7368421052631,
+      "grad_norm": 1.1659291982650757,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 98000
+    },
+    {
+      "epoch": 644.8026315789474,
+      "grad_norm": 1.242610216140747,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 98010
+    },
+    {
+      "epoch": 644.8684210526316,
+      "grad_norm": 1.096759557723999,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 98020
+    },
+    {
+      "epoch": 644.9342105263158,
+      "grad_norm": 1.0130127668380737,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 98030
+    },
+    {
+      "epoch": 645.0,
+      "grad_norm": 1.1907083988189697,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 98040
+    },
+    {
+      "epoch": 645.0657894736842,
+      "grad_norm": 1.232130765914917,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 98050
+    },
+    {
+      "epoch": 645.1315789473684,
+      "grad_norm": 1.1238503456115723,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 98060
+    },
+    {
+      "epoch": 645.1973684210526,
+      "grad_norm": 0.9487515091896057,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 98070
+    },
+    {
+      "epoch": 645.2631578947369,
+      "grad_norm": 0.9860408902168274,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 98080
+    },
+    {
+      "epoch": 645.328947368421,
+      "grad_norm": 0.7561169862747192,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 98090
+    },
+    {
+      "epoch": 645.3947368421053,
+      "grad_norm": 0.9736877679824829,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 98100
+    },
+    {
+      "epoch": 645.4605263157895,
+      "grad_norm": 0.6840393543243408,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 98110
+    },
+    {
+      "epoch": 645.5263157894736,
+      "grad_norm": 1.2354158163070679,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 98120
+    },
+    {
+      "epoch": 645.5921052631579,
+      "grad_norm": 1.306350827217102,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 98130
+    },
+    {
+      "epoch": 645.6578947368421,
+      "grad_norm": 1.1757615804672241,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 98140
+    },
+    {
+      "epoch": 645.7236842105264,
+      "grad_norm": 1.1148210763931274,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 98150
+    },
+    {
+      "epoch": 645.7894736842105,
+      "grad_norm": 1.281851887702942,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 98160
+    },
+    {
+      "epoch": 645.8552631578947,
+      "grad_norm": 0.8599856495857239,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 98170
+    },
+    {
+      "epoch": 645.921052631579,
+      "grad_norm": 0.9599358439445496,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 98180
+    },
+    {
+      "epoch": 645.9868421052631,
+      "grad_norm": 1.0771713256835938,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 98190
+    },
+    {
+      "epoch": 646.0526315789474,
+      "grad_norm": 0.7740371227264404,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 98200
+    },
+    {
+      "epoch": 646.1184210526316,
+      "grad_norm": 1.125326156616211,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 98210
+    },
+    {
+      "epoch": 646.1842105263158,
+      "grad_norm": 1.2309529781341553,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 98220
+    },
+    {
+      "epoch": 646.25,
+      "grad_norm": 1.0347673892974854,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 98230
+    },
+    {
+      "epoch": 646.3157894736842,
+      "grad_norm": 1.0844132900238037,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 98240
+    },
+    {
+      "epoch": 646.3815789473684,
+      "grad_norm": 0.7745186686515808,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 98250
+    },
+    {
+      "epoch": 646.4473684210526,
+      "grad_norm": 1.0603491067886353,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 98260
+    },
+    {
+      "epoch": 646.5131578947369,
+      "grad_norm": 1.0565520524978638,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 98270
+    },
+    {
+      "epoch": 646.578947368421,
+      "grad_norm": 1.1339266300201416,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 98280
+    },
+    {
+      "epoch": 646.6447368421053,
+      "grad_norm": 1.2667466402053833,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 98290
+    },
+    {
+      "epoch": 646.7105263157895,
+      "grad_norm": 1.1827952861785889,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 98300
+    },
+    {
+      "epoch": 646.7763157894736,
+      "grad_norm": 1.1136367321014404,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 98310
+    },
+    {
+      "epoch": 646.8421052631579,
+      "grad_norm": 0.8930481672286987,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 98320
+    },
+    {
+      "epoch": 646.9078947368421,
+      "grad_norm": 1.1677948236465454,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 98330
+    },
+    {
+      "epoch": 646.9736842105264,
+      "grad_norm": 1.003639817237854,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 98340
+    },
+    {
+      "epoch": 647.0394736842105,
+      "grad_norm": 0.9792555570602417,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 98350
+    },
+    {
+      "epoch": 647.1052631578947,
+      "grad_norm": 1.3798192739486694,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 98360
+    },
+    {
+      "epoch": 647.171052631579,
+      "grad_norm": 0.9844328165054321,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 98370
+    },
+    {
+      "epoch": 647.2368421052631,
+      "grad_norm": 1.2482646703720093,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 98380
+    },
+    {
+      "epoch": 647.3026315789474,
+      "grad_norm": 0.796341598033905,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 98390
+    },
+    {
+      "epoch": 647.3684210526316,
+      "grad_norm": 0.8844649791717529,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 98400
+    },
+    {
+      "epoch": 647.4342105263158,
+      "grad_norm": 1.4429231882095337,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 98410
+    },
+    {
+      "epoch": 647.5,
+      "grad_norm": 1.1434082984924316,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 98420
+    },
+    {
+      "epoch": 647.5657894736842,
+      "grad_norm": 1.153358817100525,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 98430
+    },
+    {
+      "epoch": 647.6315789473684,
+      "grad_norm": 1.1155855655670166,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 98440
+    },
+    {
+      "epoch": 647.6973684210526,
+      "grad_norm": 0.8420267701148987,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 98450
+    },
+    {
+      "epoch": 647.7631578947369,
+      "grad_norm": 0.9819120764732361,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 98460
+    },
+    {
+      "epoch": 647.828947368421,
+      "grad_norm": 1.6169120073318481,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 98470
+    },
+    {
+      "epoch": 647.8947368421053,
+      "grad_norm": 1.1635892391204834,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 98480
+    },
+    {
+      "epoch": 647.9605263157895,
+      "grad_norm": 1.0484620332717896,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 98490
+    },
+    {
+      "epoch": 648.0263157894736,
+      "grad_norm": 0.8859636783599854,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 98500
+    },
+    {
+      "epoch": 648.0921052631579,
+      "grad_norm": 0.9191065430641174,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 98510
+    },
+    {
+      "epoch": 648.1578947368421,
+      "grad_norm": 1.2865121364593506,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 98520
+    },
+    {
+      "epoch": 648.2236842105264,
+      "grad_norm": 1.1271616220474243,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 98530
+    },
+    {
+      "epoch": 648.2894736842105,
+      "grad_norm": 1.0781117677688599,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 98540
+    },
+    {
+      "epoch": 648.3552631578947,
+      "grad_norm": 0.8963111639022827,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 98550
+    },
+    {
+      "epoch": 648.421052631579,
+      "grad_norm": 0.8486378192901611,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 98560
+    },
+    {
+      "epoch": 648.4868421052631,
+      "grad_norm": 0.7802874445915222,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 98570
+    },
+    {
+      "epoch": 648.5526315789474,
+      "grad_norm": 0.9380501508712769,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 98580
+    },
+    {
+      "epoch": 648.6184210526316,
+      "grad_norm": 1.1357520818710327,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 98590
+    },
+    {
+      "epoch": 648.6842105263158,
+      "grad_norm": 0.9585915207862854,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 98600
+    },
+    {
+      "epoch": 648.75,
+      "grad_norm": 1.170572280883789,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 98610
+    },
+    {
+      "epoch": 648.8157894736842,
+      "grad_norm": 0.7782623767852783,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 98620
+    },
+    {
+      "epoch": 648.8815789473684,
+      "grad_norm": 1.2383381128311157,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 98630
+    },
+    {
+      "epoch": 648.9473684210526,
+      "grad_norm": 1.1335386037826538,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 98640
+    },
+    {
+      "epoch": 649.0131578947369,
+      "grad_norm": 1.0189485549926758,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 98650
+    },
+    {
+      "epoch": 649.078947368421,
+      "grad_norm": 1.0290753841400146,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 98660
+    },
+    {
+      "epoch": 649.1447368421053,
+      "grad_norm": 1.2437363862991333,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 98670
+    },
+    {
+      "epoch": 649.2105263157895,
+      "grad_norm": 0.781124472618103,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 98680
+    },
+    {
+      "epoch": 649.2763157894736,
+      "grad_norm": 1.2116012573242188,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 98690
+    },
+    {
+      "epoch": 649.3421052631579,
+      "grad_norm": 0.9548416137695312,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 98700
+    },
+    {
+      "epoch": 649.4078947368421,
+      "grad_norm": 0.7918787002563477,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 98710
+    },
+    {
+      "epoch": 649.4736842105264,
+      "grad_norm": 0.8988084197044373,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 98720
+    },
+    {
+      "epoch": 649.5394736842105,
+      "grad_norm": 0.9593010544776917,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 98730
+    },
+    {
+      "epoch": 649.6052631578947,
+      "grad_norm": 1.205809235572815,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 98740
+    },
+    {
+      "epoch": 649.671052631579,
+      "grad_norm": 1.1642436981201172,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 98750
+    },
+    {
+      "epoch": 649.7368421052631,
+      "grad_norm": 0.8834133744239807,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 98760
+    },
+    {
+      "epoch": 649.8026315789474,
+      "grad_norm": 1.1076525449752808,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 98770
+    },
+    {
+      "epoch": 649.8684210526316,
+      "grad_norm": 1.2798569202423096,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 98780
+    },
+    {
+      "epoch": 649.9342105263158,
+      "grad_norm": 0.937585175037384,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 98790
+    },
+    {
+      "epoch": 650.0,
+      "grad_norm": 1.0356471538543701,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 98800
+    },
+    {
+      "epoch": 650.0657894736842,
+      "grad_norm": 1.1217914819717407,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 98810
+    },
+    {
+      "epoch": 650.1315789473684,
+      "grad_norm": 0.9628757238388062,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 98820
+    },
+    {
+      "epoch": 650.1973684210526,
+      "grad_norm": 1.3987805843353271,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 98830
+    },
+    {
+      "epoch": 650.2631578947369,
+      "grad_norm": 1.4442050457000732,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 98840
+    },
+    {
+      "epoch": 650.328947368421,
+      "grad_norm": 1.1787656545639038,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 98850
+    },
+    {
+      "epoch": 650.3947368421053,
+      "grad_norm": 1.3678606748580933,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 98860
+    },
+    {
+      "epoch": 650.4605263157895,
+      "grad_norm": 1.1847083568572998,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 98870
+    },
+    {
+      "epoch": 650.5263157894736,
+      "grad_norm": 1.195621132850647,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 98880
+    },
+    {
+      "epoch": 650.5921052631579,
+      "grad_norm": 1.2353485822677612,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 98890
+    },
+    {
+      "epoch": 650.6578947368421,
+      "grad_norm": 0.8940504789352417,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 98900
+    },
+    {
+      "epoch": 650.7236842105264,
+      "grad_norm": 1.0895377397537231,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 98910
+    },
+    {
+      "epoch": 650.7894736842105,
+      "grad_norm": 1.0993573665618896,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 98920
+    },
+    {
+      "epoch": 650.8552631578947,
+      "grad_norm": 1.2110718488693237,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 98930
+    },
+    {
+      "epoch": 650.921052631579,
+      "grad_norm": 0.9158070683479309,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 98940
+    },
+    {
+      "epoch": 650.9868421052631,
+      "grad_norm": 1.1650333404541016,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 98950
+    },
+    {
+      "epoch": 651.0526315789474,
+      "grad_norm": 1.2735977172851562,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 98960
+    },
+    {
+      "epoch": 651.1184210526316,
+      "grad_norm": 1.1164957284927368,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 98970
+    },
+    {
+      "epoch": 651.1842105263158,
+      "grad_norm": 1.1054691076278687,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 98980
+    },
+    {
+      "epoch": 651.25,
+      "grad_norm": 1.2364662885665894,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 98990
+    },
+    {
+      "epoch": 651.3157894736842,
+      "grad_norm": 1.0556890964508057,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 99000
+    },
+    {
+      "epoch": 651.3815789473684,
+      "grad_norm": 1.3006346225738525,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 99010
+    },
+    {
+      "epoch": 651.4473684210526,
+      "grad_norm": 1.3828588724136353,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 99020
+    },
+    {
+      "epoch": 651.5131578947369,
+      "grad_norm": 1.1058744192123413,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 99030
+    },
+    {
+      "epoch": 651.578947368421,
+      "grad_norm": 1.0962953567504883,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 99040
+    },
+    {
+      "epoch": 651.6447368421053,
+      "grad_norm": 1.1675875186920166,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 99050
+    },
+    {
+      "epoch": 651.7105263157895,
+      "grad_norm": 1.486832857131958,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 99060
+    },
+    {
+      "epoch": 651.7763157894736,
+      "grad_norm": 0.9721072912216187,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 99070
+    },
+    {
+      "epoch": 651.8421052631579,
+      "grad_norm": 1.1138964891433716,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 99080
+    },
+    {
+      "epoch": 651.9078947368421,
+      "grad_norm": 0.9725443124771118,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 99090
+    },
+    {
+      "epoch": 651.9736842105264,
+      "grad_norm": 1.2599581480026245,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 99100
+    },
+    {
+      "epoch": 652.0394736842105,
+      "grad_norm": 0.8848244547843933,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 99110
+    },
+    {
+      "epoch": 652.1052631578947,
+      "grad_norm": 1.0993847846984863,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 99120
+    },
+    {
+      "epoch": 652.171052631579,
+      "grad_norm": 1.1078648567199707,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 99130
+    },
+    {
+      "epoch": 652.2368421052631,
+      "grad_norm": 1.1154159307479858,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 99140
+    },
+    {
+      "epoch": 652.3026315789474,
+      "grad_norm": 1.2033315896987915,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 99150
+    },
+    {
+      "epoch": 652.3684210526316,
+      "grad_norm": 0.8731604218482971,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 99160
+    },
+    {
+      "epoch": 652.4342105263158,
+      "grad_norm": 1.3480206727981567,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 99170
+    },
+    {
+      "epoch": 652.5,
+      "grad_norm": 1.2460421323776245,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 99180
+    },
+    {
+      "epoch": 652.5657894736842,
+      "grad_norm": 1.5589008331298828,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 99190
+    },
+    {
+      "epoch": 652.6315789473684,
+      "grad_norm": 1.400813341140747,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 99200
+    },
+    {
+      "epoch": 652.6973684210526,
+      "grad_norm": 1.5220674276351929,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 99210
+    },
+    {
+      "epoch": 652.7631578947369,
+      "grad_norm": 1.3454396724700928,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 99220
+    },
+    {
+      "epoch": 652.828947368421,
+      "grad_norm": 1.3195019960403442,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 99230
+    },
+    {
+      "epoch": 652.8947368421053,
+      "grad_norm": 1.0843956470489502,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 99240
+    },
+    {
+      "epoch": 652.9605263157895,
+      "grad_norm": 1.2075053453445435,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 99250
+    },
+    {
+      "epoch": 653.0263157894736,
+      "grad_norm": 1.3347957134246826,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 99260
+    },
+    {
+      "epoch": 653.0921052631579,
+      "grad_norm": 1.1405830383300781,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 99270
+    },
+    {
+      "epoch": 653.1578947368421,
+      "grad_norm": 0.9808728098869324,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 99280
+    },
+    {
+      "epoch": 653.2236842105264,
+      "grad_norm": 1.1668258905410767,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 99290
+    },
+    {
+      "epoch": 653.2894736842105,
+      "grad_norm": 1.2372167110443115,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 99300
+    },
+    {
+      "epoch": 653.3552631578947,
+      "grad_norm": 1.0494773387908936,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 99310
+    },
+    {
+      "epoch": 653.421052631579,
+      "grad_norm": 1.203221082687378,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 99320
+    },
+    {
+      "epoch": 653.4868421052631,
+      "grad_norm": 1.0435203313827515,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 99330
+    },
+    {
+      "epoch": 653.5526315789474,
+      "grad_norm": 1.1974724531173706,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 99340
+    },
+    {
+      "epoch": 653.6184210526316,
+      "grad_norm": 0.9907897710800171,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 99350
+    },
+    {
+      "epoch": 653.6842105263158,
+      "grad_norm": 0.9244227409362793,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 99360
+    },
+    {
+      "epoch": 653.75,
+      "grad_norm": 0.8877261877059937,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 99370
+    },
+    {
+      "epoch": 653.8157894736842,
+      "grad_norm": 0.8094327449798584,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 99380
+    },
+    {
+      "epoch": 653.8815789473684,
+      "grad_norm": 1.055003046989441,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 99390
+    },
+    {
+      "epoch": 653.9473684210526,
+      "grad_norm": 1.1381503343582153,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 99400
+    },
+    {
+      "epoch": 654.0131578947369,
+      "grad_norm": 0.7804838418960571,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 99410
+    },
+    {
+      "epoch": 654.078947368421,
+      "grad_norm": 1.3052879571914673,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 99420
+    },
+    {
+      "epoch": 654.1447368421053,
+      "grad_norm": 0.9438977241516113,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 99430
+    },
+    {
+      "epoch": 654.2105263157895,
+      "grad_norm": 1.324198603630066,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 99440
+    },
+    {
+      "epoch": 654.2763157894736,
+      "grad_norm": 1.0651750564575195,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 99450
+    },
+    {
+      "epoch": 654.3421052631579,
+      "grad_norm": 1.2382241487503052,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 99460
+    },
+    {
+      "epoch": 654.4078947368421,
+      "grad_norm": 1.0247857570648193,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 99470
+    },
+    {
+      "epoch": 654.4736842105264,
+      "grad_norm": 0.8664435744285583,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 99480
+    },
+    {
+      "epoch": 654.5394736842105,
+      "grad_norm": 1.1170216798782349,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 99490
+    },
+    {
+      "epoch": 654.6052631578947,
+      "grad_norm": 0.9370788931846619,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 99500
+    },
+    {
+      "epoch": 654.671052631579,
+      "grad_norm": 1.2386893033981323,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 99510
+    },
+    {
+      "epoch": 654.7368421052631,
+      "grad_norm": 1.4200820922851562,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 99520
+    },
+    {
+      "epoch": 654.8026315789474,
+      "grad_norm": 1.3450512886047363,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 99530
+    },
+    {
+      "epoch": 654.8684210526316,
+      "grad_norm": 1.1321581602096558,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 99540
+    },
+    {
+      "epoch": 654.9342105263158,
+      "grad_norm": 1.3326202630996704,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 99550
+    },
+    {
+      "epoch": 655.0,
+      "grad_norm": 0.9542607665061951,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 99560
+    },
+    {
+      "epoch": 655.0657894736842,
+      "grad_norm": 1.4707971811294556,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 99570
+    },
+    {
+      "epoch": 655.1315789473684,
+      "grad_norm": 1.3422001600265503,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 99580
+    },
+    {
+      "epoch": 655.1973684210526,
+      "grad_norm": 1.2660077810287476,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 99590
+    },
+    {
+      "epoch": 655.2631578947369,
+      "grad_norm": 1.3600062131881714,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 99600
+    },
+    {
+      "epoch": 655.328947368421,
+      "grad_norm": 1.2667269706726074,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 99610
+    },
+    {
+      "epoch": 655.3947368421053,
+      "grad_norm": 1.327304482460022,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 99620
+    },
+    {
+      "epoch": 655.4605263157895,
+      "grad_norm": 1.309645414352417,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 99630
+    },
+    {
+      "epoch": 655.5263157894736,
+      "grad_norm": 0.8108488917350769,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 99640
+    },
+    {
+      "epoch": 655.5921052631579,
+      "grad_norm": 0.985109806060791,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 99650
+    },
+    {
+      "epoch": 655.6578947368421,
+      "grad_norm": 0.8888941407203674,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 99660
+    },
+    {
+      "epoch": 655.7236842105264,
+      "grad_norm": 1.006456971168518,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 99670
+    },
+    {
+      "epoch": 655.7894736842105,
+      "grad_norm": 1.1159669160842896,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 99680
+    },
+    {
+      "epoch": 655.8552631578947,
+      "grad_norm": 0.9909751415252686,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 99690
+    },
+    {
+      "epoch": 655.921052631579,
+      "grad_norm": 1.045996904373169,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 99700
+    },
+    {
+      "epoch": 655.9868421052631,
+      "grad_norm": 1.0817979574203491,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 99710
+    },
+    {
+      "epoch": 656.0526315789474,
+      "grad_norm": 1.2081537246704102,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 99720
+    },
+    {
+      "epoch": 656.1184210526316,
+      "grad_norm": 0.9195104241371155,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 99730
+    },
+    {
+      "epoch": 656.1842105263158,
+      "grad_norm": 1.1696057319641113,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 99740
+    },
+    {
+      "epoch": 656.25,
+      "grad_norm": 1.0604538917541504,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 99750
+    },
+    {
+      "epoch": 656.3157894736842,
+      "grad_norm": 1.1938915252685547,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 99760
+    },
+    {
+      "epoch": 656.3815789473684,
+      "grad_norm": 1.0641862154006958,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 99770
+    },
+    {
+      "epoch": 656.4473684210526,
+      "grad_norm": 0.7114510536193848,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 99780
+    },
+    {
+      "epoch": 656.5131578947369,
+      "grad_norm": 1.3345415592193604,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 99790
+    },
+    {
+      "epoch": 656.578947368421,
+      "grad_norm": 1.2074189186096191,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 99800
+    },
+    {
+      "epoch": 656.6447368421053,
+      "grad_norm": 1.2852963209152222,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 99810
+    },
+    {
+      "epoch": 656.7105263157895,
+      "grad_norm": 0.9789184331893921,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 99820
+    },
+    {
+      "epoch": 656.7763157894736,
+      "grad_norm": 1.1512428522109985,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 99830
+    },
+    {
+      "epoch": 656.8421052631579,
+      "grad_norm": 0.9603427052497864,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 99840
+    },
+    {
+      "epoch": 656.9078947368421,
+      "grad_norm": 0.7649758458137512,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 99850
+    },
+    {
+      "epoch": 656.9736842105264,
+      "grad_norm": 1.2759963274002075,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 99860
+    },
+    {
+      "epoch": 657.0394736842105,
+      "grad_norm": 0.8978010416030884,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 99870
+    },
+    {
+      "epoch": 657.1052631578947,
+      "grad_norm": 1.0761810541152954,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 99880
+    },
+    {
+      "epoch": 657.171052631579,
+      "grad_norm": 1.0069093704223633,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 99890
+    },
+    {
+      "epoch": 657.2368421052631,
+      "grad_norm": 1.188727855682373,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 99900
+    },
+    {
+      "epoch": 657.3026315789474,
+      "grad_norm": 1.0814998149871826,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 99910
+    },
+    {
+      "epoch": 657.3684210526316,
+      "grad_norm": 1.085719347000122,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 99920
+    },
+    {
+      "epoch": 657.4342105263158,
+      "grad_norm": 1.0952857732772827,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 99930
+    },
+    {
+      "epoch": 657.5,
+      "grad_norm": 0.7331225275993347,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 99940
+    },
+    {
+      "epoch": 657.5657894736842,
+      "grad_norm": 1.202723503112793,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 99950
+    },
+    {
+      "epoch": 657.6315789473684,
+      "grad_norm": 1.0389039516448975,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 99960
+    },
+    {
+      "epoch": 657.6973684210526,
+      "grad_norm": 0.9595016241073608,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 99970
+    },
+    {
+      "epoch": 657.7631578947369,
+      "grad_norm": 0.8788406848907471,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 99980
+    },
+    {
+      "epoch": 657.828947368421,
+      "grad_norm": 1.0672800540924072,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 99990
+    },
+    {
+      "epoch": 657.8947368421053,
+      "grad_norm": 1.2291871309280396,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 100000
+    },
+    {
+      "epoch": 657.9605263157895,
+      "grad_norm": 1.470028042793274,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 100010
+    },
+    {
+      "epoch": 658.0263157894736,
+      "grad_norm": 1.5565192699432373,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 100020
+    },
+    {
+      "epoch": 658.0921052631579,
+      "grad_norm": 1.5067261457443237,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 100030
+    },
+    {
+      "epoch": 658.1578947368421,
+      "grad_norm": 1.442273497581482,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 100040
+    },
+    {
+      "epoch": 658.2236842105264,
+      "grad_norm": 1.4687352180480957,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 100050
+    },
+    {
+      "epoch": 658.2894736842105,
+      "grad_norm": 1.2669801712036133,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 100060
+    },
+    {
+      "epoch": 658.3552631578947,
+      "grad_norm": 0.9612995386123657,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 100070
+    },
+    {
+      "epoch": 658.421052631579,
+      "grad_norm": 1.0884873867034912,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 100080
+    },
+    {
+      "epoch": 658.4868421052631,
+      "grad_norm": 1.34589421749115,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 100090
+    },
+    {
+      "epoch": 658.5526315789474,
+      "grad_norm": 1.1699200868606567,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 100100
+    },
+    {
+      "epoch": 658.6184210526316,
+      "grad_norm": 0.9280645847320557,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 100110
+    },
+    {
+      "epoch": 658.6842105263158,
+      "grad_norm": 1.307676911354065,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 100120
+    },
+    {
+      "epoch": 658.75,
+      "grad_norm": 1.0661935806274414,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 100130
+    },
+    {
+      "epoch": 658.8157894736842,
+      "grad_norm": 0.9746012687683105,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 100140
+    },
+    {
+      "epoch": 658.8815789473684,
+      "grad_norm": 0.9165105223655701,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 100150
+    },
+    {
+      "epoch": 658.9473684210526,
+      "grad_norm": 1.2205272912979126,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 100160
+    },
+    {
+      "epoch": 659.0131578947369,
+      "grad_norm": 0.8076446652412415,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 100170
+    },
+    {
+      "epoch": 659.078947368421,
+      "grad_norm": 0.9755529165267944,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 100180
+    },
+    {
+      "epoch": 659.1447368421053,
+      "grad_norm": 1.021377682685852,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 100190
+    },
+    {
+      "epoch": 659.2105263157895,
+      "grad_norm": 1.2089534997940063,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 100200
+    },
+    {
+      "epoch": 659.2763157894736,
+      "grad_norm": 1.0684491395950317,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 100210
+    },
+    {
+      "epoch": 659.3421052631579,
+      "grad_norm": 1.0578230619430542,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 100220
+    },
+    {
+      "epoch": 659.4078947368421,
+      "grad_norm": 1.1062239408493042,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 100230
+    },
+    {
+      "epoch": 659.4736842105264,
+      "grad_norm": 1.1410943269729614,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 100240
+    },
+    {
+      "epoch": 659.5394736842105,
+      "grad_norm": 1.0092296600341797,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 100250
+    },
+    {
+      "epoch": 659.6052631578947,
+      "grad_norm": 1.231850266456604,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 100260
+    },
+    {
+      "epoch": 659.671052631579,
+      "grad_norm": 1.1187002658843994,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 100270
+    },
+    {
+      "epoch": 659.7368421052631,
+      "grad_norm": 1.5265558958053589,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 100280
+    },
+    {
+      "epoch": 659.8026315789474,
+      "grad_norm": 1.3116233348846436,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 100290
+    },
+    {
+      "epoch": 659.8684210526316,
+      "grad_norm": 1.1131277084350586,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 100300
+    },
+    {
+      "epoch": 659.9342105263158,
+      "grad_norm": 0.9748517274856567,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 100310
+    },
+    {
+      "epoch": 660.0,
+      "grad_norm": 0.9195544719696045,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 100320
+    },
+    {
+      "epoch": 660.0657894736842,
+      "grad_norm": 1.3856765031814575,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 100330
+    },
+    {
+      "epoch": 660.1315789473684,
+      "grad_norm": 1.0203367471694946,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 100340
+    },
+    {
+      "epoch": 660.1973684210526,
+      "grad_norm": 1.344549536705017,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 100350
+    },
+    {
+      "epoch": 660.2631578947369,
+      "grad_norm": 1.2398855686187744,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 100360
+    },
+    {
+      "epoch": 660.328947368421,
+      "grad_norm": 1.2883145809173584,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 100370
+    },
+    {
+      "epoch": 660.3947368421053,
+      "grad_norm": 1.2523137331008911,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 100380
+    },
+    {
+      "epoch": 660.4605263157895,
+      "grad_norm": 0.7788773775100708,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 100390
+    },
+    {
+      "epoch": 660.5263157894736,
+      "grad_norm": 1.165600061416626,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 100400
+    },
+    {
+      "epoch": 660.5921052631579,
+      "grad_norm": 0.8959761261940002,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 100410
+    },
+    {
+      "epoch": 660.6578947368421,
+      "grad_norm": 0.9459083676338196,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 100420
+    },
+    {
+      "epoch": 660.7236842105264,
+      "grad_norm": 1.0725597143173218,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 100430
+    },
+    {
+      "epoch": 660.7894736842105,
+      "grad_norm": 1.0163500308990479,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 100440
+    },
+    {
+      "epoch": 660.8552631578947,
+      "grad_norm": 1.0939759016036987,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 100450
+    },
+    {
+      "epoch": 660.921052631579,
+      "grad_norm": 1.2891944646835327,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 100460
+    },
+    {
+      "epoch": 660.9868421052631,
+      "grad_norm": 0.9678558111190796,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 100470
+    },
+    {
+      "epoch": 661.0526315789474,
+      "grad_norm": 0.6295179724693298,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 100480
+    },
+    {
+      "epoch": 661.1184210526316,
+      "grad_norm": 1.092713475227356,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 100490
+    },
+    {
+      "epoch": 661.1842105263158,
+      "grad_norm": 0.7096255421638489,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 100500
+    },
+    {
+      "epoch": 661.25,
+      "grad_norm": 1.1376711130142212,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 100510
+    },
+    {
+      "epoch": 661.3157894736842,
+      "grad_norm": 1.0104148387908936,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 100520
+    },
+    {
+      "epoch": 661.3815789473684,
+      "grad_norm": 0.9783036112785339,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 100530
+    },
+    {
+      "epoch": 661.4473684210526,
+      "grad_norm": 1.1273771524429321,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 100540
+    },
+    {
+      "epoch": 661.5131578947369,
+      "grad_norm": 1.2987968921661377,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 100550
+    },
+    {
+      "epoch": 661.578947368421,
+      "grad_norm": 1.0228668451309204,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 100560
+    },
+    {
+      "epoch": 661.6447368421053,
+      "grad_norm": 0.901421844959259,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 100570
+    },
+    {
+      "epoch": 661.7105263157895,
+      "grad_norm": 1.1462628841400146,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 100580
+    },
+    {
+      "epoch": 661.7763157894736,
+      "grad_norm": 0.7442108392715454,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 100590
+    },
+    {
+      "epoch": 661.8421052631579,
+      "grad_norm": 1.2875144481658936,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 100600
+    },
+    {
+      "epoch": 661.9078947368421,
+      "grad_norm": 1.1649748086929321,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 100610
+    },
+    {
+      "epoch": 661.9736842105264,
+      "grad_norm": 1.0540515184402466,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 100620
+    },
+    {
+      "epoch": 662.0394736842105,
+      "grad_norm": 1.0686025619506836,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 100630
+    },
+    {
+      "epoch": 662.1052631578947,
+      "grad_norm": 0.9818456768989563,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 100640
+    },
+    {
+      "epoch": 662.171052631579,
+      "grad_norm": 0.9439074397087097,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 100650
+    },
+    {
+      "epoch": 662.2368421052631,
+      "grad_norm": 0.9941381216049194,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 100660
+    },
+    {
+      "epoch": 662.3026315789474,
+      "grad_norm": 1.1093741655349731,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 100670
+    },
+    {
+      "epoch": 662.3684210526316,
+      "grad_norm": 1.1444611549377441,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 100680
+    },
+    {
+      "epoch": 662.4342105263158,
+      "grad_norm": 1.0933109521865845,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 100690
+    },
+    {
+      "epoch": 662.5,
+      "grad_norm": 1.2394286394119263,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 100700
+    },
+    {
+      "epoch": 662.5657894736842,
+      "grad_norm": 1.3169022798538208,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 100710
+    },
+    {
+      "epoch": 662.6315789473684,
+      "grad_norm": 0.8559953570365906,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 100720
+    },
+    {
+      "epoch": 662.6973684210526,
+      "grad_norm": 1.2426526546478271,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 100730
+    },
+    {
+      "epoch": 662.7631578947369,
+      "grad_norm": 1.1118953227996826,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 100740
+    },
+    {
+      "epoch": 662.828947368421,
+      "grad_norm": 0.955522894859314,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 100750
+    },
+    {
+      "epoch": 662.8947368421053,
+      "grad_norm": 1.0989824533462524,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 100760
+    },
+    {
+      "epoch": 662.9605263157895,
+      "grad_norm": 1.0677485466003418,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 100770
+    },
+    {
+      "epoch": 663.0263157894736,
+      "grad_norm": 0.7644389271736145,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 100780
+    },
+    {
+      "epoch": 663.0921052631579,
+      "grad_norm": 1.139380931854248,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 100790
+    },
+    {
+      "epoch": 663.1578947368421,
+      "grad_norm": 0.9676651954650879,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 100800
+    },
+    {
+      "epoch": 663.2236842105264,
+      "grad_norm": 1.2177777290344238,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 100810
+    },
+    {
+      "epoch": 663.2894736842105,
+      "grad_norm": 1.339358925819397,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 100820
+    },
+    {
+      "epoch": 663.3552631578947,
+      "grad_norm": 1.3607467412948608,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 100830
+    },
+    {
+      "epoch": 663.421052631579,
+      "grad_norm": 1.0772532224655151,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 100840
+    },
+    {
+      "epoch": 663.4868421052631,
+      "grad_norm": 1.1806657314300537,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 100850
+    },
+    {
+      "epoch": 663.5526315789474,
+      "grad_norm": 0.8904920220375061,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 100860
+    },
+    {
+      "epoch": 663.6184210526316,
+      "grad_norm": 1.1465924978256226,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 100870
+    },
+    {
+      "epoch": 663.6842105263158,
+      "grad_norm": 1.0607256889343262,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 100880
+    },
+    {
+      "epoch": 663.75,
+      "grad_norm": 1.294001579284668,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 100890
+    },
+    {
+      "epoch": 663.8157894736842,
+      "grad_norm": 1.168131947517395,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 100900
+    },
+    {
+      "epoch": 663.8815789473684,
+      "grad_norm": 0.9382193684577942,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 100910
+    },
+    {
+      "epoch": 663.9473684210526,
+      "grad_norm": 1.172577142715454,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 100920
+    },
+    {
+      "epoch": 664.0131578947369,
+      "grad_norm": 1.1000980138778687,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 100930
+    },
+    {
+      "epoch": 664.078947368421,
+      "grad_norm": 1.3834381103515625,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 100940
+    },
+    {
+      "epoch": 664.1447368421053,
+      "grad_norm": 1.354061484336853,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 100950
+    },
+    {
+      "epoch": 664.2105263157895,
+      "grad_norm": 1.3133666515350342,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 100960
+    },
+    {
+      "epoch": 664.2763157894736,
+      "grad_norm": 0.8378098011016846,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 100970
+    },
+    {
+      "epoch": 664.3421052631579,
+      "grad_norm": 0.8867501616477966,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 100980
+    },
+    {
+      "epoch": 664.4078947368421,
+      "grad_norm": 1.0622562170028687,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 100990
+    },
+    {
+      "epoch": 664.4736842105264,
+      "grad_norm": 0.6690018773078918,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 101000
+    },
+    {
+      "epoch": 664.5394736842105,
+      "grad_norm": 0.7047531008720398,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 101010
+    },
+    {
+      "epoch": 664.6052631578947,
+      "grad_norm": 1.0882129669189453,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 101020
+    },
+    {
+      "epoch": 664.671052631579,
+      "grad_norm": 0.8649616241455078,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 101030
+    },
+    {
+      "epoch": 664.7368421052631,
+      "grad_norm": 0.8693536520004272,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 101040
+    },
+    {
+      "epoch": 664.8026315789474,
+      "grad_norm": 1.3576804399490356,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 101050
+    },
+    {
+      "epoch": 664.8684210526316,
+      "grad_norm": 1.0976057052612305,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 101060
+    },
+    {
+      "epoch": 664.9342105263158,
+      "grad_norm": 1.4118653535842896,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 101070
+    },
+    {
+      "epoch": 665.0,
+      "grad_norm": 0.913599967956543,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 101080
+    },
+    {
+      "epoch": 665.0657894736842,
+      "grad_norm": 1.0430513620376587,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 101090
+    },
+    {
+      "epoch": 665.1315789473684,
+      "grad_norm": 1.3427150249481201,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 101100
+    },
+    {
+      "epoch": 665.1973684210526,
+      "grad_norm": 1.5117545127868652,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 101110
+    },
+    {
+      "epoch": 665.2631578947369,
+      "grad_norm": 0.9170265197753906,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 101120
+    },
+    {
+      "epoch": 665.328947368421,
+      "grad_norm": 1.115020513534546,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 101130
+    },
+    {
+      "epoch": 665.3947368421053,
+      "grad_norm": 0.7325767874717712,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 101140
+    },
+    {
+      "epoch": 665.4605263157895,
+      "grad_norm": 1.1029810905456543,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 101150
+    },
+    {
+      "epoch": 665.5263157894736,
+      "grad_norm": 1.1868432760238647,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 101160
+    },
+    {
+      "epoch": 665.5921052631579,
+      "grad_norm": 0.9201452136039734,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 101170
+    },
+    {
+      "epoch": 665.6578947368421,
+      "grad_norm": 1.1828328371047974,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 101180
+    },
+    {
+      "epoch": 665.7236842105264,
+      "grad_norm": 1.1864783763885498,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 101190
+    },
+    {
+      "epoch": 665.7894736842105,
+      "grad_norm": 1.2099254131317139,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 101200
+    },
+    {
+      "epoch": 665.8552631578947,
+      "grad_norm": 1.1975241899490356,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 101210
+    },
+    {
+      "epoch": 665.921052631579,
+      "grad_norm": 1.1442780494689941,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 101220
+    },
+    {
+      "epoch": 665.9868421052631,
+      "grad_norm": 1.2307586669921875,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 101230
+    },
+    {
+      "epoch": 666.0526315789474,
+      "grad_norm": 0.816516101360321,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 101240
+    },
+    {
+      "epoch": 666.1184210526316,
+      "grad_norm": 1.2257626056671143,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 101250
+    },
+    {
+      "epoch": 666.1842105263158,
+      "grad_norm": 1.1634641885757446,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 101260
+    },
+    {
+      "epoch": 666.25,
+      "grad_norm": 1.2208856344223022,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 101270
+    },
+    {
+      "epoch": 666.3157894736842,
+      "grad_norm": 0.8225875496864319,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 101280
+    },
+    {
+      "epoch": 666.3815789473684,
+      "grad_norm": 1.0692633390426636,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 101290
+    },
+    {
+      "epoch": 666.4473684210526,
+      "grad_norm": 1.2944279909133911,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 101300
+    },
+    {
+      "epoch": 666.5131578947369,
+      "grad_norm": 1.2648913860321045,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 101310
+    },
+    {
+      "epoch": 666.578947368421,
+      "grad_norm": 1.435753345489502,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 101320
+    },
+    {
+      "epoch": 666.6447368421053,
+      "grad_norm": 1.2368513345718384,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 101330
+    },
+    {
+      "epoch": 666.7105263157895,
+      "grad_norm": 1.1920435428619385,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 101340
+    },
+    {
+      "epoch": 666.7763157894736,
+      "grad_norm": 1.2127162218093872,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 101350
+    },
+    {
+      "epoch": 666.8421052631579,
+      "grad_norm": 1.2593845129013062,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 101360
+    },
+    {
+      "epoch": 666.9078947368421,
+      "grad_norm": 1.0275031328201294,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 101370
+    },
+    {
+      "epoch": 666.9736842105264,
+      "grad_norm": 1.224514126777649,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 101380
+    },
+    {
+      "epoch": 667.0394736842105,
+      "grad_norm": 1.1807442903518677,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 101390
+    },
+    {
+      "epoch": 667.1052631578947,
+      "grad_norm": 0.8523136377334595,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 101400
+    },
+    {
+      "epoch": 667.171052631579,
+      "grad_norm": 1.0002000331878662,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 101410
+    },
+    {
+      "epoch": 667.2368421052631,
+      "grad_norm": 1.0773143768310547,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 101420
+    },
+    {
+      "epoch": 667.3026315789474,
+      "grad_norm": 1.2809265851974487,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 101430
+    },
+    {
+      "epoch": 667.3684210526316,
+      "grad_norm": 1.272909164428711,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 101440
+    },
+    {
+      "epoch": 667.4342105263158,
+      "grad_norm": 1.0183367729187012,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 101450
+    },
+    {
+      "epoch": 667.5,
+      "grad_norm": 1.0988458395004272,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 101460
+    },
+    {
+      "epoch": 667.5657894736842,
+      "grad_norm": 1.024822473526001,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 101470
+    },
+    {
+      "epoch": 667.6315789473684,
+      "grad_norm": 1.2332916259765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 101480
+    },
+    {
+      "epoch": 667.6973684210526,
+      "grad_norm": 1.0866150856018066,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 101490
+    },
+    {
+      "epoch": 667.7631578947369,
+      "grad_norm": 1.0552276372909546,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 101500
+    },
+    {
+      "epoch": 667.828947368421,
+      "grad_norm": 1.0964107513427734,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 101510
+    },
+    {
+      "epoch": 667.8947368421053,
+      "grad_norm": 1.264575719833374,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 101520
+    },
+    {
+      "epoch": 667.9605263157895,
+      "grad_norm": 1.4266283512115479,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 101530
+    },
+    {
+      "epoch": 668.0263157894736,
+      "grad_norm": 0.7991641759872437,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 101540
+    },
+    {
+      "epoch": 668.0921052631579,
+      "grad_norm": 1.0575485229492188,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 101550
+    },
+    {
+      "epoch": 668.1578947368421,
+      "grad_norm": 1.040031909942627,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 101560
+    },
+    {
+      "epoch": 668.2236842105264,
+      "grad_norm": 0.8988039493560791,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 101570
+    },
+    {
+      "epoch": 668.2894736842105,
+      "grad_norm": 1.1382830142974854,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 101580
+    },
+    {
+      "epoch": 668.3552631578947,
+      "grad_norm": 1.2425233125686646,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 101590
+    },
+    {
+      "epoch": 668.421052631579,
+      "grad_norm": 1.2461456060409546,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 101600
+    },
+    {
+      "epoch": 668.4868421052631,
+      "grad_norm": 0.828624427318573,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 101610
+    },
+    {
+      "epoch": 668.5526315789474,
+      "grad_norm": 0.6338958144187927,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 101620
+    },
+    {
+      "epoch": 668.6184210526316,
+      "grad_norm": 0.9959868788719177,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 101630
+    },
+    {
+      "epoch": 668.6842105263158,
+      "grad_norm": 0.9457986354827881,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 101640
+    },
+    {
+      "epoch": 668.75,
+      "grad_norm": 1.2366830110549927,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 101650
+    },
+    {
+      "epoch": 668.8157894736842,
+      "grad_norm": 1.192163348197937,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 101660
+    },
+    {
+      "epoch": 668.8815789473684,
+      "grad_norm": 0.9253160953521729,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 101670
+    },
+    {
+      "epoch": 668.9473684210526,
+      "grad_norm": 0.8594035506248474,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 101680
+    },
+    {
+      "epoch": 669.0131578947369,
+      "grad_norm": 1.1185978651046753,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 101690
+    },
+    {
+      "epoch": 669.078947368421,
+      "grad_norm": 0.9524565935134888,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 101700
+    },
+    {
+      "epoch": 669.1447368421053,
+      "grad_norm": 1.0104931592941284,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 101710
+    },
+    {
+      "epoch": 669.2105263157895,
+      "grad_norm": 1.1492105722427368,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 101720
+    },
+    {
+      "epoch": 669.2763157894736,
+      "grad_norm": 1.1701027154922485,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 101730
+    },
+    {
+      "epoch": 669.3421052631579,
+      "grad_norm": 1.386569619178772,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 101740
+    },
+    {
+      "epoch": 669.4078947368421,
+      "grad_norm": 1.1431201696395874,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 101750
+    },
+    {
+      "epoch": 669.4736842105264,
+      "grad_norm": 1.031060814857483,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 101760
+    },
+    {
+      "epoch": 669.5394736842105,
+      "grad_norm": 0.8447512984275818,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 101770
+    },
+    {
+      "epoch": 669.6052631578947,
+      "grad_norm": 1.1162457466125488,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 101780
+    },
+    {
+      "epoch": 669.671052631579,
+      "grad_norm": 1.321597933769226,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 101790
+    },
+    {
+      "epoch": 669.7368421052631,
+      "grad_norm": 1.5164763927459717,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 101800
+    },
+    {
+      "epoch": 669.8026315789474,
+      "grad_norm": 1.3314334154129028,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 101810
+    },
+    {
+      "epoch": 669.8684210526316,
+      "grad_norm": 1.0497831106185913,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 101820
+    },
+    {
+      "epoch": 669.9342105263158,
+      "grad_norm": 1.1639678478240967,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 101830
+    },
+    {
+      "epoch": 670.0,
+      "grad_norm": 1.2581437826156616,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 101840
+    },
+    {
+      "epoch": 670.0657894736842,
+      "grad_norm": 1.3124868869781494,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 101850
+    },
+    {
+      "epoch": 670.1315789473684,
+      "grad_norm": 1.232029676437378,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 101860
+    },
+    {
+      "epoch": 670.1973684210526,
+      "grad_norm": 1.5936830043792725,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 101870
+    },
+    {
+      "epoch": 670.2631578947369,
+      "grad_norm": 1.1136298179626465,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 101880
+    },
+    {
+      "epoch": 670.328947368421,
+      "grad_norm": 1.2041126489639282,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 101890
+    },
+    {
+      "epoch": 670.3947368421053,
+      "grad_norm": 1.0076220035552979,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 101900
+    },
+    {
+      "epoch": 670.4605263157895,
+      "grad_norm": 1.001729965209961,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 101910
+    },
+    {
+      "epoch": 670.5263157894736,
+      "grad_norm": 1.1370996236801147,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 101920
+    },
+    {
+      "epoch": 670.5921052631579,
+      "grad_norm": 1.3312040567398071,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 101930
+    },
+    {
+      "epoch": 670.6578947368421,
+      "grad_norm": 1.0733331441879272,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 101940
+    },
+    {
+      "epoch": 670.7236842105264,
+      "grad_norm": 1.2168687582015991,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 101950
+    },
+    {
+      "epoch": 670.7894736842105,
+      "grad_norm": 1.0879305601119995,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 101960
+    },
+    {
+      "epoch": 670.8552631578947,
+      "grad_norm": 1.084376573562622,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 101970
+    },
+    {
+      "epoch": 670.921052631579,
+      "grad_norm": 1.1504342555999756,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 101980
+    },
+    {
+      "epoch": 670.9868421052631,
+      "grad_norm": 1.2068742513656616,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 101990
+    },
+    {
+      "epoch": 671.0526315789474,
+      "grad_norm": 1.264833688735962,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 102000
+    },
+    {
+      "epoch": 671.1184210526316,
+      "grad_norm": 0.7933449149131775,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 102010
+    },
+    {
+      "epoch": 671.1842105263158,
+      "grad_norm": 0.9847843647003174,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 102020
+    },
+    {
+      "epoch": 671.25,
+      "grad_norm": 1.2704505920410156,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 102030
+    },
+    {
+      "epoch": 671.3157894736842,
+      "grad_norm": 1.4738961458206177,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 102040
+    },
+    {
+      "epoch": 671.3815789473684,
+      "grad_norm": 1.338627576828003,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 102050
+    },
+    {
+      "epoch": 671.4473684210526,
+      "grad_norm": 1.2926578521728516,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 102060
+    },
+    {
+      "epoch": 671.5131578947369,
+      "grad_norm": 1.442939043045044,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 102070
+    },
+    {
+      "epoch": 671.578947368421,
+      "grad_norm": 1.1999456882476807,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 102080
+    },
+    {
+      "epoch": 671.6447368421053,
+      "grad_norm": 1.2807422876358032,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 102090
+    },
+    {
+      "epoch": 671.7105263157895,
+      "grad_norm": 1.0529870986938477,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 102100
+    },
+    {
+      "epoch": 671.7763157894736,
+      "grad_norm": 1.044051170349121,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 102110
+    },
+    {
+      "epoch": 671.8421052631579,
+      "grad_norm": 0.934209406375885,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 102120
+    },
+    {
+      "epoch": 671.9078947368421,
+      "grad_norm": 1.350449562072754,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 102130
+    },
+    {
+      "epoch": 671.9736842105264,
+      "grad_norm": 1.2345993518829346,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 102140
+    },
+    {
+      "epoch": 672.0394736842105,
+      "grad_norm": 1.41543447971344,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 102150
+    },
+    {
+      "epoch": 672.1052631578947,
+      "grad_norm": 0.7719045877456665,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 102160
+    },
+    {
+      "epoch": 672.171052631579,
+      "grad_norm": 1.0525262355804443,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 102170
+    },
+    {
+      "epoch": 672.2368421052631,
+      "grad_norm": 1.1869667768478394,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 102180
+    },
+    {
+      "epoch": 672.3026315789474,
+      "grad_norm": 1.0475441217422485,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 102190
+    },
+    {
+      "epoch": 672.3684210526316,
+      "grad_norm": 1.173707127571106,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 102200
+    },
+    {
+      "epoch": 672.4342105263158,
+      "grad_norm": 1.1738572120666504,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 102210
+    },
+    {
+      "epoch": 672.5,
+      "grad_norm": 1.0514370203018188,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 102220
+    },
+    {
+      "epoch": 672.5657894736842,
+      "grad_norm": 1.058618426322937,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 102230
+    },
+    {
+      "epoch": 672.6315789473684,
+      "grad_norm": 1.6183222532272339,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 102240
+    },
+    {
+      "epoch": 672.6973684210526,
+      "grad_norm": 1.2186968326568604,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 102250
+    },
+    {
+      "epoch": 672.7631578947369,
+      "grad_norm": 0.9119374752044678,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 102260
+    },
+    {
+      "epoch": 672.828947368421,
+      "grad_norm": 1.2468377351760864,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 102270
+    },
+    {
+      "epoch": 672.8947368421053,
+      "grad_norm": 0.8254573941230774,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 102280
+    },
+    {
+      "epoch": 672.9605263157895,
+      "grad_norm": 0.8752316832542419,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 102290
+    },
+    {
+      "epoch": 673.0263157894736,
+      "grad_norm": 0.8799929618835449,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 102300
+    },
+    {
+      "epoch": 673.0921052631579,
+      "grad_norm": 1.0996203422546387,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 102310
+    },
+    {
+      "epoch": 673.1578947368421,
+      "grad_norm": 1.4787687063217163,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 102320
+    },
+    {
+      "epoch": 673.2236842105264,
+      "grad_norm": 0.9628654718399048,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 102330
+    },
+    {
+      "epoch": 673.2894736842105,
+      "grad_norm": 1.1669740676879883,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 102340
+    },
+    {
+      "epoch": 673.3552631578947,
+      "grad_norm": 0.9935758709907532,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 102350
+    },
+    {
+      "epoch": 673.421052631579,
+      "grad_norm": 1.2170686721801758,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 102360
+    },
+    {
+      "epoch": 673.4868421052631,
+      "grad_norm": 1.3537840843200684,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 102370
+    },
+    {
+      "epoch": 673.5526315789474,
+      "grad_norm": 1.2032088041305542,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 102380
+    },
+    {
+      "epoch": 673.6184210526316,
+      "grad_norm": 1.0807600021362305,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 102390
+    },
+    {
+      "epoch": 673.6842105263158,
+      "grad_norm": 1.3539401292800903,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 102400
+    },
+    {
+      "epoch": 673.75,
+      "grad_norm": 0.9600739479064941,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 102410
+    },
+    {
+      "epoch": 673.8157894736842,
+      "grad_norm": 0.8437861204147339,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 102420
+    },
+    {
+      "epoch": 673.8815789473684,
+      "grad_norm": 1.0616252422332764,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 102430
+    },
+    {
+      "epoch": 673.9473684210526,
+      "grad_norm": 1.1971665620803833,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 102440
+    },
+    {
+      "epoch": 674.0131578947369,
+      "grad_norm": 1.0292773246765137,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 102450
+    },
+    {
+      "epoch": 674.078947368421,
+      "grad_norm": 1.0635182857513428,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 102460
+    },
+    {
+      "epoch": 674.1447368421053,
+      "grad_norm": 1.1411423683166504,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 102470
+    },
+    {
+      "epoch": 674.2105263157895,
+      "grad_norm": 1.0148522853851318,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 102480
+    },
+    {
+      "epoch": 674.2763157894736,
+      "grad_norm": 1.2934777736663818,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 102490
+    },
+    {
+      "epoch": 674.3421052631579,
+      "grad_norm": 1.151720404624939,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 102500
+    },
+    {
+      "epoch": 674.4078947368421,
+      "grad_norm": 0.8841307759284973,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 102510
+    },
+    {
+      "epoch": 674.4736842105264,
+      "grad_norm": 0.8646332621574402,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 102520
+    },
+    {
+      "epoch": 674.5394736842105,
+      "grad_norm": 0.8606105446815491,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 102530
+    },
+    {
+      "epoch": 674.6052631578947,
+      "grad_norm": 0.7609214186668396,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 102540
+    },
+    {
+      "epoch": 674.671052631579,
+      "grad_norm": 1.0096818208694458,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 102550
+    },
+    {
+      "epoch": 674.7368421052631,
+      "grad_norm": 1.0138216018676758,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 102560
+    },
+    {
+      "epoch": 674.8026315789474,
+      "grad_norm": 1.2651227712631226,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 102570
+    },
+    {
+      "epoch": 674.8684210526316,
+      "grad_norm": 1.3024859428405762,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 102580
+    },
+    {
+      "epoch": 674.9342105263158,
+      "grad_norm": 0.9451261758804321,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 102590
+    },
+    {
+      "epoch": 675.0,
+      "grad_norm": 1.1993613243103027,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 102600
+    },
+    {
+      "epoch": 675.0657894736842,
+      "grad_norm": 1.2021538019180298,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 102610
+    },
+    {
+      "epoch": 675.1315789473684,
+      "grad_norm": 0.7005917429924011,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 102620
+    },
+    {
+      "epoch": 675.1973684210526,
+      "grad_norm": 0.9240332245826721,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 102630
+    },
+    {
+      "epoch": 675.2631578947369,
+      "grad_norm": 1.2300150394439697,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 102640
+    },
+    {
+      "epoch": 675.328947368421,
+      "grad_norm": 1.2758888006210327,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 102650
+    },
+    {
+      "epoch": 675.3947368421053,
+      "grad_norm": 1.0356707572937012,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 102660
+    },
+    {
+      "epoch": 675.4605263157895,
+      "grad_norm": 0.9405995011329651,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 102670
+    },
+    {
+      "epoch": 675.5263157894736,
+      "grad_norm": 1.1242586374282837,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 102680
+    },
+    {
+      "epoch": 675.5921052631579,
+      "grad_norm": 1.0802112817764282,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 102690
+    },
+    {
+      "epoch": 675.6578947368421,
+      "grad_norm": 0.9207102656364441,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 102700
+    },
+    {
+      "epoch": 675.7236842105264,
+      "grad_norm": 0.9859259128570557,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 102710
+    },
+    {
+      "epoch": 675.7894736842105,
+      "grad_norm": 1.63200044631958,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 102720
+    },
+    {
+      "epoch": 675.8552631578947,
+      "grad_norm": 1.0292948484420776,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 102730
+    },
+    {
+      "epoch": 675.921052631579,
+      "grad_norm": 0.9589559435844421,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 102740
+    },
+    {
+      "epoch": 675.9868421052631,
+      "grad_norm": 1.2861278057098389,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 102750
+    },
+    {
+      "epoch": 676.0526315789474,
+      "grad_norm": 1.23167884349823,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 102760
+    },
+    {
+      "epoch": 676.1184210526316,
+      "grad_norm": 1.2074878215789795,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 102770
+    },
+    {
+      "epoch": 676.1842105263158,
+      "grad_norm": 1.1100069284439087,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 102780
+    },
+    {
+      "epoch": 676.25,
+      "grad_norm": 1.2543814182281494,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 102790
+    },
+    {
+      "epoch": 676.3157894736842,
+      "grad_norm": 1.0590333938598633,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 102800
+    },
+    {
+      "epoch": 676.3815789473684,
+      "grad_norm": 1.1635903120040894,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 102810
+    },
+    {
+      "epoch": 676.4473684210526,
+      "grad_norm": 1.191482663154602,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 102820
+    },
+    {
+      "epoch": 676.5131578947369,
+      "grad_norm": 1.2780073881149292,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 102830
+    },
+    {
+      "epoch": 676.578947368421,
+      "grad_norm": 1.0318385362625122,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 102840
+    },
+    {
+      "epoch": 676.6447368421053,
+      "grad_norm": 1.411617636680603,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 102850
+    },
+    {
+      "epoch": 676.7105263157895,
+      "grad_norm": 1.007462739944458,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 102860
+    },
+    {
+      "epoch": 676.7763157894736,
+      "grad_norm": 1.653554916381836,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 102870
+    },
+    {
+      "epoch": 676.8421052631579,
+      "grad_norm": 1.147131085395813,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 102880
+    },
+    {
+      "epoch": 676.9078947368421,
+      "grad_norm": 0.8459810018539429,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 102890
+    },
+    {
+      "epoch": 676.9736842105264,
+      "grad_norm": 1.1392914056777954,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 102900
+    },
+    {
+      "epoch": 677.0394736842105,
+      "grad_norm": 1.1632038354873657,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 102910
+    },
+    {
+      "epoch": 677.1052631578947,
+      "grad_norm": 1.4160902500152588,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 102920
+    },
+    {
+      "epoch": 677.171052631579,
+      "grad_norm": 1.1566230058670044,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 102930
+    },
+    {
+      "epoch": 677.2368421052631,
+      "grad_norm": 0.9785585403442383,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 102940
+    },
+    {
+      "epoch": 677.3026315789474,
+      "grad_norm": 0.8845245242118835,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 102950
+    },
+    {
+      "epoch": 677.3684210526316,
+      "grad_norm": 0.8199566006660461,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 102960
+    },
+    {
+      "epoch": 677.4342105263158,
+      "grad_norm": 0.8677225708961487,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 102970
+    },
+    {
+      "epoch": 677.5,
+      "grad_norm": 1.06745445728302,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 102980
+    },
+    {
+      "epoch": 677.5657894736842,
+      "grad_norm": 1.011222004890442,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 102990
+    },
+    {
+      "epoch": 677.6315789473684,
+      "grad_norm": 1.0868799686431885,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 103000
+    },
+    {
+      "epoch": 677.6973684210526,
+      "grad_norm": 0.8394410610198975,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 103010
+    },
+    {
+      "epoch": 677.7631578947369,
+      "grad_norm": 1.1427472829818726,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 103020
+    },
+    {
+      "epoch": 677.828947368421,
+      "grad_norm": 1.2967380285263062,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 103030
+    },
+    {
+      "epoch": 677.8947368421053,
+      "grad_norm": 1.249349594116211,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 103040
+    },
+    {
+      "epoch": 677.9605263157895,
+      "grad_norm": 1.3048033714294434,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 103050
+    },
+    {
+      "epoch": 678.0263157894736,
+      "grad_norm": 1.421112298965454,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 103060
+    },
+    {
+      "epoch": 678.0921052631579,
+      "grad_norm": 1.029354214668274,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 103070
+    },
+    {
+      "epoch": 678.1578947368421,
+      "grad_norm": 0.6975602507591248,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 103080
+    },
+    {
+      "epoch": 678.2236842105264,
+      "grad_norm": 1.1515804529190063,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 103090
+    },
+    {
+      "epoch": 678.2894736842105,
+      "grad_norm": 1.4265718460083008,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 103100
+    },
+    {
+      "epoch": 678.3552631578947,
+      "grad_norm": 1.0537846088409424,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 103110
+    },
+    {
+      "epoch": 678.421052631579,
+      "grad_norm": 1.3082852363586426,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 103120
+    },
+    {
+      "epoch": 678.4868421052631,
+      "grad_norm": 1.1465588808059692,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 103130
+    },
+    {
+      "epoch": 678.5526315789474,
+      "grad_norm": 1.254515290260315,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 103140
+    },
+    {
+      "epoch": 678.6184210526316,
+      "grad_norm": 0.8402789831161499,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 103150
+    },
+    {
+      "epoch": 678.6842105263158,
+      "grad_norm": 1.0215173959732056,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 103160
+    },
+    {
+      "epoch": 678.75,
+      "grad_norm": 1.0827964544296265,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 103170
+    },
+    {
+      "epoch": 678.8157894736842,
+      "grad_norm": 1.0201925039291382,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 103180
+    },
+    {
+      "epoch": 678.8815789473684,
+      "grad_norm": 0.9389650821685791,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 103190
+    },
+    {
+      "epoch": 678.9473684210526,
+      "grad_norm": 0.8603837490081787,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 103200
+    },
+    {
+      "epoch": 679.0131578947369,
+      "grad_norm": 1.0995327234268188,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 103210
+    },
+    {
+      "epoch": 679.078947368421,
+      "grad_norm": 1.0585341453552246,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 103220
+    },
+    {
+      "epoch": 679.1447368421053,
+      "grad_norm": 1.1244828701019287,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 103230
+    },
+    {
+      "epoch": 679.2105263157895,
+      "grad_norm": 1.1929583549499512,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 103240
+    },
+    {
+      "epoch": 679.2763157894736,
+      "grad_norm": 1.2523797750473022,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 103250
+    },
+    {
+      "epoch": 679.3421052631579,
+      "grad_norm": 1.0820051431655884,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 103260
+    },
+    {
+      "epoch": 679.4078947368421,
+      "grad_norm": 1.3185031414031982,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 103270
+    },
+    {
+      "epoch": 679.4736842105264,
+      "grad_norm": 1.0388109683990479,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 103280
+    },
+    {
+      "epoch": 679.5394736842105,
+      "grad_norm": 1.443453073501587,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 103290
+    },
+    {
+      "epoch": 679.6052631578947,
+      "grad_norm": 0.7869113087654114,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 103300
+    },
+    {
+      "epoch": 679.671052631579,
+      "grad_norm": 1.2342416048049927,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 103310
+    },
+    {
+      "epoch": 679.7368421052631,
+      "grad_norm": 1.1199147701263428,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 103320
+    },
+    {
+      "epoch": 679.8026315789474,
+      "grad_norm": 0.9589998126029968,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 103330
+    },
+    {
+      "epoch": 679.8684210526316,
+      "grad_norm": 1.0581178665161133,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 103340
+    },
+    {
+      "epoch": 679.9342105263158,
+      "grad_norm": 0.9367907643318176,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 103350
+    },
+    {
+      "epoch": 680.0,
+      "grad_norm": 0.9200935363769531,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 103360
+    },
+    {
+      "epoch": 680.0657894736842,
+      "grad_norm": 1.1545464992523193,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 103370
+    },
+    {
+      "epoch": 680.1315789473684,
+      "grad_norm": 0.9586318731307983,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 103380
+    },
+    {
+      "epoch": 680.1973684210526,
+      "grad_norm": 1.0395398139953613,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 103390
+    },
+    {
+      "epoch": 680.2631578947369,
+      "grad_norm": 1.193278431892395,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 103400
+    },
+    {
+      "epoch": 680.328947368421,
+      "grad_norm": 0.9284635186195374,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 103410
+    },
+    {
+      "epoch": 680.3947368421053,
+      "grad_norm": 1.0416210889816284,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 103420
+    },
+    {
+      "epoch": 680.4605263157895,
+      "grad_norm": 0.880230188369751,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 103430
+    },
+    {
+      "epoch": 680.5263157894736,
+      "grad_norm": 1.139072299003601,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 103440
+    },
+    {
+      "epoch": 680.5921052631579,
+      "grad_norm": 0.859987735748291,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 103450
+    },
+    {
+      "epoch": 680.6578947368421,
+      "grad_norm": 1.0812146663665771,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 103460
+    },
+    {
+      "epoch": 680.7236842105264,
+      "grad_norm": 0.973387598991394,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 103470
+    },
+    {
+      "epoch": 680.7894736842105,
+      "grad_norm": 1.28573739528656,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 103480
+    },
+    {
+      "epoch": 680.8552631578947,
+      "grad_norm": 1.2095247507095337,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 103490
+    },
+    {
+      "epoch": 680.921052631579,
+      "grad_norm": 1.1945544481277466,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 103500
+    },
+    {
+      "epoch": 680.9868421052631,
+      "grad_norm": 1.0555740594863892,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 103510
+    },
+    {
+      "epoch": 681.0526315789474,
+      "grad_norm": 1.0830953121185303,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 103520
+    },
+    {
+      "epoch": 681.1184210526316,
+      "grad_norm": 1.1533199548721313,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 103530
+    },
+    {
+      "epoch": 681.1842105263158,
+      "grad_norm": 1.2779477834701538,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 103540
+    },
+    {
+      "epoch": 681.25,
+      "grad_norm": 1.3541412353515625,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 103550
+    },
+    {
+      "epoch": 681.3157894736842,
+      "grad_norm": 1.2908533811569214,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 103560
+    },
+    {
+      "epoch": 681.3815789473684,
+      "grad_norm": 1.2065902948379517,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 103570
+    },
+    {
+      "epoch": 681.4473684210526,
+      "grad_norm": 0.755403995513916,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 103580
+    },
+    {
+      "epoch": 681.5131578947369,
+      "grad_norm": 1.087852954864502,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 103590
+    },
+    {
+      "epoch": 681.578947368421,
+      "grad_norm": 0.9179571270942688,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 103600
+    },
+    {
+      "epoch": 681.6447368421053,
+      "grad_norm": 1.2405117750167847,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 103610
+    },
+    {
+      "epoch": 681.7105263157895,
+      "grad_norm": 0.9832314252853394,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 103620
+    },
+    {
+      "epoch": 681.7763157894736,
+      "grad_norm": 1.0479704141616821,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 103630
+    },
+    {
+      "epoch": 681.8421052631579,
+      "grad_norm": 1.2243036031723022,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 103640
+    },
+    {
+      "epoch": 681.9078947368421,
+      "grad_norm": 0.8835751414299011,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 103650
+    },
+    {
+      "epoch": 681.9736842105264,
+      "grad_norm": 0.982793927192688,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 103660
+    },
+    {
+      "epoch": 682.0394736842105,
+      "grad_norm": 1.2806956768035889,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 103670
+    },
+    {
+      "epoch": 682.1052631578947,
+      "grad_norm": 1.2741878032684326,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 103680
+    },
+    {
+      "epoch": 682.171052631579,
+      "grad_norm": 0.92679762840271,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 103690
+    },
+    {
+      "epoch": 682.2368421052631,
+      "grad_norm": 1.367266058921814,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 103700
+    },
+    {
+      "epoch": 682.3026315789474,
+      "grad_norm": 0.7617621421813965,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 103710
+    },
+    {
+      "epoch": 682.3684210526316,
+      "grad_norm": 0.740490734577179,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 103720
+    },
+    {
+      "epoch": 682.4342105263158,
+      "grad_norm": 1.2656317949295044,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 103730
+    },
+    {
+      "epoch": 682.5,
+      "grad_norm": 1.286602258682251,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 103740
+    },
+    {
+      "epoch": 682.5657894736842,
+      "grad_norm": 1.0341092348098755,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 103750
+    },
+    {
+      "epoch": 682.6315789473684,
+      "grad_norm": 1.1948630809783936,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 103760
+    },
+    {
+      "epoch": 682.6973684210526,
+      "grad_norm": 1.1949267387390137,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 103770
+    },
+    {
+      "epoch": 682.7631578947369,
+      "grad_norm": 1.2278507947921753,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 103780
+    },
+    {
+      "epoch": 682.828947368421,
+      "grad_norm": 1.1813137531280518,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 103790
+    },
+    {
+      "epoch": 682.8947368421053,
+      "grad_norm": 1.0005921125411987,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 103800
+    },
+    {
+      "epoch": 682.9605263157895,
+      "grad_norm": 0.7229200005531311,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 103810
+    },
+    {
+      "epoch": 683.0263157894736,
+      "grad_norm": 0.7326030135154724,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 103820
+    },
+    {
+      "epoch": 683.0921052631579,
+      "grad_norm": 0.5357984900474548,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 103830
+    },
+    {
+      "epoch": 683.1578947368421,
+      "grad_norm": 0.856106162071228,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 103840
+    },
+    {
+      "epoch": 683.2236842105264,
+      "grad_norm": 0.9525607228279114,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 103850
+    },
+    {
+      "epoch": 683.2894736842105,
+      "grad_norm": 0.6489855647087097,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 103860
+    },
+    {
+      "epoch": 683.3552631578947,
+      "grad_norm": 0.7851767539978027,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 103870
+    },
+    {
+      "epoch": 683.421052631579,
+      "grad_norm": 1.2372000217437744,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 103880
+    },
+    {
+      "epoch": 683.4868421052631,
+      "grad_norm": 1.1379144191741943,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 103890
+    },
+    {
+      "epoch": 683.5526315789474,
+      "grad_norm": 1.1280021667480469,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 103900
+    },
+    {
+      "epoch": 683.6184210526316,
+      "grad_norm": 1.049955129623413,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 103910
+    },
+    {
+      "epoch": 683.6842105263158,
+      "grad_norm": 1.1678504943847656,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 103920
+    },
+    {
+      "epoch": 683.75,
+      "grad_norm": 1.31575345993042,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 103930
+    },
+    {
+      "epoch": 683.8157894736842,
+      "grad_norm": 1.1992285251617432,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 103940
+    },
+    {
+      "epoch": 683.8815789473684,
+      "grad_norm": 1.1740840673446655,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 103950
+    },
+    {
+      "epoch": 683.9473684210526,
+      "grad_norm": 1.3050742149353027,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 103960
+    },
+    {
+      "epoch": 684.0131578947369,
+      "grad_norm": 1.0441380739212036,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 103970
+    },
+    {
+      "epoch": 684.078947368421,
+      "grad_norm": 0.8772562146186829,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 103980
+    },
+    {
+      "epoch": 684.1447368421053,
+      "grad_norm": 1.1009024381637573,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 103990
+    },
+    {
+      "epoch": 684.2105263157895,
+      "grad_norm": 1.1562106609344482,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 104000
+    },
+    {
+      "epoch": 684.2763157894736,
+      "grad_norm": 1.0368518829345703,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 104010
+    },
+    {
+      "epoch": 684.3421052631579,
+      "grad_norm": 0.992682933807373,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 104020
+    },
+    {
+      "epoch": 684.4078947368421,
+      "grad_norm": 0.7431752681732178,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 104030
+    },
+    {
+      "epoch": 684.4736842105264,
+      "grad_norm": 1.2687865495681763,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 104040
+    },
+    {
+      "epoch": 684.5394736842105,
+      "grad_norm": 1.1064850091934204,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 104050
+    },
+    {
+      "epoch": 684.6052631578947,
+      "grad_norm": 1.0989872217178345,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 104060
+    },
+    {
+      "epoch": 684.671052631579,
+      "grad_norm": 0.931109607219696,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 104070
+    },
+    {
+      "epoch": 684.7368421052631,
+      "grad_norm": 1.1777174472808838,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 104080
+    },
+    {
+      "epoch": 684.8026315789474,
+      "grad_norm": 1.218532919883728,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 104090
+    },
+    {
+      "epoch": 684.8684210526316,
+      "grad_norm": 1.0839264392852783,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 104100
+    },
+    {
+      "epoch": 684.9342105263158,
+      "grad_norm": 1.5299314260482788,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 104110
+    },
+    {
+      "epoch": 685.0,
+      "grad_norm": 0.85472172498703,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 104120
+    },
+    {
+      "epoch": 685.0657894736842,
+      "grad_norm": 1.038342833518982,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 104130
+    },
+    {
+      "epoch": 685.1315789473684,
+      "grad_norm": 1.3401973247528076,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 104140
+    },
+    {
+      "epoch": 685.1973684210526,
+      "grad_norm": 1.3317383527755737,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 104150
+    },
+    {
+      "epoch": 685.2631578947369,
+      "grad_norm": 1.0047595500946045,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 104160
+    },
+    {
+      "epoch": 685.328947368421,
+      "grad_norm": 1.0131652355194092,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 104170
+    },
+    {
+      "epoch": 685.3947368421053,
+      "grad_norm": 0.9121875166893005,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 104180
+    },
+    {
+      "epoch": 685.4605263157895,
+      "grad_norm": 1.079406976699829,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 104190
+    },
+    {
+      "epoch": 685.5263157894736,
+      "grad_norm": 1.008108139038086,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 104200
+    },
+    {
+      "epoch": 685.5921052631579,
+      "grad_norm": 0.8241605758666992,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 104210
+    },
+    {
+      "epoch": 685.6578947368421,
+      "grad_norm": 0.8410286903381348,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 104220
+    },
+    {
+      "epoch": 685.7236842105264,
+      "grad_norm": 0.8288211226463318,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 104230
+    },
+    {
+      "epoch": 685.7894736842105,
+      "grad_norm": 0.8989503383636475,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 104240
+    },
+    {
+      "epoch": 685.8552631578947,
+      "grad_norm": 1.1565848588943481,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 104250
+    },
+    {
+      "epoch": 685.921052631579,
+      "grad_norm": 1.376077651977539,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 104260
+    },
+    {
+      "epoch": 685.9868421052631,
+      "grad_norm": 0.8136918544769287,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 104270
+    },
+    {
+      "epoch": 686.0526315789474,
+      "grad_norm": 1.2481287717819214,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 104280
+    },
+    {
+      "epoch": 686.1184210526316,
+      "grad_norm": 1.1149044036865234,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 104290
+    },
+    {
+      "epoch": 686.1842105263158,
+      "grad_norm": 1.1859511137008667,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 104300
+    },
+    {
+      "epoch": 686.25,
+      "grad_norm": 1.0750795602798462,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 104310
+    },
+    {
+      "epoch": 686.3157894736842,
+      "grad_norm": 1.4063059091567993,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 104320
+    },
+    {
+      "epoch": 686.3815789473684,
+      "grad_norm": 1.4898396730422974,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 104330
+    },
+    {
+      "epoch": 686.4473684210526,
+      "grad_norm": 1.1081674098968506,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 104340
+    },
+    {
+      "epoch": 686.5131578947369,
+      "grad_norm": 1.1555651426315308,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 104350
+    },
+    {
+      "epoch": 686.578947368421,
+      "grad_norm": 1.2798792123794556,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 104360
+    },
+    {
+      "epoch": 686.6447368421053,
+      "grad_norm": 0.9958081841468811,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 104370
+    },
+    {
+      "epoch": 686.7105263157895,
+      "grad_norm": 1.3393999338150024,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 104380
+    },
+    {
+      "epoch": 686.7763157894736,
+      "grad_norm": 1.0677820444107056,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 104390
+    },
+    {
+      "epoch": 686.8421052631579,
+      "grad_norm": 1.1201198101043701,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 104400
+    },
+    {
+      "epoch": 686.9078947368421,
+      "grad_norm": 0.9454928040504456,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 104410
+    },
+    {
+      "epoch": 686.9736842105264,
+      "grad_norm": 1.056349515914917,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 104420
+    },
+    {
+      "epoch": 687.0394736842105,
+      "grad_norm": 0.84330153465271,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 104430
+    },
+    {
+      "epoch": 687.1052631578947,
+      "grad_norm": 0.9636723399162292,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 104440
+    },
+    {
+      "epoch": 687.171052631579,
+      "grad_norm": 1.4594964981079102,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 104450
+    },
+    {
+      "epoch": 687.2368421052631,
+      "grad_norm": 0.8333598375320435,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 104460
+    },
+    {
+      "epoch": 687.3026315789474,
+      "grad_norm": 0.9445642828941345,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 104470
+    },
+    {
+      "epoch": 687.3684210526316,
+      "grad_norm": 1.2898783683776855,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 104480
+    },
+    {
+      "epoch": 687.4342105263158,
+      "grad_norm": 1.1669278144836426,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 104490
+    },
+    {
+      "epoch": 687.5,
+      "grad_norm": 0.9558044075965881,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 104500
+    },
+    {
+      "epoch": 687.5657894736842,
+      "grad_norm": 1.2216863632202148,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 104510
+    },
+    {
+      "epoch": 687.6315789473684,
+      "grad_norm": 1.3271684646606445,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 104520
+    },
+    {
+      "epoch": 687.6973684210526,
+      "grad_norm": 1.0442551374435425,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 104530
+    },
+    {
+      "epoch": 687.7631578947369,
+      "grad_norm": 0.6984831094741821,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 104540
+    },
+    {
+      "epoch": 687.828947368421,
+      "grad_norm": 1.0056344270706177,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 104550
+    },
+    {
+      "epoch": 687.8947368421053,
+      "grad_norm": 1.0924835205078125,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 104560
+    },
+    {
+      "epoch": 687.9605263157895,
+      "grad_norm": 1.1241261959075928,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 104570
+    },
+    {
+      "epoch": 688.0263157894736,
+      "grad_norm": 0.9026856422424316,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 104580
+    },
+    {
+      "epoch": 688.0921052631579,
+      "grad_norm": 0.7907440662384033,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 104590
+    },
+    {
+      "epoch": 688.1578947368421,
+      "grad_norm": 1.012350082397461,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 104600
+    },
+    {
+      "epoch": 688.2236842105264,
+      "grad_norm": 1.0234317779541016,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 104610
+    },
+    {
+      "epoch": 688.2894736842105,
+      "grad_norm": 0.9007984399795532,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 104620
+    },
+    {
+      "epoch": 688.3552631578947,
+      "grad_norm": 1.339667797088623,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 104630
+    },
+    {
+      "epoch": 688.421052631579,
+      "grad_norm": 1.140592336654663,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 104640
+    },
+    {
+      "epoch": 688.4868421052631,
+      "grad_norm": 1.2752139568328857,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 104650
+    },
+    {
+      "epoch": 688.5526315789474,
+      "grad_norm": 1.20328950881958,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 104660
+    },
+    {
+      "epoch": 688.6184210526316,
+      "grad_norm": 1.0453274250030518,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 104670
+    },
+    {
+      "epoch": 688.6842105263158,
+      "grad_norm": 1.0228806734085083,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 104680
+    },
+    {
+      "epoch": 688.75,
+      "grad_norm": 1.1263914108276367,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 104690
+    },
+    {
+      "epoch": 688.8157894736842,
+      "grad_norm": 1.1457710266113281,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 104700
+    },
+    {
+      "epoch": 688.8815789473684,
+      "grad_norm": 1.1575244665145874,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 104710
+    },
+    {
+      "epoch": 688.9473684210526,
+      "grad_norm": 1.1374318599700928,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 104720
+    },
+    {
+      "epoch": 689.0131578947369,
+      "grad_norm": 1.1628844738006592,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 104730
+    },
+    {
+      "epoch": 689.078947368421,
+      "grad_norm": 1.267177939414978,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 104740
+    },
+    {
+      "epoch": 689.1447368421053,
+      "grad_norm": 1.0953381061553955,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 104750
+    },
+    {
+      "epoch": 689.2105263157895,
+      "grad_norm": 0.6190617680549622,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 104760
+    },
+    {
+      "epoch": 689.2763157894736,
+      "grad_norm": 1.0876604318618774,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 104770
+    },
+    {
+      "epoch": 689.3421052631579,
+      "grad_norm": 1.1195549964904785,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 104780
+    },
+    {
+      "epoch": 689.4078947368421,
+      "grad_norm": 1.1582069396972656,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 104790
+    },
+    {
+      "epoch": 689.4736842105264,
+      "grad_norm": 1.3708412647247314,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 104800
+    },
+    {
+      "epoch": 689.5394736842105,
+      "grad_norm": 1.0523473024368286,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 104810
+    },
+    {
+      "epoch": 689.6052631578947,
+      "grad_norm": 1.0299116373062134,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 104820
+    },
+    {
+      "epoch": 689.671052631579,
+      "grad_norm": 1.1898280382156372,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 104830
+    },
+    {
+      "epoch": 689.7368421052631,
+      "grad_norm": 1.1457937955856323,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 104840
+    },
+    {
+      "epoch": 689.8026315789474,
+      "grad_norm": 1.2077155113220215,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 104850
+    },
+    {
+      "epoch": 689.8684210526316,
+      "grad_norm": 1.064317226409912,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 104860
+    },
+    {
+      "epoch": 689.9342105263158,
+      "grad_norm": 1.3832712173461914,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 104870
+    },
+    {
+      "epoch": 690.0,
+      "grad_norm": 1.2490742206573486,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 104880
+    },
+    {
+      "epoch": 690.0657894736842,
+      "grad_norm": 0.9746840000152588,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 104890
+    },
+    {
+      "epoch": 690.1315789473684,
+      "grad_norm": 1.291522741317749,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 104900
+    },
+    {
+      "epoch": 690.1973684210526,
+      "grad_norm": 1.422899603843689,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 104910
+    },
+    {
+      "epoch": 690.2631578947369,
+      "grad_norm": 1.692259669303894,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 104920
+    },
+    {
+      "epoch": 690.328947368421,
+      "grad_norm": 1.206990361213684,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 104930
+    },
+    {
+      "epoch": 690.3947368421053,
+      "grad_norm": 1.1877973079681396,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 104940
+    },
+    {
+      "epoch": 690.4605263157895,
+      "grad_norm": 0.9999209642410278,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 104950
+    },
+    {
+      "epoch": 690.5263157894736,
+      "grad_norm": 0.9556674361228943,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 104960
+    },
+    {
+      "epoch": 690.5921052631579,
+      "grad_norm": 1.3551650047302246,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 104970
+    },
+    {
+      "epoch": 690.6578947368421,
+      "grad_norm": 0.9526252746582031,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 104980
+    },
+    {
+      "epoch": 690.7236842105264,
+      "grad_norm": 1.16046941280365,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 104990
+    },
+    {
+      "epoch": 690.7894736842105,
+      "grad_norm": 1.410556674003601,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 105000
+    },
+    {
+      "epoch": 690.8552631578947,
+      "grad_norm": 1.1336820125579834,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 105010
+    },
+    {
+      "epoch": 690.921052631579,
+      "grad_norm": 1.0210304260253906,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 105020
+    },
+    {
+      "epoch": 690.9868421052631,
+      "grad_norm": 0.7669985890388489,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 105030
+    },
+    {
+      "epoch": 691.0526315789474,
+      "grad_norm": 1.2497632503509521,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 105040
+    },
+    {
+      "epoch": 691.1184210526316,
+      "grad_norm": 0.9652649760246277,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 105050
+    },
+    {
+      "epoch": 691.1842105263158,
+      "grad_norm": 1.0892058610916138,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 105060
+    },
+    {
+      "epoch": 691.25,
+      "grad_norm": 1.201240062713623,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 105070
+    },
+    {
+      "epoch": 691.3157894736842,
+      "grad_norm": 1.2225768566131592,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 105080
+    },
+    {
+      "epoch": 691.3815789473684,
+      "grad_norm": 1.4184614419937134,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 105090
+    },
+    {
+      "epoch": 691.4473684210526,
+      "grad_norm": 0.9505151510238647,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 105100
+    },
+    {
+      "epoch": 691.5131578947369,
+      "grad_norm": 0.7423208355903625,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 105110
+    },
+    {
+      "epoch": 691.578947368421,
+      "grad_norm": 1.2950553894042969,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 105120
+    },
+    {
+      "epoch": 691.6447368421053,
+      "grad_norm": 1.050185203552246,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 105130
+    },
+    {
+      "epoch": 691.7105263157895,
+      "grad_norm": 1.229802131652832,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 105140
+    },
+    {
+      "epoch": 691.7763157894736,
+      "grad_norm": 1.34513258934021,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 105150
+    },
+    {
+      "epoch": 691.8421052631579,
+      "grad_norm": 0.9867190718650818,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 105160
+    },
+    {
+      "epoch": 691.9078947368421,
+      "grad_norm": 1.0777831077575684,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 105170
+    },
+    {
+      "epoch": 691.9736842105264,
+      "grad_norm": 0.8535577058792114,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 105180
+    },
+    {
+      "epoch": 692.0394736842105,
+      "grad_norm": 1.4894434213638306,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 105190
+    },
+    {
+      "epoch": 692.1052631578947,
+      "grad_norm": 1.5268137454986572,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 105200
+    },
+    {
+      "epoch": 692.171052631579,
+      "grad_norm": 0.9532687067985535,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 105210
+    },
+    {
+      "epoch": 692.2368421052631,
+      "grad_norm": 1.5903687477111816,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 105220
+    },
+    {
+      "epoch": 692.3026315789474,
+      "grad_norm": 1.4972559213638306,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 105230
+    },
+    {
+      "epoch": 692.3684210526316,
+      "grad_norm": 1.5704691410064697,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 105240
+    },
+    {
+      "epoch": 692.4342105263158,
+      "grad_norm": 1.3427348136901855,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 105250
+    },
+    {
+      "epoch": 692.5,
+      "grad_norm": 1.3334673643112183,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 105260
+    },
+    {
+      "epoch": 692.5657894736842,
+      "grad_norm": 1.3983582258224487,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 105270
+    },
+    {
+      "epoch": 692.6315789473684,
+      "grad_norm": 1.0505462884902954,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 105280
+    },
+    {
+      "epoch": 692.6973684210526,
+      "grad_norm": 1.273409128189087,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 105290
+    },
+    {
+      "epoch": 692.7631578947369,
+      "grad_norm": 1.1249748468399048,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 105300
+    },
+    {
+      "epoch": 692.828947368421,
+      "grad_norm": 1.0098894834518433,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 105310
+    },
+    {
+      "epoch": 692.8947368421053,
+      "grad_norm": 0.9849072694778442,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 105320
+    },
+    {
+      "epoch": 692.9605263157895,
+      "grad_norm": 1.1056095361709595,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 105330
+    },
+    {
+      "epoch": 693.0263157894736,
+      "grad_norm": 1.139883041381836,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 105340
+    },
+    {
+      "epoch": 693.0921052631579,
+      "grad_norm": 1.1363656520843506,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 105350
+    },
+    {
+      "epoch": 693.1578947368421,
+      "grad_norm": 1.1857936382293701,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 105360
+    },
+    {
+      "epoch": 693.2236842105264,
+      "grad_norm": 0.88065105676651,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 105370
+    },
+    {
+      "epoch": 693.2894736842105,
+      "grad_norm": 0.8013133406639099,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 105380
+    },
+    {
+      "epoch": 693.3552631578947,
+      "grad_norm": 1.1217681169509888,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 105390
+    },
+    {
+      "epoch": 693.421052631579,
+      "grad_norm": 1.2208178043365479,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 105400
+    },
+    {
+      "epoch": 693.4868421052631,
+      "grad_norm": 1.139967918395996,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 105410
+    },
+    {
+      "epoch": 693.5526315789474,
+      "grad_norm": 1.063003659248352,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 105420
+    },
+    {
+      "epoch": 693.6184210526316,
+      "grad_norm": 1.4261283874511719,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 105430
+    },
+    {
+      "epoch": 693.6842105263158,
+      "grad_norm": 1.0778611898422241,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 105440
+    },
+    {
+      "epoch": 693.75,
+      "grad_norm": 1.0890920162200928,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 105450
+    },
+    {
+      "epoch": 693.8157894736842,
+      "grad_norm": 1.4902729988098145,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 105460
+    },
+    {
+      "epoch": 693.8815789473684,
+      "grad_norm": 1.185309886932373,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 105470
+    },
+    {
+      "epoch": 693.9473684210526,
+      "grad_norm": 1.2872449159622192,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 105480
+    },
+    {
+      "epoch": 694.0131578947369,
+      "grad_norm": 1.1898432970046997,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 105490
+    },
+    {
+      "epoch": 694.078947368421,
+      "grad_norm": 1.2668589353561401,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 105500
+    },
+    {
+      "epoch": 694.1447368421053,
+      "grad_norm": 1.3832660913467407,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 105510
+    },
+    {
+      "epoch": 694.2105263157895,
+      "grad_norm": 1.297754168510437,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 105520
+    },
+    {
+      "epoch": 694.2763157894736,
+      "grad_norm": 1.1962448358535767,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 105530
+    },
+    {
+      "epoch": 694.3421052631579,
+      "grad_norm": 1.3533624410629272,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 105540
+    },
+    {
+      "epoch": 694.4078947368421,
+      "grad_norm": 1.2034482955932617,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 105550
+    },
+    {
+      "epoch": 694.4736842105264,
+      "grad_norm": 1.1491233110427856,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 105560
+    },
+    {
+      "epoch": 694.5394736842105,
+      "grad_norm": 1.1336445808410645,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 105570
+    },
+    {
+      "epoch": 694.6052631578947,
+      "grad_norm": 1.028238296508789,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 105580
+    },
+    {
+      "epoch": 694.671052631579,
+      "grad_norm": 1.0678424835205078,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 105590
+    },
+    {
+      "epoch": 694.7368421052631,
+      "grad_norm": 1.087962031364441,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 105600
+    },
+    {
+      "epoch": 694.8026315789474,
+      "grad_norm": 1.0385785102844238,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 105610
+    },
+    {
+      "epoch": 694.8684210526316,
+      "grad_norm": 1.196866512298584,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 105620
+    },
+    {
+      "epoch": 694.9342105263158,
+      "grad_norm": 1.0130568742752075,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 105630
+    },
+    {
+      "epoch": 695.0,
+      "grad_norm": 0.7995595932006836,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 105640
+    },
+    {
+      "epoch": 695.0657894736842,
+      "grad_norm": 1.008529782295227,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 105650
+    },
+    {
+      "epoch": 695.1315789473684,
+      "grad_norm": 0.806501030921936,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 105660
+    },
+    {
+      "epoch": 695.1973684210526,
+      "grad_norm": 0.8974860906600952,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 105670
+    },
+    {
+      "epoch": 695.2631578947369,
+      "grad_norm": 1.2004984617233276,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 105680
+    },
+    {
+      "epoch": 695.328947368421,
+      "grad_norm": 1.1275862455368042,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 105690
+    },
+    {
+      "epoch": 695.3947368421053,
+      "grad_norm": 0.9173672199249268,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 105700
+    },
+    {
+      "epoch": 695.4605263157895,
+      "grad_norm": 0.8183532953262329,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 105710
+    },
+    {
+      "epoch": 695.5263157894736,
+      "grad_norm": 0.8390060663223267,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 105720
+    },
+    {
+      "epoch": 695.5921052631579,
+      "grad_norm": 1.1580220460891724,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 105730
+    },
+    {
+      "epoch": 695.6578947368421,
+      "grad_norm": 0.9561638832092285,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 105740
+    },
+    {
+      "epoch": 695.7236842105264,
+      "grad_norm": 1.305795431137085,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 105750
+    },
+    {
+      "epoch": 695.7894736842105,
+      "grad_norm": 1.044447422027588,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 105760
+    },
+    {
+      "epoch": 695.8552631578947,
+      "grad_norm": 1.3991165161132812,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 105770
+    },
+    {
+      "epoch": 695.921052631579,
+      "grad_norm": 0.8383837342262268,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 105780
+    },
+    {
+      "epoch": 695.9868421052631,
+      "grad_norm": 0.9691088795661926,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 105790
+    },
+    {
+      "epoch": 696.0526315789474,
+      "grad_norm": 1.0391768217086792,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 105800
+    },
+    {
+      "epoch": 696.1184210526316,
+      "grad_norm": 0.8894152641296387,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 105810
+    },
+    {
+      "epoch": 696.1842105263158,
+      "grad_norm": 0.7640244364738464,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 105820
+    },
+    {
+      "epoch": 696.25,
+      "grad_norm": 0.6968852877616882,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 105830
+    },
+    {
+      "epoch": 696.3157894736842,
+      "grad_norm": 0.7941989302635193,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 105840
+    },
+    {
+      "epoch": 696.3815789473684,
+      "grad_norm": 1.2468845844268799,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 105850
+    },
+    {
+      "epoch": 696.4473684210526,
+      "grad_norm": 1.1141244173049927,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 105860
+    },
+    {
+      "epoch": 696.5131578947369,
+      "grad_norm": 0.9799408912658691,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 105870
+    },
+    {
+      "epoch": 696.578947368421,
+      "grad_norm": 0.7694076299667358,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 105880
+    },
+    {
+      "epoch": 696.6447368421053,
+      "grad_norm": 1.0681445598602295,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 105890
+    },
+    {
+      "epoch": 696.7105263157895,
+      "grad_norm": 0.7481711506843567,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 105900
+    },
+    {
+      "epoch": 696.7763157894736,
+      "grad_norm": 0.9661341309547424,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 105910
+    },
+    {
+      "epoch": 696.8421052631579,
+      "grad_norm": 1.1508374214172363,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 105920
+    },
+    {
+      "epoch": 696.9078947368421,
+      "grad_norm": 1.2137349843978882,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 105930
+    },
+    {
+      "epoch": 696.9736842105264,
+      "grad_norm": 1.2096316814422607,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 105940
+    },
+    {
+      "epoch": 697.0394736842105,
+      "grad_norm": 1.1053109169006348,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 105950
+    },
+    {
+      "epoch": 697.1052631578947,
+      "grad_norm": 1.0717750787734985,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 105960
+    },
+    {
+      "epoch": 697.171052631579,
+      "grad_norm": 1.2490699291229248,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 105970
+    },
+    {
+      "epoch": 697.2368421052631,
+      "grad_norm": 1.2428292036056519,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 105980
+    },
+    {
+      "epoch": 697.3026315789474,
+      "grad_norm": 1.023769736289978,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 105990
+    },
+    {
+      "epoch": 697.3684210526316,
+      "grad_norm": 1.1357252597808838,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 106000
+    },
+    {
+      "epoch": 697.4342105263158,
+      "grad_norm": 1.178993582725525,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 106010
+    },
+    {
+      "epoch": 697.5,
+      "grad_norm": 0.642010509967804,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 106020
+    },
+    {
+      "epoch": 697.5657894736842,
+      "grad_norm": 0.8353654742240906,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 106030
+    },
+    {
+      "epoch": 697.6315789473684,
+      "grad_norm": 0.9867295622825623,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 106040
+    },
+    {
+      "epoch": 697.6973684210526,
+      "grad_norm": 1.1230249404907227,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 106050
+    },
+    {
+      "epoch": 697.7631578947369,
+      "grad_norm": 1.3012531995773315,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 106060
+    },
+    {
+      "epoch": 697.828947368421,
+      "grad_norm": 1.0835860967636108,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 106070
+    },
+    {
+      "epoch": 697.8947368421053,
+      "grad_norm": 1.1090937852859497,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 106080
+    },
+    {
+      "epoch": 697.9605263157895,
+      "grad_norm": 1.16708505153656,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 106090
+    },
+    {
+      "epoch": 698.0263157894736,
+      "grad_norm": 1.0519673824310303,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 106100
+    },
+    {
+      "epoch": 698.0921052631579,
+      "grad_norm": 1.2105119228363037,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 106110
+    },
+    {
+      "epoch": 698.1578947368421,
+      "grad_norm": 1.2713031768798828,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 106120
+    },
+    {
+      "epoch": 698.2236842105264,
+      "grad_norm": 0.7641732692718506,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 106130
+    },
+    {
+      "epoch": 698.2894736842105,
+      "grad_norm": 1.30423903465271,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 106140
+    },
+    {
+      "epoch": 698.3552631578947,
+      "grad_norm": 1.076094388961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 106150
+    },
+    {
+      "epoch": 698.421052631579,
+      "grad_norm": 1.1429264545440674,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 106160
+    },
+    {
+      "epoch": 698.4868421052631,
+      "grad_norm": 1.1242903470993042,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 106170
+    },
+    {
+      "epoch": 698.5526315789474,
+      "grad_norm": 0.9787787199020386,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 106180
+    },
+    {
+      "epoch": 698.6184210526316,
+      "grad_norm": 1.1341170072555542,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 106190
+    },
+    {
+      "epoch": 698.6842105263158,
+      "grad_norm": 0.9176144003868103,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 106200
+    },
+    {
+      "epoch": 698.75,
+      "grad_norm": 1.141721487045288,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 106210
+    },
+    {
+      "epoch": 698.8157894736842,
+      "grad_norm": 0.9301637411117554,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 106220
+    },
+    {
+      "epoch": 698.8815789473684,
+      "grad_norm": 1.2631868124008179,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 106230
+    },
+    {
+      "epoch": 698.9473684210526,
+      "grad_norm": 0.9568750262260437,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 106240
+    },
+    {
+      "epoch": 699.0131578947369,
+      "grad_norm": 1.0320165157318115,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 106250
+    },
+    {
+      "epoch": 699.078947368421,
+      "grad_norm": 0.9375261664390564,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 106260
+    },
+    {
+      "epoch": 699.1447368421053,
+      "grad_norm": 1.236198902130127,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 106270
+    },
+    {
+      "epoch": 699.2105263157895,
+      "grad_norm": 0.9655860662460327,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 106280
+    },
+    {
+      "epoch": 699.2763157894736,
+      "grad_norm": 0.9262257814407349,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 106290
+    },
+    {
+      "epoch": 699.3421052631579,
+      "grad_norm": 1.303225040435791,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 106300
+    },
+    {
+      "epoch": 699.4078947368421,
+      "grad_norm": 1.1554994583129883,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 106310
+    },
+    {
+      "epoch": 699.4736842105264,
+      "grad_norm": 1.3132860660552979,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 106320
+    },
+    {
+      "epoch": 699.5394736842105,
+      "grad_norm": 0.9363493919372559,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 106330
+    },
+    {
+      "epoch": 699.6052631578947,
+      "grad_norm": 1.1558855772018433,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 106340
+    },
+    {
+      "epoch": 699.671052631579,
+      "grad_norm": 1.206477403640747,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 106350
+    },
+    {
+      "epoch": 699.7368421052631,
+      "grad_norm": 1.2269618511199951,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 106360
+    },
+    {
+      "epoch": 699.8026315789474,
+      "grad_norm": 1.058182716369629,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 106370
+    },
+    {
+      "epoch": 699.8684210526316,
+      "grad_norm": 1.0236554145812988,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 106380
+    },
+    {
+      "epoch": 699.9342105263158,
+      "grad_norm": 1.0067118406295776,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 106390
+    },
+    {
+      "epoch": 700.0,
+      "grad_norm": 1.0427459478378296,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 106400
+    },
+    {
+      "epoch": 700.0657894736842,
+      "grad_norm": 1.3142248392105103,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 106410
+    },
+    {
+      "epoch": 700.1315789473684,
+      "grad_norm": 0.9090234041213989,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 106420
+    },
+    {
+      "epoch": 700.1973684210526,
+      "grad_norm": 0.9273867607116699,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 106430
+    },
+    {
+      "epoch": 700.2631578947369,
+      "grad_norm": 0.958476722240448,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 106440
+    },
+    {
+      "epoch": 700.328947368421,
+      "grad_norm": 1.0677157640457153,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 106450
+    },
+    {
+      "epoch": 700.3947368421053,
+      "grad_norm": 1.2199952602386475,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 106460
+    },
+    {
+      "epoch": 700.4605263157895,
+      "grad_norm": 1.244336724281311,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 106470
+    },
+    {
+      "epoch": 700.5263157894736,
+      "grad_norm": 1.1296188831329346,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 106480
+    },
+    {
+      "epoch": 700.5921052631579,
+      "grad_norm": 1.4885672330856323,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 106490
+    },
+    {
+      "epoch": 700.6578947368421,
+      "grad_norm": 1.0214803218841553,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 106500
+    },
+    {
+      "epoch": 700.7236842105264,
+      "grad_norm": 1.2202999591827393,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 106510
+    },
+    {
+      "epoch": 700.7894736842105,
+      "grad_norm": 0.9407332539558411,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 106520
+    },
+    {
+      "epoch": 700.8552631578947,
+      "grad_norm": 0.8566510677337646,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 106530
+    },
+    {
+      "epoch": 700.921052631579,
+      "grad_norm": 0.9324583411216736,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 106540
+    },
+    {
+      "epoch": 700.9868421052631,
+      "grad_norm": 1.044553279876709,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 106550
+    },
+    {
+      "epoch": 701.0526315789474,
+      "grad_norm": 1.0946259498596191,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 106560
+    },
+    {
+      "epoch": 701.1184210526316,
+      "grad_norm": 1.0644162893295288,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 106570
+    },
+    {
+      "epoch": 701.1842105263158,
+      "grad_norm": 1.1168071031570435,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 106580
+    },
+    {
+      "epoch": 701.25,
+      "grad_norm": 1.2700804471969604,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 106590
+    },
+    {
+      "epoch": 701.3157894736842,
+      "grad_norm": 1.0045099258422852,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 106600
+    },
+    {
+      "epoch": 701.3815789473684,
+      "grad_norm": 1.2132641077041626,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 106610
+    },
+    {
+      "epoch": 701.4473684210526,
+      "grad_norm": 0.9734482169151306,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 106620
+    },
+    {
+      "epoch": 701.5131578947369,
+      "grad_norm": 0.9646698832511902,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 106630
+    },
+    {
+      "epoch": 701.578947368421,
+      "grad_norm": 1.0157219171524048,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 106640
+    },
+    {
+      "epoch": 701.6447368421053,
+      "grad_norm": 0.8552806973457336,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 106650
+    },
+    {
+      "epoch": 701.7105263157895,
+      "grad_norm": 0.9953233003616333,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 106660
+    },
+    {
+      "epoch": 701.7763157894736,
+      "grad_norm": 1.2354696989059448,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 106670
+    },
+    {
+      "epoch": 701.8421052631579,
+      "grad_norm": 1.019411563873291,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 106680
+    },
+    {
+      "epoch": 701.9078947368421,
+      "grad_norm": 1.2094115018844604,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 106690
+    },
+    {
+      "epoch": 701.9736842105264,
+      "grad_norm": 1.2663511037826538,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 106700
+    },
+    {
+      "epoch": 702.0394736842105,
+      "grad_norm": 1.2119556665420532,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 106710
+    },
+    {
+      "epoch": 702.1052631578947,
+      "grad_norm": 1.0109612941741943,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 106720
+    },
+    {
+      "epoch": 702.171052631579,
+      "grad_norm": 1.126444697380066,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 106730
+    },
+    {
+      "epoch": 702.2368421052631,
+      "grad_norm": 1.0462061166763306,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 106740
+    },
+    {
+      "epoch": 702.3026315789474,
+      "grad_norm": 1.3411667346954346,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 106750
+    },
+    {
+      "epoch": 702.3684210526316,
+      "grad_norm": 1.215775728225708,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 106760
+    },
+    {
+      "epoch": 702.4342105263158,
+      "grad_norm": 1.4780688285827637,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 106770
+    },
+    {
+      "epoch": 702.5,
+      "grad_norm": 1.3486248254776,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 106780
+    },
+    {
+      "epoch": 702.5657894736842,
+      "grad_norm": 1.1051580905914307,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 106790
+    },
+    {
+      "epoch": 702.6315789473684,
+      "grad_norm": 1.1159751415252686,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 106800
+    },
+    {
+      "epoch": 702.6973684210526,
+      "grad_norm": 0.8592166900634766,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 106810
+    },
+    {
+      "epoch": 702.7631578947369,
+      "grad_norm": 1.0782039165496826,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 106820
+    },
+    {
+      "epoch": 702.828947368421,
+      "grad_norm": 1.10999596118927,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 106830
+    },
+    {
+      "epoch": 702.8947368421053,
+      "grad_norm": 1.2560298442840576,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 106840
+    },
+    {
+      "epoch": 702.9605263157895,
+      "grad_norm": 1.123194694519043,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 106850
+    },
+    {
+      "epoch": 703.0263157894736,
+      "grad_norm": 1.0512282848358154,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 106860
+    },
+    {
+      "epoch": 703.0921052631579,
+      "grad_norm": 1.1006485223770142,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 106870
+    },
+    {
+      "epoch": 703.1578947368421,
+      "grad_norm": 0.8787376284599304,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 106880
+    },
+    {
+      "epoch": 703.2236842105264,
+      "grad_norm": 1.178478479385376,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 106890
+    },
+    {
+      "epoch": 703.2894736842105,
+      "grad_norm": 1.2921770811080933,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 106900
+    },
+    {
+      "epoch": 703.3552631578947,
+      "grad_norm": 1.2834266424179077,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 106910
+    },
+    {
+      "epoch": 703.421052631579,
+      "grad_norm": 1.2144252061843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 106920
+    },
+    {
+      "epoch": 703.4868421052631,
+      "grad_norm": 1.4216984510421753,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 106930
+    },
+    {
+      "epoch": 703.5526315789474,
+      "grad_norm": 1.094687581062317,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 106940
+    },
+    {
+      "epoch": 703.6184210526316,
+      "grad_norm": 0.8012235164642334,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 106950
+    },
+    {
+      "epoch": 703.6842105263158,
+      "grad_norm": 1.1857279539108276,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 106960
+    },
+    {
+      "epoch": 703.75,
+      "grad_norm": 1.2665016651153564,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 106970
+    },
+    {
+      "epoch": 703.8157894736842,
+      "grad_norm": 0.9577199220657349,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 106980
+    },
+    {
+      "epoch": 703.8815789473684,
+      "grad_norm": 0.9756731986999512,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 106990
+    },
+    {
+      "epoch": 703.9473684210526,
+      "grad_norm": 1.3572697639465332,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 107000
+    },
+    {
+      "epoch": 704.0131578947369,
+      "grad_norm": 0.9276780486106873,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 107010
+    },
+    {
+      "epoch": 704.078947368421,
+      "grad_norm": 0.9649149775505066,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 107020
+    },
+    {
+      "epoch": 704.1447368421053,
+      "grad_norm": 0.9579582810401917,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 107030
+    },
+    {
+      "epoch": 704.2105263157895,
+      "grad_norm": 1.0553480386734009,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 107040
+    },
+    {
+      "epoch": 704.2763157894736,
+      "grad_norm": 1.0931501388549805,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 107050
+    },
+    {
+      "epoch": 704.3421052631579,
+      "grad_norm": 1.0343230962753296,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 107060
+    },
+    {
+      "epoch": 704.4078947368421,
+      "grad_norm": 1.1571636199951172,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 107070
+    },
+    {
+      "epoch": 704.4736842105264,
+      "grad_norm": 1.0774343013763428,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 107080
+    },
+    {
+      "epoch": 704.5394736842105,
+      "grad_norm": 0.9952540397644043,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 107090
+    },
+    {
+      "epoch": 704.6052631578947,
+      "grad_norm": 1.2612237930297852,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 107100
+    },
+    {
+      "epoch": 704.671052631579,
+      "grad_norm": 1.1099755764007568,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 107110
+    },
+    {
+      "epoch": 704.7368421052631,
+      "grad_norm": 1.1933823823928833,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 107120
+    },
+    {
+      "epoch": 704.8026315789474,
+      "grad_norm": 0.8420557379722595,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 107130
+    },
+    {
+      "epoch": 704.8684210526316,
+      "grad_norm": 0.7003465294837952,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 107140
+    },
+    {
+      "epoch": 704.9342105263158,
+      "grad_norm": 1.202143907546997,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 107150
+    },
+    {
+      "epoch": 705.0,
+      "grad_norm": 0.8362906575202942,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 107160
+    },
+    {
+      "epoch": 705.0657894736842,
+      "grad_norm": 1.445649266242981,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 107170
+    },
+    {
+      "epoch": 705.1315789473684,
+      "grad_norm": 1.4059619903564453,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 107180
+    },
+    {
+      "epoch": 705.1973684210526,
+      "grad_norm": 1.178699016571045,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 107190
+    },
+    {
+      "epoch": 705.2631578947369,
+      "grad_norm": 1.397766351699829,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 107200
+    },
+    {
+      "epoch": 705.328947368421,
+      "grad_norm": 1.5875097513198853,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 107210
+    },
+    {
+      "epoch": 705.3947368421053,
+      "grad_norm": 1.234609842300415,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 107220
+    },
+    {
+      "epoch": 705.4605263157895,
+      "grad_norm": 1.3890899419784546,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 107230
+    },
+    {
+      "epoch": 705.5263157894736,
+      "grad_norm": 1.17513906955719,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 107240
+    },
+    {
+      "epoch": 705.5921052631579,
+      "grad_norm": 0.8957661986351013,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 107250
+    },
+    {
+      "epoch": 705.6578947368421,
+      "grad_norm": 0.950888991355896,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 107260
+    },
+    {
+      "epoch": 705.7236842105264,
+      "grad_norm": 1.2250102758407593,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 107270
+    },
+    {
+      "epoch": 705.7894736842105,
+      "grad_norm": 1.09373939037323,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 107280
+    },
+    {
+      "epoch": 705.8552631578947,
+      "grad_norm": 1.0987354516983032,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 107290
+    },
+    {
+      "epoch": 705.921052631579,
+      "grad_norm": 0.9803661704063416,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 107300
+    },
+    {
+      "epoch": 705.9868421052631,
+      "grad_norm": 0.8128421306610107,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 107310
+    },
+    {
+      "epoch": 706.0526315789474,
+      "grad_norm": 0.9702147841453552,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 107320
+    },
+    {
+      "epoch": 706.1184210526316,
+      "grad_norm": 1.0280282497406006,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 107330
+    },
+    {
+      "epoch": 706.1842105263158,
+      "grad_norm": 1.0692169666290283,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 107340
+    },
+    {
+      "epoch": 706.25,
+      "grad_norm": 1.2259516716003418,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 107350
+    },
+    {
+      "epoch": 706.3157894736842,
+      "grad_norm": 1.2134467363357544,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 107360
+    },
+    {
+      "epoch": 706.3815789473684,
+      "grad_norm": 1.4001606702804565,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 107370
+    },
+    {
+      "epoch": 706.4473684210526,
+      "grad_norm": 0.8867390155792236,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 107380
+    },
+    {
+      "epoch": 706.5131578947369,
+      "grad_norm": 1.1521034240722656,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 107390
+    },
+    {
+      "epoch": 706.578947368421,
+      "grad_norm": 1.07185959815979,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 107400
+    },
+    {
+      "epoch": 706.6447368421053,
+      "grad_norm": 0.9920993447303772,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 107410
+    },
+    {
+      "epoch": 706.7105263157895,
+      "grad_norm": 1.0084408521652222,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 107420
+    },
+    {
+      "epoch": 706.7763157894736,
+      "grad_norm": 1.267495036125183,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 107430
+    },
+    {
+      "epoch": 706.8421052631579,
+      "grad_norm": 1.3698124885559082,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 107440
+    },
+    {
+      "epoch": 706.9078947368421,
+      "grad_norm": 1.062820315361023,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 107450
+    },
+    {
+      "epoch": 706.9736842105264,
+      "grad_norm": 1.5721125602722168,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 107460
+    },
+    {
+      "epoch": 707.0394736842105,
+      "grad_norm": 1.1249845027923584,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 107470
+    },
+    {
+      "epoch": 707.1052631578947,
+      "grad_norm": 1.3193303346633911,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 107480
+    },
+    {
+      "epoch": 707.171052631579,
+      "grad_norm": 1.1248865127563477,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 107490
+    },
+    {
+      "epoch": 707.2368421052631,
+      "grad_norm": 1.0584726333618164,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 107500
+    },
+    {
+      "epoch": 707.3026315789474,
+      "grad_norm": 1.1499197483062744,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 107510
+    },
+    {
+      "epoch": 707.3684210526316,
+      "grad_norm": 1.3799810409545898,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 107520
+    },
+    {
+      "epoch": 707.4342105263158,
+      "grad_norm": 1.3305660486221313,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 107530
+    },
+    {
+      "epoch": 707.5,
+      "grad_norm": 1.3056739568710327,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 107540
+    },
+    {
+      "epoch": 707.5657894736842,
+      "grad_norm": 0.9053949117660522,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 107550
+    },
+    {
+      "epoch": 707.6315789473684,
+      "grad_norm": 0.6791939735412598,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 107560
+    },
+    {
+      "epoch": 707.6973684210526,
+      "grad_norm": 1.0320830345153809,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 107570
+    },
+    {
+      "epoch": 707.7631578947369,
+      "grad_norm": 0.8447424173355103,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 107580
+    },
+    {
+      "epoch": 707.828947368421,
+      "grad_norm": 0.9377061724662781,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 107590
+    },
+    {
+      "epoch": 707.8947368421053,
+      "grad_norm": 1.2281724214553833,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 107600
+    },
+    {
+      "epoch": 707.9605263157895,
+      "grad_norm": 1.1441106796264648,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 107610
+    },
+    {
+      "epoch": 708.0263157894736,
+      "grad_norm": 1.3841345310211182,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 107620
+    },
+    {
+      "epoch": 708.0921052631579,
+      "grad_norm": 1.2582640647888184,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 107630
+    },
+    {
+      "epoch": 708.1578947368421,
+      "grad_norm": 1.3789803981781006,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 107640
+    },
+    {
+      "epoch": 708.2236842105264,
+      "grad_norm": 1.2278392314910889,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 107650
+    },
+    {
+      "epoch": 708.2894736842105,
+      "grad_norm": 1.1727294921875,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 107660
+    },
+    {
+      "epoch": 708.3552631578947,
+      "grad_norm": 0.9205034375190735,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 107670
+    },
+    {
+      "epoch": 708.421052631579,
+      "grad_norm": 0.9984912872314453,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 107680
+    },
+    {
+      "epoch": 708.4868421052631,
+      "grad_norm": 0.7507139444351196,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 107690
+    },
+    {
+      "epoch": 708.5526315789474,
+      "grad_norm": 1.054671287536621,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 107700
+    },
+    {
+      "epoch": 708.6184210526316,
+      "grad_norm": 0.9641456604003906,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 107710
+    },
+    {
+      "epoch": 708.6842105263158,
+      "grad_norm": 1.3411998748779297,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 107720
+    },
+    {
+      "epoch": 708.75,
+      "grad_norm": 0.993855357170105,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 107730
+    },
+    {
+      "epoch": 708.8157894736842,
+      "grad_norm": 0.6834460496902466,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 107740
+    },
+    {
+      "epoch": 708.8815789473684,
+      "grad_norm": 0.7746645212173462,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 107750
+    },
+    {
+      "epoch": 708.9473684210526,
+      "grad_norm": 0.9106453061103821,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 107760
+    },
+    {
+      "epoch": 709.0131578947369,
+      "grad_norm": 1.1743462085723877,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 107770
+    },
+    {
+      "epoch": 709.078947368421,
+      "grad_norm": 1.0879472494125366,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 107780
+    },
+    {
+      "epoch": 709.1447368421053,
+      "grad_norm": 0.9474685192108154,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 107790
+    },
+    {
+      "epoch": 709.2105263157895,
+      "grad_norm": 1.2010844945907593,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 107800
+    },
+    {
+      "epoch": 709.2763157894736,
+      "grad_norm": 1.3946421146392822,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 107810
+    },
+    {
+      "epoch": 709.3421052631579,
+      "grad_norm": 1.2751479148864746,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 107820
+    },
+    {
+      "epoch": 709.4078947368421,
+      "grad_norm": 1.0376313924789429,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 107830
+    },
+    {
+      "epoch": 709.4736842105264,
+      "grad_norm": 0.96278977394104,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 107840
+    },
+    {
+      "epoch": 709.5394736842105,
+      "grad_norm": 0.7817369699478149,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 107850
+    },
+    {
+      "epoch": 709.6052631578947,
+      "grad_norm": 1.010752558708191,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 107860
+    },
+    {
+      "epoch": 709.671052631579,
+      "grad_norm": 1.0647060871124268,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 107870
+    },
+    {
+      "epoch": 709.7368421052631,
+      "grad_norm": 0.7408843636512756,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 107880
+    },
+    {
+      "epoch": 709.8026315789474,
+      "grad_norm": 0.9281281232833862,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 107890
+    },
+    {
+      "epoch": 709.8684210526316,
+      "grad_norm": 1.3812792301177979,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 107900
+    },
+    {
+      "epoch": 709.9342105263158,
+      "grad_norm": 1.2687792778015137,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 107910
+    },
+    {
+      "epoch": 710.0,
+      "grad_norm": 0.8348847031593323,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 107920
+    },
+    {
+      "epoch": 710.0657894736842,
+      "grad_norm": 1.199946641921997,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 107930
+    },
+    {
+      "epoch": 710.1315789473684,
+      "grad_norm": 0.9090976119041443,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 107940
+    },
+    {
+      "epoch": 710.1973684210526,
+      "grad_norm": 1.042677879333496,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 107950
+    },
+    {
+      "epoch": 710.2631578947369,
+      "grad_norm": 1.2170066833496094,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 107960
+    },
+    {
+      "epoch": 710.328947368421,
+      "grad_norm": 1.3705898523330688,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 107970
+    },
+    {
+      "epoch": 710.3947368421053,
+      "grad_norm": 1.2710314989089966,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 107980
+    },
+    {
+      "epoch": 710.4605263157895,
+      "grad_norm": 1.175453543663025,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 107990
+    },
+    {
+      "epoch": 710.5263157894736,
+      "grad_norm": 0.9529065489768982,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 108000
+    },
+    {
+      "epoch": 710.5921052631579,
+      "grad_norm": 1.1746102571487427,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 108010
+    },
+    {
+      "epoch": 710.6578947368421,
+      "grad_norm": 1.0855262279510498,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 108020
+    },
+    {
+      "epoch": 710.7236842105264,
+      "grad_norm": 1.1520380973815918,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 108030
+    },
+    {
+      "epoch": 710.7894736842105,
+      "grad_norm": 0.9170784950256348,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 108040
+    },
+    {
+      "epoch": 710.8552631578947,
+      "grad_norm": 1.1136395931243896,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 108050
+    },
+    {
+      "epoch": 710.921052631579,
+      "grad_norm": 1.207287073135376,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 108060
+    },
+    {
+      "epoch": 710.9868421052631,
+      "grad_norm": 1.001600980758667,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 108070
+    },
+    {
+      "epoch": 711.0526315789474,
+      "grad_norm": 0.9510000348091125,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 108080
+    },
+    {
+      "epoch": 711.1184210526316,
+      "grad_norm": 0.7180987000465393,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 108090
+    },
+    {
+      "epoch": 711.1842105263158,
+      "grad_norm": 0.977580189704895,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 108100
+    },
+    {
+      "epoch": 711.25,
+      "grad_norm": 0.843988299369812,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 108110
+    },
+    {
+      "epoch": 711.3157894736842,
+      "grad_norm": 1.3870481252670288,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 108120
+    },
+    {
+      "epoch": 711.3815789473684,
+      "grad_norm": 1.0824686288833618,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 108130
+    },
+    {
+      "epoch": 711.4473684210526,
+      "grad_norm": 1.089383602142334,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 108140
+    },
+    {
+      "epoch": 711.5131578947369,
+      "grad_norm": 0.9979451298713684,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 108150
+    },
+    {
+      "epoch": 711.578947368421,
+      "grad_norm": 0.9893368482589722,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 108160
+    },
+    {
+      "epoch": 711.6447368421053,
+      "grad_norm": 1.0383979082107544,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 108170
+    },
+    {
+      "epoch": 711.7105263157895,
+      "grad_norm": 1.280486822128296,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 108180
+    },
+    {
+      "epoch": 711.7763157894736,
+      "grad_norm": 1.3183673620224,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 108190
+    },
+    {
+      "epoch": 711.8421052631579,
+      "grad_norm": 1.2419326305389404,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 108200
+    },
+    {
+      "epoch": 711.9078947368421,
+      "grad_norm": 1.3799538612365723,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 108210
+    },
+    {
+      "epoch": 711.9736842105264,
+      "grad_norm": 0.9520628452301025,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 108220
+    },
+    {
+      "epoch": 712.0394736842105,
+      "grad_norm": 0.9893721342086792,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 108230
+    },
+    {
+      "epoch": 712.1052631578947,
+      "grad_norm": 1.260324239730835,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 108240
+    },
+    {
+      "epoch": 712.171052631579,
+      "grad_norm": 1.377862572669983,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 108250
+    },
+    {
+      "epoch": 712.2368421052631,
+      "grad_norm": 1.4345449209213257,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 108260
+    },
+    {
+      "epoch": 712.3026315789474,
+      "grad_norm": 1.0738847255706787,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 108270
+    },
+    {
+      "epoch": 712.3684210526316,
+      "grad_norm": 1.1670902967453003,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 108280
+    },
+    {
+      "epoch": 712.4342105263158,
+      "grad_norm": 1.362013578414917,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 108290
+    },
+    {
+      "epoch": 712.5,
+      "grad_norm": 1.2504547834396362,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 108300
+    },
+    {
+      "epoch": 712.5657894736842,
+      "grad_norm": 1.6184645891189575,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 108310
+    },
+    {
+      "epoch": 712.6315789473684,
+      "grad_norm": 1.2186678647994995,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 108320
+    },
+    {
+      "epoch": 712.6973684210526,
+      "grad_norm": 1.0406875610351562,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 108330
+    },
+    {
+      "epoch": 712.7631578947369,
+      "grad_norm": 0.9314107894897461,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 108340
+    },
+    {
+      "epoch": 712.828947368421,
+      "grad_norm": 0.996528148651123,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 108350
+    },
+    {
+      "epoch": 712.8947368421053,
+      "grad_norm": 0.9486305117607117,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 108360
+    },
+    {
+      "epoch": 712.9605263157895,
+      "grad_norm": 0.9168351292610168,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 108370
+    },
+    {
+      "epoch": 713.0263157894736,
+      "grad_norm": 0.9039626717567444,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 108380
+    },
+    {
+      "epoch": 713.0921052631579,
+      "grad_norm": 0.9132272601127625,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 108390
+    },
+    {
+      "epoch": 713.1578947368421,
+      "grad_norm": 0.9673351049423218,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 108400
+    },
+    {
+      "epoch": 713.2236842105264,
+      "grad_norm": 0.8779450058937073,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 108410
+    },
+    {
+      "epoch": 713.2894736842105,
+      "grad_norm": 1.4076229333877563,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 108420
+    },
+    {
+      "epoch": 713.3552631578947,
+      "grad_norm": 1.1083792448043823,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 108430
+    },
+    {
+      "epoch": 713.421052631579,
+      "grad_norm": 1.1614620685577393,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 108440
+    },
+    {
+      "epoch": 713.4868421052631,
+      "grad_norm": 1.1655492782592773,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 108450
+    },
+    {
+      "epoch": 713.5526315789474,
+      "grad_norm": 0.9747868776321411,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 108460
+    },
+    {
+      "epoch": 713.6184210526316,
+      "grad_norm": 1.1151975393295288,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 108470
+    },
+    {
+      "epoch": 713.6842105263158,
+      "grad_norm": 1.0362054109573364,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 108480
+    },
+    {
+      "epoch": 713.75,
+      "grad_norm": 1.196274757385254,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 108490
+    },
+    {
+      "epoch": 713.8157894736842,
+      "grad_norm": 0.6207993030548096,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 108500
+    },
+    {
+      "epoch": 713.8815789473684,
+      "grad_norm": 0.8537484407424927,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 108510
+    },
+    {
+      "epoch": 713.9473684210526,
+      "grad_norm": 0.7432962656021118,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 108520
+    },
+    {
+      "epoch": 714.0131578947369,
+      "grad_norm": 1.1963402032852173,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 108530
+    },
+    {
+      "epoch": 714.078947368421,
+      "grad_norm": 1.0920330286026,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 108540
+    },
+    {
+      "epoch": 714.1447368421053,
+      "grad_norm": 0.5513865351676941,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 108550
+    },
+    {
+      "epoch": 714.2105263157895,
+      "grad_norm": 1.2096713781356812,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 108560
+    },
+    {
+      "epoch": 714.2763157894736,
+      "grad_norm": 0.7468422651290894,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 108570
+    },
+    {
+      "epoch": 714.3421052631579,
+      "grad_norm": 1.296449899673462,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 108580
+    },
+    {
+      "epoch": 714.4078947368421,
+      "grad_norm": 1.2958561182022095,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 108590
+    },
+    {
+      "epoch": 714.4736842105264,
+      "grad_norm": 1.0128918886184692,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 108600
+    },
+    {
+      "epoch": 714.5394736842105,
+      "grad_norm": 0.9160696864128113,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 108610
+    },
+    {
+      "epoch": 714.6052631578947,
+      "grad_norm": 0.9371721148490906,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 108620
+    },
+    {
+      "epoch": 714.671052631579,
+      "grad_norm": 0.8780196905136108,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 108630
+    },
+    {
+      "epoch": 714.7368421052631,
+      "grad_norm": 0.8234481811523438,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 108640
+    },
+    {
+      "epoch": 714.8026315789474,
+      "grad_norm": 0.9354177713394165,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 108650
+    },
+    {
+      "epoch": 714.8684210526316,
+      "grad_norm": 1.4554506540298462,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 108660
+    },
+    {
+      "epoch": 714.9342105263158,
+      "grad_norm": 0.8353884220123291,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 108670
+    },
+    {
+      "epoch": 715.0,
+      "grad_norm": 1.3016752004623413,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 108680
+    },
+    {
+      "epoch": 715.0657894736842,
+      "grad_norm": 1.0093426704406738,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 108690
+    },
+    {
+      "epoch": 715.1315789473684,
+      "grad_norm": 0.8471440672874451,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 108700
+    },
+    {
+      "epoch": 715.1973684210526,
+      "grad_norm": 1.381130576133728,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 108710
+    },
+    {
+      "epoch": 715.2631578947369,
+      "grad_norm": 1.0571134090423584,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 108720
+    },
+    {
+      "epoch": 715.328947368421,
+      "grad_norm": 1.1896861791610718,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 108730
+    },
+    {
+      "epoch": 715.3947368421053,
+      "grad_norm": 0.7548841834068298,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 108740
+    },
+    {
+      "epoch": 715.4605263157895,
+      "grad_norm": 0.652093231678009,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 108750
+    },
+    {
+      "epoch": 715.5263157894736,
+      "grad_norm": 0.6288610696792603,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 108760
+    },
+    {
+      "epoch": 715.5921052631579,
+      "grad_norm": 0.9039884209632874,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 108770
+    },
+    {
+      "epoch": 715.6578947368421,
+      "grad_norm": 0.7594443559646606,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 108780
+    },
+    {
+      "epoch": 715.7236842105264,
+      "grad_norm": 1.3326084613800049,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 108790
+    },
+    {
+      "epoch": 715.7894736842105,
+      "grad_norm": 0.7681350708007812,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 108800
+    },
+    {
+      "epoch": 715.8552631578947,
+      "grad_norm": 0.6910384893417358,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 108810
+    },
+    {
+      "epoch": 715.921052631579,
+      "grad_norm": 0.9494452476501465,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 108820
+    },
+    {
+      "epoch": 715.9868421052631,
+      "grad_norm": 1.248620867729187,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 108830
+    },
+    {
+      "epoch": 716.0526315789474,
+      "grad_norm": 1.2778465747833252,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 108840
+    },
+    {
+      "epoch": 716.1184210526316,
+      "grad_norm": 1.229967474937439,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 108850
+    },
+    {
+      "epoch": 716.1842105263158,
+      "grad_norm": 1.2221544981002808,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 108860
+    },
+    {
+      "epoch": 716.25,
+      "grad_norm": 0.9114399552345276,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 108870
+    },
+    {
+      "epoch": 716.3157894736842,
+      "grad_norm": 1.2433297634124756,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 108880
+    },
+    {
+      "epoch": 716.3815789473684,
+      "grad_norm": 1.0392719507217407,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 108890
+    },
+    {
+      "epoch": 716.4473684210526,
+      "grad_norm": 1.3560720682144165,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 108900
+    },
+    {
+      "epoch": 716.5131578947369,
+      "grad_norm": 1.7873417139053345,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 108910
+    },
+    {
+      "epoch": 716.578947368421,
+      "grad_norm": 1.2388242483139038,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 108920
+    },
+    {
+      "epoch": 716.6447368421053,
+      "grad_norm": 1.307459831237793,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 108930
+    },
+    {
+      "epoch": 716.7105263157895,
+      "grad_norm": 1.3190605640411377,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 108940
+    },
+    {
+      "epoch": 716.7763157894736,
+      "grad_norm": 1.2416794300079346,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 108950
+    },
+    {
+      "epoch": 716.8421052631579,
+      "grad_norm": 1.4050918817520142,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 108960
+    },
+    {
+      "epoch": 716.9078947368421,
+      "grad_norm": 0.8353002667427063,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 108970
+    },
+    {
+      "epoch": 716.9736842105264,
+      "grad_norm": 1.285539984703064,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 108980
+    },
+    {
+      "epoch": 717.0394736842105,
+      "grad_norm": 0.93118816614151,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 108990
+    },
+    {
+      "epoch": 717.1052631578947,
+      "grad_norm": 1.2005360126495361,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 109000
+    },
+    {
+      "epoch": 717.171052631579,
+      "grad_norm": 1.0540269613265991,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 109010
+    },
+    {
+      "epoch": 717.2368421052631,
+      "grad_norm": 1.0876415967941284,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 109020
+    },
+    {
+      "epoch": 717.3026315789474,
+      "grad_norm": 1.1824675798416138,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 109030
+    },
+    {
+      "epoch": 717.3684210526316,
+      "grad_norm": 0.9963378310203552,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 109040
+    },
+    {
+      "epoch": 717.4342105263158,
+      "grad_norm": 1.1265755891799927,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 109050
+    },
+    {
+      "epoch": 717.5,
+      "grad_norm": 0.9112656712532043,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 109060
+    },
+    {
+      "epoch": 717.5657894736842,
+      "grad_norm": 0.6385452151298523,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 109070
+    },
+    {
+      "epoch": 717.6315789473684,
+      "grad_norm": 1.0386406183242798,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 109080
+    },
+    {
+      "epoch": 717.6973684210526,
+      "grad_norm": 0.5495930314064026,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 109090
+    },
+    {
+      "epoch": 717.7631578947369,
+      "grad_norm": 1.1156800985336304,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 109100
+    },
+    {
+      "epoch": 717.828947368421,
+      "grad_norm": 1.2771714925765991,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 109110
+    },
+    {
+      "epoch": 717.8947368421053,
+      "grad_norm": 1.0908472537994385,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 109120
+    },
+    {
+      "epoch": 717.9605263157895,
+      "grad_norm": 0.8165896534919739,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 109130
+    },
+    {
+      "epoch": 718.0263157894736,
+      "grad_norm": 1.1818957328796387,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 109140
+    },
+    {
+      "epoch": 718.0921052631579,
+      "grad_norm": 1.2370299100875854,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 109150
+    },
+    {
+      "epoch": 718.1578947368421,
+      "grad_norm": 1.1179412603378296,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 109160
+    },
+    {
+      "epoch": 718.2236842105264,
+      "grad_norm": 1.040429711341858,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 109170
+    },
+    {
+      "epoch": 718.2894736842105,
+      "grad_norm": 1.090136170387268,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 109180
+    },
+    {
+      "epoch": 718.3552631578947,
+      "grad_norm": 1.2007248401641846,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 109190
+    },
+    {
+      "epoch": 718.421052631579,
+      "grad_norm": 1.092756986618042,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 109200
+    },
+    {
+      "epoch": 718.4868421052631,
+      "grad_norm": 0.8665720820426941,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 109210
+    },
+    {
+      "epoch": 718.5526315789474,
+      "grad_norm": 1.117387056350708,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 109220
+    },
+    {
+      "epoch": 718.6184210526316,
+      "grad_norm": 1.2737882137298584,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 109230
+    },
+    {
+      "epoch": 718.6842105263158,
+      "grad_norm": 0.8910441398620605,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 109240
+    },
+    {
+      "epoch": 718.75,
+      "grad_norm": 1.0049889087677002,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 109250
+    },
+    {
+      "epoch": 718.8157894736842,
+      "grad_norm": 1.1752616167068481,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 109260
+    },
+    {
+      "epoch": 718.8815789473684,
+      "grad_norm": 0.8606551885604858,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 109270
+    },
+    {
+      "epoch": 718.9473684210526,
+      "grad_norm": 0.8159741163253784,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 109280
+    },
+    {
+      "epoch": 719.0131578947369,
+      "grad_norm": 1.0030544996261597,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 109290
+    },
+    {
+      "epoch": 719.078947368421,
+      "grad_norm": 0.8993750810623169,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 109300
+    },
+    {
+      "epoch": 719.1447368421053,
+      "grad_norm": 1.2870763540267944,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 109310
+    },
+    {
+      "epoch": 719.2105263157895,
+      "grad_norm": 0.9233137369155884,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 109320
+    },
+    {
+      "epoch": 719.2763157894736,
+      "grad_norm": 0.928329586982727,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 109330
+    },
+    {
+      "epoch": 719.3421052631579,
+      "grad_norm": 1.117807388305664,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 109340
+    },
+    {
+      "epoch": 719.4078947368421,
+      "grad_norm": 1.040723204612732,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 109350
+    },
+    {
+      "epoch": 719.4736842105264,
+      "grad_norm": 1.066552996635437,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 109360
+    },
+    {
+      "epoch": 719.5394736842105,
+      "grad_norm": 0.8651625514030457,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 109370
+    },
+    {
+      "epoch": 719.6052631578947,
+      "grad_norm": 1.3908514976501465,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 109380
+    },
+    {
+      "epoch": 719.671052631579,
+      "grad_norm": 1.2088861465454102,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 109390
+    },
+    {
+      "epoch": 719.7368421052631,
+      "grad_norm": 1.2017627954483032,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 109400
+    },
+    {
+      "epoch": 719.8026315789474,
+      "grad_norm": 1.2163819074630737,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 109410
+    },
+    {
+      "epoch": 719.8684210526316,
+      "grad_norm": 0.8632364273071289,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 109420
+    },
+    {
+      "epoch": 719.9342105263158,
+      "grad_norm": 1.2523313760757446,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 109430
+    },
+    {
+      "epoch": 720.0,
+      "grad_norm": 1.2396034002304077,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 109440
+    },
+    {
+      "epoch": 720.0657894736842,
+      "grad_norm": 1.3992606401443481,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 109450
+    },
+    {
+      "epoch": 720.1315789473684,
+      "grad_norm": 1.1208072900772095,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 109460
+    },
+    {
+      "epoch": 720.1973684210526,
+      "grad_norm": 1.3922314643859863,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 109470
+    },
+    {
+      "epoch": 720.2631578947369,
+      "grad_norm": 1.2266921997070312,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 109480
+    },
+    {
+      "epoch": 720.328947368421,
+      "grad_norm": 0.8874131441116333,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 109490
+    },
+    {
+      "epoch": 720.3947368421053,
+      "grad_norm": 0.9725931882858276,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 109500
+    },
+    {
+      "epoch": 720.4605263157895,
+      "grad_norm": 1.0284030437469482,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 109510
+    },
+    {
+      "epoch": 720.5263157894736,
+      "grad_norm": 0.9212563633918762,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 109520
+    },
+    {
+      "epoch": 720.5921052631579,
+      "grad_norm": 1.033521294593811,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 109530
+    },
+    {
+      "epoch": 720.6578947368421,
+      "grad_norm": 0.933686375617981,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 109540
+    },
+    {
+      "epoch": 720.7236842105264,
+      "grad_norm": 1.2563905715942383,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 109550
+    },
+    {
+      "epoch": 720.7894736842105,
+      "grad_norm": 1.322197675704956,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 109560
+    },
+    {
+      "epoch": 720.8552631578947,
+      "grad_norm": 1.221634864807129,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 109570
+    },
+    {
+      "epoch": 720.921052631579,
+      "grad_norm": 1.2580702304840088,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 109580
+    },
+    {
+      "epoch": 720.9868421052631,
+      "grad_norm": 0.6149132251739502,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 109590
+    },
+    {
+      "epoch": 721.0526315789474,
+      "grad_norm": 1.1910752058029175,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 109600
+    },
+    {
+      "epoch": 721.1184210526316,
+      "grad_norm": 0.7073631286621094,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 109610
+    },
+    {
+      "epoch": 721.1842105263158,
+      "grad_norm": 1.3418338298797607,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 109620
+    },
+    {
+      "epoch": 721.25,
+      "grad_norm": 1.2676738500595093,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 109630
+    },
+    {
+      "epoch": 721.3157894736842,
+      "grad_norm": 1.0815623998641968,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 109640
+    },
+    {
+      "epoch": 721.3815789473684,
+      "grad_norm": 0.6487982273101807,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 109650
+    },
+    {
+      "epoch": 721.4473684210526,
+      "grad_norm": 1.0579499006271362,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 109660
+    },
+    {
+      "epoch": 721.5131578947369,
+      "grad_norm": 1.0880662202835083,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 109670
+    },
+    {
+      "epoch": 721.578947368421,
+      "grad_norm": 1.020585298538208,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 109680
+    },
+    {
+      "epoch": 721.6447368421053,
+      "grad_norm": 0.9497626423835754,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 109690
+    },
+    {
+      "epoch": 721.7105263157895,
+      "grad_norm": 1.3067365884780884,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 109700
+    },
+    {
+      "epoch": 721.7763157894736,
+      "grad_norm": 1.3649437427520752,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 109710
+    },
+    {
+      "epoch": 721.8421052631579,
+      "grad_norm": 1.4548853635787964,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 109720
+    },
+    {
+      "epoch": 721.9078947368421,
+      "grad_norm": 1.0740547180175781,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 109730
+    },
+    {
+      "epoch": 721.9736842105264,
+      "grad_norm": 1.0808460712432861,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 109740
+    },
+    {
+      "epoch": 722.0394736842105,
+      "grad_norm": 0.738896906375885,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 109750
+    },
+    {
+      "epoch": 722.1052631578947,
+      "grad_norm": 1.3755104541778564,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 109760
+    },
+    {
+      "epoch": 722.171052631579,
+      "grad_norm": 1.0921390056610107,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 109770
+    },
+    {
+      "epoch": 722.2368421052631,
+      "grad_norm": 1.5075722932815552,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 109780
+    },
+    {
+      "epoch": 722.3026315789474,
+      "grad_norm": 1.332755446434021,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 109790
+    },
+    {
+      "epoch": 722.3684210526316,
+      "grad_norm": 0.9924520254135132,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 109800
+    },
+    {
+      "epoch": 722.4342105263158,
+      "grad_norm": 1.1538974046707153,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 109810
+    },
+    {
+      "epoch": 722.5,
+      "grad_norm": 0.8574804663658142,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 109820
+    },
+    {
+      "epoch": 722.5657894736842,
+      "grad_norm": 1.1876112222671509,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 109830
+    },
+    {
+      "epoch": 722.6315789473684,
+      "grad_norm": 1.1805399656295776,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 109840
+    },
+    {
+      "epoch": 722.6973684210526,
+      "grad_norm": 1.1132640838623047,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 109850
+    },
+    {
+      "epoch": 722.7631578947369,
+      "grad_norm": 1.2226154804229736,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 109860
+    },
+    {
+      "epoch": 722.828947368421,
+      "grad_norm": 1.0198978185653687,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 109870
+    },
+    {
+      "epoch": 722.8947368421053,
+      "grad_norm": 0.758260190486908,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 109880
+    },
+    {
+      "epoch": 722.9605263157895,
+      "grad_norm": 1.051596999168396,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 109890
+    },
+    {
+      "epoch": 723.0263157894736,
+      "grad_norm": 1.2912297248840332,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 109900
+    },
+    {
+      "epoch": 723.0921052631579,
+      "grad_norm": 1.3294340372085571,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 109910
+    },
+    {
+      "epoch": 723.1578947368421,
+      "grad_norm": 1.156272053718567,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 109920
+    },
+    {
+      "epoch": 723.2236842105264,
+      "grad_norm": 0.9503754377365112,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 109930
+    },
+    {
+      "epoch": 723.2894736842105,
+      "grad_norm": 1.5646758079528809,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 109940
+    },
+    {
+      "epoch": 723.3552631578947,
+      "grad_norm": 1.065543532371521,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 109950
+    },
+    {
+      "epoch": 723.421052631579,
+      "grad_norm": 1.1717690229415894,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 109960
+    },
+    {
+      "epoch": 723.4868421052631,
+      "grad_norm": 0.8114318251609802,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 109970
+    },
+    {
+      "epoch": 723.5526315789474,
+      "grad_norm": 1.155068278312683,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 109980
+    },
+    {
+      "epoch": 723.6184210526316,
+      "grad_norm": 1.220639705657959,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 109990
+    },
+    {
+      "epoch": 723.6842105263158,
+      "grad_norm": 0.996774435043335,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 110000
+    },
+    {
+      "epoch": 723.75,
+      "grad_norm": 1.1410325765609741,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 110010
+    },
+    {
+      "epoch": 723.8157894736842,
+      "grad_norm": 1.0671559572219849,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 110020
+    },
+    {
+      "epoch": 723.8815789473684,
+      "grad_norm": 1.0174076557159424,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 110030
+    },
+    {
+      "epoch": 723.9473684210526,
+      "grad_norm": 1.0742820501327515,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 110040
+    },
+    {
+      "epoch": 724.0131578947369,
+      "grad_norm": 0.996814489364624,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 110050
+    },
+    {
+      "epoch": 724.078947368421,
+      "grad_norm": 1.142468810081482,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 110060
+    },
+    {
+      "epoch": 724.1447368421053,
+      "grad_norm": 1.2725366353988647,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 110070
+    },
+    {
+      "epoch": 724.2105263157895,
+      "grad_norm": 1.052403211593628,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 110080
+    },
+    {
+      "epoch": 724.2763157894736,
+      "grad_norm": 1.1229180097579956,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 110090
+    },
+    {
+      "epoch": 724.3421052631579,
+      "grad_norm": 0.9793723821640015,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 110100
+    },
+    {
+      "epoch": 724.4078947368421,
+      "grad_norm": 0.8089613318443298,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 110110
+    },
+    {
+      "epoch": 724.4736842105264,
+      "grad_norm": 1.0036156177520752,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 110120
+    },
+    {
+      "epoch": 724.5394736842105,
+      "grad_norm": 1.4579956531524658,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 110130
+    },
+    {
+      "epoch": 724.6052631578947,
+      "grad_norm": 1.225488543510437,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 110140
+    },
+    {
+      "epoch": 724.671052631579,
+      "grad_norm": 0.904151201248169,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 110150
+    },
+    {
+      "epoch": 724.7368421052631,
+      "grad_norm": 0.8515450954437256,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 110160
+    },
+    {
+      "epoch": 724.8026315789474,
+      "grad_norm": 0.9605996608734131,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 110170
+    },
+    {
+      "epoch": 724.8684210526316,
+      "grad_norm": 0.8468832969665527,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 110180
+    },
+    {
+      "epoch": 724.9342105263158,
+      "grad_norm": 1.0464105606079102,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 110190
+    },
+    {
+      "epoch": 725.0,
+      "grad_norm": 0.886685848236084,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 110200
+    },
+    {
+      "epoch": 725.0657894736842,
+      "grad_norm": 1.0475046634674072,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 110210
+    },
+    {
+      "epoch": 725.1315789473684,
+      "grad_norm": 1.0469486713409424,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 110220
+    },
+    {
+      "epoch": 725.1973684210526,
+      "grad_norm": 1.3123482465744019,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 110230
+    },
+    {
+      "epoch": 725.2631578947369,
+      "grad_norm": 1.2292877435684204,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 110240
+    },
+    {
+      "epoch": 725.328947368421,
+      "grad_norm": 1.54679536819458,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 110250
+    },
+    {
+      "epoch": 725.3947368421053,
+      "grad_norm": 1.5144623517990112,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 110260
+    },
+    {
+      "epoch": 725.4605263157895,
+      "grad_norm": 1.4638532400131226,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 110270
+    },
+    {
+      "epoch": 725.5263157894736,
+      "grad_norm": 1.037933349609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 110280
+    },
+    {
+      "epoch": 725.5921052631579,
+      "grad_norm": 0.8391295671463013,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 110290
+    },
+    {
+      "epoch": 725.6578947368421,
+      "grad_norm": 1.3161094188690186,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 110300
+    },
+    {
+      "epoch": 725.7236842105264,
+      "grad_norm": 1.3753902912139893,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 110310
+    },
+    {
+      "epoch": 725.7894736842105,
+      "grad_norm": 1.369210124015808,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 110320
+    },
+    {
+      "epoch": 725.8552631578947,
+      "grad_norm": 0.8835994005203247,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 110330
+    },
+    {
+      "epoch": 725.921052631579,
+      "grad_norm": 1.2269823551177979,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 110340
+    },
+    {
+      "epoch": 725.9868421052631,
+      "grad_norm": 1.24722158908844,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 110350
+    },
+    {
+      "epoch": 726.0526315789474,
+      "grad_norm": 1.065712571144104,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 110360
+    },
+    {
+      "epoch": 726.1184210526316,
+      "grad_norm": 1.5890785455703735,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 110370
+    },
+    {
+      "epoch": 726.1842105263158,
+      "grad_norm": 0.9685258269309998,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 110380
+    },
+    {
+      "epoch": 726.25,
+      "grad_norm": 1.2705445289611816,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 110390
+    },
+    {
+      "epoch": 726.3157894736842,
+      "grad_norm": 1.1773557662963867,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 110400
+    },
+    {
+      "epoch": 726.3815789473684,
+      "grad_norm": 1.0990701913833618,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 110410
+    },
+    {
+      "epoch": 726.4473684210526,
+      "grad_norm": 1.3985167741775513,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 110420
+    },
+    {
+      "epoch": 726.5131578947369,
+      "grad_norm": 1.0265055894851685,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 110430
+    },
+    {
+      "epoch": 726.578947368421,
+      "grad_norm": 0.6000217199325562,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 110440
+    },
+    {
+      "epoch": 726.6447368421053,
+      "grad_norm": 1.1134717464447021,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 110450
+    },
+    {
+      "epoch": 726.7105263157895,
+      "grad_norm": 1.1539477109909058,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 110460
+    },
+    {
+      "epoch": 726.7763157894736,
+      "grad_norm": 1.103757619857788,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 110470
+    },
+    {
+      "epoch": 726.8421052631579,
+      "grad_norm": 0.9701244235038757,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 110480
+    },
+    {
+      "epoch": 726.9078947368421,
+      "grad_norm": 1.1969598531723022,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 110490
+    },
+    {
+      "epoch": 726.9736842105264,
+      "grad_norm": 1.2793946266174316,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 110500
+    },
+    {
+      "epoch": 727.0394736842105,
+      "grad_norm": 1.4208569526672363,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 110510
+    },
+    {
+      "epoch": 727.1052631578947,
+      "grad_norm": 1.100865364074707,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 110520
+    },
+    {
+      "epoch": 727.171052631579,
+      "grad_norm": 1.3485084772109985,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 110530
+    },
+    {
+      "epoch": 727.2368421052631,
+      "grad_norm": 1.2594733238220215,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 110540
+    },
+    {
+      "epoch": 727.3026315789474,
+      "grad_norm": 1.0424339771270752,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 110550
+    },
+    {
+      "epoch": 727.3684210526316,
+      "grad_norm": 0.9489828944206238,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 110560
+    },
+    {
+      "epoch": 727.4342105263158,
+      "grad_norm": 1.0845290422439575,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 110570
+    },
+    {
+      "epoch": 727.5,
+      "grad_norm": 1.0237302780151367,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 110580
+    },
+    {
+      "epoch": 727.5657894736842,
+      "grad_norm": 0.6209089756011963,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 110590
+    },
+    {
+      "epoch": 727.6315789473684,
+      "grad_norm": 0.8243921995162964,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 110600
+    },
+    {
+      "epoch": 727.6973684210526,
+      "grad_norm": 1.0948792695999146,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 110610
+    },
+    {
+      "epoch": 727.7631578947369,
+      "grad_norm": 1.0297280550003052,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 110620
+    },
+    {
+      "epoch": 727.828947368421,
+      "grad_norm": 1.0834107398986816,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 110630
+    },
+    {
+      "epoch": 727.8947368421053,
+      "grad_norm": 1.2613449096679688,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 110640
+    },
+    {
+      "epoch": 727.9605263157895,
+      "grad_norm": 1.1736477613449097,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 110650
+    },
+    {
+      "epoch": 728.0263157894736,
+      "grad_norm": 0.9048214554786682,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 110660
+    },
+    {
+      "epoch": 728.0921052631579,
+      "grad_norm": 0.9805360436439514,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 110670
+    },
+    {
+      "epoch": 728.1578947368421,
+      "grad_norm": 0.9766744375228882,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 110680
+    },
+    {
+      "epoch": 728.2236842105264,
+      "grad_norm": 1.171241283416748,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 110690
+    },
+    {
+      "epoch": 728.2894736842105,
+      "grad_norm": 1.0495202541351318,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 110700
+    },
+    {
+      "epoch": 728.3552631578947,
+      "grad_norm": 1.1976003646850586,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 110710
+    },
+    {
+      "epoch": 728.421052631579,
+      "grad_norm": 1.323385238647461,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 110720
+    },
+    {
+      "epoch": 728.4868421052631,
+      "grad_norm": 0.8279552459716797,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 110730
+    },
+    {
+      "epoch": 728.5526315789474,
+      "grad_norm": 1.1523821353912354,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 110740
+    },
+    {
+      "epoch": 728.6184210526316,
+      "grad_norm": 1.1618291139602661,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 110750
+    },
+    {
+      "epoch": 728.6842105263158,
+      "grad_norm": 1.517456293106079,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 110760
+    },
+    {
+      "epoch": 728.75,
+      "grad_norm": 1.259458303451538,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 110770
+    },
+    {
+      "epoch": 728.8157894736842,
+      "grad_norm": 1.2021534442901611,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 110780
+    },
+    {
+      "epoch": 728.8815789473684,
+      "grad_norm": 0.7205750346183777,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 110790
+    },
+    {
+      "epoch": 728.9473684210526,
+      "grad_norm": 0.7980080246925354,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 110800
+    },
+    {
+      "epoch": 729.0131578947369,
+      "grad_norm": 1.2943599224090576,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 110810
+    },
+    {
+      "epoch": 729.078947368421,
+      "grad_norm": 1.2890563011169434,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 110820
+    },
+    {
+      "epoch": 729.1447368421053,
+      "grad_norm": 0.8411272764205933,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 110830
+    },
+    {
+      "epoch": 729.2105263157895,
+      "grad_norm": 1.1245949268341064,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 110840
+    },
+    {
+      "epoch": 729.2763157894736,
+      "grad_norm": 0.9203550815582275,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 110850
+    },
+    {
+      "epoch": 729.3421052631579,
+      "grad_norm": 1.0467205047607422,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 110860
+    },
+    {
+      "epoch": 729.4078947368421,
+      "grad_norm": 1.1412601470947266,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 110870
+    },
+    {
+      "epoch": 729.4736842105264,
+      "grad_norm": 0.8333537578582764,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 110880
+    },
+    {
+      "epoch": 729.5394736842105,
+      "grad_norm": 0.9673253893852234,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 110890
+    },
+    {
+      "epoch": 729.6052631578947,
+      "grad_norm": 0.9576820135116577,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 110900
+    },
+    {
+      "epoch": 729.671052631579,
+      "grad_norm": 1.1713014841079712,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 110910
+    },
+    {
+      "epoch": 729.7368421052631,
+      "grad_norm": 1.1785231828689575,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 110920
+    },
+    {
+      "epoch": 729.8026315789474,
+      "grad_norm": 1.2538374662399292,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 110930
+    },
+    {
+      "epoch": 729.8684210526316,
+      "grad_norm": 1.0921484231948853,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 110940
+    },
+    {
+      "epoch": 729.9342105263158,
+      "grad_norm": 1.2144492864608765,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 110950
+    },
+    {
+      "epoch": 730.0,
+      "grad_norm": 1.5016517639160156,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 110960
+    },
+    {
+      "epoch": 730.0657894736842,
+      "grad_norm": 1.3069437742233276,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 110970
+    },
+    {
+      "epoch": 730.1315789473684,
+      "grad_norm": 1.0793986320495605,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 110980
+    },
+    {
+      "epoch": 730.1973684210526,
+      "grad_norm": 0.8832784295082092,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 110990
+    },
+    {
+      "epoch": 730.2631578947369,
+      "grad_norm": 0.7944328188896179,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 111000
+    },
+    {
+      "epoch": 730.328947368421,
+      "grad_norm": 1.1215214729309082,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 111010
+    },
+    {
+      "epoch": 730.3947368421053,
+      "grad_norm": 1.332367181777954,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 111020
+    },
+    {
+      "epoch": 730.4605263157895,
+      "grad_norm": 1.3931553363800049,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 111030
+    },
+    {
+      "epoch": 730.5263157894736,
+      "grad_norm": 1.4220699071884155,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 111040
+    },
+    {
+      "epoch": 730.5921052631579,
+      "grad_norm": 1.1629732847213745,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 111050
+    },
+    {
+      "epoch": 730.6578947368421,
+      "grad_norm": 1.2412759065628052,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 111060
+    },
+    {
+      "epoch": 730.7236842105264,
+      "grad_norm": 1.228760838508606,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 111070
+    },
+    {
+      "epoch": 730.7894736842105,
+      "grad_norm": 1.2496362924575806,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 111080
+    },
+    {
+      "epoch": 730.8552631578947,
+      "grad_norm": 0.9673229455947876,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 111090
+    },
+    {
+      "epoch": 730.921052631579,
+      "grad_norm": 1.0482990741729736,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 111100
+    },
+    {
+      "epoch": 730.9868421052631,
+      "grad_norm": 1.0728371143341064,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 111110
+    },
+    {
+      "epoch": 731.0526315789474,
+      "grad_norm": 0.8858627080917358,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 111120
+    },
+    {
+      "epoch": 731.1184210526316,
+      "grad_norm": 1.158208966255188,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 111130
+    },
+    {
+      "epoch": 731.1842105263158,
+      "grad_norm": 1.145789623260498,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 111140
+    },
+    {
+      "epoch": 731.25,
+      "grad_norm": 1.142052412033081,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 111150
+    },
+    {
+      "epoch": 731.3157894736842,
+      "grad_norm": 1.0644676685333252,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 111160
+    },
+    {
+      "epoch": 731.3815789473684,
+      "grad_norm": 1.0623435974121094,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 111170
+    },
+    {
+      "epoch": 731.4473684210526,
+      "grad_norm": 1.053824543952942,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 111180
+    },
+    {
+      "epoch": 731.5131578947369,
+      "grad_norm": 1.2041244506835938,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 111190
+    },
+    {
+      "epoch": 731.578947368421,
+      "grad_norm": 1.0749359130859375,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 111200
+    },
+    {
+      "epoch": 731.6447368421053,
+      "grad_norm": 1.109894037246704,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 111210
+    },
+    {
+      "epoch": 731.7105263157895,
+      "grad_norm": 1.581756830215454,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 111220
+    },
+    {
+      "epoch": 731.7763157894736,
+      "grad_norm": 0.937778115272522,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 111230
+    },
+    {
+      "epoch": 731.8421052631579,
+      "grad_norm": 1.134636402130127,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 111240
+    },
+    {
+      "epoch": 731.9078947368421,
+      "grad_norm": 1.137024998664856,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 111250
+    },
+    {
+      "epoch": 731.9736842105264,
+      "grad_norm": 0.980426013469696,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 111260
+    },
+    {
+      "epoch": 732.0394736842105,
+      "grad_norm": 1.3249105215072632,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 111270
+    },
+    {
+      "epoch": 732.1052631578947,
+      "grad_norm": 1.0460824966430664,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 111280
+    },
+    {
+      "epoch": 732.171052631579,
+      "grad_norm": 1.2906813621520996,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 111290
+    },
+    {
+      "epoch": 732.2368421052631,
+      "grad_norm": 0.8858234286308289,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 111300
+    },
+    {
+      "epoch": 732.3026315789474,
+      "grad_norm": 1.170811653137207,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 111310
+    },
+    {
+      "epoch": 732.3684210526316,
+      "grad_norm": 1.3404710292816162,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 111320
+    },
+    {
+      "epoch": 732.4342105263158,
+      "grad_norm": 1.3341257572174072,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 111330
+    },
+    {
+      "epoch": 732.5,
+      "grad_norm": 1.3876317739486694,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 111340
+    },
+    {
+      "epoch": 732.5657894736842,
+      "grad_norm": 0.7830882668495178,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 111350
+    },
+    {
+      "epoch": 732.6315789473684,
+      "grad_norm": 1.436397910118103,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 111360
+    },
+    {
+      "epoch": 732.6973684210526,
+      "grad_norm": 0.7687216997146606,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 111370
+    },
+    {
+      "epoch": 732.7631578947369,
+      "grad_norm": 0.7303744554519653,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 111380
+    },
+    {
+      "epoch": 732.828947368421,
+      "grad_norm": 1.1733373403549194,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 111390
+    },
+    {
+      "epoch": 732.8947368421053,
+      "grad_norm": 0.9514874219894409,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 111400
+    },
+    {
+      "epoch": 732.9605263157895,
+      "grad_norm": 0.594891369342804,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 111410
+    },
+    {
+      "epoch": 733.0263157894736,
+      "grad_norm": 0.945779025554657,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 111420
+    },
+    {
+      "epoch": 733.0921052631579,
+      "grad_norm": 0.6920212507247925,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 111430
+    },
+    {
+      "epoch": 733.1578947368421,
+      "grad_norm": 0.796112060546875,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 111440
+    },
+    {
+      "epoch": 733.2236842105264,
+      "grad_norm": 0.9636341333389282,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 111450
+    },
+    {
+      "epoch": 733.2894736842105,
+      "grad_norm": 1.2237485647201538,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 111460
+    },
+    {
+      "epoch": 733.3552631578947,
+      "grad_norm": 1.8574937582015991,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 111470
+    },
+    {
+      "epoch": 733.421052631579,
+      "grad_norm": 1.866210699081421,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 111480
+    },
+    {
+      "epoch": 733.4868421052631,
+      "grad_norm": 1.6350382566452026,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 111490
+    },
+    {
+      "epoch": 733.5526315789474,
+      "grad_norm": 1.0349119901657104,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 111500
+    },
+    {
+      "epoch": 733.6184210526316,
+      "grad_norm": 1.5205327272415161,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 111510
+    },
+    {
+      "epoch": 733.6842105263158,
+      "grad_norm": 1.2245243787765503,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 111520
+    },
+    {
+      "epoch": 733.75,
+      "grad_norm": 1.118973731994629,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 111530
+    },
+    {
+      "epoch": 733.8157894736842,
+      "grad_norm": 1.170804500579834,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 111540
+    },
+    {
+      "epoch": 733.8815789473684,
+      "grad_norm": 1.130577802658081,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 111550
+    },
+    {
+      "epoch": 733.9473684210526,
+      "grad_norm": 1.4491939544677734,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 111560
+    },
+    {
+      "epoch": 734.0131578947369,
+      "grad_norm": 1.1310603618621826,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 111570
+    },
+    {
+      "epoch": 734.078947368421,
+      "grad_norm": 0.9815219640731812,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 111580
+    },
+    {
+      "epoch": 734.1447368421053,
+      "grad_norm": 1.254376769065857,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 111590
+    },
+    {
+      "epoch": 734.2105263157895,
+      "grad_norm": 0.9605775475502014,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 111600
+    },
+    {
+      "epoch": 734.2763157894736,
+      "grad_norm": 1.2415673732757568,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 111610
+    },
+    {
+      "epoch": 734.3421052631579,
+      "grad_norm": 1.124882698059082,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 111620
+    },
+    {
+      "epoch": 734.4078947368421,
+      "grad_norm": 0.7746277451515198,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 111630
+    },
+    {
+      "epoch": 734.4736842105264,
+      "grad_norm": 1.1267927885055542,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 111640
+    },
+    {
+      "epoch": 734.5394736842105,
+      "grad_norm": 0.798928439617157,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 111650
+    },
+    {
+      "epoch": 734.6052631578947,
+      "grad_norm": 1.297399878501892,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 111660
+    },
+    {
+      "epoch": 734.671052631579,
+      "grad_norm": 0.66205894947052,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 111670
+    },
+    {
+      "epoch": 734.7368421052631,
+      "grad_norm": 0.7813223004341125,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 111680
+    },
+    {
+      "epoch": 734.8026315789474,
+      "grad_norm": 0.9377914667129517,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 111690
+    },
+    {
+      "epoch": 734.8684210526316,
+      "grad_norm": 1.1338937282562256,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 111700
+    },
+    {
+      "epoch": 734.9342105263158,
+      "grad_norm": 1.4318599700927734,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 111710
+    },
+    {
+      "epoch": 735.0,
+      "grad_norm": 1.221707820892334,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 111720
+    },
+    {
+      "epoch": 735.0657894736842,
+      "grad_norm": 1.2542157173156738,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 111730
+    },
+    {
+      "epoch": 735.1315789473684,
+      "grad_norm": 1.02726149559021,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 111740
+    },
+    {
+      "epoch": 735.1973684210526,
+      "grad_norm": 0.7633141875267029,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 111750
+    },
+    {
+      "epoch": 735.2631578947369,
+      "grad_norm": 1.315930962562561,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 111760
+    },
+    {
+      "epoch": 735.328947368421,
+      "grad_norm": 1.430780291557312,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 111770
+    },
+    {
+      "epoch": 735.3947368421053,
+      "grad_norm": 1.1580687761306763,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 111780
+    },
+    {
+      "epoch": 735.4605263157895,
+      "grad_norm": 1.3785609006881714,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 111790
+    },
+    {
+      "epoch": 735.5263157894736,
+      "grad_norm": 1.3708674907684326,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 111800
+    },
+    {
+      "epoch": 735.5921052631579,
+      "grad_norm": 1.1644020080566406,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 111810
+    },
+    {
+      "epoch": 735.6578947368421,
+      "grad_norm": 1.1379879713058472,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 111820
+    },
+    {
+      "epoch": 735.7236842105264,
+      "grad_norm": 0.817884087562561,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 111830
+    },
+    {
+      "epoch": 735.7894736842105,
+      "grad_norm": 0.846930742263794,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 111840
+    },
+    {
+      "epoch": 735.8552631578947,
+      "grad_norm": 0.9146661162376404,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 111850
+    },
+    {
+      "epoch": 735.921052631579,
+      "grad_norm": 1.0802667140960693,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 111860
+    },
+    {
+      "epoch": 735.9868421052631,
+      "grad_norm": 0.7806511521339417,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 111870
+    },
+    {
+      "epoch": 736.0526315789474,
+      "grad_norm": 0.9969675540924072,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 111880
+    },
+    {
+      "epoch": 736.1184210526316,
+      "grad_norm": 1.1050714254379272,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 111890
+    },
+    {
+      "epoch": 736.1842105263158,
+      "grad_norm": 1.1047745943069458,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 111900
+    },
+    {
+      "epoch": 736.25,
+      "grad_norm": 1.2120915651321411,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 111910
+    },
+    {
+      "epoch": 736.3157894736842,
+      "grad_norm": 0.9838027358055115,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 111920
+    },
+    {
+      "epoch": 736.3815789473684,
+      "grad_norm": 1.0756142139434814,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 111930
+    },
+    {
+      "epoch": 736.4473684210526,
+      "grad_norm": 0.8673560619354248,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 111940
+    },
+    {
+      "epoch": 736.5131578947369,
+      "grad_norm": 0.8475706577301025,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 111950
+    },
+    {
+      "epoch": 736.578947368421,
+      "grad_norm": 0.9812502264976501,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 111960
+    },
+    {
+      "epoch": 736.6447368421053,
+      "grad_norm": 1.0416502952575684,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 111970
+    },
+    {
+      "epoch": 736.7105263157895,
+      "grad_norm": 1.1706211566925049,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 111980
+    },
+    {
+      "epoch": 736.7763157894736,
+      "grad_norm": 1.0563849210739136,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 111990
+    },
+    {
+      "epoch": 736.8421052631579,
+      "grad_norm": 1.2770580053329468,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 112000
+    },
+    {
+      "epoch": 736.9078947368421,
+      "grad_norm": 1.0203648805618286,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 112010
+    },
+    {
+      "epoch": 736.9736842105264,
+      "grad_norm": 0.937138557434082,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 112020
+    },
+    {
+      "epoch": 737.0394736842105,
+      "grad_norm": 1.422008752822876,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 112030
+    },
+    {
+      "epoch": 737.1052631578947,
+      "grad_norm": 1.0362669229507446,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 112040
+    },
+    {
+      "epoch": 737.171052631579,
+      "grad_norm": 0.7629768252372742,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 112050
+    },
+    {
+      "epoch": 737.2368421052631,
+      "grad_norm": 0.8714403510093689,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 112060
+    },
+    {
+      "epoch": 737.3026315789474,
+      "grad_norm": 1.0577564239501953,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 112070
+    },
+    {
+      "epoch": 737.3684210526316,
+      "grad_norm": 1.1749887466430664,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 112080
+    },
+    {
+      "epoch": 737.4342105263158,
+      "grad_norm": 1.071063756942749,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 112090
+    },
+    {
+      "epoch": 737.5,
+      "grad_norm": 1.1474461555480957,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 112100
+    },
+    {
+      "epoch": 737.5657894736842,
+      "grad_norm": 1.089721441268921,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 112110
+    },
+    {
+      "epoch": 737.6315789473684,
+      "grad_norm": 1.0660192966461182,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 112120
+    },
+    {
+      "epoch": 737.6973684210526,
+      "grad_norm": 0.9403685331344604,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 112130
+    },
+    {
+      "epoch": 737.7631578947369,
+      "grad_norm": 0.7262328267097473,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 112140
+    },
+    {
+      "epoch": 737.828947368421,
+      "grad_norm": 0.9046671986579895,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 112150
+    },
+    {
+      "epoch": 737.8947368421053,
+      "grad_norm": 1.0580425262451172,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 112160
+    },
+    {
+      "epoch": 737.9605263157895,
+      "grad_norm": 0.8231117129325867,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 112170
+    },
+    {
+      "epoch": 738.0263157894736,
+      "grad_norm": 0.9785894751548767,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 112180
+    },
+    {
+      "epoch": 738.0921052631579,
+      "grad_norm": 1.0782703161239624,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 112190
+    },
+    {
+      "epoch": 738.1578947368421,
+      "grad_norm": 1.0809273719787598,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 112200
+    },
+    {
+      "epoch": 738.2236842105264,
+      "grad_norm": 1.4004234075546265,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 112210
+    },
+    {
+      "epoch": 738.2894736842105,
+      "grad_norm": 1.5561052560806274,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 112220
+    },
+    {
+      "epoch": 738.3552631578947,
+      "grad_norm": 1.2708110809326172,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 112230
+    },
+    {
+      "epoch": 738.421052631579,
+      "grad_norm": 1.1724753379821777,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 112240
+    },
+    {
+      "epoch": 738.4868421052631,
+      "grad_norm": 1.3249428272247314,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 112250
+    },
+    {
+      "epoch": 738.5526315789474,
+      "grad_norm": 1.2601690292358398,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 112260
+    },
+    {
+      "epoch": 738.6184210526316,
+      "grad_norm": 1.4754626750946045,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 112270
+    },
+    {
+      "epoch": 738.6842105263158,
+      "grad_norm": 1.0365551710128784,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 112280
+    },
+    {
+      "epoch": 738.75,
+      "grad_norm": 1.195425033569336,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 112290
+    },
+    {
+      "epoch": 738.8157894736842,
+      "grad_norm": 0.8115053772926331,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 112300
+    },
+    {
+      "epoch": 738.8815789473684,
+      "grad_norm": 1.1184942722320557,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 112310
+    },
+    {
+      "epoch": 738.9473684210526,
+      "grad_norm": 0.7952884435653687,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 112320
+    },
+    {
+      "epoch": 739.0131578947369,
+      "grad_norm": 0.909673810005188,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 112330
+    },
+    {
+      "epoch": 739.078947368421,
+      "grad_norm": 0.9384917616844177,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 112340
+    },
+    {
+      "epoch": 739.1447368421053,
+      "grad_norm": 0.9624817967414856,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 112350
+    },
+    {
+      "epoch": 739.2105263157895,
+      "grad_norm": 1.2342602014541626,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 112360
+    },
+    {
+      "epoch": 739.2763157894736,
+      "grad_norm": 0.8785709142684937,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 112370
+    },
+    {
+      "epoch": 739.3421052631579,
+      "grad_norm": 1.1421648263931274,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 112380
+    },
+    {
+      "epoch": 739.4078947368421,
+      "grad_norm": 0.7869369387626648,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 112390
+    },
+    {
+      "epoch": 739.4736842105264,
+      "grad_norm": 0.9102826118469238,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 112400
+    },
+    {
+      "epoch": 739.5394736842105,
+      "grad_norm": 1.3796066045761108,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 112410
+    },
+    {
+      "epoch": 739.6052631578947,
+      "grad_norm": 0.9464783072471619,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 112420
+    },
+    {
+      "epoch": 739.671052631579,
+      "grad_norm": 1.226456642150879,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 112430
+    },
+    {
+      "epoch": 739.7368421052631,
+      "grad_norm": 0.8695907592773438,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 112440
+    },
+    {
+      "epoch": 739.8026315789474,
+      "grad_norm": 1.0778448581695557,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 112450
+    },
+    {
+      "epoch": 739.8684210526316,
+      "grad_norm": 1.033981204032898,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 112460
+    },
+    {
+      "epoch": 739.9342105263158,
+      "grad_norm": 1.3811094760894775,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 112470
+    },
+    {
+      "epoch": 740.0,
+      "grad_norm": 1.0558196306228638,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 112480
+    },
+    {
+      "epoch": 740.0657894736842,
+      "grad_norm": 1.4138654470443726,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 112490
+    },
+    {
+      "epoch": 740.1315789473684,
+      "grad_norm": 1.170053482055664,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 112500
+    },
+    {
+      "epoch": 740.1973684210526,
+      "grad_norm": 1.0176713466644287,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 112510
+    },
+    {
+      "epoch": 740.2631578947369,
+      "grad_norm": 1.2900663614273071,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 112520
+    },
+    {
+      "epoch": 740.328947368421,
+      "grad_norm": 1.0003573894500732,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 112530
+    },
+    {
+      "epoch": 740.3947368421053,
+      "grad_norm": 1.079323410987854,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 112540
+    },
+    {
+      "epoch": 740.4605263157895,
+      "grad_norm": 0.878826916217804,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 112550
+    },
+    {
+      "epoch": 740.5263157894736,
+      "grad_norm": 0.9983590841293335,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 112560
+    },
+    {
+      "epoch": 740.5921052631579,
+      "grad_norm": 1.1599466800689697,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 112570
+    },
+    {
+      "epoch": 740.6578947368421,
+      "grad_norm": 1.4644721746444702,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 112580
+    },
+    {
+      "epoch": 740.7236842105264,
+      "grad_norm": 1.0327931642532349,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 112590
+    },
+    {
+      "epoch": 740.7894736842105,
+      "grad_norm": 1.1933623552322388,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 112600
+    },
+    {
+      "epoch": 740.8552631578947,
+      "grad_norm": 1.1514612436294556,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 112610
+    },
+    {
+      "epoch": 740.921052631579,
+      "grad_norm": 1.074176549911499,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 112620
+    },
+    {
+      "epoch": 740.9868421052631,
+      "grad_norm": 1.3565348386764526,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 112630
+    },
+    {
+      "epoch": 741.0526315789474,
+      "grad_norm": 1.487715721130371,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 112640
+    },
+    {
+      "epoch": 741.1184210526316,
+      "grad_norm": 1.228196620941162,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 112650
+    },
+    {
+      "epoch": 741.1842105263158,
+      "grad_norm": 0.9726716876029968,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 112660
+    },
+    {
+      "epoch": 741.25,
+      "grad_norm": 1.1228713989257812,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 112670
+    },
+    {
+      "epoch": 741.3157894736842,
+      "grad_norm": 1.3492858409881592,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 112680
+    },
+    {
+      "epoch": 741.3815789473684,
+      "grad_norm": 1.2448452711105347,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 112690
+    },
+    {
+      "epoch": 741.4473684210526,
+      "grad_norm": 1.2772343158721924,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 112700
+    },
+    {
+      "epoch": 741.5131578947369,
+      "grad_norm": 0.9293051958084106,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 112710
+    },
+    {
+      "epoch": 741.578947368421,
+      "grad_norm": 0.7797158360481262,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 112720
+    },
+    {
+      "epoch": 741.6447368421053,
+      "grad_norm": 1.2524482011795044,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 112730
+    },
+    {
+      "epoch": 741.7105263157895,
+      "grad_norm": 1.0010104179382324,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 112740
+    },
+    {
+      "epoch": 741.7763157894736,
+      "grad_norm": 0.8157704472541809,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 112750
+    },
+    {
+      "epoch": 741.8421052631579,
+      "grad_norm": 0.7338435649871826,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 112760
+    },
+    {
+      "epoch": 741.9078947368421,
+      "grad_norm": 0.8441011905670166,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 112770
+    },
+    {
+      "epoch": 741.9736842105264,
+      "grad_norm": 0.5987163186073303,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 112780
+    },
+    {
+      "epoch": 742.0394736842105,
+      "grad_norm": 0.5863604545593262,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 112790
+    },
+    {
+      "epoch": 742.1052631578947,
+      "grad_norm": 1.1553442478179932,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 112800
+    },
+    {
+      "epoch": 742.171052631579,
+      "grad_norm": 1.0437456369400024,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 112810
+    },
+    {
+      "epoch": 742.2368421052631,
+      "grad_norm": 1.0978097915649414,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 112820
+    },
+    {
+      "epoch": 742.3026315789474,
+      "grad_norm": 1.2839778661727905,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 112830
+    },
+    {
+      "epoch": 742.3684210526316,
+      "grad_norm": 0.8955215215682983,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 112840
+    },
+    {
+      "epoch": 742.4342105263158,
+      "grad_norm": 0.9545695781707764,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 112850
+    },
+    {
+      "epoch": 742.5,
+      "grad_norm": 0.9076725840568542,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 112860
+    },
+    {
+      "epoch": 742.5657894736842,
+      "grad_norm": 0.9868484735488892,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 112870
+    },
+    {
+      "epoch": 742.6315789473684,
+      "grad_norm": 0.9689452052116394,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 112880
+    },
+    {
+      "epoch": 742.6973684210526,
+      "grad_norm": 0.9611213207244873,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 112890
+    },
+    {
+      "epoch": 742.7631578947369,
+      "grad_norm": 1.3017029762268066,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 112900
+    },
+    {
+      "epoch": 742.828947368421,
+      "grad_norm": 1.010025978088379,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 112910
+    },
+    {
+      "epoch": 742.8947368421053,
+      "grad_norm": 1.1082737445831299,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 112920
+    },
+    {
+      "epoch": 742.9605263157895,
+      "grad_norm": 1.2095134258270264,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 112930
+    },
+    {
+      "epoch": 743.0263157894736,
+      "grad_norm": 0.7735888361930847,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 112940
+    },
+    {
+      "epoch": 743.0921052631579,
+      "grad_norm": 0.8232780694961548,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 112950
+    },
+    {
+      "epoch": 743.1578947368421,
+      "grad_norm": 1.1837214231491089,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 112960
+    },
+    {
+      "epoch": 743.2236842105264,
+      "grad_norm": 1.0164154767990112,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 112970
+    },
+    {
+      "epoch": 743.2894736842105,
+      "grad_norm": 0.9214721918106079,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 112980
+    },
+    {
+      "epoch": 743.3552631578947,
+      "grad_norm": 1.178781270980835,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 112990
+    },
+    {
+      "epoch": 743.421052631579,
+      "grad_norm": 0.8632880449295044,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 113000
+    },
+    {
+      "epoch": 743.4868421052631,
+      "grad_norm": 0.8442752957344055,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 113010
+    },
+    {
+      "epoch": 743.5526315789474,
+      "grad_norm": 1.3869895935058594,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 113020
+    },
+    {
+      "epoch": 743.6184210526316,
+      "grad_norm": 0.642612636089325,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 113030
+    },
+    {
+      "epoch": 743.6842105263158,
+      "grad_norm": 0.9901149868965149,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 113040
+    },
+    {
+      "epoch": 743.75,
+      "grad_norm": 0.6913559436798096,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 113050
+    },
+    {
+      "epoch": 743.8157894736842,
+      "grad_norm": 1.3106902837753296,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 113060
+    },
+    {
+      "epoch": 743.8815789473684,
+      "grad_norm": 0.6828902959823608,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 113070
+    },
+    {
+      "epoch": 743.9473684210526,
+      "grad_norm": 1.2042633295059204,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 113080
+    },
+    {
+      "epoch": 744.0131578947369,
+      "grad_norm": 1.0320390462875366,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 113090
+    },
+    {
+      "epoch": 744.078947368421,
+      "grad_norm": 1.210471272468567,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 113100
+    },
+    {
+      "epoch": 744.1447368421053,
+      "grad_norm": 0.7670679092407227,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 113110
+    },
+    {
+      "epoch": 744.2105263157895,
+      "grad_norm": 1.2954962253570557,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 113120
+    },
+    {
+      "epoch": 744.2763157894736,
+      "grad_norm": 1.0254065990447998,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 113130
+    },
+    {
+      "epoch": 744.3421052631579,
+      "grad_norm": 0.9328817129135132,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 113140
+    },
+    {
+      "epoch": 744.4078947368421,
+      "grad_norm": 1.0787217617034912,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 113150
+    },
+    {
+      "epoch": 744.4736842105264,
+      "grad_norm": 1.0308198928833008,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 113160
+    },
+    {
+      "epoch": 744.5394736842105,
+      "grad_norm": 1.011286973953247,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 113170
+    },
+    {
+      "epoch": 744.6052631578947,
+      "grad_norm": 0.8020488023757935,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 113180
+    },
+    {
+      "epoch": 744.671052631579,
+      "grad_norm": 1.1139671802520752,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 113190
+    },
+    {
+      "epoch": 744.7368421052631,
+      "grad_norm": 1.1352884769439697,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 113200
+    },
+    {
+      "epoch": 744.8026315789474,
+      "grad_norm": 0.8926935791969299,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 113210
+    },
+    {
+      "epoch": 744.8684210526316,
+      "grad_norm": 1.2045756578445435,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 113220
+    },
+    {
+      "epoch": 744.9342105263158,
+      "grad_norm": 1.2373918294906616,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 113230
+    },
+    {
+      "epoch": 745.0,
+      "grad_norm": 0.9970014095306396,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 113240
+    },
+    {
+      "epoch": 745.0657894736842,
+      "grad_norm": 0.9644217491149902,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 113250
+    },
+    {
+      "epoch": 745.1315789473684,
+      "grad_norm": 1.1484285593032837,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 113260
+    },
+    {
+      "epoch": 745.1973684210526,
+      "grad_norm": 0.9639143347740173,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 113270
+    },
+    {
+      "epoch": 745.2631578947369,
+      "grad_norm": 1.1494964361190796,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 113280
+    },
+    {
+      "epoch": 745.328947368421,
+      "grad_norm": 1.025691032409668,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 113290
+    },
+    {
+      "epoch": 745.3947368421053,
+      "grad_norm": 1.154371976852417,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 113300
+    },
+    {
+      "epoch": 745.4605263157895,
+      "grad_norm": 1.3313137292861938,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 113310
+    },
+    {
+      "epoch": 745.5263157894736,
+      "grad_norm": 1.094545841217041,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 113320
+    },
+    {
+      "epoch": 745.5921052631579,
+      "grad_norm": 1.2096467018127441,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 113330
+    },
+    {
+      "epoch": 745.6578947368421,
+      "grad_norm": 1.1005749702453613,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 113340
+    },
+    {
+      "epoch": 745.7236842105264,
+      "grad_norm": 1.1122779846191406,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 113350
+    },
+    {
+      "epoch": 745.7894736842105,
+      "grad_norm": 1.2181123495101929,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 113360
+    },
+    {
+      "epoch": 745.8552631578947,
+      "grad_norm": 1.4004229307174683,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 113370
+    },
+    {
+      "epoch": 745.921052631579,
+      "grad_norm": 0.9129208922386169,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 113380
+    },
+    {
+      "epoch": 745.9868421052631,
+      "grad_norm": 1.1694635152816772,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 113390
+    },
+    {
+      "epoch": 746.0526315789474,
+      "grad_norm": 1.1359633207321167,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 113400
+    },
+    {
+      "epoch": 746.1184210526316,
+      "grad_norm": 1.1821686029434204,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 113410
+    },
+    {
+      "epoch": 746.1842105263158,
+      "grad_norm": 1.0882134437561035,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 113420
+    },
+    {
+      "epoch": 746.25,
+      "grad_norm": 1.2553226947784424,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 113430
+    },
+    {
+      "epoch": 746.3157894736842,
+      "grad_norm": 0.9115428924560547,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 113440
+    },
+    {
+      "epoch": 746.3815789473684,
+      "grad_norm": 1.281915307044983,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 113450
+    },
+    {
+      "epoch": 746.4473684210526,
+      "grad_norm": 1.3433202505111694,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 113460
+    },
+    {
+      "epoch": 746.5131578947369,
+      "grad_norm": 1.5758907794952393,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 113470
+    },
+    {
+      "epoch": 746.578947368421,
+      "grad_norm": 1.3677822351455688,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 113480
+    },
+    {
+      "epoch": 746.6447368421053,
+      "grad_norm": 1.343830943107605,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 113490
+    },
+    {
+      "epoch": 746.7105263157895,
+      "grad_norm": 1.2050989866256714,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 113500
+    },
+    {
+      "epoch": 746.7763157894736,
+      "grad_norm": 0.8132672905921936,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 113510
+    },
+    {
+      "epoch": 746.8421052631579,
+      "grad_norm": 1.0796236991882324,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 113520
+    },
+    {
+      "epoch": 746.9078947368421,
+      "grad_norm": 0.9713119268417358,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 113530
+    },
+    {
+      "epoch": 746.9736842105264,
+      "grad_norm": 1.068225383758545,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 113540
+    },
+    {
+      "epoch": 747.0394736842105,
+      "grad_norm": 0.9908331036567688,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 113550
+    },
+    {
+      "epoch": 747.1052631578947,
+      "grad_norm": 0.8865001797676086,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 113560
+    },
+    {
+      "epoch": 747.171052631579,
+      "grad_norm": 0.9770333766937256,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 113570
+    },
+    {
+      "epoch": 747.2368421052631,
+      "grad_norm": 0.9718663692474365,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 113580
+    },
+    {
+      "epoch": 747.3026315789474,
+      "grad_norm": 0.9248496890068054,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 113590
+    },
+    {
+      "epoch": 747.3684210526316,
+      "grad_norm": 0.9006205797195435,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 113600
+    },
+    {
+      "epoch": 747.4342105263158,
+      "grad_norm": 1.0532220602035522,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 113610
+    },
+    {
+      "epoch": 747.5,
+      "grad_norm": 1.0480901002883911,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 113620
+    },
+    {
+      "epoch": 747.5657894736842,
+      "grad_norm": 1.1913716793060303,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 113630
+    },
+    {
+      "epoch": 747.6315789473684,
+      "grad_norm": 1.3170071840286255,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 113640
+    },
+    {
+      "epoch": 747.6973684210526,
+      "grad_norm": 0.8311954140663147,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 113650
+    },
+    {
+      "epoch": 747.7631578947369,
+      "grad_norm": 1.102514624595642,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 113660
+    },
+    {
+      "epoch": 747.828947368421,
+      "grad_norm": 1.3776170015335083,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 113670
+    },
+    {
+      "epoch": 747.8947368421053,
+      "grad_norm": 1.1094131469726562,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 113680
+    },
+    {
+      "epoch": 747.9605263157895,
+      "grad_norm": 1.3613204956054688,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 113690
+    },
+    {
+      "epoch": 748.0263157894736,
+      "grad_norm": 1.2817715406417847,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 113700
+    },
+    {
+      "epoch": 748.0921052631579,
+      "grad_norm": 1.1699529886245728,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 113710
+    },
+    {
+      "epoch": 748.1578947368421,
+      "grad_norm": 1.2629177570343018,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 113720
+    },
+    {
+      "epoch": 748.2236842105264,
+      "grad_norm": 1.0418188571929932,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 113730
+    },
+    {
+      "epoch": 748.2894736842105,
+      "grad_norm": 1.2210304737091064,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 113740
+    },
+    {
+      "epoch": 748.3552631578947,
+      "grad_norm": 0.6313312649726868,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 113750
+    },
+    {
+      "epoch": 748.421052631579,
+      "grad_norm": 1.2036254405975342,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 113760
+    },
+    {
+      "epoch": 748.4868421052631,
+      "grad_norm": 0.9804714918136597,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 113770
+    },
+    {
+      "epoch": 748.5526315789474,
+      "grad_norm": 0.9679644107818604,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 113780
+    },
+    {
+      "epoch": 748.6184210526316,
+      "grad_norm": 0.7930268049240112,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 113790
+    },
+    {
+      "epoch": 748.6842105263158,
+      "grad_norm": 1.0785940885543823,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 113800
+    },
+    {
+      "epoch": 748.75,
+      "grad_norm": 1.1710313558578491,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 113810
+    },
+    {
+      "epoch": 748.8157894736842,
+      "grad_norm": 0.9718161225318909,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 113820
+    },
+    {
+      "epoch": 748.8815789473684,
+      "grad_norm": 1.169264316558838,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 113830
+    },
+    {
+      "epoch": 748.9473684210526,
+      "grad_norm": 0.9664254188537598,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 113840
+    },
+    {
+      "epoch": 749.0131578947369,
+      "grad_norm": 0.7254254817962646,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 113850
+    },
+    {
+      "epoch": 749.078947368421,
+      "grad_norm": 0.9984303116798401,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 113860
+    },
+    {
+      "epoch": 749.1447368421053,
+      "grad_norm": 1.2939634323120117,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 113870
+    },
+    {
+      "epoch": 749.2105263157895,
+      "grad_norm": 1.12336003780365,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 113880
+    },
+    {
+      "epoch": 749.2763157894736,
+      "grad_norm": 1.6549626588821411,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 113890
+    },
+    {
+      "epoch": 749.3421052631579,
+      "grad_norm": 1.5198620557785034,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 113900
+    },
+    {
+      "epoch": 749.4078947368421,
+      "grad_norm": 1.2456414699554443,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 113910
+    },
+    {
+      "epoch": 749.4736842105264,
+      "grad_norm": 1.2373472452163696,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 113920
+    },
+    {
+      "epoch": 749.5394736842105,
+      "grad_norm": 1.309753179550171,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 113930
+    },
+    {
+      "epoch": 749.6052631578947,
+      "grad_norm": 0.9735221862792969,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 113940
+    },
+    {
+      "epoch": 749.671052631579,
+      "grad_norm": 0.6393201351165771,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 113950
+    },
+    {
+      "epoch": 749.7368421052631,
+      "grad_norm": 1.0009740591049194,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 113960
+    },
+    {
+      "epoch": 749.8026315789474,
+      "grad_norm": 0.8014363646507263,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 113970
+    },
+    {
+      "epoch": 749.8684210526316,
+      "grad_norm": 0.9935765266418457,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 113980
+    },
+    {
+      "epoch": 749.9342105263158,
+      "grad_norm": 1.1239501237869263,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 113990
+    },
+    {
+      "epoch": 750.0,
+      "grad_norm": 1.2619990110397339,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 114000
+    },
+    {
+      "epoch": 750.0657894736842,
+      "grad_norm": 0.8633625507354736,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 114010
+    },
+    {
+      "epoch": 750.1315789473684,
+      "grad_norm": 1.195512294769287,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 114020
+    },
+    {
+      "epoch": 750.1973684210526,
+      "grad_norm": 1.052501916885376,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 114030
+    },
+    {
+      "epoch": 750.2631578947369,
+      "grad_norm": 0.899624228477478,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 114040
+    },
+    {
+      "epoch": 750.328947368421,
+      "grad_norm": 0.7613295316696167,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 114050
+    },
+    {
+      "epoch": 750.3947368421053,
+      "grad_norm": 1.224990725517273,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 114060
+    },
+    {
+      "epoch": 750.4605263157895,
+      "grad_norm": 1.157004475593567,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 114070
+    },
+    {
+      "epoch": 750.5263157894736,
+      "grad_norm": 1.021988034248352,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 114080
+    },
+    {
+      "epoch": 750.5921052631579,
+      "grad_norm": 1.2240087985992432,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 114090
+    },
+    {
+      "epoch": 750.6578947368421,
+      "grad_norm": 0.8899611830711365,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 114100
+    },
+    {
+      "epoch": 750.7236842105264,
+      "grad_norm": 1.063750982284546,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 114110
+    },
+    {
+      "epoch": 750.7894736842105,
+      "grad_norm": 0.6514599919319153,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 114120
+    },
+    {
+      "epoch": 750.8552631578947,
+      "grad_norm": 0.8705139756202698,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 114130
+    },
+    {
+      "epoch": 750.921052631579,
+      "grad_norm": 1.2214490175247192,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 114140
+    },
+    {
+      "epoch": 750.9868421052631,
+      "grad_norm": 1.1715788841247559,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 114150
+    },
+    {
+      "epoch": 751.0526315789474,
+      "grad_norm": 1.076686143875122,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 114160
+    },
+    {
+      "epoch": 751.1184210526316,
+      "grad_norm": 1.0638906955718994,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 114170
+    },
+    {
+      "epoch": 751.1842105263158,
+      "grad_norm": 0.9506366848945618,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 114180
+    },
+    {
+      "epoch": 751.25,
+      "grad_norm": 0.9974748492240906,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 114190
+    },
+    {
+      "epoch": 751.3157894736842,
+      "grad_norm": 1.044983983039856,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 114200
+    },
+    {
+      "epoch": 751.3815789473684,
+      "grad_norm": 0.7726526856422424,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 114210
+    },
+    {
+      "epoch": 751.4473684210526,
+      "grad_norm": 0.9929822683334351,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 114220
+    },
+    {
+      "epoch": 751.5131578947369,
+      "grad_norm": 1.2815465927124023,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 114230
+    },
+    {
+      "epoch": 751.578947368421,
+      "grad_norm": 1.2115155458450317,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 114240
+    },
+    {
+      "epoch": 751.6447368421053,
+      "grad_norm": 1.184339165687561,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 114250
+    },
+    {
+      "epoch": 751.7105263157895,
+      "grad_norm": 1.2363086938858032,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 114260
+    },
+    {
+      "epoch": 751.7763157894736,
+      "grad_norm": 1.2934776544570923,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 114270
+    },
+    {
+      "epoch": 751.8421052631579,
+      "grad_norm": 1.1832584142684937,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 114280
+    },
+    {
+      "epoch": 751.9078947368421,
+      "grad_norm": 1.02647066116333,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 114290
+    },
+    {
+      "epoch": 751.9736842105264,
+      "grad_norm": 1.5524499416351318,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 114300
+    },
+    {
+      "epoch": 752.0394736842105,
+      "grad_norm": 1.193410038948059,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 114310
+    },
+    {
+      "epoch": 752.1052631578947,
+      "grad_norm": 1.1393669843673706,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 114320
+    },
+    {
+      "epoch": 752.171052631579,
+      "grad_norm": 0.7503915429115295,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 114330
+    },
+    {
+      "epoch": 752.2368421052631,
+      "grad_norm": 1.088619351387024,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 114340
+    },
+    {
+      "epoch": 752.3026315789474,
+      "grad_norm": 0.9867507815361023,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 114350
+    },
+    {
+      "epoch": 752.3684210526316,
+      "grad_norm": 1.2592015266418457,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 114360
+    },
+    {
+      "epoch": 752.4342105263158,
+      "grad_norm": 1.365531086921692,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 114370
+    },
+    {
+      "epoch": 752.5,
+      "grad_norm": 1.318558931350708,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 114380
+    },
+    {
+      "epoch": 752.5657894736842,
+      "grad_norm": 0.9499292373657227,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 114390
+    },
+    {
+      "epoch": 752.6315789473684,
+      "grad_norm": 1.3132816553115845,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 114400
+    },
+    {
+      "epoch": 752.6973684210526,
+      "grad_norm": 1.6993619203567505,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 114410
+    },
+    {
+      "epoch": 752.7631578947369,
+      "grad_norm": 1.5415050983428955,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 114420
+    },
+    {
+      "epoch": 752.828947368421,
+      "grad_norm": 1.364233136177063,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 114430
+    },
+    {
+      "epoch": 752.8947368421053,
+      "grad_norm": 1.6695607900619507,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 114440
+    },
+    {
+      "epoch": 752.9605263157895,
+      "grad_norm": 1.379301905632019,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 114450
+    },
+    {
+      "epoch": 753.0263157894736,
+      "grad_norm": 1.2977728843688965,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 114460
+    },
+    {
+      "epoch": 753.0921052631579,
+      "grad_norm": 1.069703459739685,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 114470
+    },
+    {
+      "epoch": 753.1578947368421,
+      "grad_norm": 1.064315915107727,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 114480
+    },
+    {
+      "epoch": 753.2236842105264,
+      "grad_norm": 1.2300894260406494,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 114490
+    },
+    {
+      "epoch": 753.2894736842105,
+      "grad_norm": 1.0416491031646729,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 114500
+    },
+    {
+      "epoch": 753.3552631578947,
+      "grad_norm": 1.2523179054260254,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 114510
+    },
+    {
+      "epoch": 753.421052631579,
+      "grad_norm": 1.2075591087341309,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 114520
+    },
+    {
+      "epoch": 753.4868421052631,
+      "grad_norm": 1.3077597618103027,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 114530
+    },
+    {
+      "epoch": 753.5526315789474,
+      "grad_norm": 0.7280270457267761,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 114540
+    },
+    {
+      "epoch": 753.6184210526316,
+      "grad_norm": 1.2432559728622437,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 114550
+    },
+    {
+      "epoch": 753.6842105263158,
+      "grad_norm": 1.0678569078445435,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 114560
+    },
+    {
+      "epoch": 753.75,
+      "grad_norm": 0.9261232614517212,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 114570
+    },
+    {
+      "epoch": 753.8157894736842,
+      "grad_norm": 1.2366281747817993,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 114580
+    },
+    {
+      "epoch": 753.8815789473684,
+      "grad_norm": 1.1832107305526733,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 114590
+    },
+    {
+      "epoch": 753.9473684210526,
+      "grad_norm": 1.4531916379928589,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 114600
+    },
+    {
+      "epoch": 754.0131578947369,
+      "grad_norm": 1.1988731622695923,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 114610
+    },
+    {
+      "epoch": 754.078947368421,
+      "grad_norm": 1.1996492147445679,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 114620
+    },
+    {
+      "epoch": 754.1447368421053,
+      "grad_norm": 1.2475368976593018,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 114630
+    },
+    {
+      "epoch": 754.2105263157895,
+      "grad_norm": 0.7737396359443665,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 114640
+    },
+    {
+      "epoch": 754.2763157894736,
+      "grad_norm": 1.1298565864562988,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 114650
+    },
+    {
+      "epoch": 754.3421052631579,
+      "grad_norm": 0.8025405406951904,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 114660
+    },
+    {
+      "epoch": 754.4078947368421,
+      "grad_norm": 1.1314117908477783,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 114670
+    },
+    {
+      "epoch": 754.4736842105264,
+      "grad_norm": 1.032278299331665,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 114680
+    },
+    {
+      "epoch": 754.5394736842105,
+      "grad_norm": 1.1019865274429321,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 114690
+    },
+    {
+      "epoch": 754.6052631578947,
+      "grad_norm": 1.3335840702056885,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 114700
+    },
+    {
+      "epoch": 754.671052631579,
+      "grad_norm": 0.6844580769538879,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 114710
+    },
+    {
+      "epoch": 754.7368421052631,
+      "grad_norm": 0.8246498107910156,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 114720
+    },
+    {
+      "epoch": 754.8026315789474,
+      "grad_norm": 1.2790837287902832,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 114730
+    },
+    {
+      "epoch": 754.8684210526316,
+      "grad_norm": 1.169844388961792,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 114740
+    },
+    {
+      "epoch": 754.9342105263158,
+      "grad_norm": 0.9610524773597717,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 114750
+    },
+    {
+      "epoch": 755.0,
+      "grad_norm": 0.9811365604400635,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 114760
+    },
+    {
+      "epoch": 755.0657894736842,
+      "grad_norm": 1.2406882047653198,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 114770
+    },
+    {
+      "epoch": 755.1315789473684,
+      "grad_norm": 1.348677635192871,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 114780
+    },
+    {
+      "epoch": 755.1973684210526,
+      "grad_norm": 1.0763145685195923,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 114790
+    },
+    {
+      "epoch": 755.2631578947369,
+      "grad_norm": 0.8977418541908264,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 114800
+    },
+    {
+      "epoch": 755.328947368421,
+      "grad_norm": 1.2123111486434937,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 114810
+    },
+    {
+      "epoch": 755.3947368421053,
+      "grad_norm": 1.1371431350708008,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 114820
+    },
+    {
+      "epoch": 755.4605263157895,
+      "grad_norm": 1.1556872129440308,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 114830
+    },
+    {
+      "epoch": 755.5263157894736,
+      "grad_norm": 1.5574454069137573,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 114840
+    },
+    {
+      "epoch": 755.5921052631579,
+      "grad_norm": 1.1535435914993286,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 114850
+    },
+    {
+      "epoch": 755.6578947368421,
+      "grad_norm": 1.2058866024017334,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 114860
+    },
+    {
+      "epoch": 755.7236842105264,
+      "grad_norm": 0.8264251947402954,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 114870
+    },
+    {
+      "epoch": 755.7894736842105,
+      "grad_norm": 1.1852251291275024,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 114880
+    },
+    {
+      "epoch": 755.8552631578947,
+      "grad_norm": 0.5516109466552734,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 114890
+    },
+    {
+      "epoch": 755.921052631579,
+      "grad_norm": 0.8325181007385254,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 114900
+    },
+    {
+      "epoch": 755.9868421052631,
+      "grad_norm": 0.7445228695869446,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 114910
+    },
+    {
+      "epoch": 756.0526315789474,
+      "grad_norm": 1.4654862880706787,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 114920
+    },
+    {
+      "epoch": 756.1184210526316,
+      "grad_norm": 1.1833246946334839,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 114930
+    },
+    {
+      "epoch": 756.1842105263158,
+      "grad_norm": 0.8253543972969055,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 114940
+    },
+    {
+      "epoch": 756.25,
+      "grad_norm": 1.0825392007827759,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 114950
+    },
+    {
+      "epoch": 756.3157894736842,
+      "grad_norm": 0.9824175834655762,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 114960
+    },
+    {
+      "epoch": 756.3815789473684,
+      "grad_norm": 1.0814141035079956,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 114970
+    },
+    {
+      "epoch": 756.4473684210526,
+      "grad_norm": 0.9042399525642395,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 114980
+    },
+    {
+      "epoch": 756.5131578947369,
+      "grad_norm": 1.2757545709609985,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 114990
+    },
+    {
+      "epoch": 756.578947368421,
+      "grad_norm": 1.336796760559082,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 115000
+    },
+    {
+      "epoch": 756.6447368421053,
+      "grad_norm": 1.2338722944259644,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 115010
+    },
+    {
+      "epoch": 756.7105263157895,
+      "grad_norm": 0.8805379271507263,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 115020
+    },
+    {
+      "epoch": 756.7763157894736,
+      "grad_norm": 1.1899020671844482,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 115030
+    },
+    {
+      "epoch": 756.8421052631579,
+      "grad_norm": 1.0832873582839966,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 115040
+    },
+    {
+      "epoch": 756.9078947368421,
+      "grad_norm": 0.9258546233177185,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 115050
+    },
+    {
+      "epoch": 756.9736842105264,
+      "grad_norm": 0.8678163290023804,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 115060
+    },
+    {
+      "epoch": 757.0394736842105,
+      "grad_norm": 0.9894230365753174,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 115070
+    },
+    {
+      "epoch": 757.1052631578947,
+      "grad_norm": 1.470314383506775,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 115080
+    },
+    {
+      "epoch": 757.171052631579,
+      "grad_norm": 1.0584030151367188,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 115090
+    },
+    {
+      "epoch": 757.2368421052631,
+      "grad_norm": 1.0429503917694092,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 115100
+    },
+    {
+      "epoch": 757.3026315789474,
+      "grad_norm": 1.2286428213119507,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 115110
+    },
+    {
+      "epoch": 757.3684210526316,
+      "grad_norm": 1.1292539834976196,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 115120
+    },
+    {
+      "epoch": 757.4342105263158,
+      "grad_norm": 0.7512161135673523,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 115130
+    },
+    {
+      "epoch": 757.5,
+      "grad_norm": 1.228721261024475,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 115140
+    },
+    {
+      "epoch": 757.5657894736842,
+      "grad_norm": 1.2214051485061646,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 115150
+    },
+    {
+      "epoch": 757.6315789473684,
+      "grad_norm": 1.1014398336410522,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 115160
+    },
+    {
+      "epoch": 757.6973684210526,
+      "grad_norm": 0.9443309307098389,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 115170
+    },
+    {
+      "epoch": 757.7631578947369,
+      "grad_norm": 0.7845607995986938,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 115180
+    },
+    {
+      "epoch": 757.828947368421,
+      "grad_norm": 1.0893877744674683,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 115190
+    },
+    {
+      "epoch": 757.8947368421053,
+      "grad_norm": 0.7762365937232971,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 115200
+    },
+    {
+      "epoch": 757.9605263157895,
+      "grad_norm": 0.9474076628684998,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 115210
+    },
+    {
+      "epoch": 758.0263157894736,
+      "grad_norm": 1.041471242904663,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 115220
+    },
+    {
+      "epoch": 758.0921052631579,
+      "grad_norm": 1.2159472703933716,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 115230
+    },
+    {
+      "epoch": 758.1578947368421,
+      "grad_norm": 1.0471974611282349,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 115240
+    },
+    {
+      "epoch": 758.2236842105264,
+      "grad_norm": 0.8706479072570801,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 115250
+    },
+    {
+      "epoch": 758.2894736842105,
+      "grad_norm": 0.847398042678833,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 115260
+    },
+    {
+      "epoch": 758.3552631578947,
+      "grad_norm": 1.139387845993042,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 115270
+    },
+    {
+      "epoch": 758.421052631579,
+      "grad_norm": 1.2678453922271729,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 115280
+    },
+    {
+      "epoch": 758.4868421052631,
+      "grad_norm": 1.1725306510925293,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 115290
+    },
+    {
+      "epoch": 758.5526315789474,
+      "grad_norm": 1.2793757915496826,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 115300
+    },
+    {
+      "epoch": 758.6184210526316,
+      "grad_norm": 1.028630018234253,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 115310
+    },
+    {
+      "epoch": 758.6842105263158,
+      "grad_norm": 1.1012976169586182,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 115320
+    },
+    {
+      "epoch": 758.75,
+      "grad_norm": 0.8987807035446167,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 115330
+    },
+    {
+      "epoch": 758.8157894736842,
+      "grad_norm": 1.493779182434082,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 115340
+    },
+    {
+      "epoch": 758.8815789473684,
+      "grad_norm": 1.4419828653335571,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 115350
+    },
+    {
+      "epoch": 758.9473684210526,
+      "grad_norm": 1.4344781637191772,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 115360
+    },
+    {
+      "epoch": 759.0131578947369,
+      "grad_norm": 0.9399645924568176,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 115370
+    },
+    {
+      "epoch": 759.078947368421,
+      "grad_norm": 0.8678160309791565,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 115380
+    },
+    {
+      "epoch": 759.1447368421053,
+      "grad_norm": 1.2153128385543823,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 115390
+    },
+    {
+      "epoch": 759.2105263157895,
+      "grad_norm": 1.31186044216156,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 115400
+    },
+    {
+      "epoch": 759.2763157894736,
+      "grad_norm": 1.637511968612671,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 115410
+    },
+    {
+      "epoch": 759.3421052631579,
+      "grad_norm": 1.3562475442886353,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 115420
+    },
+    {
+      "epoch": 759.4078947368421,
+      "grad_norm": 1.192093014717102,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 115430
+    },
+    {
+      "epoch": 759.4736842105264,
+      "grad_norm": 1.5335882902145386,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 115440
+    },
+    {
+      "epoch": 759.5394736842105,
+      "grad_norm": 1.1234040260314941,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 115450
+    },
+    {
+      "epoch": 759.6052631578947,
+      "grad_norm": 1.1577470302581787,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 115460
+    },
+    {
+      "epoch": 759.671052631579,
+      "grad_norm": 1.1491925716400146,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 115470
+    },
+    {
+      "epoch": 759.7368421052631,
+      "grad_norm": 1.103284239768982,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 115480
+    },
+    {
+      "epoch": 759.8026315789474,
+      "grad_norm": 1.1908904314041138,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 115490
+    },
+    {
+      "epoch": 759.8684210526316,
+      "grad_norm": 1.180321216583252,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 115500
+    },
+    {
+      "epoch": 759.9342105263158,
+      "grad_norm": 1.2648427486419678,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 115510
+    },
+    {
+      "epoch": 760.0,
+      "grad_norm": 1.4859260320663452,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 115520
+    },
+    {
+      "epoch": 760.0657894736842,
+      "grad_norm": 1.3933957815170288,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 115530
+    },
+    {
+      "epoch": 760.1315789473684,
+      "grad_norm": 1.4522415399551392,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 115540
+    },
+    {
+      "epoch": 760.1973684210526,
+      "grad_norm": 1.2445298433303833,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 115550
+    },
+    {
+      "epoch": 760.2631578947369,
+      "grad_norm": 1.222784399986267,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 115560
+    },
+    {
+      "epoch": 760.328947368421,
+      "grad_norm": 0.9933916330337524,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 115570
+    },
+    {
+      "epoch": 760.3947368421053,
+      "grad_norm": 1.02178955078125,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 115580
+    },
+    {
+      "epoch": 760.4605263157895,
+      "grad_norm": 1.0598787069320679,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 115590
+    },
+    {
+      "epoch": 760.5263157894736,
+      "grad_norm": 1.264055609703064,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 115600
+    },
+    {
+      "epoch": 760.5921052631579,
+      "grad_norm": 1.1147860288619995,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 115610
+    },
+    {
+      "epoch": 760.6578947368421,
+      "grad_norm": 1.159653902053833,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 115620
+    },
+    {
+      "epoch": 760.7236842105264,
+      "grad_norm": 0.6607643961906433,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 115630
+    },
+    {
+      "epoch": 760.7894736842105,
+      "grad_norm": 0.9183317422866821,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 115640
+    },
+    {
+      "epoch": 760.8552631578947,
+      "grad_norm": 1.034013032913208,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 115650
+    },
+    {
+      "epoch": 760.921052631579,
+      "grad_norm": 0.9568067789077759,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 115660
+    },
+    {
+      "epoch": 760.9868421052631,
+      "grad_norm": 1.138213872909546,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 115670
+    },
+    {
+      "epoch": 761.0526315789474,
+      "grad_norm": 1.0073471069335938,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 115680
+    },
+    {
+      "epoch": 761.1184210526316,
+      "grad_norm": 1.5636231899261475,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 115690
+    },
+    {
+      "epoch": 761.1842105263158,
+      "grad_norm": 1.0607483386993408,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 115700
+    },
+    {
+      "epoch": 761.25,
+      "grad_norm": 1.3943238258361816,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 115710
+    },
+    {
+      "epoch": 761.3157894736842,
+      "grad_norm": 0.7510659694671631,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 115720
+    },
+    {
+      "epoch": 761.3815789473684,
+      "grad_norm": 1.033464789390564,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 115730
+    },
+    {
+      "epoch": 761.4473684210526,
+      "grad_norm": 1.1388704776763916,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 115740
+    },
+    {
+      "epoch": 761.5131578947369,
+      "grad_norm": 1.0201410055160522,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 115750
+    },
+    {
+      "epoch": 761.578947368421,
+      "grad_norm": 0.690221905708313,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 115760
+    },
+    {
+      "epoch": 761.6447368421053,
+      "grad_norm": 0.7212555408477783,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 115770
+    },
+    {
+      "epoch": 761.7105263157895,
+      "grad_norm": 1.3436615467071533,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 115780
+    },
+    {
+      "epoch": 761.7763157894736,
+      "grad_norm": 1.2216511964797974,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 115790
+    },
+    {
+      "epoch": 761.8421052631579,
+      "grad_norm": 1.0771437883377075,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 115800
+    },
+    {
+      "epoch": 761.9078947368421,
+      "grad_norm": 1.3642146587371826,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 115810
+    },
+    {
+      "epoch": 761.9736842105264,
+      "grad_norm": 1.320507287979126,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 115820
+    },
+    {
+      "epoch": 762.0394736842105,
+      "grad_norm": 0.8073870539665222,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 115830
+    },
+    {
+      "epoch": 762.1052631578947,
+      "grad_norm": 1.161726951599121,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 115840
+    },
+    {
+      "epoch": 762.171052631579,
+      "grad_norm": 1.1861587762832642,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 115850
+    },
+    {
+      "epoch": 762.2368421052631,
+      "grad_norm": 0.9289435744285583,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 115860
+    },
+    {
+      "epoch": 762.3026315789474,
+      "grad_norm": 1.0481675863265991,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 115870
+    },
+    {
+      "epoch": 762.3684210526316,
+      "grad_norm": 1.2852026224136353,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 115880
+    },
+    {
+      "epoch": 762.4342105263158,
+      "grad_norm": 1.4661617279052734,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 115890
+    },
+    {
+      "epoch": 762.5,
+      "grad_norm": 1.2878676652908325,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 115900
+    },
+    {
+      "epoch": 762.5657894736842,
+      "grad_norm": 1.1841388940811157,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 115910
+    },
+    {
+      "epoch": 762.6315789473684,
+      "grad_norm": 1.1691356897354126,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 115920
+    },
+    {
+      "epoch": 762.6973684210526,
+      "grad_norm": 1.0297003984451294,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 115930
+    },
+    {
+      "epoch": 762.7631578947369,
+      "grad_norm": 1.1073315143585205,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 115940
+    },
+    {
+      "epoch": 762.828947368421,
+      "grad_norm": 1.3465665578842163,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 115950
+    },
+    {
+      "epoch": 762.8947368421053,
+      "grad_norm": 1.0311391353607178,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 115960
+    },
+    {
+      "epoch": 762.9605263157895,
+      "grad_norm": 0.8130193948745728,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 115970
+    },
+    {
+      "epoch": 763.0263157894736,
+      "grad_norm": 0.788371741771698,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 115980
+    },
+    {
+      "epoch": 763.0921052631579,
+      "grad_norm": 1.0673997402191162,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 115990
+    },
+    {
+      "epoch": 763.1578947368421,
+      "grad_norm": 0.967298150062561,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 116000
+    },
+    {
+      "epoch": 763.2236842105264,
+      "grad_norm": 0.8940585255622864,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 116010
+    },
+    {
+      "epoch": 763.2894736842105,
+      "grad_norm": 0.8840467929840088,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 116020
+    },
+    {
+      "epoch": 763.3552631578947,
+      "grad_norm": 1.0176132917404175,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 116030
+    },
+    {
+      "epoch": 763.421052631579,
+      "grad_norm": 1.144092321395874,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 116040
+    },
+    {
+      "epoch": 763.4868421052631,
+      "grad_norm": 1.1976016759872437,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 116050
+    },
+    {
+      "epoch": 763.5526315789474,
+      "grad_norm": 1.0607638359069824,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 116060
+    },
+    {
+      "epoch": 763.6184210526316,
+      "grad_norm": 1.0160369873046875,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 116070
+    },
+    {
+      "epoch": 763.6842105263158,
+      "grad_norm": 0.8879520297050476,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 116080
+    },
+    {
+      "epoch": 763.75,
+      "grad_norm": 1.279135823249817,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 116090
+    },
+    {
+      "epoch": 763.8157894736842,
+      "grad_norm": 0.8777012228965759,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 116100
+    },
+    {
+      "epoch": 763.8815789473684,
+      "grad_norm": 1.0412824153900146,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 116110
+    },
+    {
+      "epoch": 763.9473684210526,
+      "grad_norm": 1.1568728685379028,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 116120
+    },
+    {
+      "epoch": 764.0131578947369,
+      "grad_norm": 0.8233476281166077,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 116130
+    },
+    {
+      "epoch": 764.078947368421,
+      "grad_norm": 0.7248982191085815,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 116140
+    },
+    {
+      "epoch": 764.1447368421053,
+      "grad_norm": 0.712303876876831,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 116150
+    },
+    {
+      "epoch": 764.2105263157895,
+      "grad_norm": 1.0194206237792969,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 116160
+    },
+    {
+      "epoch": 764.2763157894736,
+      "grad_norm": 1.2041058540344238,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 116170
+    },
+    {
+      "epoch": 764.3421052631579,
+      "grad_norm": 0.784016489982605,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 116180
+    },
+    {
+      "epoch": 764.4078947368421,
+      "grad_norm": 1.4073185920715332,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 116190
+    },
+    {
+      "epoch": 764.4736842105264,
+      "grad_norm": 1.0075128078460693,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 116200
+    },
+    {
+      "epoch": 764.5394736842105,
+      "grad_norm": 0.6819977164268494,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 116210
+    },
+    {
+      "epoch": 764.6052631578947,
+      "grad_norm": 1.2221415042877197,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 116220
+    },
+    {
+      "epoch": 764.671052631579,
+      "grad_norm": 1.6122931241989136,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 116230
+    },
+    {
+      "epoch": 764.7368421052631,
+      "grad_norm": 1.30339515209198,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 116240
+    },
+    {
+      "epoch": 764.8026315789474,
+      "grad_norm": 1.2761629819869995,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 116250
+    },
+    {
+      "epoch": 764.8684210526316,
+      "grad_norm": 1.287692904472351,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 116260
+    },
+    {
+      "epoch": 764.9342105263158,
+      "grad_norm": 0.9369876384735107,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 116270
+    },
+    {
+      "epoch": 765.0,
+      "grad_norm": 0.9917391538619995,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 116280
+    },
+    {
+      "epoch": 765.0657894736842,
+      "grad_norm": 0.8295028805732727,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 116290
+    },
+    {
+      "epoch": 765.1315789473684,
+      "grad_norm": 1.1534699201583862,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 116300
+    },
+    {
+      "epoch": 765.1973684210526,
+      "grad_norm": 0.8612220883369446,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 116310
+    },
+    {
+      "epoch": 765.2631578947369,
+      "grad_norm": 1.1963870525360107,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 116320
+    },
+    {
+      "epoch": 765.328947368421,
+      "grad_norm": 1.3065626621246338,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 116330
+    },
+    {
+      "epoch": 765.3947368421053,
+      "grad_norm": 0.979164719581604,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 116340
+    },
+    {
+      "epoch": 765.4605263157895,
+      "grad_norm": 0.9245379567146301,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 116350
+    },
+    {
+      "epoch": 765.5263157894736,
+      "grad_norm": 0.8577090501785278,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 116360
+    },
+    {
+      "epoch": 765.5921052631579,
+      "grad_norm": 0.9338501691818237,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 116370
+    },
+    {
+      "epoch": 765.6578947368421,
+      "grad_norm": 1.0019291639328003,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 116380
+    },
+    {
+      "epoch": 765.7236842105264,
+      "grad_norm": 1.2914475202560425,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 116390
+    },
+    {
+      "epoch": 765.7894736842105,
+      "grad_norm": 1.1972627639770508,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 116400
+    },
+    {
+      "epoch": 765.8552631578947,
+      "grad_norm": 1.0993999242782593,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 116410
+    },
+    {
+      "epoch": 765.921052631579,
+      "grad_norm": 0.9448077082633972,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 116420
+    },
+    {
+      "epoch": 765.9868421052631,
+      "grad_norm": 0.893061637878418,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 116430
+    },
+    {
+      "epoch": 766.0526315789474,
+      "grad_norm": 1.3330209255218506,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 116440
+    },
+    {
+      "epoch": 766.1184210526316,
+      "grad_norm": 1.363829255104065,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 116450
+    },
+    {
+      "epoch": 766.1842105263158,
+      "grad_norm": 1.1652305126190186,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 116460
+    },
+    {
+      "epoch": 766.25,
+      "grad_norm": 1.1729694604873657,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 116470
+    },
+    {
+      "epoch": 766.3157894736842,
+      "grad_norm": 0.8965724110603333,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 116480
+    },
+    {
+      "epoch": 766.3815789473684,
+      "grad_norm": 0.8969491720199585,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 116490
+    },
+    {
+      "epoch": 766.4473684210526,
+      "grad_norm": 1.0770326852798462,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 116500
+    },
+    {
+      "epoch": 766.5131578947369,
+      "grad_norm": 0.9563101530075073,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 116510
+    },
+    {
+      "epoch": 766.578947368421,
+      "grad_norm": 1.18818199634552,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 116520
+    },
+    {
+      "epoch": 766.6447368421053,
+      "grad_norm": 0.7398898005485535,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 116530
+    },
+    {
+      "epoch": 766.7105263157895,
+      "grad_norm": 0.6611107587814331,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 116540
+    },
+    {
+      "epoch": 766.7763157894736,
+      "grad_norm": 0.8535279631614685,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 116550
+    },
+    {
+      "epoch": 766.8421052631579,
+      "grad_norm": 1.0214197635650635,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 116560
+    },
+    {
+      "epoch": 766.9078947368421,
+      "grad_norm": 1.0022668838500977,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 116570
+    },
+    {
+      "epoch": 766.9736842105264,
+      "grad_norm": 1.2458581924438477,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 116580
+    },
+    {
+      "epoch": 767.0394736842105,
+      "grad_norm": 1.3370084762573242,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 116590
+    },
+    {
+      "epoch": 767.1052631578947,
+      "grad_norm": 0.9627024531364441,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 116600
+    },
+    {
+      "epoch": 767.171052631579,
+      "grad_norm": 1.2679235935211182,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 116610
+    },
+    {
+      "epoch": 767.2368421052631,
+      "grad_norm": 0.9105582237243652,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 116620
+    },
+    {
+      "epoch": 767.3026315789474,
+      "grad_norm": 0.9305436015129089,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 116630
+    },
+    {
+      "epoch": 767.3684210526316,
+      "grad_norm": 0.8848668336868286,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 116640
+    },
+    {
+      "epoch": 767.4342105263158,
+      "grad_norm": 1.0530251264572144,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 116650
+    },
+    {
+      "epoch": 767.5,
+      "grad_norm": 0.8415973782539368,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 116660
+    },
+    {
+      "epoch": 767.5657894736842,
+      "grad_norm": 1.0781415700912476,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 116670
+    },
+    {
+      "epoch": 767.6315789473684,
+      "grad_norm": 0.8499521017074585,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 116680
+    },
+    {
+      "epoch": 767.6973684210526,
+      "grad_norm": 1.271653175354004,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 116690
+    },
+    {
+      "epoch": 767.7631578947369,
+      "grad_norm": 1.3526909351348877,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 116700
+    },
+    {
+      "epoch": 767.828947368421,
+      "grad_norm": 1.1030946969985962,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 116710
+    },
+    {
+      "epoch": 767.8947368421053,
+      "grad_norm": 1.2905566692352295,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 116720
+    },
+    {
+      "epoch": 767.9605263157895,
+      "grad_norm": 1.2587246894836426,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 116730
+    },
+    {
+      "epoch": 768.0263157894736,
+      "grad_norm": 1.263112187385559,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 116740
+    },
+    {
+      "epoch": 768.0921052631579,
+      "grad_norm": 1.2640330791473389,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 116750
+    },
+    {
+      "epoch": 768.1578947368421,
+      "grad_norm": 0.9966381788253784,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 116760
+    },
+    {
+      "epoch": 768.2236842105264,
+      "grad_norm": 1.5357413291931152,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 116770
+    },
+    {
+      "epoch": 768.2894736842105,
+      "grad_norm": 1.3589855432510376,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 116780
+    },
+    {
+      "epoch": 768.3552631578947,
+      "grad_norm": 1.3303495645523071,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 116790
+    },
+    {
+      "epoch": 768.421052631579,
+      "grad_norm": 1.531052827835083,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 116800
+    },
+    {
+      "epoch": 768.4868421052631,
+      "grad_norm": 1.2647902965545654,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 116810
+    },
+    {
+      "epoch": 768.5526315789474,
+      "grad_norm": 1.2003931999206543,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 116820
+    },
+    {
+      "epoch": 768.6184210526316,
+      "grad_norm": 1.2271924018859863,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 116830
+    },
+    {
+      "epoch": 768.6842105263158,
+      "grad_norm": 0.8011854887008667,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 116840
+    },
+    {
+      "epoch": 768.75,
+      "grad_norm": 1.0051759481430054,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 116850
+    },
+    {
+      "epoch": 768.8157894736842,
+      "grad_norm": 1.0065410137176514,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 116860
+    },
+    {
+      "epoch": 768.8815789473684,
+      "grad_norm": 1.094523310661316,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 116870
+    },
+    {
+      "epoch": 768.9473684210526,
+      "grad_norm": 0.9108685255050659,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 116880
+    },
+    {
+      "epoch": 769.0131578947369,
+      "grad_norm": 1.0537774562835693,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 116890
+    },
+    {
+      "epoch": 769.078947368421,
+      "grad_norm": 0.8847028613090515,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 116900
+    },
+    {
+      "epoch": 769.1447368421053,
+      "grad_norm": 0.8502101302146912,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 116910
+    },
+    {
+      "epoch": 769.2105263157895,
+      "grad_norm": 1.3079115152359009,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 116920
+    },
+    {
+      "epoch": 769.2763157894736,
+      "grad_norm": 0.997676432132721,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 116930
+    },
+    {
+      "epoch": 769.3421052631579,
+      "grad_norm": 0.9060066342353821,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 116940
+    },
+    {
+      "epoch": 769.4078947368421,
+      "grad_norm": 0.9685508012771606,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 116950
+    },
+    {
+      "epoch": 769.4736842105264,
+      "grad_norm": 1.053354024887085,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 116960
+    },
+    {
+      "epoch": 769.5394736842105,
+      "grad_norm": 0.901861310005188,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 116970
+    },
+    {
+      "epoch": 769.6052631578947,
+      "grad_norm": 1.1374804973602295,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 116980
+    },
+    {
+      "epoch": 769.671052631579,
+      "grad_norm": 1.060011386871338,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 116990
+    },
+    {
+      "epoch": 769.7368421052631,
+      "grad_norm": 0.8617300391197205,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 117000
+    },
+    {
+      "epoch": 769.8026315789474,
+      "grad_norm": 0.7569366097450256,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 117010
+    },
+    {
+      "epoch": 769.8684210526316,
+      "grad_norm": 0.9865979552268982,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 117020
+    },
+    {
+      "epoch": 769.9342105263158,
+      "grad_norm": 0.6194046139717102,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 117030
+    },
+    {
+      "epoch": 770.0,
+      "grad_norm": 1.2223562002182007,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 117040
+    },
+    {
+      "epoch": 770.0657894736842,
+      "grad_norm": 0.9539296627044678,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 117050
+    },
+    {
+      "epoch": 770.1315789473684,
+      "grad_norm": 0.7826547622680664,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 117060
+    },
+    {
+      "epoch": 770.1973684210526,
+      "grad_norm": 1.4138188362121582,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 117070
+    },
+    {
+      "epoch": 770.2631578947369,
+      "grad_norm": 1.2407159805297852,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 117080
+    },
+    {
+      "epoch": 770.328947368421,
+      "grad_norm": 1.1824477910995483,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 117090
+    },
+    {
+      "epoch": 770.3947368421053,
+      "grad_norm": 0.9199476838111877,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 117100
+    },
+    {
+      "epoch": 770.4605263157895,
+      "grad_norm": 0.7619976997375488,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 117110
+    },
+    {
+      "epoch": 770.5263157894736,
+      "grad_norm": 0.8713205456733704,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 117120
+    },
+    {
+      "epoch": 770.5921052631579,
+      "grad_norm": 0.819759726524353,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 117130
+    },
+    {
+      "epoch": 770.6578947368421,
+      "grad_norm": 1.4600467681884766,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 117140
+    },
+    {
+      "epoch": 770.7236842105264,
+      "grad_norm": 0.7182674407958984,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 117150
+    },
+    {
+      "epoch": 770.7894736842105,
+      "grad_norm": 1.2647074460983276,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 117160
+    },
+    {
+      "epoch": 770.8552631578947,
+      "grad_norm": 1.4599173069000244,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 117170
+    },
+    {
+      "epoch": 770.921052631579,
+      "grad_norm": 1.161414384841919,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 117180
+    },
+    {
+      "epoch": 770.9868421052631,
+      "grad_norm": 0.9526516199111938,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 117190
+    },
+    {
+      "epoch": 771.0526315789474,
+      "grad_norm": 1.2501287460327148,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 117200
+    },
+    {
+      "epoch": 771.1184210526316,
+      "grad_norm": 1.2030541896820068,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 117210
+    },
+    {
+      "epoch": 771.1842105263158,
+      "grad_norm": 1.1028085947036743,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 117220
+    },
+    {
+      "epoch": 771.25,
+      "grad_norm": 1.2376686334609985,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 117230
+    },
+    {
+      "epoch": 771.3157894736842,
+      "grad_norm": 1.1704808473587036,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 117240
+    },
+    {
+      "epoch": 771.3815789473684,
+      "grad_norm": 0.899161159992218,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 117250
+    },
+    {
+      "epoch": 771.4473684210526,
+      "grad_norm": 1.2865201234817505,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 117260
+    },
+    {
+      "epoch": 771.5131578947369,
+      "grad_norm": 1.0590287446975708,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 117270
+    },
+    {
+      "epoch": 771.578947368421,
+      "grad_norm": 0.7633146047592163,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 117280
+    },
+    {
+      "epoch": 771.6447368421053,
+      "grad_norm": 0.6437667012214661,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 117290
+    },
+    {
+      "epoch": 771.7105263157895,
+      "grad_norm": 1.379522442817688,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 117300
+    },
+    {
+      "epoch": 771.7763157894736,
+      "grad_norm": 1.1769373416900635,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 117310
+    },
+    {
+      "epoch": 771.8421052631579,
+      "grad_norm": 1.229821801185608,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 117320
+    },
+    {
+      "epoch": 771.9078947368421,
+      "grad_norm": 0.8484160304069519,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 117330
+    },
+    {
+      "epoch": 771.9736842105264,
+      "grad_norm": 0.7620605230331421,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 117340
+    },
+    {
+      "epoch": 772.0394736842105,
+      "grad_norm": 1.064082384109497,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 117350
+    },
+    {
+      "epoch": 772.1052631578947,
+      "grad_norm": 0.9610821604728699,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 117360
+    },
+    {
+      "epoch": 772.171052631579,
+      "grad_norm": 1.0888431072235107,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 117370
+    },
+    {
+      "epoch": 772.2368421052631,
+      "grad_norm": 0.9147794842720032,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 117380
+    },
+    {
+      "epoch": 772.3026315789474,
+      "grad_norm": 0.930996298789978,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 117390
+    },
+    {
+      "epoch": 772.3684210526316,
+      "grad_norm": 0.9899475574493408,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 117400
+    },
+    {
+      "epoch": 772.4342105263158,
+      "grad_norm": 1.1888105869293213,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 117410
+    },
+    {
+      "epoch": 772.5,
+      "grad_norm": 1.1909250020980835,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 117420
+    },
+    {
+      "epoch": 772.5657894736842,
+      "grad_norm": 1.144278645515442,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 117430
+    },
+    {
+      "epoch": 772.6315789473684,
+      "grad_norm": 1.2057489156723022,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 117440
+    },
+    {
+      "epoch": 772.6973684210526,
+      "grad_norm": 1.240365982055664,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 117450
+    },
+    {
+      "epoch": 772.7631578947369,
+      "grad_norm": 1.1910370588302612,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 117460
+    },
+    {
+      "epoch": 772.828947368421,
+      "grad_norm": 1.0955934524536133,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 117470
+    },
+    {
+      "epoch": 772.8947368421053,
+      "grad_norm": 1.1824718713760376,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 117480
+    },
+    {
+      "epoch": 772.9605263157895,
+      "grad_norm": 1.323282241821289,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 117490
+    },
+    {
+      "epoch": 773.0263157894736,
+      "grad_norm": 0.9789089560508728,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 117500
+    },
+    {
+      "epoch": 773.0921052631579,
+      "grad_norm": 1.0450570583343506,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 117510
+    },
+    {
+      "epoch": 773.1578947368421,
+      "grad_norm": 1.3305141925811768,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 117520
+    },
+    {
+      "epoch": 773.2236842105264,
+      "grad_norm": 1.228081226348877,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 117530
+    },
+    {
+      "epoch": 773.2894736842105,
+      "grad_norm": 1.1966365575790405,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 117540
+    },
+    {
+      "epoch": 773.3552631578947,
+      "grad_norm": 1.154728651046753,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 117550
+    },
+    {
+      "epoch": 773.421052631579,
+      "grad_norm": 1.119430422782898,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 117560
+    },
+    {
+      "epoch": 773.4868421052631,
+      "grad_norm": 1.3745265007019043,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 117570
+    },
+    {
+      "epoch": 773.5526315789474,
+      "grad_norm": 1.3830089569091797,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 117580
+    },
+    {
+      "epoch": 773.6184210526316,
+      "grad_norm": 1.2149348258972168,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 117590
+    },
+    {
+      "epoch": 773.6842105263158,
+      "grad_norm": 1.084110975265503,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 117600
+    },
+    {
+      "epoch": 773.75,
+      "grad_norm": 1.2367154359817505,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 117610
+    },
+    {
+      "epoch": 773.8157894736842,
+      "grad_norm": 1.1649274826049805,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 117620
+    },
+    {
+      "epoch": 773.8815789473684,
+      "grad_norm": 1.4875725507736206,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 117630
+    },
+    {
+      "epoch": 773.9473684210526,
+      "grad_norm": 0.9839734435081482,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 117640
+    },
+    {
+      "epoch": 774.0131578947369,
+      "grad_norm": 1.041127324104309,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 117650
+    },
+    {
+      "epoch": 774.078947368421,
+      "grad_norm": 1.2891390323638916,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 117660
+    },
+    {
+      "epoch": 774.1447368421053,
+      "grad_norm": 1.126435399055481,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 117670
+    },
+    {
+      "epoch": 774.2105263157895,
+      "grad_norm": 1.295599341392517,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 117680
+    },
+    {
+      "epoch": 774.2763157894736,
+      "grad_norm": 1.0001301765441895,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 117690
+    },
+    {
+      "epoch": 774.3421052631579,
+      "grad_norm": 0.824128270149231,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 117700
+    },
+    {
+      "epoch": 774.4078947368421,
+      "grad_norm": 1.071433663368225,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 117710
+    },
+    {
+      "epoch": 774.4736842105264,
+      "grad_norm": 0.8961533308029175,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 117720
+    },
+    {
+      "epoch": 774.5394736842105,
+      "grad_norm": 0.9159356355667114,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 117730
+    },
+    {
+      "epoch": 774.6052631578947,
+      "grad_norm": 0.8452175259590149,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 117740
+    },
+    {
+      "epoch": 774.671052631579,
+      "grad_norm": 0.9821557998657227,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 117750
+    },
+    {
+      "epoch": 774.7368421052631,
+      "grad_norm": 1.0779387950897217,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 117760
+    },
+    {
+      "epoch": 774.8026315789474,
+      "grad_norm": 1.1578203439712524,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 117770
+    },
+    {
+      "epoch": 774.8684210526316,
+      "grad_norm": 1.1006511449813843,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 117780
+    },
+    {
+      "epoch": 774.9342105263158,
+      "grad_norm": 1.2114787101745605,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 117790
+    },
+    {
+      "epoch": 775.0,
+      "grad_norm": 0.8628996014595032,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 117800
+    },
+    {
+      "epoch": 775.0657894736842,
+      "grad_norm": 1.1864206790924072,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 117810
+    },
+    {
+      "epoch": 775.1315789473684,
+      "grad_norm": 1.050344467163086,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 117820
+    },
+    {
+      "epoch": 775.1973684210526,
+      "grad_norm": 1.0661941766738892,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 117830
+    },
+    {
+      "epoch": 775.2631578947369,
+      "grad_norm": 0.982932984828949,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 117840
+    },
+    {
+      "epoch": 775.328947368421,
+      "grad_norm": 1.0128767490386963,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 117850
+    },
+    {
+      "epoch": 775.3947368421053,
+      "grad_norm": 1.2008044719696045,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 117860
+    },
+    {
+      "epoch": 775.4605263157895,
+      "grad_norm": 1.156033992767334,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 117870
+    },
+    {
+      "epoch": 775.5263157894736,
+      "grad_norm": 1.492291808128357,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 117880
+    },
+    {
+      "epoch": 775.5921052631579,
+      "grad_norm": 0.656265914440155,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 117890
+    },
+    {
+      "epoch": 775.6578947368421,
+      "grad_norm": 1.0817625522613525,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 117900
+    },
+    {
+      "epoch": 775.7236842105264,
+      "grad_norm": 0.741335391998291,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 117910
+    },
+    {
+      "epoch": 775.7894736842105,
+      "grad_norm": 0.9007006883621216,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 117920
+    },
+    {
+      "epoch": 775.8552631578947,
+      "grad_norm": 0.8411262035369873,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 117930
+    },
+    {
+      "epoch": 775.921052631579,
+      "grad_norm": 0.8096522688865662,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 117940
+    },
+    {
+      "epoch": 775.9868421052631,
+      "grad_norm": 0.6779792308807373,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 117950
+    },
+    {
+      "epoch": 776.0526315789474,
+      "grad_norm": 0.9828864336013794,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 117960
+    },
+    {
+      "epoch": 776.1184210526316,
+      "grad_norm": 1.536733627319336,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 117970
+    },
+    {
+      "epoch": 776.1842105263158,
+      "grad_norm": 1.2087494134902954,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 117980
+    },
+    {
+      "epoch": 776.25,
+      "grad_norm": 0.7931889295578003,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 117990
+    },
+    {
+      "epoch": 776.3157894736842,
+      "grad_norm": 0.8839361667633057,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 118000
+    },
+    {
+      "epoch": 776.3815789473684,
+      "grad_norm": 1.1379063129425049,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 118010
+    },
+    {
+      "epoch": 776.4473684210526,
+      "grad_norm": 1.1438642740249634,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 118020
+    },
+    {
+      "epoch": 776.5131578947369,
+      "grad_norm": 0.9373273253440857,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 118030
+    },
+    {
+      "epoch": 776.578947368421,
+      "grad_norm": 1.2530593872070312,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 118040
+    },
+    {
+      "epoch": 776.6447368421053,
+      "grad_norm": 1.2131603956222534,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 118050
+    },
+    {
+      "epoch": 776.7105263157895,
+      "grad_norm": 0.7886523008346558,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 118060
+    },
+    {
+      "epoch": 776.7763157894736,
+      "grad_norm": 0.6613753437995911,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 118070
+    },
+    {
+      "epoch": 776.8421052631579,
+      "grad_norm": 1.1990413665771484,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 118080
+    },
+    {
+      "epoch": 776.9078947368421,
+      "grad_norm": 1.0473389625549316,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 118090
+    },
+    {
+      "epoch": 776.9736842105264,
+      "grad_norm": 0.9223041534423828,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 118100
+    },
+    {
+      "epoch": 777.0394736842105,
+      "grad_norm": 0.7598844170570374,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 118110
+    },
+    {
+      "epoch": 777.1052631578947,
+      "grad_norm": 1.1341654062271118,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 118120
+    },
+    {
+      "epoch": 777.171052631579,
+      "grad_norm": 0.5702991485595703,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 118130
+    },
+    {
+      "epoch": 777.2368421052631,
+      "grad_norm": 1.0409194231033325,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 118140
+    },
+    {
+      "epoch": 777.3026315789474,
+      "grad_norm": 1.469417691230774,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 118150
+    },
+    {
+      "epoch": 777.3684210526316,
+      "grad_norm": 1.1192512512207031,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 118160
+    },
+    {
+      "epoch": 777.4342105263158,
+      "grad_norm": 1.1567856073379517,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 118170
+    },
+    {
+      "epoch": 777.5,
+      "grad_norm": 0.8657984733581543,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 118180
+    },
+    {
+      "epoch": 777.5657894736842,
+      "grad_norm": 0.8693804740905762,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 118190
+    },
+    {
+      "epoch": 777.6315789473684,
+      "grad_norm": 1.0661799907684326,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 118200
+    },
+    {
+      "epoch": 777.6973684210526,
+      "grad_norm": 0.9497994184494019,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 118210
+    },
+    {
+      "epoch": 777.7631578947369,
+      "grad_norm": 1.2563682794570923,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 118220
+    },
+    {
+      "epoch": 777.828947368421,
+      "grad_norm": 0.7826574444770813,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 118230
+    },
+    {
+      "epoch": 777.8947368421053,
+      "grad_norm": 1.0407278537750244,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 118240
+    },
+    {
+      "epoch": 777.9605263157895,
+      "grad_norm": 1.288908839225769,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 118250
+    },
+    {
+      "epoch": 778.0263157894736,
+      "grad_norm": 1.0953668355941772,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 118260
+    },
+    {
+      "epoch": 778.0921052631579,
+      "grad_norm": 1.315590500831604,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 118270
+    },
+    {
+      "epoch": 778.1578947368421,
+      "grad_norm": 1.1540820598602295,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 118280
+    },
+    {
+      "epoch": 778.2236842105264,
+      "grad_norm": 1.0182602405548096,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 118290
+    },
+    {
+      "epoch": 778.2894736842105,
+      "grad_norm": 1.1212661266326904,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 118300
+    },
+    {
+      "epoch": 778.3552631578947,
+      "grad_norm": 0.8415754437446594,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 118310
+    },
+    {
+      "epoch": 778.421052631579,
+      "grad_norm": 0.8753746747970581,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 118320
+    },
+    {
+      "epoch": 778.4868421052631,
+      "grad_norm": 1.123030662536621,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 118330
+    },
+    {
+      "epoch": 778.5526315789474,
+      "grad_norm": 1.058523416519165,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 118340
+    },
+    {
+      "epoch": 778.6184210526316,
+      "grad_norm": 1.2867323160171509,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 118350
+    },
+    {
+      "epoch": 778.6842105263158,
+      "grad_norm": 1.0465610027313232,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 118360
+    },
+    {
+      "epoch": 778.75,
+      "grad_norm": 1.1913409233093262,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 118370
+    },
+    {
+      "epoch": 778.8157894736842,
+      "grad_norm": 1.0035961866378784,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 118380
+    },
+    {
+      "epoch": 778.8815789473684,
+      "grad_norm": 1.2081618309020996,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 118390
+    },
+    {
+      "epoch": 778.9473684210526,
+      "grad_norm": 0.9412795901298523,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 118400
+    },
+    {
+      "epoch": 779.0131578947369,
+      "grad_norm": 1.1386052370071411,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 118410
+    },
+    {
+      "epoch": 779.078947368421,
+      "grad_norm": 1.215187907218933,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 118420
+    },
+    {
+      "epoch": 779.1447368421053,
+      "grad_norm": 1.0938812494277954,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 118430
+    },
+    {
+      "epoch": 779.2105263157895,
+      "grad_norm": 1.063369631767273,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 118440
+    },
+    {
+      "epoch": 779.2763157894736,
+      "grad_norm": 1.0048210620880127,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 118450
+    },
+    {
+      "epoch": 779.3421052631579,
+      "grad_norm": 1.2882784605026245,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 118460
+    },
+    {
+      "epoch": 779.4078947368421,
+      "grad_norm": 0.9867761731147766,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 118470
+    },
+    {
+      "epoch": 779.4736842105264,
+      "grad_norm": 1.0761276483535767,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 118480
+    },
+    {
+      "epoch": 779.5394736842105,
+      "grad_norm": 0.9819127321243286,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 118490
+    },
+    {
+      "epoch": 779.6052631578947,
+      "grad_norm": 1.317988395690918,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 118500
+    },
+    {
+      "epoch": 779.671052631579,
+      "grad_norm": 0.8267877101898193,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 118510
+    },
+    {
+      "epoch": 779.7368421052631,
+      "grad_norm": 1.0201878547668457,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 118520
+    },
+    {
+      "epoch": 779.8026315789474,
+      "grad_norm": 0.907877504825592,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 118530
+    },
+    {
+      "epoch": 779.8684210526316,
+      "grad_norm": 0.8943087458610535,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 118540
+    },
+    {
+      "epoch": 779.9342105263158,
+      "grad_norm": 1.0807287693023682,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 118550
+    },
+    {
+      "epoch": 780.0,
+      "grad_norm": 1.523391842842102,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 118560
+    },
+    {
+      "epoch": 780.0657894736842,
+      "grad_norm": 0.7130796313285828,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 118570
+    },
+    {
+      "epoch": 780.1315789473684,
+      "grad_norm": 1.3061423301696777,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 118580
+    },
+    {
+      "epoch": 780.1973684210526,
+      "grad_norm": 1.1210944652557373,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 118590
+    },
+    {
+      "epoch": 780.2631578947369,
+      "grad_norm": 1.2665170431137085,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 118600
+    },
+    {
+      "epoch": 780.328947368421,
+      "grad_norm": 1.1752995252609253,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 118610
+    },
+    {
+      "epoch": 780.3947368421053,
+      "grad_norm": 1.102726936340332,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 118620
+    },
+    {
+      "epoch": 780.4605263157895,
+      "grad_norm": 1.2804216146469116,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 118630
+    },
+    {
+      "epoch": 780.5263157894736,
+      "grad_norm": 1.0606364011764526,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 118640
+    },
+    {
+      "epoch": 780.5921052631579,
+      "grad_norm": 0.9021885991096497,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 118650
+    },
+    {
+      "epoch": 780.6578947368421,
+      "grad_norm": 1.3025189638137817,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 118660
+    },
+    {
+      "epoch": 780.7236842105264,
+      "grad_norm": 1.1469581127166748,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 118670
+    },
+    {
+      "epoch": 780.7894736842105,
+      "grad_norm": 1.1277015209197998,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 118680
+    },
+    {
+      "epoch": 780.8552631578947,
+      "grad_norm": 1.2387878894805908,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 118690
+    },
+    {
+      "epoch": 780.921052631579,
+      "grad_norm": 0.9669885635375977,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 118700
+    },
+    {
+      "epoch": 780.9868421052631,
+      "grad_norm": 1.0439066886901855,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 118710
+    },
+    {
+      "epoch": 781.0526315789474,
+      "grad_norm": 0.8005104660987854,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 118720
+    },
+    {
+      "epoch": 781.1184210526316,
+      "grad_norm": 1.0265685319900513,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 118730
+    },
+    {
+      "epoch": 781.1842105263158,
+      "grad_norm": 1.4636842012405396,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 118740
+    },
+    {
+      "epoch": 781.25,
+      "grad_norm": 1.312772512435913,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 118750
+    },
+    {
+      "epoch": 781.3157894736842,
+      "grad_norm": 1.1143829822540283,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 118760
+    },
+    {
+      "epoch": 781.3815789473684,
+      "grad_norm": 1.134764313697815,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 118770
+    },
+    {
+      "epoch": 781.4473684210526,
+      "grad_norm": 1.432435393333435,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 118780
+    },
+    {
+      "epoch": 781.5131578947369,
+      "grad_norm": 1.2253336906433105,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 118790
+    },
+    {
+      "epoch": 781.578947368421,
+      "grad_norm": 1.319845199584961,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 118800
+    },
+    {
+      "epoch": 781.6447368421053,
+      "grad_norm": 1.283031940460205,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 118810
+    },
+    {
+      "epoch": 781.7105263157895,
+      "grad_norm": 1.425012469291687,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 118820
+    },
+    {
+      "epoch": 781.7763157894736,
+      "grad_norm": 0.8615345358848572,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 118830
+    },
+    {
+      "epoch": 781.8421052631579,
+      "grad_norm": 1.3266713619232178,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 118840
+    },
+    {
+      "epoch": 781.9078947368421,
+      "grad_norm": 1.3085308074951172,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 118850
+    },
+    {
+      "epoch": 781.9736842105264,
+      "grad_norm": 1.1384379863739014,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 118860
+    },
+    {
+      "epoch": 782.0394736842105,
+      "grad_norm": 1.471619963645935,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 118870
+    },
+    {
+      "epoch": 782.1052631578947,
+      "grad_norm": 1.2752647399902344,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 118880
+    },
+    {
+      "epoch": 782.171052631579,
+      "grad_norm": 1.337410569190979,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 118890
+    },
+    {
+      "epoch": 782.2368421052631,
+      "grad_norm": 1.0045582056045532,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 118900
+    },
+    {
+      "epoch": 782.3026315789474,
+      "grad_norm": 1.1922293901443481,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 118910
+    },
+    {
+      "epoch": 782.3684210526316,
+      "grad_norm": 1.145751714706421,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 118920
+    },
+    {
+      "epoch": 782.4342105263158,
+      "grad_norm": 1.2008522748947144,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 118930
+    },
+    {
+      "epoch": 782.5,
+      "grad_norm": 1.2413567304611206,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 118940
+    },
+    {
+      "epoch": 782.5657894736842,
+      "grad_norm": 0.9739688634872437,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 118950
+    },
+    {
+      "epoch": 782.6315789473684,
+      "grad_norm": 0.9391964077949524,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 118960
+    },
+    {
+      "epoch": 782.6973684210526,
+      "grad_norm": 1.4125076532363892,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 118970
+    },
+    {
+      "epoch": 782.7631578947369,
+      "grad_norm": 1.2959308624267578,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 118980
+    },
+    {
+      "epoch": 782.828947368421,
+      "grad_norm": 1.301820993423462,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 118990
+    },
+    {
+      "epoch": 782.8947368421053,
+      "grad_norm": 1.200832724571228,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 119000
+    },
+    {
+      "epoch": 782.9605263157895,
+      "grad_norm": 0.8034687042236328,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 119010
+    },
+    {
+      "epoch": 783.0263157894736,
+      "grad_norm": 1.1737399101257324,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 119020
+    },
+    {
+      "epoch": 783.0921052631579,
+      "grad_norm": 1.0292094945907593,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 119030
+    },
+    {
+      "epoch": 783.1578947368421,
+      "grad_norm": 0.9858736395835876,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 119040
+    },
+    {
+      "epoch": 783.2236842105264,
+      "grad_norm": 1.2504321336746216,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 119050
+    },
+    {
+      "epoch": 783.2894736842105,
+      "grad_norm": 1.0716443061828613,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 119060
+    },
+    {
+      "epoch": 783.3552631578947,
+      "grad_norm": 0.9858033657073975,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 119070
+    },
+    {
+      "epoch": 783.421052631579,
+      "grad_norm": 1.0344947576522827,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 119080
+    },
+    {
+      "epoch": 783.4868421052631,
+      "grad_norm": 1.1620103120803833,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 119090
+    },
+    {
+      "epoch": 783.5526315789474,
+      "grad_norm": 1.2477577924728394,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 119100
+    },
+    {
+      "epoch": 783.6184210526316,
+      "grad_norm": 0.96187824010849,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 119110
+    },
+    {
+      "epoch": 783.6842105263158,
+      "grad_norm": 0.9657493829727173,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 119120
+    },
+    {
+      "epoch": 783.75,
+      "grad_norm": 1.1220613718032837,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 119130
+    },
+    {
+      "epoch": 783.8157894736842,
+      "grad_norm": 0.9884085059165955,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 119140
+    },
+    {
+      "epoch": 783.8815789473684,
+      "grad_norm": 1.1578776836395264,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 119150
+    },
+    {
+      "epoch": 783.9473684210526,
+      "grad_norm": 1.015238642692566,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 119160
+    },
+    {
+      "epoch": 784.0131578947369,
+      "grad_norm": 1.1379379034042358,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 119170
+    },
+    {
+      "epoch": 784.078947368421,
+      "grad_norm": 1.041599154472351,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 119180
+    },
+    {
+      "epoch": 784.1447368421053,
+      "grad_norm": 1.3538576364517212,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 119190
+    },
+    {
+      "epoch": 784.2105263157895,
+      "grad_norm": 1.2984648942947388,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 119200
+    },
+    {
+      "epoch": 784.2763157894736,
+      "grad_norm": 1.5270977020263672,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 119210
+    },
+    {
+      "epoch": 784.3421052631579,
+      "grad_norm": 0.9184948801994324,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 119220
+    },
+    {
+      "epoch": 784.4078947368421,
+      "grad_norm": 1.3463494777679443,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 119230
+    },
+    {
+      "epoch": 784.4736842105264,
+      "grad_norm": 1.1868815422058105,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 119240
+    },
+    {
+      "epoch": 784.5394736842105,
+      "grad_norm": 1.198378324508667,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 119250
+    },
+    {
+      "epoch": 784.6052631578947,
+      "grad_norm": 1.215132236480713,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 119260
+    },
+    {
+      "epoch": 784.671052631579,
+      "grad_norm": 1.1473467350006104,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 119270
+    },
+    {
+      "epoch": 784.7368421052631,
+      "grad_norm": 1.5379295349121094,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 119280
+    },
+    {
+      "epoch": 784.8026315789474,
+      "grad_norm": 1.4966095685958862,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 119290
+    },
+    {
+      "epoch": 784.8684210526316,
+      "grad_norm": 1.0885952711105347,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 119300
+    },
+    {
+      "epoch": 784.9342105263158,
+      "grad_norm": 0.980161726474762,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 119310
+    },
+    {
+      "epoch": 785.0,
+      "grad_norm": 0.916150689125061,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 119320
+    },
+    {
+      "epoch": 785.0657894736842,
+      "grad_norm": 1.2834523916244507,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 119330
+    },
+    {
+      "epoch": 785.1315789473684,
+      "grad_norm": 1.2613166570663452,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 119340
+    },
+    {
+      "epoch": 785.1973684210526,
+      "grad_norm": 1.0938130617141724,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 119350
+    },
+    {
+      "epoch": 785.2631578947369,
+      "grad_norm": 1.0228303670883179,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 119360
+    },
+    {
+      "epoch": 785.328947368421,
+      "grad_norm": 0.9948782920837402,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 119370
+    },
+    {
+      "epoch": 785.3947368421053,
+      "grad_norm": 1.0518125295639038,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 119380
+    },
+    {
+      "epoch": 785.4605263157895,
+      "grad_norm": 0.9449158906936646,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 119390
+    },
+    {
+      "epoch": 785.5263157894736,
+      "grad_norm": 0.914734959602356,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 119400
+    },
+    {
+      "epoch": 785.5921052631579,
+      "grad_norm": 1.529738426208496,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 119410
+    },
+    {
+      "epoch": 785.6578947368421,
+      "grad_norm": 1.1134679317474365,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 119420
+    },
+    {
+      "epoch": 785.7236842105264,
+      "grad_norm": 0.9557541012763977,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 119430
+    },
+    {
+      "epoch": 785.7894736842105,
+      "grad_norm": 1.2106385231018066,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 119440
+    },
+    {
+      "epoch": 785.8552631578947,
+      "grad_norm": 0.9082741141319275,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 119450
+    },
+    {
+      "epoch": 785.921052631579,
+      "grad_norm": 1.3996853828430176,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 119460
+    },
+    {
+      "epoch": 785.9868421052631,
+      "grad_norm": 1.2628298997879028,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 119470
+    },
+    {
+      "epoch": 786.0526315789474,
+      "grad_norm": 1.445421814918518,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 119480
+    },
+    {
+      "epoch": 786.1184210526316,
+      "grad_norm": 0.8487147092819214,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 119490
+    },
+    {
+      "epoch": 786.1842105263158,
+      "grad_norm": 0.9393507242202759,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 119500
+    },
+    {
+      "epoch": 786.25,
+      "grad_norm": 1.010326862335205,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 119510
+    },
+    {
+      "epoch": 786.3157894736842,
+      "grad_norm": 1.1670511960983276,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 119520
+    },
+    {
+      "epoch": 786.3815789473684,
+      "grad_norm": 0.9680734872817993,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 119530
+    },
+    {
+      "epoch": 786.4473684210526,
+      "grad_norm": 1.2515125274658203,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 119540
+    },
+    {
+      "epoch": 786.5131578947369,
+      "grad_norm": 1.0361891984939575,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 119550
+    },
+    {
+      "epoch": 786.578947368421,
+      "grad_norm": 0.6271700859069824,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 119560
+    },
+    {
+      "epoch": 786.6447368421053,
+      "grad_norm": 0.9543516635894775,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 119570
+    },
+    {
+      "epoch": 786.7105263157895,
+      "grad_norm": 1.0879392623901367,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 119580
+    },
+    {
+      "epoch": 786.7763157894736,
+      "grad_norm": 0.9335300922393799,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 119590
+    },
+    {
+      "epoch": 786.8421052631579,
+      "grad_norm": 1.244016408920288,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 119600
+    },
+    {
+      "epoch": 786.9078947368421,
+      "grad_norm": 1.3877922296524048,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 119610
+    },
+    {
+      "epoch": 786.9736842105264,
+      "grad_norm": 1.0246111154556274,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 119620
+    },
+    {
+      "epoch": 787.0394736842105,
+      "grad_norm": 1.1666784286499023,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 119630
+    },
+    {
+      "epoch": 787.1052631578947,
+      "grad_norm": 1.422247052192688,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 119640
+    },
+    {
+      "epoch": 787.171052631579,
+      "grad_norm": 1.2223421335220337,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 119650
+    },
+    {
+      "epoch": 787.2368421052631,
+      "grad_norm": 1.0068063735961914,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 119660
+    },
+    {
+      "epoch": 787.3026315789474,
+      "grad_norm": 1.4940046072006226,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 119670
+    },
+    {
+      "epoch": 787.3684210526316,
+      "grad_norm": 0.9652587175369263,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 119680
+    },
+    {
+      "epoch": 787.4342105263158,
+      "grad_norm": 0.7590309977531433,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 119690
+    },
+    {
+      "epoch": 787.5,
+      "grad_norm": 1.0170762538909912,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 119700
+    },
+    {
+      "epoch": 787.5657894736842,
+      "grad_norm": 1.0947462320327759,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 119710
+    },
+    {
+      "epoch": 787.6315789473684,
+      "grad_norm": 0.9564235806465149,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 119720
+    },
+    {
+      "epoch": 787.6973684210526,
+      "grad_norm": 0.8482645153999329,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 119730
+    },
+    {
+      "epoch": 787.7631578947369,
+      "grad_norm": 1.2533555030822754,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 119740
+    },
+    {
+      "epoch": 787.828947368421,
+      "grad_norm": 1.3869478702545166,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 119750
+    },
+    {
+      "epoch": 787.8947368421053,
+      "grad_norm": 1.1893796920776367,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 119760
+    },
+    {
+      "epoch": 787.9605263157895,
+      "grad_norm": 1.3825172185897827,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 119770
+    },
+    {
+      "epoch": 788.0263157894736,
+      "grad_norm": 1.1921755075454712,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 119780
+    },
+    {
+      "epoch": 788.0921052631579,
+      "grad_norm": 1.175467610359192,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 119790
+    },
+    {
+      "epoch": 788.1578947368421,
+      "grad_norm": 1.1586354970932007,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 119800
+    },
+    {
+      "epoch": 788.2236842105264,
+      "grad_norm": 0.9912821054458618,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 119810
+    },
+    {
+      "epoch": 788.2894736842105,
+      "grad_norm": 0.786872923374176,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 119820
+    },
+    {
+      "epoch": 788.3552631578947,
+      "grad_norm": 0.9939967393875122,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 119830
+    },
+    {
+      "epoch": 788.421052631579,
+      "grad_norm": 0.7620839476585388,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 119840
+    },
+    {
+      "epoch": 788.4868421052631,
+      "grad_norm": 1.0876280069351196,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 119850
+    },
+    {
+      "epoch": 788.5526315789474,
+      "grad_norm": 0.9076918959617615,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 119860
+    },
+    {
+      "epoch": 788.6184210526316,
+      "grad_norm": 1.2486871480941772,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 119870
+    },
+    {
+      "epoch": 788.6842105263158,
+      "grad_norm": 0.9349822402000427,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 119880
+    },
+    {
+      "epoch": 788.75,
+      "grad_norm": 1.160534381866455,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 119890
+    },
+    {
+      "epoch": 788.8157894736842,
+      "grad_norm": 1.1158519983291626,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 119900
+    },
+    {
+      "epoch": 788.8815789473684,
+      "grad_norm": 1.0565320253372192,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 119910
+    },
+    {
+      "epoch": 788.9473684210526,
+      "grad_norm": 1.1070889234542847,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 119920
+    },
+    {
+      "epoch": 789.0131578947369,
+      "grad_norm": 1.0946141481399536,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 119930
+    },
+    {
+      "epoch": 789.078947368421,
+      "grad_norm": 1.1808589696884155,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 119940
+    },
+    {
+      "epoch": 789.1447368421053,
+      "grad_norm": 1.0160439014434814,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 119950
+    },
+    {
+      "epoch": 789.2105263157895,
+      "grad_norm": 1.3254737854003906,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 119960
+    },
+    {
+      "epoch": 789.2763157894736,
+      "grad_norm": 1.3260244131088257,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 119970
+    },
+    {
+      "epoch": 789.3421052631579,
+      "grad_norm": 1.212821125984192,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 119980
+    },
+    {
+      "epoch": 789.4078947368421,
+      "grad_norm": 1.2427418231964111,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 119990
+    },
+    {
+      "epoch": 789.4736842105264,
+      "grad_norm": 0.9601447582244873,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 120000
+    },
+    {
+      "epoch": 789.5394736842105,
+      "grad_norm": 1.4246236085891724,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 120010
+    },
+    {
+      "epoch": 789.6052631578947,
+      "grad_norm": 1.162190556526184,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 120020
+    },
+    {
+      "epoch": 789.671052631579,
+      "grad_norm": 1.3025704622268677,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 120030
+    },
+    {
+      "epoch": 789.7368421052631,
+      "grad_norm": 1.1935677528381348,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 120040
+    },
+    {
+      "epoch": 789.8026315789474,
+      "grad_norm": 2.3716793060302734,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 120050
+    },
+    {
+      "epoch": 789.8684210526316,
+      "grad_norm": 1.407984733581543,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 120060
+    },
+    {
+      "epoch": 789.9342105263158,
+      "grad_norm": 1.2061392068862915,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 120070
+    },
+    {
+      "epoch": 790.0,
+      "grad_norm": 1.1397764682769775,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 120080
+    },
+    {
+      "epoch": 790.0657894736842,
+      "grad_norm": 1.4993038177490234,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 120090
+    },
+    {
+      "epoch": 790.1315789473684,
+      "grad_norm": 1.1924669742584229,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 120100
+    },
+    {
+      "epoch": 790.1973684210526,
+      "grad_norm": 1.2470619678497314,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 120110
+    },
+    {
+      "epoch": 790.2631578947369,
+      "grad_norm": 1.23881196975708,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 120120
+    },
+    {
+      "epoch": 790.328947368421,
+      "grad_norm": 1.0482486486434937,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 120130
+    },
+    {
+      "epoch": 790.3947368421053,
+      "grad_norm": 1.1907408237457275,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 120140
+    },
+    {
+      "epoch": 790.4605263157895,
+      "grad_norm": 1.216784119606018,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 120150
+    },
+    {
+      "epoch": 790.5263157894736,
+      "grad_norm": 0.947479248046875,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 120160
+    },
+    {
+      "epoch": 790.5921052631579,
+      "grad_norm": 1.0760152339935303,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 120170
+    },
+    {
+      "epoch": 790.6578947368421,
+      "grad_norm": 0.7800330519676208,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 120180
+    },
+    {
+      "epoch": 790.7236842105264,
+      "grad_norm": 1.2009683847427368,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 120190
+    },
+    {
+      "epoch": 790.7894736842105,
+      "grad_norm": 0.9596481919288635,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 120200
+    },
+    {
+      "epoch": 790.8552631578947,
+      "grad_norm": 1.0723578929901123,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 120210
+    },
+    {
+      "epoch": 790.921052631579,
+      "grad_norm": 0.8646221160888672,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 120220
+    },
+    {
+      "epoch": 790.9868421052631,
+      "grad_norm": 0.9819400310516357,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 120230
+    },
+    {
+      "epoch": 791.0526315789474,
+      "grad_norm": 0.9730360507965088,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 120240
+    },
+    {
+      "epoch": 791.1184210526316,
+      "grad_norm": 1.2054215669631958,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 120250
+    },
+    {
+      "epoch": 791.1842105263158,
+      "grad_norm": 1.0366238355636597,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 120260
+    },
+    {
+      "epoch": 791.25,
+      "grad_norm": 0.992953360080719,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 120270
+    },
+    {
+      "epoch": 791.3157894736842,
+      "grad_norm": 0.7910591959953308,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 120280
+    },
+    {
+      "epoch": 791.3815789473684,
+      "grad_norm": 0.9786763191223145,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 120290
+    },
+    {
+      "epoch": 791.4473684210526,
+      "grad_norm": 0.9105993509292603,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 120300
+    },
+    {
+      "epoch": 791.5131578947369,
+      "grad_norm": 1.005461573600769,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 120310
+    },
+    {
+      "epoch": 791.578947368421,
+      "grad_norm": 0.9919746518135071,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 120320
+    },
+    {
+      "epoch": 791.6447368421053,
+      "grad_norm": 0.6113591194152832,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 120330
+    },
+    {
+      "epoch": 791.7105263157895,
+      "grad_norm": 0.7895347476005554,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 120340
+    },
+    {
+      "epoch": 791.7763157894736,
+      "grad_norm": 1.1017674207687378,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 120350
+    },
+    {
+      "epoch": 791.8421052631579,
+      "grad_norm": 0.9249036908149719,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 120360
+    },
+    {
+      "epoch": 791.9078947368421,
+      "grad_norm": 1.0315765142440796,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 120370
+    },
+    {
+      "epoch": 791.9736842105264,
+      "grad_norm": 0.9553092122077942,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 120380
+    },
+    {
+      "epoch": 792.0394736842105,
+      "grad_norm": 0.9283358454704285,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 120390
+    },
+    {
+      "epoch": 792.1052631578947,
+      "grad_norm": 1.0717108249664307,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 120400
+    },
+    {
+      "epoch": 792.171052631579,
+      "grad_norm": 1.008732557296753,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 120410
+    },
+    {
+      "epoch": 792.2368421052631,
+      "grad_norm": 0.6750414967536926,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 120420
+    },
+    {
+      "epoch": 792.3026315789474,
+      "grad_norm": 0.9144452810287476,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 120430
+    },
+    {
+      "epoch": 792.3684210526316,
+      "grad_norm": 0.8606979250907898,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 120440
+    },
+    {
+      "epoch": 792.4342105263158,
+      "grad_norm": 0.9034025073051453,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 120450
+    },
+    {
+      "epoch": 792.5,
+      "grad_norm": 0.8836445808410645,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 120460
+    },
+    {
+      "epoch": 792.5657894736842,
+      "grad_norm": 1.0012311935424805,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 120470
+    },
+    {
+      "epoch": 792.6315789473684,
+      "grad_norm": 1.7558934688568115,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 120480
+    },
+    {
+      "epoch": 792.6973684210526,
+      "grad_norm": 1.1554054021835327,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 120490
+    },
+    {
+      "epoch": 792.7631578947369,
+      "grad_norm": 0.9315431714057922,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 120500
+    },
+    {
+      "epoch": 792.828947368421,
+      "grad_norm": 1.0039079189300537,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 120510
+    },
+    {
+      "epoch": 792.8947368421053,
+      "grad_norm": 0.9370085597038269,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 120520
+    },
+    {
+      "epoch": 792.9605263157895,
+      "grad_norm": 1.2092498540878296,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 120530
+    },
+    {
+      "epoch": 793.0263157894736,
+      "grad_norm": 1.0734901428222656,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 120540
+    },
+    {
+      "epoch": 793.0921052631579,
+      "grad_norm": 1.0039688348770142,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 120550
+    },
+    {
+      "epoch": 793.1578947368421,
+      "grad_norm": 0.904447078704834,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 120560
+    },
+    {
+      "epoch": 793.2236842105264,
+      "grad_norm": 0.8521278500556946,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 120570
+    },
+    {
+      "epoch": 793.2894736842105,
+      "grad_norm": 0.7753221988677979,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 120580
+    },
+    {
+      "epoch": 793.3552631578947,
+      "grad_norm": 1.0764966011047363,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 120590
+    },
+    {
+      "epoch": 793.421052631579,
+      "grad_norm": 1.086124062538147,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 120600
+    },
+    {
+      "epoch": 793.4868421052631,
+      "grad_norm": 0.8036643862724304,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 120610
+    },
+    {
+      "epoch": 793.5526315789474,
+      "grad_norm": 1.0477961301803589,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 120620
+    },
+    {
+      "epoch": 793.6184210526316,
+      "grad_norm": 0.8894155621528625,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 120630
+    },
+    {
+      "epoch": 793.6842105263158,
+      "grad_norm": 1.6042909622192383,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 120640
+    },
+    {
+      "epoch": 793.75,
+      "grad_norm": 0.8697584867477417,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 120650
+    },
+    {
+      "epoch": 793.8157894736842,
+      "grad_norm": 1.1683377027511597,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 120660
+    },
+    {
+      "epoch": 793.8815789473684,
+      "grad_norm": 1.1286085844039917,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 120670
+    },
+    {
+      "epoch": 793.9473684210526,
+      "grad_norm": 1.295102834701538,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 120680
+    },
+    {
+      "epoch": 794.0131578947369,
+      "grad_norm": 1.4578384160995483,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 120690
+    },
+    {
+      "epoch": 794.078947368421,
+      "grad_norm": 1.1735413074493408,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 120700
+    },
+    {
+      "epoch": 794.1447368421053,
+      "grad_norm": 0.7384294867515564,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 120710
+    },
+    {
+      "epoch": 794.2105263157895,
+      "grad_norm": 0.9748033285140991,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 120720
+    },
+    {
+      "epoch": 794.2763157894736,
+      "grad_norm": 1.08341646194458,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 120730
+    },
+    {
+      "epoch": 794.3421052631579,
+      "grad_norm": 1.2374165058135986,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 120740
+    },
+    {
+      "epoch": 794.4078947368421,
+      "grad_norm": 0.8181259632110596,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 120750
+    },
+    {
+      "epoch": 794.4736842105264,
+      "grad_norm": 1.0374884605407715,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 120760
+    },
+    {
+      "epoch": 794.5394736842105,
+      "grad_norm": 1.3677926063537598,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 120770
+    },
+    {
+      "epoch": 794.6052631578947,
+      "grad_norm": 0.8347147107124329,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 120780
+    },
+    {
+      "epoch": 794.671052631579,
+      "grad_norm": 0.8443951606750488,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 120790
+    },
+    {
+      "epoch": 794.7368421052631,
+      "grad_norm": 0.8409653902053833,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 120800
+    },
+    {
+      "epoch": 794.8026315789474,
+      "grad_norm": 0.8954428434371948,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 120810
+    },
+    {
+      "epoch": 794.8684210526316,
+      "grad_norm": 1.0372549295425415,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 120820
+    },
+    {
+      "epoch": 794.9342105263158,
+      "grad_norm": 0.8429449796676636,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 120830
+    },
+    {
+      "epoch": 795.0,
+      "grad_norm": 0.6495620608329773,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 120840
+    },
+    {
+      "epoch": 795.0657894736842,
+      "grad_norm": 1.2131069898605347,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 120850
+    },
+    {
+      "epoch": 795.1315789473684,
+      "grad_norm": 1.1015368700027466,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 120860
+    },
+    {
+      "epoch": 795.1973684210526,
+      "grad_norm": 0.9811515212059021,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 120870
+    },
+    {
+      "epoch": 795.2631578947369,
+      "grad_norm": 1.299547791481018,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 120880
+    },
+    {
+      "epoch": 795.328947368421,
+      "grad_norm": 1.1710044145584106,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 120890
+    },
+    {
+      "epoch": 795.3947368421053,
+      "grad_norm": 1.1657590866088867,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 120900
+    },
+    {
+      "epoch": 795.4605263157895,
+      "grad_norm": 0.8403371572494507,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 120910
+    },
+    {
+      "epoch": 795.5263157894736,
+      "grad_norm": 0.9806166887283325,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 120920
+    },
+    {
+      "epoch": 795.5921052631579,
+      "grad_norm": 1.4901376962661743,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 120930
+    },
+    {
+      "epoch": 795.6578947368421,
+      "grad_norm": 0.8995088338851929,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 120940
+    },
+    {
+      "epoch": 795.7236842105264,
+      "grad_norm": 1.037126064300537,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 120950
+    },
+    {
+      "epoch": 795.7894736842105,
+      "grad_norm": 1.0357317924499512,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 120960
+    },
+    {
+      "epoch": 795.8552631578947,
+      "grad_norm": 0.8257767558097839,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 120970
+    },
+    {
+      "epoch": 795.921052631579,
+      "grad_norm": 0.8299721479415894,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 120980
+    },
+    {
+      "epoch": 795.9868421052631,
+      "grad_norm": 1.0660762786865234,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 120990
+    },
+    {
+      "epoch": 796.0526315789474,
+      "grad_norm": 1.2139171361923218,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 121000
+    },
+    {
+      "epoch": 796.1184210526316,
+      "grad_norm": 1.1926602125167847,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 121010
+    },
+    {
+      "epoch": 796.1842105263158,
+      "grad_norm": 1.40194571018219,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 121020
+    },
+    {
+      "epoch": 796.25,
+      "grad_norm": 1.023759126663208,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 121030
+    },
+    {
+      "epoch": 796.3157894736842,
+      "grad_norm": 1.3803359270095825,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 121040
+    },
+    {
+      "epoch": 796.3815789473684,
+      "grad_norm": 1.2790648937225342,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 121050
+    },
+    {
+      "epoch": 796.4473684210526,
+      "grad_norm": 0.9783958792686462,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 121060
+    },
+    {
+      "epoch": 796.5131578947369,
+      "grad_norm": 1.3217369318008423,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 121070
+    },
+    {
+      "epoch": 796.578947368421,
+      "grad_norm": 1.2040969133377075,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 121080
+    },
+    {
+      "epoch": 796.6447368421053,
+      "grad_norm": 1.263850212097168,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 121090
+    },
+    {
+      "epoch": 796.7105263157895,
+      "grad_norm": 0.976997971534729,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 121100
+    },
+    {
+      "epoch": 796.7763157894736,
+      "grad_norm": 1.194297194480896,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 121110
+    },
+    {
+      "epoch": 796.8421052631579,
+      "grad_norm": 1.2234728336334229,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 121120
+    },
+    {
+      "epoch": 796.9078947368421,
+      "grad_norm": 1.7128822803497314,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 121130
+    },
+    {
+      "epoch": 796.9736842105264,
+      "grad_norm": 1.269959807395935,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 121140
+    },
+    {
+      "epoch": 797.0394736842105,
+      "grad_norm": 1.3292056322097778,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 121150
+    },
+    {
+      "epoch": 797.1052631578947,
+      "grad_norm": 0.8442802429199219,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 121160
+    },
+    {
+      "epoch": 797.171052631579,
+      "grad_norm": 0.9467982649803162,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 121170
+    },
+    {
+      "epoch": 797.2368421052631,
+      "grad_norm": 0.8293236494064331,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 121180
+    },
+    {
+      "epoch": 797.3026315789474,
+      "grad_norm": 0.5392968654632568,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 121190
+    },
+    {
+      "epoch": 797.3684210526316,
+      "grad_norm": 0.9286937713623047,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 121200
+    },
+    {
+      "epoch": 797.4342105263158,
+      "grad_norm": 0.6233976483345032,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 121210
+    },
+    {
+      "epoch": 797.5,
+      "grad_norm": 0.9368897080421448,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 121220
+    },
+    {
+      "epoch": 797.5657894736842,
+      "grad_norm": 0.8141809105873108,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 121230
+    },
+    {
+      "epoch": 797.6315789473684,
+      "grad_norm": 0.7223671674728394,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 121240
+    },
+    {
+      "epoch": 797.6973684210526,
+      "grad_norm": 1.1252926588058472,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 121250
+    },
+    {
+      "epoch": 797.7631578947369,
+      "grad_norm": 0.971419095993042,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 121260
+    },
+    {
+      "epoch": 797.828947368421,
+      "grad_norm": 1.5512263774871826,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 121270
+    },
+    {
+      "epoch": 797.8947368421053,
+      "grad_norm": 2.1177520751953125,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 121280
+    },
+    {
+      "epoch": 797.9605263157895,
+      "grad_norm": 1.3696571588516235,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 121290
+    },
+    {
+      "epoch": 798.0263157894736,
+      "grad_norm": 2.183379888534546,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 121300
+    },
+    {
+      "epoch": 798.0921052631579,
+      "grad_norm": 1.6594129800796509,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 121310
+    },
+    {
+      "epoch": 798.1578947368421,
+      "grad_norm": 1.3081175088882446,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 121320
+    },
+    {
+      "epoch": 798.2236842105264,
+      "grad_norm": 1.0478174686431885,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 121330
+    },
+    {
+      "epoch": 798.2894736842105,
+      "grad_norm": 1.081154227256775,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 121340
+    },
+    {
+      "epoch": 798.3552631578947,
+      "grad_norm": 1.172153353691101,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 121350
+    },
+    {
+      "epoch": 798.421052631579,
+      "grad_norm": 0.9921495318412781,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 121360
+    },
+    {
+      "epoch": 798.4868421052631,
+      "grad_norm": 0.8902541399002075,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 121370
+    },
+    {
+      "epoch": 798.5526315789474,
+      "grad_norm": 1.2591131925582886,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 121380
+    },
+    {
+      "epoch": 798.6184210526316,
+      "grad_norm": 1.258408546447754,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 121390
+    },
+    {
+      "epoch": 798.6842105263158,
+      "grad_norm": 0.9958367943763733,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 121400
+    },
+    {
+      "epoch": 798.75,
+      "grad_norm": 1.3845791816711426,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 121410
+    },
+    {
+      "epoch": 798.8157894736842,
+      "grad_norm": 1.2385156154632568,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 121420
+    },
+    {
+      "epoch": 798.8815789473684,
+      "grad_norm": 0.9330929517745972,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 121430
+    },
+    {
+      "epoch": 798.9473684210526,
+      "grad_norm": 1.2166731357574463,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 121440
+    },
+    {
+      "epoch": 799.0131578947369,
+      "grad_norm": 1.1683727502822876,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 121450
+    },
+    {
+      "epoch": 799.078947368421,
+      "grad_norm": 1.0285605192184448,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 121460
+    },
+    {
+      "epoch": 799.1447368421053,
+      "grad_norm": 0.9653977155685425,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 121470
+    },
+    {
+      "epoch": 799.2105263157895,
+      "grad_norm": 1.072914481163025,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 121480
+    },
+    {
+      "epoch": 799.2763157894736,
+      "grad_norm": 1.1126468181610107,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 121490
+    },
+    {
+      "epoch": 799.3421052631579,
+      "grad_norm": 1.0155147314071655,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 121500
+    },
+    {
+      "epoch": 799.4078947368421,
+      "grad_norm": 1.1577136516571045,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 121510
+    },
+    {
+      "epoch": 799.4736842105264,
+      "grad_norm": 1.2593013048171997,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 121520
+    },
+    {
+      "epoch": 799.5394736842105,
+      "grad_norm": 1.4376327991485596,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 121530
+    },
+    {
+      "epoch": 799.6052631578947,
+      "grad_norm": 1.1606967449188232,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 121540
+    },
+    {
+      "epoch": 799.671052631579,
+      "grad_norm": 1.1336442232131958,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 121550
+    },
+    {
+      "epoch": 799.7368421052631,
+      "grad_norm": 0.9730342626571655,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 121560
+    },
+    {
+      "epoch": 799.8026315789474,
+      "grad_norm": 1.030070185661316,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 121570
+    },
+    {
+      "epoch": 799.8684210526316,
+      "grad_norm": 0.6735216975212097,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 121580
+    },
+    {
+      "epoch": 799.9342105263158,
+      "grad_norm": 1.4537135362625122,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 121590
+    },
+    {
+      "epoch": 800.0,
+      "grad_norm": 0.9319224953651428,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 121600
+    },
+    {
+      "epoch": 800.0657894736842,
+      "grad_norm": 0.779762864112854,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 121610
+    },
+    {
+      "epoch": 800.1315789473684,
+      "grad_norm": 0.7982974648475647,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 121620
+    },
+    {
+      "epoch": 800.1973684210526,
+      "grad_norm": 0.9684889316558838,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 121630
+    },
+    {
+      "epoch": 800.2631578947369,
+      "grad_norm": 0.9351498484611511,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 121640
+    },
+    {
+      "epoch": 800.328947368421,
+      "grad_norm": 0.9390810132026672,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 121650
+    },
+    {
+      "epoch": 800.3947368421053,
+      "grad_norm": 0.9394250512123108,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 121660
+    },
+    {
+      "epoch": 800.4605263157895,
+      "grad_norm": 0.8497784733772278,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 121670
+    },
+    {
+      "epoch": 800.5263157894736,
+      "grad_norm": 0.9997221827507019,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 121680
+    },
+    {
+      "epoch": 800.5921052631579,
+      "grad_norm": 1.0118328332901,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 121690
+    },
+    {
+      "epoch": 800.6578947368421,
+      "grad_norm": 1.0781736373901367,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 121700
+    },
+    {
+      "epoch": 800.7236842105264,
+      "grad_norm": 0.9491978287696838,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 121710
+    },
+    {
+      "epoch": 800.7894736842105,
+      "grad_norm": 1.0116422176361084,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 121720
+    },
+    {
+      "epoch": 800.8552631578947,
+      "grad_norm": 1.0374714136123657,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 121730
+    },
+    {
+      "epoch": 800.921052631579,
+      "grad_norm": 0.8975595831871033,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 121740
+    },
+    {
+      "epoch": 800.9868421052631,
+      "grad_norm": 1.0354864597320557,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 121750
+    },
+    {
+      "epoch": 801.0526315789474,
+      "grad_norm": 1.3337525129318237,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 121760
+    },
+    {
+      "epoch": 801.1184210526316,
+      "grad_norm": 1.3484176397323608,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 121770
+    },
+    {
+      "epoch": 801.1842105263158,
+      "grad_norm": 0.9428836703300476,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 121780
+    },
+    {
+      "epoch": 801.25,
+      "grad_norm": 1.2294591665267944,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 121790
+    },
+    {
+      "epoch": 801.3157894736842,
+      "grad_norm": 1.2167091369628906,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 121800
+    },
+    {
+      "epoch": 801.3815789473684,
+      "grad_norm": 1.2219659090042114,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 121810
+    },
+    {
+      "epoch": 801.4473684210526,
+      "grad_norm": 1.1897832155227661,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 121820
+    },
+    {
+      "epoch": 801.5131578947369,
+      "grad_norm": 1.3615158796310425,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 121830
+    },
+    {
+      "epoch": 801.578947368421,
+      "grad_norm": 0.8604697585105896,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 121840
+    },
+    {
+      "epoch": 801.6447368421053,
+      "grad_norm": 1.149682879447937,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 121850
+    },
+    {
+      "epoch": 801.7105263157895,
+      "grad_norm": 0.803564190864563,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 121860
+    },
+    {
+      "epoch": 801.7763157894736,
+      "grad_norm": 1.223574161529541,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 121870
+    },
+    {
+      "epoch": 801.8421052631579,
+      "grad_norm": 0.9793342351913452,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 121880
+    },
+    {
+      "epoch": 801.9078947368421,
+      "grad_norm": 0.9008523225784302,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 121890
+    },
+    {
+      "epoch": 801.9736842105264,
+      "grad_norm": 0.9227433204650879,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 121900
+    },
+    {
+      "epoch": 802.0394736842105,
+      "grad_norm": 0.8591435551643372,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 121910
+    },
+    {
+      "epoch": 802.1052631578947,
+      "grad_norm": 1.1136678457260132,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 121920
+    },
+    {
+      "epoch": 802.171052631579,
+      "grad_norm": 0.9162068367004395,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 121930
+    },
+    {
+      "epoch": 802.2368421052631,
+      "grad_norm": 0.9551861882209778,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 121940
+    },
+    {
+      "epoch": 802.3026315789474,
+      "grad_norm": 0.9675922989845276,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 121950
+    },
+    {
+      "epoch": 802.3684210526316,
+      "grad_norm": 1.1318920850753784,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 121960
+    },
+    {
+      "epoch": 802.4342105263158,
+      "grad_norm": 0.9588920474052429,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 121970
+    },
+    {
+      "epoch": 802.5,
+      "grad_norm": 0.6656215190887451,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 121980
+    },
+    {
+      "epoch": 802.5657894736842,
+      "grad_norm": 1.1019165515899658,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 121990
+    },
+    {
+      "epoch": 802.6315789473684,
+      "grad_norm": 1.2802983522415161,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 122000
+    },
+    {
+      "epoch": 802.6973684210526,
+      "grad_norm": 1.1623402833938599,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 122010
+    },
+    {
+      "epoch": 802.7631578947369,
+      "grad_norm": 1.039617896080017,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 122020
+    },
+    {
+      "epoch": 802.828947368421,
+      "grad_norm": 0.8688013553619385,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 122030
+    },
+    {
+      "epoch": 802.8947368421053,
+      "grad_norm": 1.22115957736969,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 122040
+    },
+    {
+      "epoch": 802.9605263157895,
+      "grad_norm": 0.9164858460426331,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 122050
+    },
+    {
+      "epoch": 803.0263157894736,
+      "grad_norm": 0.8192176818847656,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 122060
+    },
+    {
+      "epoch": 803.0921052631579,
+      "grad_norm": 1.2811557054519653,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 122070
+    },
+    {
+      "epoch": 803.1578947368421,
+      "grad_norm": 1.3830119371414185,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 122080
+    },
+    {
+      "epoch": 803.2236842105264,
+      "grad_norm": 1.146261215209961,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 122090
+    },
+    {
+      "epoch": 803.2894736842105,
+      "grad_norm": 0.9297433495521545,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 122100
+    },
+    {
+      "epoch": 803.3552631578947,
+      "grad_norm": 0.8575513958930969,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 122110
+    },
+    {
+      "epoch": 803.421052631579,
+      "grad_norm": 1.1129348278045654,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 122120
+    },
+    {
+      "epoch": 803.4868421052631,
+      "grad_norm": 1.1914259195327759,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 122130
+    },
+    {
+      "epoch": 803.5526315789474,
+      "grad_norm": 0.8768858909606934,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 122140
+    },
+    {
+      "epoch": 803.6184210526316,
+      "grad_norm": 0.8389908671379089,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 122150
+    },
+    {
+      "epoch": 803.6842105263158,
+      "grad_norm": 1.1169975996017456,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 122160
+    },
+    {
+      "epoch": 803.75,
+      "grad_norm": 0.8531729578971863,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 122170
+    },
+    {
+      "epoch": 803.8157894736842,
+      "grad_norm": 1.1040211915969849,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 122180
+    },
+    {
+      "epoch": 803.8815789473684,
+      "grad_norm": 0.9061696529388428,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 122190
+    },
+    {
+      "epoch": 803.9473684210526,
+      "grad_norm": 0.9023482799530029,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 122200
+    },
+    {
+      "epoch": 804.0131578947369,
+      "grad_norm": 1.1508461236953735,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 122210
+    },
+    {
+      "epoch": 804.078947368421,
+      "grad_norm": 1.0282151699066162,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 122220
+    },
+    {
+      "epoch": 804.1447368421053,
+      "grad_norm": 1.2764521837234497,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 122230
+    },
+    {
+      "epoch": 804.2105263157895,
+      "grad_norm": 1.1191545724868774,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 122240
+    },
+    {
+      "epoch": 804.2763157894736,
+      "grad_norm": 0.988823413848877,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 122250
+    },
+    {
+      "epoch": 804.3421052631579,
+      "grad_norm": 1.095762848854065,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 122260
+    },
+    {
+      "epoch": 804.4078947368421,
+      "grad_norm": 1.3492151498794556,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 122270
+    },
+    {
+      "epoch": 804.4736842105264,
+      "grad_norm": 1.2106963396072388,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 122280
+    },
+    {
+      "epoch": 804.5394736842105,
+      "grad_norm": 1.2582029104232788,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 122290
+    },
+    {
+      "epoch": 804.6052631578947,
+      "grad_norm": 1.3207285404205322,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 122300
+    },
+    {
+      "epoch": 804.671052631579,
+      "grad_norm": 1.0762017965316772,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 122310
+    },
+    {
+      "epoch": 804.7368421052631,
+      "grad_norm": 0.871727466583252,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 122320
+    },
+    {
+      "epoch": 804.8026315789474,
+      "grad_norm": 0.9750553369522095,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 122330
+    },
+    {
+      "epoch": 804.8684210526316,
+      "grad_norm": 1.4661288261413574,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 122340
+    },
+    {
+      "epoch": 804.9342105263158,
+      "grad_norm": 1.1938623189926147,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 122350
+    },
+    {
+      "epoch": 805.0,
+      "grad_norm": 1.210606575012207,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 122360
+    },
+    {
+      "epoch": 805.0657894736842,
+      "grad_norm": 0.9685527086257935,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 122370
+    },
+    {
+      "epoch": 805.1315789473684,
+      "grad_norm": 0.7988755702972412,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 122380
+    },
+    {
+      "epoch": 805.1973684210526,
+      "grad_norm": 1.0523570775985718,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 122390
+    },
+    {
+      "epoch": 805.2631578947369,
+      "grad_norm": 0.9532704949378967,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 122400
+    },
+    {
+      "epoch": 805.328947368421,
+      "grad_norm": 0.774535059928894,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 122410
+    },
+    {
+      "epoch": 805.3947368421053,
+      "grad_norm": 1.0820705890655518,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 122420
+    },
+    {
+      "epoch": 805.4605263157895,
+      "grad_norm": 1.2834467887878418,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 122430
+    },
+    {
+      "epoch": 805.5263157894736,
+      "grad_norm": 1.2776272296905518,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 122440
+    },
+    {
+      "epoch": 805.5921052631579,
+      "grad_norm": 1.7061843872070312,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 122450
+    },
+    {
+      "epoch": 805.6578947368421,
+      "grad_norm": 1.2151910066604614,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 122460
+    },
+    {
+      "epoch": 805.7236842105264,
+      "grad_norm": 0.9397733211517334,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 122470
+    },
+    {
+      "epoch": 805.7894736842105,
+      "grad_norm": 1.0783069133758545,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 122480
+    },
+    {
+      "epoch": 805.8552631578947,
+      "grad_norm": 0.965991735458374,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 122490
+    },
+    {
+      "epoch": 805.921052631579,
+      "grad_norm": 1.1885125637054443,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 122500
+    },
+    {
+      "epoch": 805.9868421052631,
+      "grad_norm": 1.4462835788726807,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 122510
+    },
+    {
+      "epoch": 806.0526315789474,
+      "grad_norm": 1.4112342596054077,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 122520
+    },
+    {
+      "epoch": 806.1184210526316,
+      "grad_norm": 1.3368967771530151,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 122530
+    },
+    {
+      "epoch": 806.1842105263158,
+      "grad_norm": 0.9588228464126587,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 122540
+    },
+    {
+      "epoch": 806.25,
+      "grad_norm": 1.0012880563735962,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 122550
+    },
+    {
+      "epoch": 806.3157894736842,
+      "grad_norm": 1.0490740537643433,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 122560
+    },
+    {
+      "epoch": 806.3815789473684,
+      "grad_norm": 1.1824147701263428,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 122570
+    },
+    {
+      "epoch": 806.4473684210526,
+      "grad_norm": 1.1107277870178223,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 122580
+    },
+    {
+      "epoch": 806.5131578947369,
+      "grad_norm": 0.9902931451797485,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 122590
+    },
+    {
+      "epoch": 806.578947368421,
+      "grad_norm": 1.0864098072052002,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 122600
+    },
+    {
+      "epoch": 806.6447368421053,
+      "grad_norm": 1.0239638090133667,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 122610
+    },
+    {
+      "epoch": 806.7105263157895,
+      "grad_norm": 0.9125197529792786,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 122620
+    },
+    {
+      "epoch": 806.7763157894736,
+      "grad_norm": 0.8427323698997498,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 122630
+    },
+    {
+      "epoch": 806.8421052631579,
+      "grad_norm": 0.6944894790649414,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 122640
+    },
+    {
+      "epoch": 806.9078947368421,
+      "grad_norm": 0.6341553926467896,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 122650
+    },
+    {
+      "epoch": 806.9736842105264,
+      "grad_norm": 1.362230896949768,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 122660
+    },
+    {
+      "epoch": 807.0394736842105,
+      "grad_norm": 1.2482671737670898,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 122670
+    },
+    {
+      "epoch": 807.1052631578947,
+      "grad_norm": 1.413817048072815,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 122680
+    },
+    {
+      "epoch": 807.171052631579,
+      "grad_norm": 1.4257447719573975,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 122690
+    },
+    {
+      "epoch": 807.2368421052631,
+      "grad_norm": 1.0907763242721558,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 122700
+    },
+    {
+      "epoch": 807.3026315789474,
+      "grad_norm": 1.1896060705184937,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 122710
+    },
+    {
+      "epoch": 807.3684210526316,
+      "grad_norm": 0.9209076166152954,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 122720
+    },
+    {
+      "epoch": 807.4342105263158,
+      "grad_norm": 0.7475315928459167,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 122730
+    },
+    {
+      "epoch": 807.5,
+      "grad_norm": 1.2551970481872559,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 122740
+    },
+    {
+      "epoch": 807.5657894736842,
+      "grad_norm": 1.2004441022872925,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 122750
+    },
+    {
+      "epoch": 807.6315789473684,
+      "grad_norm": 1.0100356340408325,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 122760
+    },
+    {
+      "epoch": 807.6973684210526,
+      "grad_norm": 1.3553928136825562,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 122770
+    },
+    {
+      "epoch": 807.7631578947369,
+      "grad_norm": 0.8858333826065063,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 122780
+    },
+    {
+      "epoch": 807.828947368421,
+      "grad_norm": 1.5102218389511108,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 122790
+    },
+    {
+      "epoch": 807.8947368421053,
+      "grad_norm": 1.0069150924682617,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 122800
+    },
+    {
+      "epoch": 807.9605263157895,
+      "grad_norm": 0.7301906943321228,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 122810
+    },
+    {
+      "epoch": 808.0263157894736,
+      "grad_norm": 1.0068855285644531,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 122820
+    },
+    {
+      "epoch": 808.0921052631579,
+      "grad_norm": 0.9534813165664673,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 122830
+    },
+    {
+      "epoch": 808.1578947368421,
+      "grad_norm": 0.9146772027015686,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 122840
+    },
+    {
+      "epoch": 808.2236842105264,
+      "grad_norm": 0.9505394697189331,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 122850
+    },
+    {
+      "epoch": 808.2894736842105,
+      "grad_norm": 1.0315290689468384,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 122860
+    },
+    {
+      "epoch": 808.3552631578947,
+      "grad_norm": 1.236918568611145,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 122870
+    },
+    {
+      "epoch": 808.421052631579,
+      "grad_norm": 1.0912704467773438,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 122880
+    },
+    {
+      "epoch": 808.4868421052631,
+      "grad_norm": 0.8193346261978149,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 122890
+    },
+    {
+      "epoch": 808.5526315789474,
+      "grad_norm": 1.54730224609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 122900
+    },
+    {
+      "epoch": 808.6184210526316,
+      "grad_norm": 1.441839575767517,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 122910
+    },
+    {
+      "epoch": 808.6842105263158,
+      "grad_norm": 1.055781602859497,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 122920
+    },
+    {
+      "epoch": 808.75,
+      "grad_norm": 1.353714108467102,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 122930
+    },
+    {
+      "epoch": 808.8157894736842,
+      "grad_norm": 1.2414506673812866,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 122940
+    },
+    {
+      "epoch": 808.8815789473684,
+      "grad_norm": 1.2906715869903564,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 122950
+    },
+    {
+      "epoch": 808.9473684210526,
+      "grad_norm": 0.7939502596855164,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 122960
+    },
+    {
+      "epoch": 809.0131578947369,
+      "grad_norm": 0.9539815783500671,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 122970
+    },
+    {
+      "epoch": 809.078947368421,
+      "grad_norm": 0.7728420495986938,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 122980
+    },
+    {
+      "epoch": 809.1447368421053,
+      "grad_norm": 0.9012200236320496,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 122990
+    },
+    {
+      "epoch": 809.2105263157895,
+      "grad_norm": 0.874159574508667,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 123000
+    },
+    {
+      "epoch": 809.2763157894736,
+      "grad_norm": 1.1063437461853027,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 123010
+    },
+    {
+      "epoch": 809.3421052631579,
+      "grad_norm": 1.1049076318740845,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 123020
+    },
+    {
+      "epoch": 809.4078947368421,
+      "grad_norm": 1.4535138607025146,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 123030
+    },
+    {
+      "epoch": 809.4736842105264,
+      "grad_norm": 1.022336483001709,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 123040
+    },
+    {
+      "epoch": 809.5394736842105,
+      "grad_norm": 1.1350669860839844,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 123050
+    },
+    {
+      "epoch": 809.6052631578947,
+      "grad_norm": 1.5300946235656738,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 123060
+    },
+    {
+      "epoch": 809.671052631579,
+      "grad_norm": 1.1498500108718872,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 123070
+    },
+    {
+      "epoch": 809.7368421052631,
+      "grad_norm": 0.6810249090194702,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 123080
+    },
+    {
+      "epoch": 809.8026315789474,
+      "grad_norm": 0.9116702675819397,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 123090
+    },
+    {
+      "epoch": 809.8684210526316,
+      "grad_norm": 1.095033049583435,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 123100
+    },
+    {
+      "epoch": 809.9342105263158,
+      "grad_norm": 0.9382158517837524,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 123110
+    },
+    {
+      "epoch": 810.0,
+      "grad_norm": 1.0285440683364868,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 123120
+    },
+    {
+      "epoch": 810.0657894736842,
+      "grad_norm": 0.9891853332519531,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 123130
+    },
+    {
+      "epoch": 810.1315789473684,
+      "grad_norm": 0.9508638381958008,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 123140
+    },
+    {
+      "epoch": 810.1973684210526,
+      "grad_norm": 0.9087507724761963,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 123150
+    },
+    {
+      "epoch": 810.2631578947369,
+      "grad_norm": 1.0318292379379272,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 123160
+    },
+    {
+      "epoch": 810.328947368421,
+      "grad_norm": 0.8435489535331726,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 123170
+    },
+    {
+      "epoch": 810.3947368421053,
+      "grad_norm": 0.7418573498725891,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 123180
+    },
+    {
+      "epoch": 810.4605263157895,
+      "grad_norm": 1.2154449224472046,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 123190
+    },
+    {
+      "epoch": 810.5263157894736,
+      "grad_norm": 1.1838921308517456,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 123200
+    },
+    {
+      "epoch": 810.5921052631579,
+      "grad_norm": 0.6878968477249146,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 123210
+    },
+    {
+      "epoch": 810.6578947368421,
+      "grad_norm": 1.2046258449554443,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 123220
+    },
+    {
+      "epoch": 810.7236842105264,
+      "grad_norm": 1.1671987771987915,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 123230
+    },
+    {
+      "epoch": 810.7894736842105,
+      "grad_norm": 1.0558589696884155,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 123240
+    },
+    {
+      "epoch": 810.8552631578947,
+      "grad_norm": 1.035437822341919,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 123250
+    },
+    {
+      "epoch": 810.921052631579,
+      "grad_norm": 1.2086669206619263,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 123260
+    },
+    {
+      "epoch": 810.9868421052631,
+      "grad_norm": 1.086695909500122,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 123270
+    },
+    {
+      "epoch": 811.0526315789474,
+      "grad_norm": 1.3100385665893555,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 123280
+    },
+    {
+      "epoch": 811.1184210526316,
+      "grad_norm": 1.3624625205993652,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 123290
+    },
+    {
+      "epoch": 811.1842105263158,
+      "grad_norm": 1.1712080240249634,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 123300
+    },
+    {
+      "epoch": 811.25,
+      "grad_norm": 1.2696281671524048,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 123310
+    },
+    {
+      "epoch": 811.3157894736842,
+      "grad_norm": 0.7880839109420776,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 123320
+    },
+    {
+      "epoch": 811.3815789473684,
+      "grad_norm": 1.3143435716629028,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 123330
+    },
+    {
+      "epoch": 811.4473684210526,
+      "grad_norm": 1.0152329206466675,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 123340
+    },
+    {
+      "epoch": 811.5131578947369,
+      "grad_norm": 0.9424976110458374,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 123350
+    },
+    {
+      "epoch": 811.578947368421,
+      "grad_norm": 1.0948466062545776,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 123360
+    },
+    {
+      "epoch": 811.6447368421053,
+      "grad_norm": 0.9606415033340454,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 123370
+    },
+    {
+      "epoch": 811.7105263157895,
+      "grad_norm": 1.1549152135849,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 123380
+    },
+    {
+      "epoch": 811.7763157894736,
+      "grad_norm": 0.8370715975761414,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 123390
+    },
+    {
+      "epoch": 811.8421052631579,
+      "grad_norm": 1.1757169961929321,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 123400
+    },
+    {
+      "epoch": 811.9078947368421,
+      "grad_norm": 1.3213396072387695,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 123410
+    },
+    {
+      "epoch": 811.9736842105264,
+      "grad_norm": 1.3721901178359985,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 123420
+    },
+    {
+      "epoch": 812.0394736842105,
+      "grad_norm": 1.2278730869293213,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 123430
+    },
+    {
+      "epoch": 812.1052631578947,
+      "grad_norm": 0.8597956895828247,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 123440
+    },
+    {
+      "epoch": 812.171052631579,
+      "grad_norm": 1.0379050970077515,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 123450
+    },
+    {
+      "epoch": 812.2368421052631,
+      "grad_norm": 1.107707142829895,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 123460
+    },
+    {
+      "epoch": 812.3026315789474,
+      "grad_norm": 1.37968909740448,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 123470
+    },
+    {
+      "epoch": 812.3684210526316,
+      "grad_norm": 1.6168029308319092,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 123480
+    },
+    {
+      "epoch": 812.4342105263158,
+      "grad_norm": 1.3366464376449585,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 123490
+    },
+    {
+      "epoch": 812.5,
+      "grad_norm": 1.2438991069793701,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 123500
+    },
+    {
+      "epoch": 812.5657894736842,
+      "grad_norm": 1.4187861680984497,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 123510
+    },
+    {
+      "epoch": 812.6315789473684,
+      "grad_norm": 1.350807547569275,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 123520
+    },
+    {
+      "epoch": 812.6973684210526,
+      "grad_norm": 1.1984448432922363,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 123530
+    },
+    {
+      "epoch": 812.7631578947369,
+      "grad_norm": 0.9718255400657654,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 123540
+    },
+    {
+      "epoch": 812.828947368421,
+      "grad_norm": 0.8007934093475342,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 123550
+    },
+    {
+      "epoch": 812.8947368421053,
+      "grad_norm": 1.1120939254760742,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 123560
+    },
+    {
+      "epoch": 812.9605263157895,
+      "grad_norm": 0.8073641657829285,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 123570
+    },
+    {
+      "epoch": 813.0263157894736,
+      "grad_norm": 1.3328661918640137,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 123580
+    },
+    {
+      "epoch": 813.0921052631579,
+      "grad_norm": 1.2822130918502808,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 123590
+    },
+    {
+      "epoch": 813.1578947368421,
+      "grad_norm": 1.1979964971542358,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 123600
+    },
+    {
+      "epoch": 813.2236842105264,
+      "grad_norm": 1.2304644584655762,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 123610
+    },
+    {
+      "epoch": 813.2894736842105,
+      "grad_norm": 0.8856472373008728,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 123620
+    },
+    {
+      "epoch": 813.3552631578947,
+      "grad_norm": 1.14380943775177,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 123630
+    },
+    {
+      "epoch": 813.421052631579,
+      "grad_norm": 1.169329047203064,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 123640
+    },
+    {
+      "epoch": 813.4868421052631,
+      "grad_norm": 1.1275250911712646,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 123650
+    },
+    {
+      "epoch": 813.5526315789474,
+      "grad_norm": 0.9524691700935364,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 123660
+    },
+    {
+      "epoch": 813.6184210526316,
+      "grad_norm": 1.1175603866577148,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 123670
+    },
+    {
+      "epoch": 813.6842105263158,
+      "grad_norm": 1.0256012678146362,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 123680
+    },
+    {
+      "epoch": 813.75,
+      "grad_norm": 0.9166593551635742,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 123690
+    },
+    {
+      "epoch": 813.8157894736842,
+      "grad_norm": 1.1852658987045288,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 123700
+    },
+    {
+      "epoch": 813.8815789473684,
+      "grad_norm": 1.0424803495407104,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 123710
+    },
+    {
+      "epoch": 813.9473684210526,
+      "grad_norm": 0.9205971956253052,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 123720
+    },
+    {
+      "epoch": 814.0131578947369,
+      "grad_norm": 0.8064239621162415,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 123730
+    },
+    {
+      "epoch": 814.078947368421,
+      "grad_norm": 1.1456804275512695,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 123740
+    },
+    {
+      "epoch": 814.1447368421053,
+      "grad_norm": 1.1536784172058105,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 123750
+    },
+    {
+      "epoch": 814.2105263157895,
+      "grad_norm": 0.9535488486289978,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 123760
+    },
+    {
+      "epoch": 814.2763157894736,
+      "grad_norm": 0.8925880193710327,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 123770
+    },
+    {
+      "epoch": 814.3421052631579,
+      "grad_norm": 0.7770874500274658,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 123780
+    },
+    {
+      "epoch": 814.4078947368421,
+      "grad_norm": 0.7658482193946838,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 123790
+    },
+    {
+      "epoch": 814.4736842105264,
+      "grad_norm": 1.1080727577209473,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 123800
+    },
+    {
+      "epoch": 814.5394736842105,
+      "grad_norm": 0.9930034279823303,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 123810
+    },
+    {
+      "epoch": 814.6052631578947,
+      "grad_norm": 1.04108464717865,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 123820
+    },
+    {
+      "epoch": 814.671052631579,
+      "grad_norm": 1.4060074090957642,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 123830
+    },
+    {
+      "epoch": 814.7368421052631,
+      "grad_norm": 1.135053277015686,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 123840
+    },
+    {
+      "epoch": 814.8026315789474,
+      "grad_norm": 1.102472186088562,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 123850
+    },
+    {
+      "epoch": 814.8684210526316,
+      "grad_norm": 1.0958255529403687,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 123860
+    },
+    {
+      "epoch": 814.9342105263158,
+      "grad_norm": 1.2109631299972534,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 123870
+    },
+    {
+      "epoch": 815.0,
+      "grad_norm": 0.8327732086181641,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 123880
+    },
+    {
+      "epoch": 815.0657894736842,
+      "grad_norm": 0.830962598323822,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 123890
+    },
+    {
+      "epoch": 815.1315789473684,
+      "grad_norm": 1.0585317611694336,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 123900
+    },
+    {
+      "epoch": 815.1973684210526,
+      "grad_norm": 1.095977783203125,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 123910
+    },
+    {
+      "epoch": 815.2631578947369,
+      "grad_norm": 1.2704319953918457,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 123920
+    },
+    {
+      "epoch": 815.328947368421,
+      "grad_norm": 0.957330584526062,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 123930
+    },
+    {
+      "epoch": 815.3947368421053,
+      "grad_norm": 0.8996625542640686,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 123940
+    },
+    {
+      "epoch": 815.4605263157895,
+      "grad_norm": 1.2398024797439575,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 123950
+    },
+    {
+      "epoch": 815.5263157894736,
+      "grad_norm": 1.1726700067520142,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 123960
+    },
+    {
+      "epoch": 815.5921052631579,
+      "grad_norm": 1.1765739917755127,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 123970
+    },
+    {
+      "epoch": 815.6578947368421,
+      "grad_norm": 1.4318251609802246,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 123980
+    },
+    {
+      "epoch": 815.7236842105264,
+      "grad_norm": 1.3720372915267944,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 123990
+    },
+    {
+      "epoch": 815.7894736842105,
+      "grad_norm": 0.9442108273506165,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 124000
+    },
+    {
+      "epoch": 815.8552631578947,
+      "grad_norm": 1.4249804019927979,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 124010
+    },
+    {
+      "epoch": 815.921052631579,
+      "grad_norm": 1.5111039876937866,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 124020
+    },
+    {
+      "epoch": 815.9868421052631,
+      "grad_norm": 1.065873146057129,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 124030
+    },
+    {
+      "epoch": 816.0526315789474,
+      "grad_norm": 1.224717617034912,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 124040
+    },
+    {
+      "epoch": 816.1184210526316,
+      "grad_norm": 0.8708517551422119,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 124050
+    },
+    {
+      "epoch": 816.1842105263158,
+      "grad_norm": 1.0297471284866333,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 124060
+    },
+    {
+      "epoch": 816.25,
+      "grad_norm": 1.1556639671325684,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 124070
+    },
+    {
+      "epoch": 816.3157894736842,
+      "grad_norm": 1.0219916105270386,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 124080
+    },
+    {
+      "epoch": 816.3815789473684,
+      "grad_norm": 0.9908708930015564,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 124090
+    },
+    {
+      "epoch": 816.4473684210526,
+      "grad_norm": 0.8453224897384644,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 124100
+    },
+    {
+      "epoch": 816.5131578947369,
+      "grad_norm": 0.6190137267112732,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 124110
+    },
+    {
+      "epoch": 816.578947368421,
+      "grad_norm": 1.4555507898330688,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 124120
+    },
+    {
+      "epoch": 816.6447368421053,
+      "grad_norm": 1.2438726425170898,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 124130
+    },
+    {
+      "epoch": 816.7105263157895,
+      "grad_norm": 1.2123976945877075,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 124140
+    },
+    {
+      "epoch": 816.7763157894736,
+      "grad_norm": 1.1773884296417236,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 124150
+    },
+    {
+      "epoch": 816.8421052631579,
+      "grad_norm": 1.0776981115341187,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 124160
+    },
+    {
+      "epoch": 816.9078947368421,
+      "grad_norm": 1.0977414846420288,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 124170
+    },
+    {
+      "epoch": 816.9736842105264,
+      "grad_norm": 0.9133142232894897,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 124180
+    },
+    {
+      "epoch": 817.0394736842105,
+      "grad_norm": 1.3177136182785034,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 124190
+    },
+    {
+      "epoch": 817.1052631578947,
+      "grad_norm": 1.1052703857421875,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 124200
+    },
+    {
+      "epoch": 817.171052631579,
+      "grad_norm": 1.2231838703155518,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 124210
+    },
+    {
+      "epoch": 817.2368421052631,
+      "grad_norm": 1.0773632526397705,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 124220
+    },
+    {
+      "epoch": 817.3026315789474,
+      "grad_norm": 0.7591347098350525,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 124230
+    },
+    {
+      "epoch": 817.3684210526316,
+      "grad_norm": 1.1627017259597778,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 124240
+    },
+    {
+      "epoch": 817.4342105263158,
+      "grad_norm": 0.88426274061203,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 124250
+    },
+    {
+      "epoch": 817.5,
+      "grad_norm": 1.0880540609359741,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 124260
+    },
+    {
+      "epoch": 817.5657894736842,
+      "grad_norm": 1.1447336673736572,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 124270
+    },
+    {
+      "epoch": 817.6315789473684,
+      "grad_norm": 0.9627189636230469,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 124280
+    },
+    {
+      "epoch": 817.6973684210526,
+      "grad_norm": 0.8809316158294678,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 124290
+    },
+    {
+      "epoch": 817.7631578947369,
+      "grad_norm": 1.1471461057662964,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 124300
+    },
+    {
+      "epoch": 817.828947368421,
+      "grad_norm": 0.9346367716789246,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 124310
+    },
+    {
+      "epoch": 817.8947368421053,
+      "grad_norm": 1.247454047203064,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 124320
+    },
+    {
+      "epoch": 817.9605263157895,
+      "grad_norm": 1.3746150732040405,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 124330
+    },
+    {
+      "epoch": 818.0263157894736,
+      "grad_norm": 1.183838963508606,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 124340
+    },
+    {
+      "epoch": 818.0921052631579,
+      "grad_norm": 0.9270862936973572,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 124350
+    },
+    {
+      "epoch": 818.1578947368421,
+      "grad_norm": 0.9272816777229309,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 124360
+    },
+    {
+      "epoch": 818.2236842105264,
+      "grad_norm": 1.0511780977249146,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 124370
+    },
+    {
+      "epoch": 818.2894736842105,
+      "grad_norm": 1.2871315479278564,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 124380
+    },
+    {
+      "epoch": 818.3552631578947,
+      "grad_norm": 1.2518093585968018,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 124390
+    },
+    {
+      "epoch": 818.421052631579,
+      "grad_norm": 1.498147964477539,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 124400
+    },
+    {
+      "epoch": 818.4868421052631,
+      "grad_norm": 2.1815390586853027,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 124410
+    },
+    {
+      "epoch": 818.5526315789474,
+      "grad_norm": 1.512572169303894,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 124420
+    },
+    {
+      "epoch": 818.6184210526316,
+      "grad_norm": 1.39220130443573,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 124430
+    },
+    {
+      "epoch": 818.6842105263158,
+      "grad_norm": 1.40886390209198,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 124440
+    },
+    {
+      "epoch": 818.75,
+      "grad_norm": 1.2492808103561401,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 124450
+    },
+    {
+      "epoch": 818.8157894736842,
+      "grad_norm": 1.3465094566345215,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 124460
+    },
+    {
+      "epoch": 818.8815789473684,
+      "grad_norm": 1.4258430004119873,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 124470
+    },
+    {
+      "epoch": 818.9473684210526,
+      "grad_norm": 1.4436920881271362,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 124480
+    },
+    {
+      "epoch": 819.0131578947369,
+      "grad_norm": 1.0473710298538208,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 124490
+    },
+    {
+      "epoch": 819.078947368421,
+      "grad_norm": 1.4626469612121582,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 124500
+    },
+    {
+      "epoch": 819.1447368421053,
+      "grad_norm": 1.2455971240997314,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 124510
+    },
+    {
+      "epoch": 819.2105263157895,
+      "grad_norm": 0.8882140517234802,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 124520
+    },
+    {
+      "epoch": 819.2763157894736,
+      "grad_norm": 1.2379828691482544,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 124530
+    },
+    {
+      "epoch": 819.3421052631579,
+      "grad_norm": 0.7983285188674927,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 124540
+    },
+    {
+      "epoch": 819.4078947368421,
+      "grad_norm": 0.8106863498687744,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 124550
+    },
+    {
+      "epoch": 819.4736842105264,
+      "grad_norm": 1.1622852087020874,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 124560
+    },
+    {
+      "epoch": 819.5394736842105,
+      "grad_norm": 0.733343243598938,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 124570
+    },
+    {
+      "epoch": 819.6052631578947,
+      "grad_norm": 0.8690012693405151,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 124580
+    },
+    {
+      "epoch": 819.671052631579,
+      "grad_norm": 0.7145095467567444,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 124590
+    },
+    {
+      "epoch": 819.7368421052631,
+      "grad_norm": 1.0075265169143677,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 124600
+    },
+    {
+      "epoch": 819.8026315789474,
+      "grad_norm": 1.1366093158721924,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 124610
+    },
+    {
+      "epoch": 819.8684210526316,
+      "grad_norm": 0.9843047261238098,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 124620
+    },
+    {
+      "epoch": 819.9342105263158,
+      "grad_norm": 0.7279674410820007,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 124630
+    },
+    {
+      "epoch": 820.0,
+      "grad_norm": 1.1037843227386475,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 124640
+    },
+    {
+      "epoch": 820.0657894736842,
+      "grad_norm": 0.7516036033630371,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 124650
+    },
+    {
+      "epoch": 820.1315789473684,
+      "grad_norm": 1.1608238220214844,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 124660
+    },
+    {
+      "epoch": 820.1973684210526,
+      "grad_norm": 1.3053244352340698,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 124670
+    },
+    {
+      "epoch": 820.2631578947369,
+      "grad_norm": 1.0844759941101074,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 124680
+    },
+    {
+      "epoch": 820.328947368421,
+      "grad_norm": 0.9442291259765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 124690
+    },
+    {
+      "epoch": 820.3947368421053,
+      "grad_norm": 1.1379389762878418,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 124700
+    },
+    {
+      "epoch": 820.4605263157895,
+      "grad_norm": 0.6525788903236389,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 124710
+    },
+    {
+      "epoch": 820.5263157894736,
+      "grad_norm": 1.278929352760315,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 124720
+    },
+    {
+      "epoch": 820.5921052631579,
+      "grad_norm": 1.4533745050430298,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 124730
+    },
+    {
+      "epoch": 820.6578947368421,
+      "grad_norm": 1.0017926692962646,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 124740
+    },
+    {
+      "epoch": 820.7236842105264,
+      "grad_norm": 0.9450610876083374,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 124750
+    },
+    {
+      "epoch": 820.7894736842105,
+      "grad_norm": 0.6079602837562561,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 124760
+    },
+    {
+      "epoch": 820.8552631578947,
+      "grad_norm": 1.112248420715332,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 124770
+    },
+    {
+      "epoch": 820.921052631579,
+      "grad_norm": 1.2943469285964966,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 124780
+    },
+    {
+      "epoch": 820.9868421052631,
+      "grad_norm": 1.1327763795852661,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 124790
+    },
+    {
+      "epoch": 821.0526315789474,
+      "grad_norm": 1.0445799827575684,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 124800
+    },
+    {
+      "epoch": 821.1184210526316,
+      "grad_norm": 1.081302285194397,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 124810
+    },
+    {
+      "epoch": 821.1842105263158,
+      "grad_norm": 1.0378952026367188,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 124820
+    },
+    {
+      "epoch": 821.25,
+      "grad_norm": 0.7026200890541077,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 124830
+    },
+    {
+      "epoch": 821.3157894736842,
+      "grad_norm": 0.709734320640564,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 124840
+    },
+    {
+      "epoch": 821.3815789473684,
+      "grad_norm": 1.214118480682373,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 124850
+    },
+    {
+      "epoch": 821.4473684210526,
+      "grad_norm": 1.0140697956085205,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 124860
+    },
+    {
+      "epoch": 821.5131578947369,
+      "grad_norm": 1.1845144033432007,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 124870
+    },
+    {
+      "epoch": 821.578947368421,
+      "grad_norm": 1.131510853767395,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 124880
+    },
+    {
+      "epoch": 821.6447368421053,
+      "grad_norm": 1.0710123777389526,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 124890
+    },
+    {
+      "epoch": 821.7105263157895,
+      "grad_norm": 0.9462993741035461,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 124900
+    },
+    {
+      "epoch": 821.7763157894736,
+      "grad_norm": 1.1488789319992065,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 124910
+    },
+    {
+      "epoch": 821.8421052631579,
+      "grad_norm": 0.8153425455093384,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 124920
+    },
+    {
+      "epoch": 821.9078947368421,
+      "grad_norm": 1.136583924293518,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 124930
+    },
+    {
+      "epoch": 821.9736842105264,
+      "grad_norm": 1.1596461534500122,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 124940
+    },
+    {
+      "epoch": 822.0394736842105,
+      "grad_norm": 1.1383651494979858,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 124950
+    },
+    {
+      "epoch": 822.1052631578947,
+      "grad_norm": 1.0268265008926392,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 124960
+    },
+    {
+      "epoch": 822.171052631579,
+      "grad_norm": 1.239668607711792,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 124970
+    },
+    {
+      "epoch": 822.2368421052631,
+      "grad_norm": 1.1452547311782837,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 124980
+    },
+    {
+      "epoch": 822.3026315789474,
+      "grad_norm": 0.804125189781189,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 124990
+    },
+    {
+      "epoch": 822.3684210526316,
+      "grad_norm": 1.5917028188705444,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 125000
+    },
+    {
+      "epoch": 822.4342105263158,
+      "grad_norm": 1.280947208404541,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 125010
+    },
+    {
+      "epoch": 822.5,
+      "grad_norm": 0.9528753161430359,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 125020
+    },
+    {
+      "epoch": 822.5657894736842,
+      "grad_norm": 0.9198108911514282,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 125030
+    },
+    {
+      "epoch": 822.6315789473684,
+      "grad_norm": 0.8453339338302612,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 125040
+    },
+    {
+      "epoch": 822.6973684210526,
+      "grad_norm": 1.1574821472167969,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125050
+    },
+    {
+      "epoch": 822.7631578947369,
+      "grad_norm": 1.1364651918411255,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 125060
+    },
+    {
+      "epoch": 822.828947368421,
+      "grad_norm": 0.9139201641082764,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 125070
+    },
+    {
+      "epoch": 822.8947368421053,
+      "grad_norm": 0.8016061782836914,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 125080
+    },
+    {
+      "epoch": 822.9605263157895,
+      "grad_norm": 1.3205960988998413,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125090
+    },
+    {
+      "epoch": 823.0263157894736,
+      "grad_norm": 0.8638978600502014,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 125100
+    },
+    {
+      "epoch": 823.0921052631579,
+      "grad_norm": 0.964411199092865,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125110
+    },
+    {
+      "epoch": 823.1578947368421,
+      "grad_norm": 1.1022132635116577,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 125120
+    },
+    {
+      "epoch": 823.2236842105264,
+      "grad_norm": 1.2288185358047485,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 125130
+    },
+    {
+      "epoch": 823.2894736842105,
+      "grad_norm": 1.1592532396316528,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 125140
+    },
+    {
+      "epoch": 823.3552631578947,
+      "grad_norm": 0.9944072961807251,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 125150
+    },
+    {
+      "epoch": 823.421052631579,
+      "grad_norm": 0.8100464344024658,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 125160
+    },
+    {
+      "epoch": 823.4868421052631,
+      "grad_norm": 0.9860333204269409,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 125170
+    },
+    {
+      "epoch": 823.5526315789474,
+      "grad_norm": 1.0454355478286743,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 125180
+    },
+    {
+      "epoch": 823.6184210526316,
+      "grad_norm": 1.0982762575149536,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 125190
+    },
+    {
+      "epoch": 823.6842105263158,
+      "grad_norm": 1.372549057006836,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 125200
+    },
+    {
+      "epoch": 823.75,
+      "grad_norm": 1.4132963418960571,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 125210
+    },
+    {
+      "epoch": 823.8157894736842,
+      "grad_norm": 1.2512835264205933,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 125220
+    },
+    {
+      "epoch": 823.8815789473684,
+      "grad_norm": 1.114558219909668,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125230
+    },
+    {
+      "epoch": 823.9473684210526,
+      "grad_norm": 1.0687578916549683,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 125240
+    },
+    {
+      "epoch": 824.0131578947369,
+      "grad_norm": 1.5718590021133423,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 125250
+    },
+    {
+      "epoch": 824.078947368421,
+      "grad_norm": 1.0337969064712524,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 125260
+    },
+    {
+      "epoch": 824.1447368421053,
+      "grad_norm": 1.0111421346664429,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 125270
+    },
+    {
+      "epoch": 824.2105263157895,
+      "grad_norm": 1.1369024515151978,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 125280
+    },
+    {
+      "epoch": 824.2763157894736,
+      "grad_norm": 0.7964611649513245,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 125290
+    },
+    {
+      "epoch": 824.3421052631579,
+      "grad_norm": 1.2055116891860962,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125300
+    },
+    {
+      "epoch": 824.4078947368421,
+      "grad_norm": 1.0731807947158813,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 125310
+    },
+    {
+      "epoch": 824.4736842105264,
+      "grad_norm": 1.1437373161315918,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 125320
+    },
+    {
+      "epoch": 824.5394736842105,
+      "grad_norm": 1.1605925559997559,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 125330
+    },
+    {
+      "epoch": 824.6052631578947,
+      "grad_norm": 1.286482334136963,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 125340
+    },
+    {
+      "epoch": 824.671052631579,
+      "grad_norm": 0.9673534035682678,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 125350
+    },
+    {
+      "epoch": 824.7368421052631,
+      "grad_norm": 1.3695217370986938,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 125360
+    },
+    {
+      "epoch": 824.8026315789474,
+      "grad_norm": 1.1328647136688232,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 125370
+    },
+    {
+      "epoch": 824.8684210526316,
+      "grad_norm": 1.0601879358291626,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 125380
+    },
+    {
+      "epoch": 824.9342105263158,
+      "grad_norm": 0.9178951978683472,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 125390
+    },
+    {
+      "epoch": 825.0,
+      "grad_norm": 0.9925839900970459,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125400
+    },
+    {
+      "epoch": 825.0657894736842,
+      "grad_norm": 1.2685902118682861,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 125410
+    },
+    {
+      "epoch": 825.1315789473684,
+      "grad_norm": 1.3963377475738525,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 125420
+    },
+    {
+      "epoch": 825.1973684210526,
+      "grad_norm": 1.2463805675506592,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 125430
+    },
+    {
+      "epoch": 825.2631578947369,
+      "grad_norm": 0.8058486580848694,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 125440
+    },
+    {
+      "epoch": 825.328947368421,
+      "grad_norm": 1.0413463115692139,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 125450
+    },
+    {
+      "epoch": 825.3947368421053,
+      "grad_norm": 1.3171190023422241,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 125460
+    },
+    {
+      "epoch": 825.4605263157895,
+      "grad_norm": 1.252544641494751,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 125470
+    },
+    {
+      "epoch": 825.5263157894736,
+      "grad_norm": 1.0242067575454712,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 125480
+    },
+    {
+      "epoch": 825.5921052631579,
+      "grad_norm": 0.9414500594139099,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 125490
+    },
+    {
+      "epoch": 825.6578947368421,
+      "grad_norm": 1.213117003440857,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 125500
+    },
+    {
+      "epoch": 825.7236842105264,
+      "grad_norm": 1.1076947450637817,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 125510
+    },
+    {
+      "epoch": 825.7894736842105,
+      "grad_norm": 1.1053651571273804,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 125520
+    },
+    {
+      "epoch": 825.8552631578947,
+      "grad_norm": 0.7674948573112488,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 125530
+    },
+    {
+      "epoch": 825.921052631579,
+      "grad_norm": 1.0354715585708618,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 125540
+    },
+    {
+      "epoch": 825.9868421052631,
+      "grad_norm": 0.8403087854385376,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 125550
+    },
+    {
+      "epoch": 826.0526315789474,
+      "grad_norm": 1.164014458656311,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 125560
+    },
+    {
+      "epoch": 826.1184210526316,
+      "grad_norm": 0.902004599571228,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 125570
+    },
+    {
+      "epoch": 826.1842105263158,
+      "grad_norm": 1.306944489479065,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 125580
+    },
+    {
+      "epoch": 826.25,
+      "grad_norm": 1.1703041791915894,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 125590
+    },
+    {
+      "epoch": 826.3157894736842,
+      "grad_norm": 1.309346079826355,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 125600
+    },
+    {
+      "epoch": 826.3815789473684,
+      "grad_norm": 0.9344452619552612,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 125610
+    },
+    {
+      "epoch": 826.4473684210526,
+      "grad_norm": 1.1551661491394043,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 125620
+    },
+    {
+      "epoch": 826.5131578947369,
+      "grad_norm": 0.8393580913543701,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 125630
+    },
+    {
+      "epoch": 826.578947368421,
+      "grad_norm": 1.1995357275009155,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 125640
+    },
+    {
+      "epoch": 826.6447368421053,
+      "grad_norm": 1.1376581192016602,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 125650
+    },
+    {
+      "epoch": 826.7105263157895,
+      "grad_norm": 1.103240966796875,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 125660
+    },
+    {
+      "epoch": 826.7763157894736,
+      "grad_norm": 0.8122648000717163,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125670
+    },
+    {
+      "epoch": 826.8421052631579,
+      "grad_norm": 0.9855820536613464,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 125680
+    },
+    {
+      "epoch": 826.9078947368421,
+      "grad_norm": 1.3283075094223022,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 125690
+    },
+    {
+      "epoch": 826.9736842105264,
+      "grad_norm": 1.198071837425232,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 125700
+    },
+    {
+      "epoch": 827.0394736842105,
+      "grad_norm": 1.1746746301651,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 125710
+    },
+    {
+      "epoch": 827.1052631578947,
+      "grad_norm": 1.1474931240081787,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 125720
+    },
+    {
+      "epoch": 827.171052631579,
+      "grad_norm": 0.8816200494766235,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125730
+    },
+    {
+      "epoch": 827.2368421052631,
+      "grad_norm": 1.0781453847885132,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 125740
+    },
+    {
+      "epoch": 827.3026315789474,
+      "grad_norm": 0.9956436157226562,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 125750
+    },
+    {
+      "epoch": 827.3684210526316,
+      "grad_norm": 0.9633864164352417,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 125760
+    },
+    {
+      "epoch": 827.4342105263158,
+      "grad_norm": 0.7504329085350037,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 125770
+    },
+    {
+      "epoch": 827.5,
+      "grad_norm": 0.7806046009063721,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 125780
+    },
+    {
+      "epoch": 827.5657894736842,
+      "grad_norm": 1.3514283895492554,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125790
+    },
+    {
+      "epoch": 827.6315789473684,
+      "grad_norm": 0.7963207960128784,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 125800
+    },
+    {
+      "epoch": 827.6973684210526,
+      "grad_norm": 1.0604304075241089,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 125810
+    },
+    {
+      "epoch": 827.7631578947369,
+      "grad_norm": 1.343090295791626,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 125820
+    },
+    {
+      "epoch": 827.828947368421,
+      "grad_norm": 1.3664075136184692,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 125830
+    },
+    {
+      "epoch": 827.8947368421053,
+      "grad_norm": 1.1324230432510376,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 125840
+    },
+    {
+      "epoch": 827.9605263157895,
+      "grad_norm": 1.1731352806091309,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 125850
+    },
+    {
+      "epoch": 828.0263157894736,
+      "grad_norm": 1.25088369846344,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 125860
+    },
+    {
+      "epoch": 828.0921052631579,
+      "grad_norm": 0.9906530976295471,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 125870
+    },
+    {
+      "epoch": 828.1578947368421,
+      "grad_norm": 0.8870477676391602,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 125880
+    },
+    {
+      "epoch": 828.2236842105264,
+      "grad_norm": 0.904126763343811,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 125890
+    },
+    {
+      "epoch": 828.2894736842105,
+      "grad_norm": 1.1464385986328125,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 125900
+    },
+    {
+      "epoch": 828.3552631578947,
+      "grad_norm": 0.9195071458816528,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 125910
+    },
+    {
+      "epoch": 828.421052631579,
+      "grad_norm": 1.2309972047805786,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 125920
+    },
+    {
+      "epoch": 828.4868421052631,
+      "grad_norm": 0.7952999472618103,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 125930
+    },
+    {
+      "epoch": 828.5526315789474,
+      "grad_norm": 1.1572850942611694,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 125940
+    },
+    {
+      "epoch": 828.6184210526316,
+      "grad_norm": 1.0016225576400757,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 125950
+    },
+    {
+      "epoch": 828.6842105263158,
+      "grad_norm": 1.1131701469421387,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 125960
+    },
+    {
+      "epoch": 828.75,
+      "grad_norm": 1.075382113456726,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 125970
+    },
+    {
+      "epoch": 828.8157894736842,
+      "grad_norm": 1.0743154287338257,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 125980
+    },
+    {
+      "epoch": 828.8815789473684,
+      "grad_norm": 0.8774272203445435,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 125990
+    },
+    {
+      "epoch": 828.9473684210526,
+      "grad_norm": 0.7976047396659851,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 126000
+    },
+    {
+      "epoch": 829.0131578947369,
+      "grad_norm": 0.9983826875686646,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 126010
+    },
+    {
+      "epoch": 829.078947368421,
+      "grad_norm": 1.0128384828567505,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 126020
+    },
+    {
+      "epoch": 829.1447368421053,
+      "grad_norm": 1.2371277809143066,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 126030
+    },
+    {
+      "epoch": 829.2105263157895,
+      "grad_norm": 1.121740460395813,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 126040
+    },
+    {
+      "epoch": 829.2763157894736,
+      "grad_norm": 1.0720088481903076,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 126050
+    },
+    {
+      "epoch": 829.3421052631579,
+      "grad_norm": 0.8553463816642761,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 126060
+    },
+    {
+      "epoch": 829.4078947368421,
+      "grad_norm": 0.9494578838348389,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 126070
+    },
+    {
+      "epoch": 829.4736842105264,
+      "grad_norm": 0.7240245342254639,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 126080
+    },
+    {
+      "epoch": 829.5394736842105,
+      "grad_norm": 0.9537514448165894,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 126090
+    },
+    {
+      "epoch": 829.6052631578947,
+      "grad_norm": 0.9568378329277039,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 126100
+    },
+    {
+      "epoch": 829.671052631579,
+      "grad_norm": 1.081212043762207,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 126110
+    },
+    {
+      "epoch": 829.7368421052631,
+      "grad_norm": 0.7988075017929077,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 126120
+    },
+    {
+      "epoch": 829.8026315789474,
+      "grad_norm": 0.926447868347168,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 126130
+    },
+    {
+      "epoch": 829.8684210526316,
+      "grad_norm": 0.8456553816795349,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 126140
+    },
+    {
+      "epoch": 829.9342105263158,
+      "grad_norm": 0.9388708472251892,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 126150
+    },
+    {
+      "epoch": 830.0,
+      "grad_norm": 1.1751067638397217,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 126160
+    },
+    {
+      "epoch": 830.0657894736842,
+      "grad_norm": 0.8441213369369507,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 126170
+    },
+    {
+      "epoch": 830.1315789473684,
+      "grad_norm": 0.8216992020606995,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 126180
+    },
+    {
+      "epoch": 830.1973684210526,
+      "grad_norm": 1.0158005952835083,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 126190
+    },
+    {
+      "epoch": 830.2631578947369,
+      "grad_norm": 1.3386634588241577,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 126200
+    },
+    {
+      "epoch": 830.328947368421,
+      "grad_norm": 1.026985764503479,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 126210
+    },
+    {
+      "epoch": 830.3947368421053,
+      "grad_norm": 1.194299578666687,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 126220
+    },
+    {
+      "epoch": 830.4605263157895,
+      "grad_norm": 1.0776842832565308,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 126230
+    },
+    {
+      "epoch": 830.5263157894736,
+      "grad_norm": 1.1848527193069458,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 126240
+    },
+    {
+      "epoch": 830.5921052631579,
+      "grad_norm": 1.341431736946106,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 126250
+    },
+    {
+      "epoch": 830.6578947368421,
+      "grad_norm": 1.150313138961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 126260
+    },
+    {
+      "epoch": 830.7236842105264,
+      "grad_norm": 1.0594727993011475,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 126270
+    },
+    {
+      "epoch": 830.7894736842105,
+      "grad_norm": 0.9882242679595947,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 126280
+    },
+    {
+      "epoch": 830.8552631578947,
+      "grad_norm": 1.1118570566177368,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 126290
+    },
+    {
+      "epoch": 830.921052631579,
+      "grad_norm": 1.4360451698303223,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 126300
+    },
+    {
+      "epoch": 830.9868421052631,
+      "grad_norm": 1.0150532722473145,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 126310
+    },
+    {
+      "epoch": 831.0526315789474,
+      "grad_norm": 1.010431170463562,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 126320
+    },
+    {
+      "epoch": 831.1184210526316,
+      "grad_norm": 1.2884689569473267,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 126330
+    },
+    {
+      "epoch": 831.1842105263158,
+      "grad_norm": 1.212661623954773,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 126340
+    },
+    {
+      "epoch": 831.25,
+      "grad_norm": 1.3419666290283203,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 126350
+    },
+    {
+      "epoch": 831.3157894736842,
+      "grad_norm": 1.1499714851379395,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 126360
+    },
+    {
+      "epoch": 831.3815789473684,
+      "grad_norm": 1.0944184064865112,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 126370
+    },
+    {
+      "epoch": 831.4473684210526,
+      "grad_norm": 1.3042875528335571,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 126380
+    },
+    {
+      "epoch": 831.5131578947369,
+      "grad_norm": 1.3170181512832642,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 126390
+    },
+    {
+      "epoch": 831.578947368421,
+      "grad_norm": 1.0547919273376465,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 126400
+    },
+    {
+      "epoch": 831.6447368421053,
+      "grad_norm": 1.129002571105957,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 126410
+    },
+    {
+      "epoch": 831.7105263157895,
+      "grad_norm": 1.3356235027313232,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 126420
+    },
+    {
+      "epoch": 831.7763157894736,
+      "grad_norm": 1.2888990640640259,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 126430
+    },
+    {
+      "epoch": 831.8421052631579,
+      "grad_norm": 1.0148605108261108,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 126440
+    },
+    {
+      "epoch": 831.9078947368421,
+      "grad_norm": 0.9079254865646362,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 126450
+    },
+    {
+      "epoch": 831.9736842105264,
+      "grad_norm": 1.0899399518966675,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 126460
+    },
+    {
+      "epoch": 832.0394736842105,
+      "grad_norm": 0.915304958820343,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 126470
+    },
+    {
+      "epoch": 832.1052631578947,
+      "grad_norm": 1.0055623054504395,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 126480
+    },
+    {
+      "epoch": 832.171052631579,
+      "grad_norm": 0.8581055998802185,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 126490
+    },
+    {
+      "epoch": 832.2368421052631,
+      "grad_norm": 0.8408956527709961,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 126500
+    },
+    {
+      "epoch": 832.3026315789474,
+      "grad_norm": 0.7524006366729736,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 126510
+    },
+    {
+      "epoch": 832.3684210526316,
+      "grad_norm": 0.799748420715332,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 126520
+    },
+    {
+      "epoch": 832.4342105263158,
+      "grad_norm": 0.9387583136558533,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 126530
+    },
+    {
+      "epoch": 832.5,
+      "grad_norm": 0.9900626540184021,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 126540
+    },
+    {
+      "epoch": 832.5657894736842,
+      "grad_norm": 1.16420578956604,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 126550
+    },
+    {
+      "epoch": 832.6315789473684,
+      "grad_norm": 1.023986577987671,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 126560
+    },
+    {
+      "epoch": 832.6973684210526,
+      "grad_norm": 0.950329601764679,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 126570
+    },
+    {
+      "epoch": 832.7631578947369,
+      "grad_norm": 1.0447871685028076,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 126580
+    },
+    {
+      "epoch": 832.828947368421,
+      "grad_norm": 0.6059691905975342,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 126590
+    },
+    {
+      "epoch": 832.8947368421053,
+      "grad_norm": 1.1729786396026611,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 126600
+    },
+    {
+      "epoch": 832.9605263157895,
+      "grad_norm": 1.0164276361465454,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 126610
+    },
+    {
+      "epoch": 833.0263157894736,
+      "grad_norm": 1.1366938352584839,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 126620
+    },
+    {
+      "epoch": 833.0921052631579,
+      "grad_norm": 1.0589661598205566,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 126630
+    },
+    {
+      "epoch": 833.1578947368421,
+      "grad_norm": 0.9996610879898071,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 126640
+    },
+    {
+      "epoch": 833.2236842105264,
+      "grad_norm": 1.13376784324646,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 126650
+    },
+    {
+      "epoch": 833.2894736842105,
+      "grad_norm": 1.0717856884002686,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 126660
+    },
+    {
+      "epoch": 833.3552631578947,
+      "grad_norm": 0.7824698090553284,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 126670
+    },
+    {
+      "epoch": 833.421052631579,
+      "grad_norm": 1.0761967897415161,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 126680
+    },
+    {
+      "epoch": 833.4868421052631,
+      "grad_norm": 1.2236111164093018,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 126690
+    },
+    {
+      "epoch": 833.5526315789474,
+      "grad_norm": 1.143056035041809,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 126700
+    },
+    {
+      "epoch": 833.6184210526316,
+      "grad_norm": 1.0561918020248413,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 126710
+    },
+    {
+      "epoch": 833.6842105263158,
+      "grad_norm": 1.1367357969284058,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 126720
+    },
+    {
+      "epoch": 833.75,
+      "grad_norm": 0.7675701379776001,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 126730
+    },
+    {
+      "epoch": 833.8157894736842,
+      "grad_norm": 1.025897741317749,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 126740
+    },
+    {
+      "epoch": 833.8815789473684,
+      "grad_norm": 0.9761922359466553,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 126750
+    },
+    {
+      "epoch": 833.9473684210526,
+      "grad_norm": 0.9018344879150391,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 126760
+    },
+    {
+      "epoch": 834.0131578947369,
+      "grad_norm": 1.203495979309082,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 126770
+    },
+    {
+      "epoch": 834.078947368421,
+      "grad_norm": 1.1950404644012451,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 126780
+    },
+    {
+      "epoch": 834.1447368421053,
+      "grad_norm": 0.943814218044281,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 126790
+    },
+    {
+      "epoch": 834.2105263157895,
+      "grad_norm": 1.0648579597473145,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 126800
+    },
+    {
+      "epoch": 834.2763157894736,
+      "grad_norm": 1.0198975801467896,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 126810
+    },
+    {
+      "epoch": 834.3421052631579,
+      "grad_norm": 1.195588231086731,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 126820
+    },
+    {
+      "epoch": 834.4078947368421,
+      "grad_norm": 0.9909479022026062,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 126830
+    },
+    {
+      "epoch": 834.4736842105264,
+      "grad_norm": 0.8295359015464783,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 126840
+    },
+    {
+      "epoch": 834.5394736842105,
+      "grad_norm": 1.1163146495819092,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 126850
+    },
+    {
+      "epoch": 834.6052631578947,
+      "grad_norm": 0.9339954853057861,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 126860
+    },
+    {
+      "epoch": 834.671052631579,
+      "grad_norm": 1.0907495021820068,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 126870
+    },
+    {
+      "epoch": 834.7368421052631,
+      "grad_norm": 0.7249103784561157,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 126880
+    },
+    {
+      "epoch": 834.8026315789474,
+      "grad_norm": 0.9385533928871155,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 126890
+    },
+    {
+      "epoch": 834.8684210526316,
+      "grad_norm": 1.03643000125885,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 126900
+    },
+    {
+      "epoch": 834.9342105263158,
+      "grad_norm": 1.1739977598190308,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 126910
+    },
+    {
+      "epoch": 835.0,
+      "grad_norm": 0.7921239733695984,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 126920
+    },
+    {
+      "epoch": 835.0657894736842,
+      "grad_norm": 0.7429852485656738,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 126930
+    },
+    {
+      "epoch": 835.1315789473684,
+      "grad_norm": 0.8239948153495789,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 126940
+    },
+    {
+      "epoch": 835.1973684210526,
+      "grad_norm": 0.9873680472373962,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 126950
+    },
+    {
+      "epoch": 835.2631578947369,
+      "grad_norm": 0.8534329533576965,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 126960
+    },
+    {
+      "epoch": 835.328947368421,
+      "grad_norm": 1.498870849609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 126970
+    },
+    {
+      "epoch": 835.3947368421053,
+      "grad_norm": 0.9700299501419067,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 126980
+    },
+    {
+      "epoch": 835.4605263157895,
+      "grad_norm": 0.9935576915740967,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 126990
+    },
+    {
+      "epoch": 835.5263157894736,
+      "grad_norm": 1.1261264085769653,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 127000
+    },
+    {
+      "epoch": 835.5921052631579,
+      "grad_norm": 1.2722344398498535,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 127010
+    },
+    {
+      "epoch": 835.6578947368421,
+      "grad_norm": 1.2571996450424194,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 127020
+    },
+    {
+      "epoch": 835.7236842105264,
+      "grad_norm": 1.4331727027893066,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 127030
+    },
+    {
+      "epoch": 835.7894736842105,
+      "grad_norm": 1.5384089946746826,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 127040
+    },
+    {
+      "epoch": 835.8552631578947,
+      "grad_norm": 1.356110692024231,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 127050
+    },
+    {
+      "epoch": 835.921052631579,
+      "grad_norm": 0.8997991681098938,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 127060
+    },
+    {
+      "epoch": 835.9868421052631,
+      "grad_norm": 0.8456292152404785,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 127070
+    },
+    {
+      "epoch": 836.0526315789474,
+      "grad_norm": 1.1669671535491943,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 127080
+    },
+    {
+      "epoch": 836.1184210526316,
+      "grad_norm": 1.0110538005828857,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 127090
+    },
+    {
+      "epoch": 836.1842105263158,
+      "grad_norm": 1.2209850549697876,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 127100
+    },
+    {
+      "epoch": 836.25,
+      "grad_norm": 1.2250803709030151,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 127110
+    },
+    {
+      "epoch": 836.3157894736842,
+      "grad_norm": 0.9316526055335999,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 127120
+    },
+    {
+      "epoch": 836.3815789473684,
+      "grad_norm": 0.9690991044044495,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 127130
+    },
+    {
+      "epoch": 836.4473684210526,
+      "grad_norm": 0.7517539858818054,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 127140
+    },
+    {
+      "epoch": 836.5131578947369,
+      "grad_norm": 1.0882537364959717,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 127150
+    },
+    {
+      "epoch": 836.578947368421,
+      "grad_norm": 1.2422233819961548,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 127160
+    },
+    {
+      "epoch": 836.6447368421053,
+      "grad_norm": 1.1308391094207764,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 127170
+    },
+    {
+      "epoch": 836.7105263157895,
+      "grad_norm": 1.181129813194275,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 127180
+    },
+    {
+      "epoch": 836.7763157894736,
+      "grad_norm": 1.0255203247070312,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 127190
+    },
+    {
+      "epoch": 836.8421052631579,
+      "grad_norm": 1.0835888385772705,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 127200
+    },
+    {
+      "epoch": 836.9078947368421,
+      "grad_norm": 1.4097410440444946,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 127210
+    },
+    {
+      "epoch": 836.9736842105264,
+      "grad_norm": 0.8870857954025269,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 127220
+    },
+    {
+      "epoch": 837.0394736842105,
+      "grad_norm": 0.9946476221084595,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 127230
+    },
+    {
+      "epoch": 837.1052631578947,
+      "grad_norm": 1.268383264541626,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 127240
+    },
+    {
+      "epoch": 837.171052631579,
+      "grad_norm": 1.1119439601898193,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 127250
+    },
+    {
+      "epoch": 837.2368421052631,
+      "grad_norm": 1.3202515840530396,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 127260
+    },
+    {
+      "epoch": 837.3026315789474,
+      "grad_norm": 1.216822624206543,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 127270
+    },
+    {
+      "epoch": 837.3684210526316,
+      "grad_norm": 1.5065245628356934,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 127280
+    },
+    {
+      "epoch": 837.4342105263158,
+      "grad_norm": 1.358711838722229,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 127290
+    },
+    {
+      "epoch": 837.5,
+      "grad_norm": 1.0266478061676025,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 127300
+    },
+    {
+      "epoch": 837.5657894736842,
+      "grad_norm": 1.2015262842178345,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 127310
+    },
+    {
+      "epoch": 837.6315789473684,
+      "grad_norm": 0.8070095777511597,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 127320
+    },
+    {
+      "epoch": 837.6973684210526,
+      "grad_norm": 1.3548043966293335,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 127330
+    },
+    {
+      "epoch": 837.7631578947369,
+      "grad_norm": 1.2841241359710693,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 127340
+    },
+    {
+      "epoch": 837.828947368421,
+      "grad_norm": 1.2051345109939575,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 127350
+    },
+    {
+      "epoch": 837.8947368421053,
+      "grad_norm": 1.303202748298645,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 127360
+    },
+    {
+      "epoch": 837.9605263157895,
+      "grad_norm": 0.9011682271957397,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 127370
+    },
+    {
+      "epoch": 838.0263157894736,
+      "grad_norm": 1.1267516613006592,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 127380
+    },
+    {
+      "epoch": 838.0921052631579,
+      "grad_norm": 1.045426368713379,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 127390
+    },
+    {
+      "epoch": 838.1578947368421,
+      "grad_norm": 1.0214307308197021,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 127400
+    },
+    {
+      "epoch": 838.2236842105264,
+      "grad_norm": 1.4649877548217773,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 127410
+    },
+    {
+      "epoch": 838.2894736842105,
+      "grad_norm": 1.314589262008667,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 127420
+    },
+    {
+      "epoch": 838.3552631578947,
+      "grad_norm": 1.1437108516693115,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 127430
+    },
+    {
+      "epoch": 838.421052631579,
+      "grad_norm": 1.1604773998260498,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 127440
+    },
+    {
+      "epoch": 838.4868421052631,
+      "grad_norm": 1.3208593130111694,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 127450
+    },
+    {
+      "epoch": 838.5526315789474,
+      "grad_norm": 1.03213369846344,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 127460
+    },
+    {
+      "epoch": 838.6184210526316,
+      "grad_norm": 1.0011709928512573,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 127470
+    },
+    {
+      "epoch": 838.6842105263158,
+      "grad_norm": 1.0269711017608643,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 127480
+    },
+    {
+      "epoch": 838.75,
+      "grad_norm": 1.070455551147461,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 127490
+    },
+    {
+      "epoch": 838.8157894736842,
+      "grad_norm": 0.789607584476471,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 127500
+    },
+    {
+      "epoch": 838.8815789473684,
+      "grad_norm": 0.8468894958496094,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 127510
+    },
+    {
+      "epoch": 838.9473684210526,
+      "grad_norm": 0.7091095447540283,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 127520
+    },
+    {
+      "epoch": 839.0131578947369,
+      "grad_norm": 1.0333023071289062,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 127530
+    },
+    {
+      "epoch": 839.078947368421,
+      "grad_norm": 1.031185269355774,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 127540
+    },
+    {
+      "epoch": 839.1447368421053,
+      "grad_norm": 0.9632649421691895,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 127550
+    },
+    {
+      "epoch": 839.2105263157895,
+      "grad_norm": 0.7596598863601685,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 127560
+    },
+    {
+      "epoch": 839.2763157894736,
+      "grad_norm": 0.7665521502494812,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 127570
+    },
+    {
+      "epoch": 839.3421052631579,
+      "grad_norm": 0.8565236926078796,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 127580
+    },
+    {
+      "epoch": 839.4078947368421,
+      "grad_norm": 1.3483563661575317,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 127590
+    },
+    {
+      "epoch": 839.4736842105264,
+      "grad_norm": 1.4640027284622192,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 127600
+    },
+    {
+      "epoch": 839.5394736842105,
+      "grad_norm": 1.413806438446045,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 127610
+    },
+    {
+      "epoch": 839.6052631578947,
+      "grad_norm": 1.3078571557998657,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 127620
+    },
+    {
+      "epoch": 839.671052631579,
+      "grad_norm": 0.8377428650856018,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 127630
+    },
+    {
+      "epoch": 839.7368421052631,
+      "grad_norm": 1.2343833446502686,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 127640
+    },
+    {
+      "epoch": 839.8026315789474,
+      "grad_norm": 0.7730600833892822,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 127650
+    },
+    {
+      "epoch": 839.8684210526316,
+      "grad_norm": 0.9938012957572937,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 127660
+    },
+    {
+      "epoch": 839.9342105263158,
+      "grad_norm": 0.7523202896118164,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 127670
+    },
+    {
+      "epoch": 840.0,
+      "grad_norm": 0.7400522828102112,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 127680
+    },
+    {
+      "epoch": 840.0657894736842,
+      "grad_norm": 0.9633582234382629,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 127690
+    },
+    {
+      "epoch": 840.1315789473684,
+      "grad_norm": 1.1232542991638184,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 127700
+    },
+    {
+      "epoch": 840.1973684210526,
+      "grad_norm": 1.1191346645355225,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 127710
+    },
+    {
+      "epoch": 840.2631578947369,
+      "grad_norm": 1.43672513961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 127720
+    },
+    {
+      "epoch": 840.328947368421,
+      "grad_norm": 0.8802378177642822,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 127730
+    },
+    {
+      "epoch": 840.3947368421053,
+      "grad_norm": 1.0515570640563965,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 127740
+    },
+    {
+      "epoch": 840.4605263157895,
+      "grad_norm": 0.9687642455101013,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 127750
+    },
+    {
+      "epoch": 840.5263157894736,
+      "grad_norm": 0.8646338582038879,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 127760
+    },
+    {
+      "epoch": 840.5921052631579,
+      "grad_norm": 0.8392863869667053,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 127770
+    },
+    {
+      "epoch": 840.6578947368421,
+      "grad_norm": 0.8507763147354126,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 127780
+    },
+    {
+      "epoch": 840.7236842105264,
+      "grad_norm": 0.9287623167037964,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 127790
+    },
+    {
+      "epoch": 840.7894736842105,
+      "grad_norm": 0.7224423289299011,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 127800
+    },
+    {
+      "epoch": 840.8552631578947,
+      "grad_norm": 0.9184572100639343,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 127810
+    },
+    {
+      "epoch": 840.921052631579,
+      "grad_norm": 1.360131859779358,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 127820
+    },
+    {
+      "epoch": 840.9868421052631,
+      "grad_norm": 1.099462628364563,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 127830
+    },
+    {
+      "epoch": 841.0526315789474,
+      "grad_norm": 1.0513808727264404,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 127840
+    },
+    {
+      "epoch": 841.1184210526316,
+      "grad_norm": 1.0300471782684326,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 127850
+    },
+    {
+      "epoch": 841.1842105263158,
+      "grad_norm": 1.2935221195220947,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 127860
+    },
+    {
+      "epoch": 841.25,
+      "grad_norm": 1.2731430530548096,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 127870
+    },
+    {
+      "epoch": 841.3157894736842,
+      "grad_norm": 0.963191032409668,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 127880
+    },
+    {
+      "epoch": 841.3815789473684,
+      "grad_norm": 1.1909409761428833,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 127890
+    },
+    {
+      "epoch": 841.4473684210526,
+      "grad_norm": 1.1539998054504395,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 127900
+    },
+    {
+      "epoch": 841.5131578947369,
+      "grad_norm": 1.3037039041519165,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 127910
+    },
+    {
+      "epoch": 841.578947368421,
+      "grad_norm": 1.0298134088516235,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 127920
+    },
+    {
+      "epoch": 841.6447368421053,
+      "grad_norm": 0.9822731018066406,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 127930
+    },
+    {
+      "epoch": 841.7105263157895,
+      "grad_norm": 0.9702892303466797,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 127940
+    },
+    {
+      "epoch": 841.7763157894736,
+      "grad_norm": 0.8605859279632568,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 127950
+    },
+    {
+      "epoch": 841.8421052631579,
+      "grad_norm": 1.282814383506775,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 127960
+    },
+    {
+      "epoch": 841.9078947368421,
+      "grad_norm": 0.9928876161575317,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 127970
+    },
+    {
+      "epoch": 841.9736842105264,
+      "grad_norm": 1.6558021306991577,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 127980
+    },
+    {
+      "epoch": 842.0394736842105,
+      "grad_norm": 0.8925629258155823,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 127990
+    },
+    {
+      "epoch": 842.1052631578947,
+      "grad_norm": 0.8495152592658997,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 128000
+    },
+    {
+      "epoch": 842.171052631579,
+      "grad_norm": 0.8504860997200012,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 128010
+    },
+    {
+      "epoch": 842.2368421052631,
+      "grad_norm": 1.1367663145065308,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 128020
+    },
+    {
+      "epoch": 842.3026315789474,
+      "grad_norm": 0.9067413210868835,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 128030
+    },
+    {
+      "epoch": 842.3684210526316,
+      "grad_norm": 1.0044593811035156,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 128040
+    },
+    {
+      "epoch": 842.4342105263158,
+      "grad_norm": 1.2345927953720093,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 128050
+    },
+    {
+      "epoch": 842.5,
+      "grad_norm": 1.3313305377960205,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 128060
+    },
+    {
+      "epoch": 842.5657894736842,
+      "grad_norm": 1.3206311464309692,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 128070
+    },
+    {
+      "epoch": 842.6315789473684,
+      "grad_norm": 0.9550430774688721,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 128080
+    },
+    {
+      "epoch": 842.6973684210526,
+      "grad_norm": 1.156746745109558,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 128090
+    },
+    {
+      "epoch": 842.7631578947369,
+      "grad_norm": 1.0178402662277222,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 128100
+    },
+    {
+      "epoch": 842.828947368421,
+      "grad_norm": 1.0084766149520874,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 128110
+    },
+    {
+      "epoch": 842.8947368421053,
+      "grad_norm": 1.0070236921310425,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 128120
+    },
+    {
+      "epoch": 842.9605263157895,
+      "grad_norm": 0.9018028378486633,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 128130
+    },
+    {
+      "epoch": 843.0263157894736,
+      "grad_norm": 0.7280605435371399,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 128140
+    },
+    {
+      "epoch": 843.0921052631579,
+      "grad_norm": 1.2799079418182373,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 128150
+    },
+    {
+      "epoch": 843.1578947368421,
+      "grad_norm": 1.1699326038360596,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 128160
+    },
+    {
+      "epoch": 843.2236842105264,
+      "grad_norm": 1.3281631469726562,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 128170
+    },
+    {
+      "epoch": 843.2894736842105,
+      "grad_norm": 1.5978515148162842,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 128180
+    },
+    {
+      "epoch": 843.3552631578947,
+      "grad_norm": 1.2502210140228271,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 128190
+    },
+    {
+      "epoch": 843.421052631579,
+      "grad_norm": 1.4333372116088867,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 128200
+    },
+    {
+      "epoch": 843.4868421052631,
+      "grad_norm": 0.8813333511352539,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 128210
+    },
+    {
+      "epoch": 843.5526315789474,
+      "grad_norm": 1.101881980895996,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 128220
+    },
+    {
+      "epoch": 843.6184210526316,
+      "grad_norm": 0.9753175377845764,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 128230
+    },
+    {
+      "epoch": 843.6842105263158,
+      "grad_norm": 1.0207269191741943,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 128240
+    },
+    {
+      "epoch": 843.75,
+      "grad_norm": 0.8684105277061462,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 128250
+    },
+    {
+      "epoch": 843.8157894736842,
+      "grad_norm": 0.8891376852989197,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 128260
+    },
+    {
+      "epoch": 843.8815789473684,
+      "grad_norm": 0.7383885979652405,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 128270
+    },
+    {
+      "epoch": 843.9473684210526,
+      "grad_norm": 0.9492682814598083,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 128280
+    },
+    {
+      "epoch": 844.0131578947369,
+      "grad_norm": 0.815500020980835,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 128290
+    },
+    {
+      "epoch": 844.078947368421,
+      "grad_norm": 0.8752801418304443,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 128300
+    },
+    {
+      "epoch": 844.1447368421053,
+      "grad_norm": 1.1680850982666016,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 128310
+    },
+    {
+      "epoch": 844.2105263157895,
+      "grad_norm": 1.0762124061584473,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 128320
+    },
+    {
+      "epoch": 844.2763157894736,
+      "grad_norm": 1.2257249355316162,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 128330
+    },
+    {
+      "epoch": 844.3421052631579,
+      "grad_norm": 1.2510292530059814,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 128340
+    },
+    {
+      "epoch": 844.4078947368421,
+      "grad_norm": 1.0363690853118896,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 128350
+    },
+    {
+      "epoch": 844.4736842105264,
+      "grad_norm": 1.1200023889541626,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 128360
+    },
+    {
+      "epoch": 844.5394736842105,
+      "grad_norm": 1.0740878582000732,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 128370
+    },
+    {
+      "epoch": 844.6052631578947,
+      "grad_norm": 1.1896917819976807,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 128380
+    },
+    {
+      "epoch": 844.671052631579,
+      "grad_norm": 1.0779705047607422,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 128390
+    },
+    {
+      "epoch": 844.7368421052631,
+      "grad_norm": 0.555634617805481,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 128400
+    },
+    {
+      "epoch": 844.8026315789474,
+      "grad_norm": 0.9556017518043518,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 128410
+    },
+    {
+      "epoch": 844.8684210526316,
+      "grad_norm": 0.8419582843780518,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 128420
+    },
+    {
+      "epoch": 844.9342105263158,
+      "grad_norm": 1.0701279640197754,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 128430
+    },
+    {
+      "epoch": 845.0,
+      "grad_norm": 1.3151577711105347,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 128440
+    },
+    {
+      "epoch": 845.0657894736842,
+      "grad_norm": 1.086671233177185,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 128450
+    },
+    {
+      "epoch": 845.1315789473684,
+      "grad_norm": 1.2113654613494873,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 128460
+    },
+    {
+      "epoch": 845.1973684210526,
+      "grad_norm": 1.2643672227859497,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 128470
+    },
+    {
+      "epoch": 845.2631578947369,
+      "grad_norm": 1.0714949369430542,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 128480
+    },
+    {
+      "epoch": 845.328947368421,
+      "grad_norm": 0.7351527214050293,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 128490
+    },
+    {
+      "epoch": 845.3947368421053,
+      "grad_norm": 1.0877636671066284,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 128500
+    },
+    {
+      "epoch": 845.4605263157895,
+      "grad_norm": 0.6363227963447571,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 128510
+    },
+    {
+      "epoch": 845.5263157894736,
+      "grad_norm": 0.7102157473564148,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 128520
+    },
+    {
+      "epoch": 845.5921052631579,
+      "grad_norm": 0.9936819076538086,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 128530
+    },
+    {
+      "epoch": 845.6578947368421,
+      "grad_norm": 0.7047979831695557,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 128540
+    },
+    {
+      "epoch": 845.7236842105264,
+      "grad_norm": 1.017305850982666,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 128550
+    },
+    {
+      "epoch": 845.7894736842105,
+      "grad_norm": 1.3176766633987427,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 128560
+    },
+    {
+      "epoch": 845.8552631578947,
+      "grad_norm": 1.1096950769424438,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 128570
+    },
+    {
+      "epoch": 845.921052631579,
+      "grad_norm": 1.1976255178451538,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 128580
+    },
+    {
+      "epoch": 845.9868421052631,
+      "grad_norm": 1.0794117450714111,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 128590
+    },
+    {
+      "epoch": 846.0526315789474,
+      "grad_norm": 0.8604280948638916,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 128600
+    },
+    {
+      "epoch": 846.1184210526316,
+      "grad_norm": 0.8841588497161865,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 128610
+    },
+    {
+      "epoch": 846.1842105263158,
+      "grad_norm": 0.9039338827133179,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 128620
+    },
+    {
+      "epoch": 846.25,
+      "grad_norm": 0.7028020024299622,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 128630
+    },
+    {
+      "epoch": 846.3157894736842,
+      "grad_norm": 1.018131136894226,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 128640
+    },
+    {
+      "epoch": 846.3815789473684,
+      "grad_norm": 1.4011059999465942,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 128650
+    },
+    {
+      "epoch": 846.4473684210526,
+      "grad_norm": 1.176498293876648,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 128660
+    },
+    {
+      "epoch": 846.5131578947369,
+      "grad_norm": 1.150488257408142,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 128670
+    },
+    {
+      "epoch": 846.578947368421,
+      "grad_norm": 1.1439628601074219,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 128680
+    },
+    {
+      "epoch": 846.6447368421053,
+      "grad_norm": 1.0313776731491089,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 128690
+    },
+    {
+      "epoch": 846.7105263157895,
+      "grad_norm": 0.8566111922264099,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 128700
+    },
+    {
+      "epoch": 846.7763157894736,
+      "grad_norm": 1.240422010421753,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 128710
+    },
+    {
+      "epoch": 846.8421052631579,
+      "grad_norm": 0.8956599831581116,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 128720
+    },
+    {
+      "epoch": 846.9078947368421,
+      "grad_norm": 1.075857400894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 128730
+    },
+    {
+      "epoch": 846.9736842105264,
+      "grad_norm": 1.1086944341659546,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 128740
+    },
+    {
+      "epoch": 847.0394736842105,
+      "grad_norm": 0.7189145684242249,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 128750
+    },
+    {
+      "epoch": 847.1052631578947,
+      "grad_norm": 1.0675982236862183,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 128760
+    },
+    {
+      "epoch": 847.171052631579,
+      "grad_norm": 1.0329749584197998,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 128770
+    },
+    {
+      "epoch": 847.2368421052631,
+      "grad_norm": 0.8450028896331787,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 128780
+    },
+    {
+      "epoch": 847.3026315789474,
+      "grad_norm": 1.0242254734039307,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 128790
+    },
+    {
+      "epoch": 847.3684210526316,
+      "grad_norm": 1.0955355167388916,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 128800
+    },
+    {
+      "epoch": 847.4342105263158,
+      "grad_norm": 0.9389364123344421,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 128810
+    },
+    {
+      "epoch": 847.5,
+      "grad_norm": 0.7814875245094299,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 128820
+    },
+    {
+      "epoch": 847.5657894736842,
+      "grad_norm": 1.0729169845581055,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 128830
+    },
+    {
+      "epoch": 847.6315789473684,
+      "grad_norm": 1.7011371850967407,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 128840
+    },
+    {
+      "epoch": 847.6973684210526,
+      "grad_norm": 1.1857264041900635,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 128850
+    },
+    {
+      "epoch": 847.7631578947369,
+      "grad_norm": 1.284214973449707,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 128860
+    },
+    {
+      "epoch": 847.828947368421,
+      "grad_norm": 1.111433982849121,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 128870
+    },
+    {
+      "epoch": 847.8947368421053,
+      "grad_norm": 0.7958429455757141,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 128880
+    },
+    {
+      "epoch": 847.9605263157895,
+      "grad_norm": 0.8999338150024414,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 128890
+    },
+    {
+      "epoch": 848.0263157894736,
+      "grad_norm": 1.1462279558181763,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 128900
+    },
+    {
+      "epoch": 848.0921052631579,
+      "grad_norm": 1.2270578145980835,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 128910
+    },
+    {
+      "epoch": 848.1578947368421,
+      "grad_norm": 1.2196954488754272,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 128920
+    },
+    {
+      "epoch": 848.2236842105264,
+      "grad_norm": 1.0865422487258911,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 128930
+    },
+    {
+      "epoch": 848.2894736842105,
+      "grad_norm": 0.8074581623077393,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 128940
+    },
+    {
+      "epoch": 848.3552631578947,
+      "grad_norm": 1.2923625707626343,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 128950
+    },
+    {
+      "epoch": 848.421052631579,
+      "grad_norm": 1.020020842552185,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 128960
+    },
+    {
+      "epoch": 848.4868421052631,
+      "grad_norm": 0.9365369081497192,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 128970
+    },
+    {
+      "epoch": 848.5526315789474,
+      "grad_norm": 1.216340184211731,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 128980
+    },
+    {
+      "epoch": 848.6184210526316,
+      "grad_norm": 1.2271922826766968,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 128990
+    },
+    {
+      "epoch": 848.6842105263158,
+      "grad_norm": 1.1109647750854492,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 129000
+    },
+    {
+      "epoch": 848.75,
+      "grad_norm": 1.0088245868682861,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 129010
+    },
+    {
+      "epoch": 848.8157894736842,
+      "grad_norm": 1.1272104978561401,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 129020
+    },
+    {
+      "epoch": 848.8815789473684,
+      "grad_norm": 1.0856529474258423,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 129030
+    },
+    {
+      "epoch": 848.9473684210526,
+      "grad_norm": 0.8239693641662598,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 129040
+    },
+    {
+      "epoch": 849.0131578947369,
+      "grad_norm": 1.0656987428665161,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 129050
+    },
+    {
+      "epoch": 849.078947368421,
+      "grad_norm": 1.2168710231781006,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 129060
+    },
+    {
+      "epoch": 849.1447368421053,
+      "grad_norm": 1.1486916542053223,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 129070
+    },
+    {
+      "epoch": 849.2105263157895,
+      "grad_norm": 1.0890756845474243,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 129080
+    },
+    {
+      "epoch": 849.2763157894736,
+      "grad_norm": 1.3131881952285767,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 129090
+    },
+    {
+      "epoch": 849.3421052631579,
+      "grad_norm": 0.8249554634094238,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 129100
+    },
+    {
+      "epoch": 849.4078947368421,
+      "grad_norm": 0.9621846079826355,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 129110
+    },
+    {
+      "epoch": 849.4736842105264,
+      "grad_norm": 0.8949623703956604,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 129120
+    },
+    {
+      "epoch": 849.5394736842105,
+      "grad_norm": 0.9129248261451721,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 129130
+    },
+    {
+      "epoch": 849.6052631578947,
+      "grad_norm": 1.3539313077926636,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 129140
+    },
+    {
+      "epoch": 849.671052631579,
+      "grad_norm": 1.445449948310852,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 129150
+    },
+    {
+      "epoch": 849.7368421052631,
+      "grad_norm": 1.0846349000930786,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 129160
+    },
+    {
+      "epoch": 849.8026315789474,
+      "grad_norm": 1.1930545568466187,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 129170
+    },
+    {
+      "epoch": 849.8684210526316,
+      "grad_norm": 1.0064821243286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 129180
+    },
+    {
+      "epoch": 849.9342105263158,
+      "grad_norm": 0.9630594849586487,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 129190
+    },
+    {
+      "epoch": 850.0,
+      "grad_norm": 1.0653307437896729,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 129200
+    },
+    {
+      "epoch": 850.0657894736842,
+      "grad_norm": 0.947507381439209,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 129210
+    },
+    {
+      "epoch": 850.1315789473684,
+      "grad_norm": 0.9130157232284546,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 129220
+    },
+    {
+      "epoch": 850.1973684210526,
+      "grad_norm": 0.8457781076431274,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 129230
+    },
+    {
+      "epoch": 850.2631578947369,
+      "grad_norm": 0.7123493552207947,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 129240
+    },
+    {
+      "epoch": 850.328947368421,
+      "grad_norm": 0.9977707862854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 129250
+    },
+    {
+      "epoch": 850.3947368421053,
+      "grad_norm": 0.7980003356933594,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 129260
+    },
+    {
+      "epoch": 850.4605263157895,
+      "grad_norm": 1.2628130912780762,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 129270
+    },
+    {
+      "epoch": 850.5263157894736,
+      "grad_norm": 1.2784351110458374,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 129280
+    },
+    {
+      "epoch": 850.5921052631579,
+      "grad_norm": 0.8135380148887634,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 129290
+    },
+    {
+      "epoch": 850.6578947368421,
+      "grad_norm": 1.2080432176589966,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 129300
+    },
+    {
+      "epoch": 850.7236842105264,
+      "grad_norm": 1.1570539474487305,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 129310
+    },
+    {
+      "epoch": 850.7894736842105,
+      "grad_norm": 0.8722181916236877,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 129320
+    },
+    {
+      "epoch": 850.8552631578947,
+      "grad_norm": 0.9665051698684692,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 129330
+    },
+    {
+      "epoch": 850.921052631579,
+      "grad_norm": 1.156315565109253,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 129340
+    },
+    {
+      "epoch": 850.9868421052631,
+      "grad_norm": 1.3095405101776123,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 129350
+    },
+    {
+      "epoch": 851.0526315789474,
+      "grad_norm": 1.2862575054168701,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 129360
+    },
+    {
+      "epoch": 851.1184210526316,
+      "grad_norm": 1.2178730964660645,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 129370
+    },
+    {
+      "epoch": 851.1842105263158,
+      "grad_norm": 1.331790804862976,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 129380
+    },
+    {
+      "epoch": 851.25,
+      "grad_norm": 1.1525365114212036,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 129390
+    },
+    {
+      "epoch": 851.3157894736842,
+      "grad_norm": 1.2628202438354492,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 129400
+    },
+    {
+      "epoch": 851.3815789473684,
+      "grad_norm": 0.9051490426063538,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 129410
+    },
+    {
+      "epoch": 851.4473684210526,
+      "grad_norm": 0.9689465165138245,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 129420
+    },
+    {
+      "epoch": 851.5131578947369,
+      "grad_norm": 0.8615492582321167,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 129430
+    },
+    {
+      "epoch": 851.578947368421,
+      "grad_norm": 0.9767460823059082,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 129440
+    },
+    {
+      "epoch": 851.6447368421053,
+      "grad_norm": 0.9622799158096313,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 129450
+    },
+    {
+      "epoch": 851.7105263157895,
+      "grad_norm": 1.0682460069656372,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 129460
+    },
+    {
+      "epoch": 851.7763157894736,
+      "grad_norm": 0.5867149233818054,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 129470
+    },
+    {
+      "epoch": 851.8421052631579,
+      "grad_norm": 0.9323984384536743,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 129480
+    },
+    {
+      "epoch": 851.9078947368421,
+      "grad_norm": 1.0186138153076172,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 129490
+    },
+    {
+      "epoch": 851.9736842105264,
+      "grad_norm": 1.1094682216644287,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 129500
+    },
+    {
+      "epoch": 852.0394736842105,
+      "grad_norm": 1.0459119081497192,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 129510
+    },
+    {
+      "epoch": 852.1052631578947,
+      "grad_norm": 0.9179003238677979,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 129520
+    },
+    {
+      "epoch": 852.171052631579,
+      "grad_norm": 1.0056957006454468,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 129530
+    },
+    {
+      "epoch": 852.2368421052631,
+      "grad_norm": 0.6981986165046692,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 129540
+    },
+    {
+      "epoch": 852.3026315789474,
+      "grad_norm": 1.2573353052139282,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 129550
+    },
+    {
+      "epoch": 852.3684210526316,
+      "grad_norm": 0.9952992796897888,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 129560
+    },
+    {
+      "epoch": 852.4342105263158,
+      "grad_norm": 0.9949916005134583,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 129570
+    },
+    {
+      "epoch": 852.5,
+      "grad_norm": 1.2157580852508545,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 129580
+    },
+    {
+      "epoch": 852.5657894736842,
+      "grad_norm": 1.0513194799423218,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 129590
+    },
+    {
+      "epoch": 852.6315789473684,
+      "grad_norm": 1.3504682779312134,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 129600
+    },
+    {
+      "epoch": 852.6973684210526,
+      "grad_norm": 0.9804072380065918,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 129610
+    },
+    {
+      "epoch": 852.7631578947369,
+      "grad_norm": 1.3309262990951538,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 129620
+    },
+    {
+      "epoch": 852.828947368421,
+      "grad_norm": 1.3208234310150146,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 129630
+    },
+    {
+      "epoch": 852.8947368421053,
+      "grad_norm": 1.8901704549789429,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 129640
+    },
+    {
+      "epoch": 852.9605263157895,
+      "grad_norm": 1.7452239990234375,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 129650
+    },
+    {
+      "epoch": 853.0263157894736,
+      "grad_norm": 1.2675105333328247,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 129660
+    },
+    {
+      "epoch": 853.0921052631579,
+      "grad_norm": 0.9307915568351746,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 129670
+    },
+    {
+      "epoch": 853.1578947368421,
+      "grad_norm": 1.136796236038208,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 129680
+    },
+    {
+      "epoch": 853.2236842105264,
+      "grad_norm": 1.2172702550888062,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 129690
+    },
+    {
+      "epoch": 853.2894736842105,
+      "grad_norm": 1.3970834016799927,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 129700
+    },
+    {
+      "epoch": 853.3552631578947,
+      "grad_norm": 1.3139710426330566,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 129710
+    },
+    {
+      "epoch": 853.421052631579,
+      "grad_norm": 1.085668683052063,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 129720
+    },
+    {
+      "epoch": 853.4868421052631,
+      "grad_norm": 1.054573655128479,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 129730
+    },
+    {
+      "epoch": 853.5526315789474,
+      "grad_norm": 1.22200345993042,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 129740
+    },
+    {
+      "epoch": 853.6184210526316,
+      "grad_norm": 1.0596305131912231,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 129750
+    },
+    {
+      "epoch": 853.6842105263158,
+      "grad_norm": 1.0495727062225342,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 129760
+    },
+    {
+      "epoch": 853.75,
+      "grad_norm": 1.4085642099380493,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 129770
+    },
+    {
+      "epoch": 853.8157894736842,
+      "grad_norm": 1.1661970615386963,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 129780
+    },
+    {
+      "epoch": 853.8815789473684,
+      "grad_norm": 1.3750501871109009,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 129790
+    },
+    {
+      "epoch": 853.9473684210526,
+      "grad_norm": 1.349413275718689,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 129800
+    },
+    {
+      "epoch": 854.0131578947369,
+      "grad_norm": 1.1775285005569458,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 129810
+    },
+    {
+      "epoch": 854.078947368421,
+      "grad_norm": 1.127249836921692,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 129820
+    },
+    {
+      "epoch": 854.1447368421053,
+      "grad_norm": 1.1303577423095703,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 129830
+    },
+    {
+      "epoch": 854.2105263157895,
+      "grad_norm": 1.0692039728164673,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 129840
+    },
+    {
+      "epoch": 854.2763157894736,
+      "grad_norm": 1.2041025161743164,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 129850
+    },
+    {
+      "epoch": 854.3421052631579,
+      "grad_norm": 0.8241481184959412,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 129860
+    },
+    {
+      "epoch": 854.4078947368421,
+      "grad_norm": 0.9176292419433594,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 129870
+    },
+    {
+      "epoch": 854.4736842105264,
+      "grad_norm": 1.2800862789154053,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 129880
+    },
+    {
+      "epoch": 854.5394736842105,
+      "grad_norm": 0.7797620892524719,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 129890
+    },
+    {
+      "epoch": 854.6052631578947,
+      "grad_norm": 1.4259511232376099,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 129900
+    },
+    {
+      "epoch": 854.671052631579,
+      "grad_norm": 1.1848866939544678,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 129910
+    },
+    {
+      "epoch": 854.7368421052631,
+      "grad_norm": 0.8558827042579651,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 129920
+    },
+    {
+      "epoch": 854.8026315789474,
+      "grad_norm": 0.9844433069229126,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 129930
+    },
+    {
+      "epoch": 854.8684210526316,
+      "grad_norm": 1.2203624248504639,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 129940
+    },
+    {
+      "epoch": 854.9342105263158,
+      "grad_norm": 1.2477107048034668,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 129950
+    },
+    {
+      "epoch": 855.0,
+      "grad_norm": 1.1950006484985352,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 129960
+    },
+    {
+      "epoch": 855.0657894736842,
+      "grad_norm": 0.8927521109580994,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 129970
+    },
+    {
+      "epoch": 855.1315789473684,
+      "grad_norm": 1.3346248865127563,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 129980
+    },
+    {
+      "epoch": 855.1973684210526,
+      "grad_norm": 1.2857704162597656,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 129990
+    },
+    {
+      "epoch": 855.2631578947369,
+      "grad_norm": 1.3358838558197021,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 130000
+    },
+    {
+      "epoch": 855.328947368421,
+      "grad_norm": 1.1128495931625366,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 130010
+    },
+    {
+      "epoch": 855.3947368421053,
+      "grad_norm": 1.34784734249115,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 130020
+    },
+    {
+      "epoch": 855.4605263157895,
+      "grad_norm": 1.322873592376709,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 130030
+    },
+    {
+      "epoch": 855.5263157894736,
+      "grad_norm": 0.7924045324325562,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 130040
+    },
+    {
+      "epoch": 855.5921052631579,
+      "grad_norm": 0.9343613982200623,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 130050
+    },
+    {
+      "epoch": 855.6578947368421,
+      "grad_norm": 0.847163736820221,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 130060
+    },
+    {
+      "epoch": 855.7236842105264,
+      "grad_norm": 0.8668592572212219,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 130070
+    },
+    {
+      "epoch": 855.7894736842105,
+      "grad_norm": 1.1902350187301636,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 130080
+    },
+    {
+      "epoch": 855.8552631578947,
+      "grad_norm": 0.9959059357643127,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 130090
+    },
+    {
+      "epoch": 855.921052631579,
+      "grad_norm": 0.9965035319328308,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 130100
+    },
+    {
+      "epoch": 855.9868421052631,
+      "grad_norm": 0.8111009001731873,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 130110
+    },
+    {
+      "epoch": 856.0526315789474,
+      "grad_norm": 0.8355821967124939,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 130120
+    },
+    {
+      "epoch": 856.1184210526316,
+      "grad_norm": 0.9395269155502319,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 130130
+    },
+    {
+      "epoch": 856.1842105263158,
+      "grad_norm": 0.7983924150466919,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 130140
+    },
+    {
+      "epoch": 856.25,
+      "grad_norm": 1.0904862880706787,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 130150
+    },
+    {
+      "epoch": 856.3157894736842,
+      "grad_norm": 0.8742513656616211,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 130160
+    },
+    {
+      "epoch": 856.3815789473684,
+      "grad_norm": 1.308323860168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 130170
+    },
+    {
+      "epoch": 856.4473684210526,
+      "grad_norm": 0.660823404788971,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 130180
+    },
+    {
+      "epoch": 856.5131578947369,
+      "grad_norm": 0.7933557033538818,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 130190
+    },
+    {
+      "epoch": 856.578947368421,
+      "grad_norm": 1.5122641324996948,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 130200
+    },
+    {
+      "epoch": 856.6447368421053,
+      "grad_norm": 1.1528301239013672,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 130210
+    },
+    {
+      "epoch": 856.7105263157895,
+      "grad_norm": 0.9372194409370422,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 130220
+    },
+    {
+      "epoch": 856.7763157894736,
+      "grad_norm": 1.312410831451416,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 130230
+    },
+    {
+      "epoch": 856.8421052631579,
+      "grad_norm": 0.9367921948432922,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 130240
+    },
+    {
+      "epoch": 856.9078947368421,
+      "grad_norm": 1.0946214199066162,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 130250
+    },
+    {
+      "epoch": 856.9736842105264,
+      "grad_norm": 1.2462278604507446,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 130260
+    },
+    {
+      "epoch": 857.0394736842105,
+      "grad_norm": 0.9840075373649597,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 130270
+    },
+    {
+      "epoch": 857.1052631578947,
+      "grad_norm": 1.2592564821243286,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 130280
+    },
+    {
+      "epoch": 857.171052631579,
+      "grad_norm": 1.078933835029602,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 130290
+    },
+    {
+      "epoch": 857.2368421052631,
+      "grad_norm": 1.3728272914886475,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 130300
+    },
+    {
+      "epoch": 857.3026315789474,
+      "grad_norm": 1.05517578125,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 130310
+    },
+    {
+      "epoch": 857.3684210526316,
+      "grad_norm": 1.3198639154434204,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 130320
+    },
+    {
+      "epoch": 857.4342105263158,
+      "grad_norm": 1.3628894090652466,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 130330
+    },
+    {
+      "epoch": 857.5,
+      "grad_norm": 1.2717257738113403,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 130340
+    },
+    {
+      "epoch": 857.5657894736842,
+      "grad_norm": 1.429996371269226,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 130350
+    },
+    {
+      "epoch": 857.6315789473684,
+      "grad_norm": 0.9314458966255188,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 130360
+    },
+    {
+      "epoch": 857.6973684210526,
+      "grad_norm": 1.0293179750442505,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 130370
+    },
+    {
+      "epoch": 857.7631578947369,
+      "grad_norm": 0.7930873036384583,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 130380
+    },
+    {
+      "epoch": 857.828947368421,
+      "grad_norm": 1.0124858617782593,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 130390
+    },
+    {
+      "epoch": 857.8947368421053,
+      "grad_norm": 0.6720251441001892,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 130400
+    },
+    {
+      "epoch": 857.9605263157895,
+      "grad_norm": 1.2420949935913086,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 130410
+    },
+    {
+      "epoch": 858.0263157894736,
+      "grad_norm": 1.295690655708313,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 130420
+    },
+    {
+      "epoch": 858.0921052631579,
+      "grad_norm": 0.9636945128440857,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 130430
+    },
+    {
+      "epoch": 858.1578947368421,
+      "grad_norm": 0.9246024489402771,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 130440
+    },
+    {
+      "epoch": 858.2236842105264,
+      "grad_norm": 1.116898536682129,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 130450
+    },
+    {
+      "epoch": 858.2894736842105,
+      "grad_norm": 1.295310378074646,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 130460
+    },
+    {
+      "epoch": 858.3552631578947,
+      "grad_norm": 1.0627254247665405,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 130470
+    },
+    {
+      "epoch": 858.421052631579,
+      "grad_norm": 0.8008264899253845,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 130480
+    },
+    {
+      "epoch": 858.4868421052631,
+      "grad_norm": 1.239762783050537,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 130490
+    },
+    {
+      "epoch": 858.5526315789474,
+      "grad_norm": 0.9881368279457092,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 130500
+    },
+    {
+      "epoch": 858.6184210526316,
+      "grad_norm": 1.0925045013427734,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 130510
+    },
+    {
+      "epoch": 858.6842105263158,
+      "grad_norm": 0.9376081824302673,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 130520
+    },
+    {
+      "epoch": 858.75,
+      "grad_norm": 0.8281996250152588,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 130530
+    },
+    {
+      "epoch": 858.8157894736842,
+      "grad_norm": 0.5148002505302429,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 130540
+    },
+    {
+      "epoch": 858.8815789473684,
+      "grad_norm": 0.9477270245552063,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 130550
+    },
+    {
+      "epoch": 858.9473684210526,
+      "grad_norm": 1.0532981157302856,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 130560
+    },
+    {
+      "epoch": 859.0131578947369,
+      "grad_norm": 1.0642023086547852,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 130570
+    },
+    {
+      "epoch": 859.078947368421,
+      "grad_norm": 1.0734174251556396,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 130580
+    },
+    {
+      "epoch": 859.1447368421053,
+      "grad_norm": 1.0683674812316895,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 130590
+    },
+    {
+      "epoch": 859.2105263157895,
+      "grad_norm": 0.8307974934577942,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 130600
+    },
+    {
+      "epoch": 859.2763157894736,
+      "grad_norm": 0.9671717882156372,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 130610
+    },
+    {
+      "epoch": 859.3421052631579,
+      "grad_norm": 0.8904266953468323,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 130620
+    },
+    {
+      "epoch": 859.4078947368421,
+      "grad_norm": 1.097537875175476,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 130630
+    },
+    {
+      "epoch": 859.4736842105264,
+      "grad_norm": 0.6235876679420471,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 130640
+    },
+    {
+      "epoch": 859.5394736842105,
+      "grad_norm": 1.1195687055587769,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 130650
+    },
+    {
+      "epoch": 859.6052631578947,
+      "grad_norm": 1.0346473455429077,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 130660
+    },
+    {
+      "epoch": 859.671052631579,
+      "grad_norm": 1.2493486404418945,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 130670
+    },
+    {
+      "epoch": 859.7368421052631,
+      "grad_norm": 1.1328728199005127,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 130680
+    },
+    {
+      "epoch": 859.8026315789474,
+      "grad_norm": 1.057497501373291,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 130690
+    },
+    {
+      "epoch": 859.8684210526316,
+      "grad_norm": 1.1181135177612305,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 130700
+    },
+    {
+      "epoch": 859.9342105263158,
+      "grad_norm": 1.242897391319275,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 130710
+    },
+    {
+      "epoch": 860.0,
+      "grad_norm": 1.0586951971054077,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 130720
+    },
+    {
+      "epoch": 860.0657894736842,
+      "grad_norm": 1.2389774322509766,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 130730
+    },
+    {
+      "epoch": 860.1315789473684,
+      "grad_norm": 1.0785973072052002,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 130740
+    },
+    {
+      "epoch": 860.1973684210526,
+      "grad_norm": 1.104970932006836,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 130750
+    },
+    {
+      "epoch": 860.2631578947369,
+      "grad_norm": 0.8328081965446472,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 130760
+    },
+    {
+      "epoch": 860.328947368421,
+      "grad_norm": 0.955934464931488,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 130770
+    },
+    {
+      "epoch": 860.3947368421053,
+      "grad_norm": 1.1552305221557617,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 130780
+    },
+    {
+      "epoch": 860.4605263157895,
+      "grad_norm": 1.000320553779602,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 130790
+    },
+    {
+      "epoch": 860.5263157894736,
+      "grad_norm": 0.9015398025512695,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 130800
+    },
+    {
+      "epoch": 860.5921052631579,
+      "grad_norm": 0.6215592622756958,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 130810
+    },
+    {
+      "epoch": 860.6578947368421,
+      "grad_norm": 0.9939150810241699,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 130820
+    },
+    {
+      "epoch": 860.7236842105264,
+      "grad_norm": 1.01790452003479,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 130830
+    },
+    {
+      "epoch": 860.7894736842105,
+      "grad_norm": 1.468485951423645,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 130840
+    },
+    {
+      "epoch": 860.8552631578947,
+      "grad_norm": 1.1106871366500854,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 130850
+    },
+    {
+      "epoch": 860.921052631579,
+      "grad_norm": 0.834820568561554,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 130860
+    },
+    {
+      "epoch": 860.9868421052631,
+      "grad_norm": 0.9590187072753906,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 130870
+    },
+    {
+      "epoch": 861.0526315789474,
+      "grad_norm": 1.1876466274261475,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 130880
+    },
+    {
+      "epoch": 861.1184210526316,
+      "grad_norm": 0.968468427658081,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 130890
+    },
+    {
+      "epoch": 861.1842105263158,
+      "grad_norm": 0.8767639398574829,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 130900
+    },
+    {
+      "epoch": 861.25,
+      "grad_norm": 0.9714983105659485,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 130910
+    },
+    {
+      "epoch": 861.3157894736842,
+      "grad_norm": 1.2602828741073608,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 130920
+    },
+    {
+      "epoch": 861.3815789473684,
+      "grad_norm": 1.2543740272521973,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 130930
+    },
+    {
+      "epoch": 861.4473684210526,
+      "grad_norm": 1.088036060333252,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 130940
+    },
+    {
+      "epoch": 861.5131578947369,
+      "grad_norm": 1.3729690313339233,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 130950
+    },
+    {
+      "epoch": 861.578947368421,
+      "grad_norm": 1.1072582006454468,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 130960
+    },
+    {
+      "epoch": 861.6447368421053,
+      "grad_norm": 1.1380093097686768,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 130970
+    },
+    {
+      "epoch": 861.7105263157895,
+      "grad_norm": 1.0508384704589844,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 130980
+    },
+    {
+      "epoch": 861.7763157894736,
+      "grad_norm": 1.1020636558532715,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 130990
+    },
+    {
+      "epoch": 861.8421052631579,
+      "grad_norm": 1.138003945350647,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 131000
+    },
+    {
+      "epoch": 861.9078947368421,
+      "grad_norm": 1.2972642183303833,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 131010
+    },
+    {
+      "epoch": 861.9736842105264,
+      "grad_norm": 0.911027729511261,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 131020
+    },
+    {
+      "epoch": 862.0394736842105,
+      "grad_norm": 0.9824381470680237,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 131030
+    },
+    {
+      "epoch": 862.1052631578947,
+      "grad_norm": 1.227348804473877,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 131040
+    },
+    {
+      "epoch": 862.171052631579,
+      "grad_norm": 0.9096339344978333,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 131050
+    },
+    {
+      "epoch": 862.2368421052631,
+      "grad_norm": 1.2303411960601807,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 131060
+    },
+    {
+      "epoch": 862.3026315789474,
+      "grad_norm": 0.8208591938018799,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 131070
+    },
+    {
+      "epoch": 862.3684210526316,
+      "grad_norm": 1.1471190452575684,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 131080
+    },
+    {
+      "epoch": 862.4342105263158,
+      "grad_norm": 0.8580765128135681,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 131090
+    },
+    {
+      "epoch": 862.5,
+      "grad_norm": 0.9216636419296265,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 131100
+    },
+    {
+      "epoch": 862.5657894736842,
+      "grad_norm": 1.3407820463180542,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 131110
+    },
+    {
+      "epoch": 862.6315789473684,
+      "grad_norm": 1.359041452407837,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 131120
+    },
+    {
+      "epoch": 862.6973684210526,
+      "grad_norm": 1.231490969657898,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 131130
+    },
+    {
+      "epoch": 862.7631578947369,
+      "grad_norm": 1.0086109638214111,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 131140
+    },
+    {
+      "epoch": 862.828947368421,
+      "grad_norm": 0.9800273180007935,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 131150
+    },
+    {
+      "epoch": 862.8947368421053,
+      "grad_norm": 1.3657548427581787,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 131160
+    },
+    {
+      "epoch": 862.9605263157895,
+      "grad_norm": 1.1648259162902832,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 131170
+    },
+    {
+      "epoch": 863.0263157894736,
+      "grad_norm": 1.3625643253326416,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 131180
+    },
+    {
+      "epoch": 863.0921052631579,
+      "grad_norm": 0.833634614944458,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 131190
+    },
+    {
+      "epoch": 863.1578947368421,
+      "grad_norm": 0.7656792402267456,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 131200
+    },
+    {
+      "epoch": 863.2236842105264,
+      "grad_norm": 1.2618483304977417,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 131210
+    },
+    {
+      "epoch": 863.2894736842105,
+      "grad_norm": 1.133224606513977,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 131220
+    },
+    {
+      "epoch": 863.3552631578947,
+      "grad_norm": 1.1126824617385864,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 131230
+    },
+    {
+      "epoch": 863.421052631579,
+      "grad_norm": 1.0469779968261719,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 131240
+    },
+    {
+      "epoch": 863.4868421052631,
+      "grad_norm": 1.0890519618988037,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 131250
+    },
+    {
+      "epoch": 863.5526315789474,
+      "grad_norm": 1.1236612796783447,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 131260
+    },
+    {
+      "epoch": 863.6184210526316,
+      "grad_norm": 1.1163815259933472,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 131270
+    },
+    {
+      "epoch": 863.6842105263158,
+      "grad_norm": 0.9706016182899475,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 131280
+    },
+    {
+      "epoch": 863.75,
+      "grad_norm": 0.7366560697555542,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 131290
+    },
+    {
+      "epoch": 863.8157894736842,
+      "grad_norm": 1.2588547468185425,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 131300
+    },
+    {
+      "epoch": 863.8815789473684,
+      "grad_norm": 0.7898572087287903,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 131310
+    },
+    {
+      "epoch": 863.9473684210526,
+      "grad_norm": 1.2470197677612305,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 131320
+    },
+    {
+      "epoch": 864.0131578947369,
+      "grad_norm": 1.060998797416687,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 131330
+    },
+    {
+      "epoch": 864.078947368421,
+      "grad_norm": 0.6772517561912537,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 131340
+    },
+    {
+      "epoch": 864.1447368421053,
+      "grad_norm": 1.205532193183899,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 131350
+    },
+    {
+      "epoch": 864.2105263157895,
+      "grad_norm": 1.150336742401123,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 131360
+    },
+    {
+      "epoch": 864.2763157894736,
+      "grad_norm": 0.8620902895927429,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 131370
+    },
+    {
+      "epoch": 864.3421052631579,
+      "grad_norm": 1.0216280221939087,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 131380
+    },
+    {
+      "epoch": 864.4078947368421,
+      "grad_norm": 1.0433382987976074,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 131390
+    },
+    {
+      "epoch": 864.4736842105264,
+      "grad_norm": 1.3569798469543457,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 131400
+    },
+    {
+      "epoch": 864.5394736842105,
+      "grad_norm": 1.1768606901168823,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 131410
+    },
+    {
+      "epoch": 864.6052631578947,
+      "grad_norm": 1.2419917583465576,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 131420
+    },
+    {
+      "epoch": 864.671052631579,
+      "grad_norm": 1.128338098526001,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 131430
+    },
+    {
+      "epoch": 864.7368421052631,
+      "grad_norm": 1.1112442016601562,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 131440
+    },
+    {
+      "epoch": 864.8026315789474,
+      "grad_norm": 1.0152097940444946,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 131450
+    },
+    {
+      "epoch": 864.8684210526316,
+      "grad_norm": 0.9926842451095581,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 131460
+    },
+    {
+      "epoch": 864.9342105263158,
+      "grad_norm": 0.9706729054450989,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 131470
+    },
+    {
+      "epoch": 865.0,
+      "grad_norm": 0.9864161610603333,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 131480
+    },
+    {
+      "epoch": 865.0657894736842,
+      "grad_norm": 1.234376072883606,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 131490
+    },
+    {
+      "epoch": 865.1315789473684,
+      "grad_norm": 0.8708391189575195,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 131500
+    },
+    {
+      "epoch": 865.1973684210526,
+      "grad_norm": 1.2367292642593384,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 131510
+    },
+    {
+      "epoch": 865.2631578947369,
+      "grad_norm": 1.1945315599441528,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 131520
+    },
+    {
+      "epoch": 865.328947368421,
+      "grad_norm": 1.3687530755996704,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 131530
+    },
+    {
+      "epoch": 865.3947368421053,
+      "grad_norm": 1.2335397005081177,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 131540
+    },
+    {
+      "epoch": 865.4605263157895,
+      "grad_norm": 0.9434360265731812,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 131550
+    },
+    {
+      "epoch": 865.5263157894736,
+      "grad_norm": 1.020836591720581,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 131560
+    },
+    {
+      "epoch": 865.5921052631579,
+      "grad_norm": 1.0214914083480835,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 131570
+    },
+    {
+      "epoch": 865.6578947368421,
+      "grad_norm": 1.2193523645401,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 131580
+    },
+    {
+      "epoch": 865.7236842105264,
+      "grad_norm": 1.0998847484588623,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 131590
+    },
+    {
+      "epoch": 865.7894736842105,
+      "grad_norm": 1.1614649295806885,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 131600
+    },
+    {
+      "epoch": 865.8552631578947,
+      "grad_norm": 1.3147051334381104,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 131610
+    },
+    {
+      "epoch": 865.921052631579,
+      "grad_norm": 0.9823493957519531,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 131620
+    },
+    {
+      "epoch": 865.9868421052631,
+      "grad_norm": 0.8709498643875122,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 131630
+    },
+    {
+      "epoch": 866.0526315789474,
+      "grad_norm": 1.4504932165145874,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 131640
+    },
+    {
+      "epoch": 866.1184210526316,
+      "grad_norm": 0.9600393772125244,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 131650
+    },
+    {
+      "epoch": 866.1842105263158,
+      "grad_norm": 0.9422047734260559,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 131660
+    },
+    {
+      "epoch": 866.25,
+      "grad_norm": 0.9015654921531677,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 131670
+    },
+    {
+      "epoch": 866.3157894736842,
+      "grad_norm": 1.2097015380859375,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 131680
+    },
+    {
+      "epoch": 866.3815789473684,
+      "grad_norm": 1.0797349214553833,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 131690
+    },
+    {
+      "epoch": 866.4473684210526,
+      "grad_norm": 1.1002148389816284,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 131700
+    },
+    {
+      "epoch": 866.5131578947369,
+      "grad_norm": 0.8693884611129761,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 131710
+    },
+    {
+      "epoch": 866.578947368421,
+      "grad_norm": 1.2665678262710571,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 131720
+    },
+    {
+      "epoch": 866.6447368421053,
+      "grad_norm": 1.0744692087173462,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 131730
+    },
+    {
+      "epoch": 866.7105263157895,
+      "grad_norm": 1.1713916063308716,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 131740
+    },
+    {
+      "epoch": 866.7763157894736,
+      "grad_norm": 1.3538668155670166,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 131750
+    },
+    {
+      "epoch": 866.8421052631579,
+      "grad_norm": 0.8203177452087402,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 131760
+    },
+    {
+      "epoch": 866.9078947368421,
+      "grad_norm": 1.1548283100128174,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 131770
+    },
+    {
+      "epoch": 866.9736842105264,
+      "grad_norm": 1.0071430206298828,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 131780
+    },
+    {
+      "epoch": 867.0394736842105,
+      "grad_norm": 1.1559392213821411,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 131790
+    },
+    {
+      "epoch": 867.1052631578947,
+      "grad_norm": 1.0900834798812866,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 131800
+    },
+    {
+      "epoch": 867.171052631579,
+      "grad_norm": 1.1441543102264404,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 131810
+    },
+    {
+      "epoch": 867.2368421052631,
+      "grad_norm": 1.2302360534667969,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 131820
+    },
+    {
+      "epoch": 867.3026315789474,
+      "grad_norm": 1.0534881353378296,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 131830
+    },
+    {
+      "epoch": 867.3684210526316,
+      "grad_norm": 0.8932411074638367,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 131840
+    },
+    {
+      "epoch": 867.4342105263158,
+      "grad_norm": 0.6699873805046082,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 131850
+    },
+    {
+      "epoch": 867.5,
+      "grad_norm": 0.7454432845115662,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 131860
+    },
+    {
+      "epoch": 867.5657894736842,
+      "grad_norm": 1.1074482202529907,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 131870
+    },
+    {
+      "epoch": 867.6315789473684,
+      "grad_norm": 1.2334349155426025,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 131880
+    },
+    {
+      "epoch": 867.6973684210526,
+      "grad_norm": 0.9547383189201355,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 131890
+    },
+    {
+      "epoch": 867.7631578947369,
+      "grad_norm": 1.069530963897705,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 131900
+    },
+    {
+      "epoch": 867.828947368421,
+      "grad_norm": 1.2506078481674194,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 131910
+    },
+    {
+      "epoch": 867.8947368421053,
+      "grad_norm": 0.9932133555412292,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 131920
+    },
+    {
+      "epoch": 867.9605263157895,
+      "grad_norm": 1.2245303392410278,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 131930
+    },
+    {
+      "epoch": 868.0263157894736,
+      "grad_norm": 1.5292211771011353,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 131940
+    },
+    {
+      "epoch": 868.0921052631579,
+      "grad_norm": 1.3680943250656128,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 131950
+    },
+    {
+      "epoch": 868.1578947368421,
+      "grad_norm": 1.1934568881988525,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 131960
+    },
+    {
+      "epoch": 868.2236842105264,
+      "grad_norm": 0.9152090549468994,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 131970
+    },
+    {
+      "epoch": 868.2894736842105,
+      "grad_norm": 0.8579458594322205,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 131980
+    },
+    {
+      "epoch": 868.3552631578947,
+      "grad_norm": 0.9107769131660461,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 131990
+    },
+    {
+      "epoch": 868.421052631579,
+      "grad_norm": 1.1969023942947388,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 132000
+    },
+    {
+      "epoch": 868.4868421052631,
+      "grad_norm": 0.9776577949523926,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 132010
+    },
+    {
+      "epoch": 868.5526315789474,
+      "grad_norm": 1.3878238201141357,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 132020
+    },
+    {
+      "epoch": 868.6184210526316,
+      "grad_norm": 0.997664213180542,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 132030
+    },
+    {
+      "epoch": 868.6842105263158,
+      "grad_norm": 1.237343430519104,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 132040
+    },
+    {
+      "epoch": 868.75,
+      "grad_norm": 0.9770588874816895,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 132050
+    },
+    {
+      "epoch": 868.8157894736842,
+      "grad_norm": 1.209973692893982,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 132060
+    },
+    {
+      "epoch": 868.8815789473684,
+      "grad_norm": 1.109294056892395,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 132070
+    },
+    {
+      "epoch": 868.9473684210526,
+      "grad_norm": 0.8828414082527161,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 132080
+    },
+    {
+      "epoch": 869.0131578947369,
+      "grad_norm": 0.8478983044624329,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 132090
+    },
+    {
+      "epoch": 869.078947368421,
+      "grad_norm": 1.0058600902557373,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 132100
+    },
+    {
+      "epoch": 869.1447368421053,
+      "grad_norm": 1.5055391788482666,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 132110
+    },
+    {
+      "epoch": 869.2105263157895,
+      "grad_norm": 1.3304319381713867,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 132120
+    },
+    {
+      "epoch": 869.2763157894736,
+      "grad_norm": 1.3461639881134033,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 132130
+    },
+    {
+      "epoch": 869.3421052631579,
+      "grad_norm": 1.318153977394104,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 132140
+    },
+    {
+      "epoch": 869.4078947368421,
+      "grad_norm": 1.3171056509017944,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 132150
+    },
+    {
+      "epoch": 869.4736842105264,
+      "grad_norm": 1.2487008571624756,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 132160
+    },
+    {
+      "epoch": 869.5394736842105,
+      "grad_norm": 0.9375091195106506,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 132170
+    },
+    {
+      "epoch": 869.6052631578947,
+      "grad_norm": 1.3363336324691772,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 132180
+    },
+    {
+      "epoch": 869.671052631579,
+      "grad_norm": 1.0731987953186035,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 132190
+    },
+    {
+      "epoch": 869.7368421052631,
+      "grad_norm": 1.6114320755004883,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 132200
+    },
+    {
+      "epoch": 869.8026315789474,
+      "grad_norm": 1.3398220539093018,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 132210
+    },
+    {
+      "epoch": 869.8684210526316,
+      "grad_norm": 0.9430725574493408,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 132220
+    },
+    {
+      "epoch": 869.9342105263158,
+      "grad_norm": 0.9981265068054199,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 132230
+    },
+    {
+      "epoch": 870.0,
+      "grad_norm": 1.1120591163635254,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 132240
+    },
+    {
+      "epoch": 870.0657894736842,
+      "grad_norm": 1.0576080083847046,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 132250
+    },
+    {
+      "epoch": 870.1315789473684,
+      "grad_norm": 1.1692156791687012,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 132260
+    },
+    {
+      "epoch": 870.1973684210526,
+      "grad_norm": 1.4510517120361328,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 132270
+    },
+    {
+      "epoch": 870.2631578947369,
+      "grad_norm": 0.8253922462463379,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 132280
+    },
+    {
+      "epoch": 870.328947368421,
+      "grad_norm": 1.1040948629379272,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 132290
+    },
+    {
+      "epoch": 870.3947368421053,
+      "grad_norm": 1.4819238185882568,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 132300
+    },
+    {
+      "epoch": 870.4605263157895,
+      "grad_norm": 0.8338831663131714,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 132310
+    },
+    {
+      "epoch": 870.5263157894736,
+      "grad_norm": 1.5496101379394531,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 132320
+    },
+    {
+      "epoch": 870.5921052631579,
+      "grad_norm": 1.3076415061950684,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 132330
+    },
+    {
+      "epoch": 870.6578947368421,
+      "grad_norm": 1.3276525735855103,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 132340
+    },
+    {
+      "epoch": 870.7236842105264,
+      "grad_norm": 1.18726646900177,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 132350
+    },
+    {
+      "epoch": 870.7894736842105,
+      "grad_norm": 1.3741918802261353,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 132360
+    },
+    {
+      "epoch": 870.8552631578947,
+      "grad_norm": 0.9850346446037292,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 132370
+    },
+    {
+      "epoch": 870.921052631579,
+      "grad_norm": 0.5921350717544556,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 132380
+    },
+    {
+      "epoch": 870.9868421052631,
+      "grad_norm": 1.010513424873352,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 132390
+    },
+    {
+      "epoch": 871.0526315789474,
+      "grad_norm": 1.3716336488723755,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 132400
+    },
+    {
+      "epoch": 871.1184210526316,
+      "grad_norm": 0.9653910994529724,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 132410
+    },
+    {
+      "epoch": 871.1842105263158,
+      "grad_norm": 1.3690600395202637,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 132420
+    },
+    {
+      "epoch": 871.25,
+      "grad_norm": 1.072267770767212,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 132430
+    },
+    {
+      "epoch": 871.3157894736842,
+      "grad_norm": 1.4082567691802979,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 132440
+    },
+    {
+      "epoch": 871.3815789473684,
+      "grad_norm": 1.0181151628494263,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 132450
+    },
+    {
+      "epoch": 871.4473684210526,
+      "grad_norm": 0.8527328968048096,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 132460
+    },
+    {
+      "epoch": 871.5131578947369,
+      "grad_norm": 1.0901014804840088,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 132470
+    },
+    {
+      "epoch": 871.578947368421,
+      "grad_norm": 1.1375715732574463,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 132480
+    },
+    {
+      "epoch": 871.6447368421053,
+      "grad_norm": 1.0507901906967163,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 132490
+    },
+    {
+      "epoch": 871.7105263157895,
+      "grad_norm": 0.819428563117981,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 132500
+    },
+    {
+      "epoch": 871.7763157894736,
+      "grad_norm": 0.7174791693687439,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 132510
+    },
+    {
+      "epoch": 871.8421052631579,
+      "grad_norm": 1.1212362051010132,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 132520
+    },
+    {
+      "epoch": 871.9078947368421,
+      "grad_norm": 1.310887098312378,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 132530
+    },
+    {
+      "epoch": 871.9736842105264,
+      "grad_norm": 0.9797291159629822,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 132540
+    },
+    {
+      "epoch": 872.0394736842105,
+      "grad_norm": 1.042771816253662,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 132550
+    },
+    {
+      "epoch": 872.1052631578947,
+      "grad_norm": 1.1746186017990112,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 132560
+    },
+    {
+      "epoch": 872.171052631579,
+      "grad_norm": 1.1553187370300293,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 132570
+    },
+    {
+      "epoch": 872.2368421052631,
+      "grad_norm": 0.9846141338348389,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 132580
+    },
+    {
+      "epoch": 872.3026315789474,
+      "grad_norm": 1.0329734086990356,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 132590
+    },
+    {
+      "epoch": 872.3684210526316,
+      "grad_norm": 1.0318963527679443,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 132600
+    },
+    {
+      "epoch": 872.4342105263158,
+      "grad_norm": 0.97999507188797,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 132610
+    },
+    {
+      "epoch": 872.5,
+      "grad_norm": 0.8947683572769165,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 132620
+    },
+    {
+      "epoch": 872.5657894736842,
+      "grad_norm": 0.7579332590103149,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 132630
+    },
+    {
+      "epoch": 872.6315789473684,
+      "grad_norm": 0.8904950022697449,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 132640
+    },
+    {
+      "epoch": 872.6973684210526,
+      "grad_norm": 1.2979966402053833,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 132650
+    },
+    {
+      "epoch": 872.7631578947369,
+      "grad_norm": 0.7980130314826965,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 132660
+    },
+    {
+      "epoch": 872.828947368421,
+      "grad_norm": 0.6705465912818909,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 132670
+    },
+    {
+      "epoch": 872.8947368421053,
+      "grad_norm": 0.5814407467842102,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 132680
+    },
+    {
+      "epoch": 872.9605263157895,
+      "grad_norm": 0.9677609205245972,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 132690
+    },
+    {
+      "epoch": 873.0263157894736,
+      "grad_norm": 0.7871490716934204,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 132700
+    },
+    {
+      "epoch": 873.0921052631579,
+      "grad_norm": 1.0843896865844727,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 132710
+    },
+    {
+      "epoch": 873.1578947368421,
+      "grad_norm": 1.4221380949020386,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 132720
+    },
+    {
+      "epoch": 873.2236842105264,
+      "grad_norm": 0.7746054530143738,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 132730
+    },
+    {
+      "epoch": 873.2894736842105,
+      "grad_norm": 1.0406986474990845,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 132740
+    },
+    {
+      "epoch": 873.3552631578947,
+      "grad_norm": 1.3779292106628418,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 132750
+    },
+    {
+      "epoch": 873.421052631579,
+      "grad_norm": 1.179019808769226,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 132760
+    },
+    {
+      "epoch": 873.4868421052631,
+      "grad_norm": 0.9589442014694214,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 132770
+    },
+    {
+      "epoch": 873.5526315789474,
+      "grad_norm": 0.8380019664764404,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 132780
+    },
+    {
+      "epoch": 873.6184210526316,
+      "grad_norm": 1.0861836671829224,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 132790
+    },
+    {
+      "epoch": 873.6842105263158,
+      "grad_norm": 0.7795262932777405,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 132800
+    },
+    {
+      "epoch": 873.75,
+      "grad_norm": 1.4798517227172852,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 132810
+    },
+    {
+      "epoch": 873.8157894736842,
+      "grad_norm": 0.8960368037223816,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 132820
+    },
+    {
+      "epoch": 873.8815789473684,
+      "grad_norm": 1.1715494394302368,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 132830
+    },
+    {
+      "epoch": 873.9473684210526,
+      "grad_norm": 1.0302726030349731,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 132840
+    },
+    {
+      "epoch": 874.0131578947369,
+      "grad_norm": 0.8323149085044861,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 132850
+    },
+    {
+      "epoch": 874.078947368421,
+      "grad_norm": 1.0214720964431763,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 132860
+    },
+    {
+      "epoch": 874.1447368421053,
+      "grad_norm": 1.2060658931732178,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 132870
+    },
+    {
+      "epoch": 874.2105263157895,
+      "grad_norm": 1.1269394159317017,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 132880
+    },
+    {
+      "epoch": 874.2763157894736,
+      "grad_norm": 1.2563400268554688,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 132890
+    },
+    {
+      "epoch": 874.3421052631579,
+      "grad_norm": 0.6155217885971069,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 132900
+    },
+    {
+      "epoch": 874.4078947368421,
+      "grad_norm": 1.0993677377700806,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 132910
+    },
+    {
+      "epoch": 874.4736842105264,
+      "grad_norm": 0.8634936809539795,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 132920
+    },
+    {
+      "epoch": 874.5394736842105,
+      "grad_norm": 1.18904709815979,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 132930
+    },
+    {
+      "epoch": 874.6052631578947,
+      "grad_norm": 0.8078858852386475,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 132940
+    },
+    {
+      "epoch": 874.671052631579,
+      "grad_norm": 0.8726778626441956,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 132950
+    },
+    {
+      "epoch": 874.7368421052631,
+      "grad_norm": 0.8983378410339355,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 132960
+    },
+    {
+      "epoch": 874.8026315789474,
+      "grad_norm": 0.6301218271255493,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 132970
+    },
+    {
+      "epoch": 874.8684210526316,
+      "grad_norm": 1.143539547920227,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 132980
+    },
+    {
+      "epoch": 874.9342105263158,
+      "grad_norm": 0.7233585119247437,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 132990
+    },
+    {
+      "epoch": 875.0,
+      "grad_norm": 0.7568589448928833,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 133000
+    },
+    {
+      "epoch": 875.0657894736842,
+      "grad_norm": 1.0015631914138794,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 133010
+    },
+    {
+      "epoch": 875.1315789473684,
+      "grad_norm": 1.3138691186904907,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 133020
+    },
+    {
+      "epoch": 875.1973684210526,
+      "grad_norm": 1.1672554016113281,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 133030
+    },
+    {
+      "epoch": 875.2631578947369,
+      "grad_norm": 1.2994216680526733,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 133040
+    },
+    {
+      "epoch": 875.328947368421,
+      "grad_norm": 1.1386960744857788,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 133050
+    },
+    {
+      "epoch": 875.3947368421053,
+      "grad_norm": 0.8103448152542114,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 133060
+    },
+    {
+      "epoch": 875.4605263157895,
+      "grad_norm": 1.0586600303649902,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 133070
+    },
+    {
+      "epoch": 875.5263157894736,
+      "grad_norm": 1.086573839187622,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 133080
+    },
+    {
+      "epoch": 875.5921052631579,
+      "grad_norm": 0.9082050323486328,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 133090
+    },
+    {
+      "epoch": 875.6578947368421,
+      "grad_norm": 1.2547775506973267,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 133100
+    },
+    {
+      "epoch": 875.7236842105264,
+      "grad_norm": 1.1525391340255737,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 133110
+    },
+    {
+      "epoch": 875.7894736842105,
+      "grad_norm": 1.0197055339813232,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 133120
+    },
+    {
+      "epoch": 875.8552631578947,
+      "grad_norm": 1.1045457124710083,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 133130
+    },
+    {
+      "epoch": 875.921052631579,
+      "grad_norm": 1.19781494140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 133140
+    },
+    {
+      "epoch": 875.9868421052631,
+      "grad_norm": 1.0859558582305908,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 133150
+    },
+    {
+      "epoch": 876.0526315789474,
+      "grad_norm": 1.278917670249939,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 133160
+    },
+    {
+      "epoch": 876.1184210526316,
+      "grad_norm": 0.9689110517501831,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 133170
+    },
+    {
+      "epoch": 876.1842105263158,
+      "grad_norm": 1.1780256032943726,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 133180
+    },
+    {
+      "epoch": 876.25,
+      "grad_norm": 1.1999872922897339,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 133190
+    },
+    {
+      "epoch": 876.3157894736842,
+      "grad_norm": 1.1459710597991943,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 133200
+    },
+    {
+      "epoch": 876.3815789473684,
+      "grad_norm": 1.1914170980453491,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 133210
+    },
+    {
+      "epoch": 876.4473684210526,
+      "grad_norm": 1.8968595266342163,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 133220
+    },
+    {
+      "epoch": 876.5131578947369,
+      "grad_norm": 1.641607642173767,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 133230
+    },
+    {
+      "epoch": 876.578947368421,
+      "grad_norm": 1.1843949556350708,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 133240
+    },
+    {
+      "epoch": 876.6447368421053,
+      "grad_norm": 1.2296756505966187,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 133250
+    },
+    {
+      "epoch": 876.7105263157895,
+      "grad_norm": 1.4620548486709595,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 133260
+    },
+    {
+      "epoch": 876.7763157894736,
+      "grad_norm": 1.4632091522216797,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 133270
+    },
+    {
+      "epoch": 876.8421052631579,
+      "grad_norm": 1.3262276649475098,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 133280
+    },
+    {
+      "epoch": 876.9078947368421,
+      "grad_norm": 1.4855178594589233,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 133290
+    },
+    {
+      "epoch": 876.9736842105264,
+      "grad_norm": 1.28376042842865,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 133300
+    },
+    {
+      "epoch": 877.0394736842105,
+      "grad_norm": 1.2845184803009033,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 133310
+    },
+    {
+      "epoch": 877.1052631578947,
+      "grad_norm": 1.154748797416687,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 133320
+    },
+    {
+      "epoch": 877.171052631579,
+      "grad_norm": 1.3325797319412231,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 133330
+    },
+    {
+      "epoch": 877.2368421052631,
+      "grad_norm": 1.2600542306900024,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 133340
+    },
+    {
+      "epoch": 877.3026315789474,
+      "grad_norm": 1.4029486179351807,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 133350
+    },
+    {
+      "epoch": 877.3684210526316,
+      "grad_norm": 1.3744686841964722,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 133360
+    },
+    {
+      "epoch": 877.4342105263158,
+      "grad_norm": 1.225288987159729,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 133370
+    },
+    {
+      "epoch": 877.5,
+      "grad_norm": 1.006150245666504,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 133380
+    },
+    {
+      "epoch": 877.5657894736842,
+      "grad_norm": 0.9903148412704468,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 133390
+    },
+    {
+      "epoch": 877.6315789473684,
+      "grad_norm": 1.0531034469604492,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 133400
+    },
+    {
+      "epoch": 877.6973684210526,
+      "grad_norm": 1.0496103763580322,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 133410
+    },
+    {
+      "epoch": 877.7631578947369,
+      "grad_norm": 1.3180263042449951,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 133420
+    },
+    {
+      "epoch": 877.828947368421,
+      "grad_norm": 1.455904245376587,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 133430
+    },
+    {
+      "epoch": 877.8947368421053,
+      "grad_norm": 0.9219629168510437,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 133440
+    },
+    {
+      "epoch": 877.9605263157895,
+      "grad_norm": 1.1099201440811157,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 133450
+    },
+    {
+      "epoch": 878.0263157894736,
+      "grad_norm": 1.0845720767974854,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 133460
+    },
+    {
+      "epoch": 878.0921052631579,
+      "grad_norm": 1.3005256652832031,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 133470
+    },
+    {
+      "epoch": 878.1578947368421,
+      "grad_norm": 1.2297483682632446,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 133480
+    },
+    {
+      "epoch": 878.2236842105264,
+      "grad_norm": 1.2901902198791504,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 133490
+    },
+    {
+      "epoch": 878.2894736842105,
+      "grad_norm": 1.3723936080932617,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 133500
+    },
+    {
+      "epoch": 878.3552631578947,
+      "grad_norm": 1.08962082862854,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 133510
+    },
+    {
+      "epoch": 878.421052631579,
+      "grad_norm": 1.2790974378585815,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 133520
+    },
+    {
+      "epoch": 878.4868421052631,
+      "grad_norm": 1.027013897895813,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 133530
+    },
+    {
+      "epoch": 878.5526315789474,
+      "grad_norm": 1.4130576848983765,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 133540
+    },
+    {
+      "epoch": 878.6184210526316,
+      "grad_norm": 1.0892243385314941,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 133550
+    },
+    {
+      "epoch": 878.6842105263158,
+      "grad_norm": 1.1556276082992554,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 133560
+    },
+    {
+      "epoch": 878.75,
+      "grad_norm": 1.008480191230774,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 133570
+    },
+    {
+      "epoch": 878.8157894736842,
+      "grad_norm": 0.7436794638633728,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 133580
+    },
+    {
+      "epoch": 878.8815789473684,
+      "grad_norm": 1.1126219034194946,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 133590
+    },
+    {
+      "epoch": 878.9473684210526,
+      "grad_norm": 1.165494441986084,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 133600
+    },
+    {
+      "epoch": 879.0131578947369,
+      "grad_norm": 1.1165974140167236,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 133610
+    },
+    {
+      "epoch": 879.078947368421,
+      "grad_norm": 1.1761646270751953,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 133620
+    },
+    {
+      "epoch": 879.1447368421053,
+      "grad_norm": 1.1715446710586548,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 133630
+    },
+    {
+      "epoch": 879.2105263157895,
+      "grad_norm": 1.2596334218978882,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 133640
+    },
+    {
+      "epoch": 879.2763157894736,
+      "grad_norm": 0.70237797498703,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 133650
+    },
+    {
+      "epoch": 879.3421052631579,
+      "grad_norm": 0.857401967048645,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 133660
+    },
+    {
+      "epoch": 879.4078947368421,
+      "grad_norm": 0.7992585301399231,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 133670
+    },
+    {
+      "epoch": 879.4736842105264,
+      "grad_norm": 0.7546584606170654,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 133680
+    },
+    {
+      "epoch": 879.5394736842105,
+      "grad_norm": 1.1993327140808105,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 133690
+    },
+    {
+      "epoch": 879.6052631578947,
+      "grad_norm": 1.0572268962860107,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 133700
+    },
+    {
+      "epoch": 879.671052631579,
+      "grad_norm": 1.2813950777053833,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 133710
+    },
+    {
+      "epoch": 879.7368421052631,
+      "grad_norm": 0.7247679233551025,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 133720
+    },
+    {
+      "epoch": 879.8026315789474,
+      "grad_norm": 0.8819660544395447,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 133730
+    },
+    {
+      "epoch": 879.8684210526316,
+      "grad_norm": 0.9946263432502747,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 133740
+    },
+    {
+      "epoch": 879.9342105263158,
+      "grad_norm": 0.8821104168891907,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 133750
+    },
+    {
+      "epoch": 880.0,
+      "grad_norm": 0.6893178820610046,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 133760
+    },
+    {
+      "epoch": 880.0657894736842,
+      "grad_norm": 1.0600444078445435,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 133770
+    },
+    {
+      "epoch": 880.1315789473684,
+      "grad_norm": 1.0986602306365967,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 133780
+    },
+    {
+      "epoch": 880.1973684210526,
+      "grad_norm": 0.8387561440467834,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 133790
+    },
+    {
+      "epoch": 880.2631578947369,
+      "grad_norm": 0.8584687113761902,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 133800
+    },
+    {
+      "epoch": 880.328947368421,
+      "grad_norm": 0.9088717699050903,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 133810
+    },
+    {
+      "epoch": 880.3947368421053,
+      "grad_norm": 0.9743188619613647,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 133820
+    },
+    {
+      "epoch": 880.4605263157895,
+      "grad_norm": 0.6925018429756165,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 133830
+    },
+    {
+      "epoch": 880.5263157894736,
+      "grad_norm": 0.9171350002288818,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 133840
+    },
+    {
+      "epoch": 880.5921052631579,
+      "grad_norm": 0.8944483399391174,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 133850
+    },
+    {
+      "epoch": 880.6578947368421,
+      "grad_norm": 0.859816312789917,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 133860
+    },
+    {
+      "epoch": 880.7236842105264,
+      "grad_norm": 0.8540825247764587,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 133870
+    },
+    {
+      "epoch": 880.7894736842105,
+      "grad_norm": 0.8680028319358826,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 133880
+    },
+    {
+      "epoch": 880.8552631578947,
+      "grad_norm": 0.6797283291816711,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 133890
+    },
+    {
+      "epoch": 880.921052631579,
+      "grad_norm": 1.095505952835083,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 133900
+    },
+    {
+      "epoch": 880.9868421052631,
+      "grad_norm": 0.8483147025108337,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 133910
+    },
+    {
+      "epoch": 881.0526315789474,
+      "grad_norm": 1.0879533290863037,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 133920
+    },
+    {
+      "epoch": 881.1184210526316,
+      "grad_norm": 1.1705760955810547,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 133930
+    },
+    {
+      "epoch": 881.1842105263158,
+      "grad_norm": 1.3736196756362915,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 133940
+    },
+    {
+      "epoch": 881.25,
+      "grad_norm": 1.5451818704605103,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 133950
+    },
+    {
+      "epoch": 881.3157894736842,
+      "grad_norm": 1.1037559509277344,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 133960
+    },
+    {
+      "epoch": 881.3815789473684,
+      "grad_norm": 1.054043173789978,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 133970
+    },
+    {
+      "epoch": 881.4473684210526,
+      "grad_norm": 1.1496354341506958,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 133980
+    },
+    {
+      "epoch": 881.5131578947369,
+      "grad_norm": 0.895317792892456,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 133990
+    },
+    {
+      "epoch": 881.578947368421,
+      "grad_norm": 0.9729178547859192,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 134000
+    },
+    {
+      "epoch": 881.6447368421053,
+      "grad_norm": 0.8765199184417725,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 134010
+    },
+    {
+      "epoch": 881.7105263157895,
+      "grad_norm": 1.216593861579895,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 134020
+    },
+    {
+      "epoch": 881.7763157894736,
+      "grad_norm": 0.9411874413490295,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 134030
+    },
+    {
+      "epoch": 881.8421052631579,
+      "grad_norm": 1.250806450843811,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 134040
+    },
+    {
+      "epoch": 881.9078947368421,
+      "grad_norm": 1.06573486328125,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 134050
+    },
+    {
+      "epoch": 881.9736842105264,
+      "grad_norm": 1.1607677936553955,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 134060
+    },
+    {
+      "epoch": 882.0394736842105,
+      "grad_norm": 1.0911805629730225,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 134070
+    },
+    {
+      "epoch": 882.1052631578947,
+      "grad_norm": 0.9352615475654602,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 134080
+    },
+    {
+      "epoch": 882.171052631579,
+      "grad_norm": 1.1933131217956543,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 134090
+    },
+    {
+      "epoch": 882.2368421052631,
+      "grad_norm": 1.0852186679840088,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 134100
+    },
+    {
+      "epoch": 882.3026315789474,
+      "grad_norm": 1.5493214130401611,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 134110
+    },
+    {
+      "epoch": 882.3684210526316,
+      "grad_norm": 1.0032248497009277,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 134120
+    },
+    {
+      "epoch": 882.4342105263158,
+      "grad_norm": 0.861934244632721,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 134130
+    },
+    {
+      "epoch": 882.5,
+      "grad_norm": 1.3053967952728271,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 134140
+    },
+    {
+      "epoch": 882.5657894736842,
+      "grad_norm": 1.0918253660202026,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 134150
+    },
+    {
+      "epoch": 882.6315789473684,
+      "grad_norm": 1.0111234188079834,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 134160
+    },
+    {
+      "epoch": 882.6973684210526,
+      "grad_norm": 1.3519738912582397,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 134170
+    },
+    {
+      "epoch": 882.7631578947369,
+      "grad_norm": 1.0138260126113892,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 134180
+    },
+    {
+      "epoch": 882.828947368421,
+      "grad_norm": 1.0171897411346436,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 134190
+    },
+    {
+      "epoch": 882.8947368421053,
+      "grad_norm": 0.7954549789428711,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 134200
+    },
+    {
+      "epoch": 882.9605263157895,
+      "grad_norm": 0.9768205285072327,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 134210
+    },
+    {
+      "epoch": 883.0263157894736,
+      "grad_norm": 1.317543625831604,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 134220
+    },
+    {
+      "epoch": 883.0921052631579,
+      "grad_norm": 1.5799720287322998,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 134230
+    },
+    {
+      "epoch": 883.1578947368421,
+      "grad_norm": 1.1066539287567139,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 134240
+    },
+    {
+      "epoch": 883.2236842105264,
+      "grad_norm": 0.8365485668182373,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 134250
+    },
+    {
+      "epoch": 883.2894736842105,
+      "grad_norm": 1.3858592510223389,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 134260
+    },
+    {
+      "epoch": 883.3552631578947,
+      "grad_norm": 0.8898154497146606,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 134270
+    },
+    {
+      "epoch": 883.421052631579,
+      "grad_norm": 0.8104990124702454,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 134280
+    },
+    {
+      "epoch": 883.4868421052631,
+      "grad_norm": 1.3037996292114258,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 134290
+    },
+    {
+      "epoch": 883.5526315789474,
+      "grad_norm": 1.1328245401382446,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 134300
+    },
+    {
+      "epoch": 883.6184210526316,
+      "grad_norm": 1.0135562419891357,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 134310
+    },
+    {
+      "epoch": 883.6842105263158,
+      "grad_norm": 0.8955759406089783,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 134320
+    },
+    {
+      "epoch": 883.75,
+      "grad_norm": 0.7221687436103821,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 134330
+    },
+    {
+      "epoch": 883.8157894736842,
+      "grad_norm": 0.8941370844841003,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 134340
+    },
+    {
+      "epoch": 883.8815789473684,
+      "grad_norm": 1.185426115989685,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 134350
+    },
+    {
+      "epoch": 883.9473684210526,
+      "grad_norm": 1.2640587091445923,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 134360
+    },
+    {
+      "epoch": 884.0131578947369,
+      "grad_norm": 1.3449172973632812,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 134370
+    },
+    {
+      "epoch": 884.078947368421,
+      "grad_norm": 1.0787019729614258,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 134380
+    },
+    {
+      "epoch": 884.1447368421053,
+      "grad_norm": 1.3704967498779297,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 134390
+    },
+    {
+      "epoch": 884.2105263157895,
+      "grad_norm": 1.1371135711669922,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 134400
+    },
+    {
+      "epoch": 884.2763157894736,
+      "grad_norm": 1.1591137647628784,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 134410
+    },
+    {
+      "epoch": 884.3421052631579,
+      "grad_norm": 1.0808539390563965,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 134420
+    },
+    {
+      "epoch": 884.4078947368421,
+      "grad_norm": 0.9412270784378052,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 134430
+    },
+    {
+      "epoch": 884.4736842105264,
+      "grad_norm": 0.8797613978385925,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 134440
+    },
+    {
+      "epoch": 884.5394736842105,
+      "grad_norm": 1.1469194889068604,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 134450
+    },
+    {
+      "epoch": 884.6052631578947,
+      "grad_norm": 1.2293258905410767,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 134460
+    },
+    {
+      "epoch": 884.671052631579,
+      "grad_norm": 0.9875232577323914,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 134470
+    },
+    {
+      "epoch": 884.7368421052631,
+      "grad_norm": 0.9115191698074341,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 134480
+    },
+    {
+      "epoch": 884.8026315789474,
+      "grad_norm": 1.4155162572860718,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 134490
+    },
+    {
+      "epoch": 884.8684210526316,
+      "grad_norm": 1.0972641706466675,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 134500
+    },
+    {
+      "epoch": 884.9342105263158,
+      "grad_norm": 1.0389680862426758,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 134510
+    },
+    {
+      "epoch": 885.0,
+      "grad_norm": 1.0352895259857178,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 134520
+    },
+    {
+      "epoch": 885.0657894736842,
+      "grad_norm": 0.6757728457450867,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 134530
+    },
+    {
+      "epoch": 885.1315789473684,
+      "grad_norm": 1.3461501598358154,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 134540
+    },
+    {
+      "epoch": 885.1973684210526,
+      "grad_norm": 0.9551567435264587,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 134550
+    },
+    {
+      "epoch": 885.2631578947369,
+      "grad_norm": 1.348388910293579,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 134560
+    },
+    {
+      "epoch": 885.328947368421,
+      "grad_norm": 1.1484230756759644,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 134570
+    },
+    {
+      "epoch": 885.3947368421053,
+      "grad_norm": 1.0391778945922852,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 134580
+    },
+    {
+      "epoch": 885.4605263157895,
+      "grad_norm": 0.9618095755577087,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 134590
+    },
+    {
+      "epoch": 885.5263157894736,
+      "grad_norm": 0.9640491604804993,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 134600
+    },
+    {
+      "epoch": 885.5921052631579,
+      "grad_norm": 1.0603336095809937,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 134610
+    },
+    {
+      "epoch": 885.6578947368421,
+      "grad_norm": 0.8449630737304688,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 134620
+    },
+    {
+      "epoch": 885.7236842105264,
+      "grad_norm": 0.8843861222267151,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 134630
+    },
+    {
+      "epoch": 885.7894736842105,
+      "grad_norm": 1.1363091468811035,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 134640
+    },
+    {
+      "epoch": 885.8552631578947,
+      "grad_norm": 1.238927960395813,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 134650
+    },
+    {
+      "epoch": 885.921052631579,
+      "grad_norm": 1.1133671998977661,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 134660
+    },
+    {
+      "epoch": 885.9868421052631,
+      "grad_norm": 1.23615300655365,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 134670
+    },
+    {
+      "epoch": 886.0526315789474,
+      "grad_norm": 1.0656203031539917,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 134680
+    },
+    {
+      "epoch": 886.1184210526316,
+      "grad_norm": 1.0231038331985474,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 134690
+    },
+    {
+      "epoch": 886.1842105263158,
+      "grad_norm": 1.6583019495010376,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 134700
+    },
+    {
+      "epoch": 886.25,
+      "grad_norm": 1.5105453729629517,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 134710
+    },
+    {
+      "epoch": 886.3157894736842,
+      "grad_norm": 1.3457199335098267,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 134720
+    },
+    {
+      "epoch": 886.3815789473684,
+      "grad_norm": 1.0339312553405762,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 134730
+    },
+    {
+      "epoch": 886.4473684210526,
+      "grad_norm": 0.9141952395439148,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 134740
+    },
+    {
+      "epoch": 886.5131578947369,
+      "grad_norm": 1.3014699220657349,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 134750
+    },
+    {
+      "epoch": 886.578947368421,
+      "grad_norm": 0.8726618885993958,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 134760
+    },
+    {
+      "epoch": 886.6447368421053,
+      "grad_norm": 1.2617624998092651,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 134770
+    },
+    {
+      "epoch": 886.7105263157895,
+      "grad_norm": 0.9956019520759583,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 134780
+    },
+    {
+      "epoch": 886.7763157894736,
+      "grad_norm": 1.3004852533340454,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 134790
+    },
+    {
+      "epoch": 886.8421052631579,
+      "grad_norm": 1.200190782546997,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 134800
+    },
+    {
+      "epoch": 886.9078947368421,
+      "grad_norm": 1.1957974433898926,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 134810
+    },
+    {
+      "epoch": 886.9736842105264,
+      "grad_norm": 1.3454686403274536,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 134820
+    },
+    {
+      "epoch": 887.0394736842105,
+      "grad_norm": 0.9715621471405029,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 134830
+    },
+    {
+      "epoch": 887.1052631578947,
+      "grad_norm": 1.2138034105300903,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 134840
+    },
+    {
+      "epoch": 887.171052631579,
+      "grad_norm": 1.0837821960449219,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 134850
+    },
+    {
+      "epoch": 887.2368421052631,
+      "grad_norm": 0.9954550266265869,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 134860
+    },
+    {
+      "epoch": 887.3026315789474,
+      "grad_norm": 1.319032907485962,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 134870
+    },
+    {
+      "epoch": 887.3684210526316,
+      "grad_norm": 1.0620183944702148,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 134880
+    },
+    {
+      "epoch": 887.4342105263158,
+      "grad_norm": 1.06571626663208,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 134890
+    },
+    {
+      "epoch": 887.5,
+      "grad_norm": 1.1749192476272583,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 134900
+    },
+    {
+      "epoch": 887.5657894736842,
+      "grad_norm": 1.475423812866211,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 134910
+    },
+    {
+      "epoch": 887.6315789473684,
+      "grad_norm": 1.3862122297286987,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 134920
+    },
+    {
+      "epoch": 887.6973684210526,
+      "grad_norm": 1.3569873571395874,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 134930
+    },
+    {
+      "epoch": 887.7631578947369,
+      "grad_norm": 1.0890201330184937,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 134940
+    },
+    {
+      "epoch": 887.828947368421,
+      "grad_norm": 1.0563850402832031,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 134950
+    },
+    {
+      "epoch": 887.8947368421053,
+      "grad_norm": 1.4732695817947388,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 134960
+    },
+    {
+      "epoch": 887.9605263157895,
+      "grad_norm": 0.7517303824424744,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 134970
+    },
+    {
+      "epoch": 888.0263157894736,
+      "grad_norm": 0.9856480956077576,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 134980
+    },
+    {
+      "epoch": 888.0921052631579,
+      "grad_norm": 0.9564664363861084,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 134990
+    },
+    {
+      "epoch": 888.1578947368421,
+      "grad_norm": 0.8272193670272827,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 135000
+    },
+    {
+      "epoch": 888.2236842105264,
+      "grad_norm": 1.3324556350708008,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 135010
+    },
+    {
+      "epoch": 888.2894736842105,
+      "grad_norm": 1.2476513385772705,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 135020
+    },
+    {
+      "epoch": 888.3552631578947,
+      "grad_norm": 1.0319244861602783,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 135030
+    },
+    {
+      "epoch": 888.421052631579,
+      "grad_norm": 0.6704224944114685,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 135040
+    },
+    {
+      "epoch": 888.4868421052631,
+      "grad_norm": 0.8636611700057983,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 135050
+    },
+    {
+      "epoch": 888.5526315789474,
+      "grad_norm": 1.0018370151519775,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 135060
+    },
+    {
+      "epoch": 888.6184210526316,
+      "grad_norm": 1.3011027574539185,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 135070
+    },
+    {
+      "epoch": 888.6842105263158,
+      "grad_norm": 0.8524222373962402,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 135080
+    },
+    {
+      "epoch": 888.75,
+      "grad_norm": 1.3557988405227661,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 135090
+    },
+    {
+      "epoch": 888.8157894736842,
+      "grad_norm": 0.79920494556427,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 135100
+    },
+    {
+      "epoch": 888.8815789473684,
+      "grad_norm": 0.8106809258460999,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 135110
+    },
+    {
+      "epoch": 888.9473684210526,
+      "grad_norm": 0.9653734564781189,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 135120
+    },
+    {
+      "epoch": 889.0131578947369,
+      "grad_norm": 1.4216874837875366,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 135130
+    },
+    {
+      "epoch": 889.078947368421,
+      "grad_norm": 1.148635983467102,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 135140
+    },
+    {
+      "epoch": 889.1447368421053,
+      "grad_norm": 1.2149169445037842,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 135150
+    },
+    {
+      "epoch": 889.2105263157895,
+      "grad_norm": 1.0104612112045288,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 135160
+    },
+    {
+      "epoch": 889.2763157894736,
+      "grad_norm": 0.984620988368988,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 135170
+    },
+    {
+      "epoch": 889.3421052631579,
+      "grad_norm": 1.2821751832962036,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 135180
+    },
+    {
+      "epoch": 889.4078947368421,
+      "grad_norm": 1.0690802335739136,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 135190
+    },
+    {
+      "epoch": 889.4736842105264,
+      "grad_norm": 1.02140474319458,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 135200
+    },
+    {
+      "epoch": 889.5394736842105,
+      "grad_norm": 1.0375328063964844,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 135210
+    },
+    {
+      "epoch": 889.6052631578947,
+      "grad_norm": 0.9176158308982849,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 135220
+    },
+    {
+      "epoch": 889.671052631579,
+      "grad_norm": 0.9485794305801392,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 135230
+    },
+    {
+      "epoch": 889.7368421052631,
+      "grad_norm": 0.9150336980819702,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 135240
+    },
+    {
+      "epoch": 889.8026315789474,
+      "grad_norm": 0.9094645380973816,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 135250
+    },
+    {
+      "epoch": 889.8684210526316,
+      "grad_norm": 0.6861184239387512,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 135260
+    },
+    {
+      "epoch": 889.9342105263158,
+      "grad_norm": 0.9334322810173035,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 135270
+    },
+    {
+      "epoch": 890.0,
+      "grad_norm": 1.377838134765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 135280
+    },
+    {
+      "epoch": 890.0657894736842,
+      "grad_norm": 1.133252739906311,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 135290
+    },
+    {
+      "epoch": 890.1315789473684,
+      "grad_norm": 1.1310118436813354,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 135300
+    },
+    {
+      "epoch": 890.1973684210526,
+      "grad_norm": 0.7168204188346863,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 135310
+    },
+    {
+      "epoch": 890.2631578947369,
+      "grad_norm": 0.8758770227432251,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 135320
+    },
+    {
+      "epoch": 890.328947368421,
+      "grad_norm": 1.1734673976898193,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 135330
+    },
+    {
+      "epoch": 890.3947368421053,
+      "grad_norm": 1.19082772731781,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 135340
+    },
+    {
+      "epoch": 890.4605263157895,
+      "grad_norm": 0.8569977283477783,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 135350
+    },
+    {
+      "epoch": 890.5263157894736,
+      "grad_norm": 1.1632040739059448,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 135360
+    },
+    {
+      "epoch": 890.5921052631579,
+      "grad_norm": 1.2661454677581787,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 135370
+    },
+    {
+      "epoch": 890.6578947368421,
+      "grad_norm": 0.9945191144943237,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 135380
+    },
+    {
+      "epoch": 890.7236842105264,
+      "grad_norm": 1.382722020149231,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 135390
+    },
+    {
+      "epoch": 890.7894736842105,
+      "grad_norm": 1.317020297050476,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 135400
+    },
+    {
+      "epoch": 890.8552631578947,
+      "grad_norm": 1.1985100507736206,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 135410
+    },
+    {
+      "epoch": 890.921052631579,
+      "grad_norm": 0.7880237698554993,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 135420
+    },
+    {
+      "epoch": 890.9868421052631,
+      "grad_norm": 1.2420897483825684,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 135430
+    },
+    {
+      "epoch": 891.0526315789474,
+      "grad_norm": 1.693311333656311,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 135440
+    },
+    {
+      "epoch": 891.1184210526316,
+      "grad_norm": 1.6864817142486572,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 135450
+    },
+    {
+      "epoch": 891.1842105263158,
+      "grad_norm": 1.1330416202545166,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 135460
+    },
+    {
+      "epoch": 891.25,
+      "grad_norm": 1.0997774600982666,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 135470
+    },
+    {
+      "epoch": 891.3157894736842,
+      "grad_norm": 0.924976646900177,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 135480
+    },
+    {
+      "epoch": 891.3815789473684,
+      "grad_norm": 1.182472825050354,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 135490
+    },
+    {
+      "epoch": 891.4473684210526,
+      "grad_norm": 1.1451120376586914,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 135500
+    },
+    {
+      "epoch": 891.5131578947369,
+      "grad_norm": 0.876240611076355,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 135510
+    },
+    {
+      "epoch": 891.578947368421,
+      "grad_norm": 1.0297558307647705,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 135520
+    },
+    {
+      "epoch": 891.6447368421053,
+      "grad_norm": 0.6283572912216187,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 135530
+    },
+    {
+      "epoch": 891.7105263157895,
+      "grad_norm": 1.2140889167785645,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 135540
+    },
+    {
+      "epoch": 891.7763157894736,
+      "grad_norm": 1.0670727491378784,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 135550
+    },
+    {
+      "epoch": 891.8421052631579,
+      "grad_norm": 1.0269482135772705,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 135560
+    },
+    {
+      "epoch": 891.9078947368421,
+      "grad_norm": 1.1649383306503296,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 135570
+    },
+    {
+      "epoch": 891.9736842105264,
+      "grad_norm": 1.085970163345337,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 135580
+    },
+    {
+      "epoch": 892.0394736842105,
+      "grad_norm": 1.1776764392852783,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 135590
+    },
+    {
+      "epoch": 892.1052631578947,
+      "grad_norm": 1.2964081764221191,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 135600
+    },
+    {
+      "epoch": 892.171052631579,
+      "grad_norm": 1.1058318614959717,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 135610
+    },
+    {
+      "epoch": 892.2368421052631,
+      "grad_norm": 0.9314119815826416,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 135620
+    },
+    {
+      "epoch": 892.3026315789474,
+      "grad_norm": 1.4460657835006714,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 135630
+    },
+    {
+      "epoch": 892.3684210526316,
+      "grad_norm": 0.9084354043006897,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 135640
+    },
+    {
+      "epoch": 892.4342105263158,
+      "grad_norm": 1.1502026319503784,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 135650
+    },
+    {
+      "epoch": 892.5,
+      "grad_norm": 1.2194117307662964,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 135660
+    },
+    {
+      "epoch": 892.5657894736842,
+      "grad_norm": 0.9721207618713379,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 135670
+    },
+    {
+      "epoch": 892.6315789473684,
+      "grad_norm": 0.7545431852340698,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 135680
+    },
+    {
+      "epoch": 892.6973684210526,
+      "grad_norm": 0.8242561221122742,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 135690
+    },
+    {
+      "epoch": 892.7631578947369,
+      "grad_norm": 0.8406139612197876,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 135700
+    },
+    {
+      "epoch": 892.828947368421,
+      "grad_norm": 1.0956982374191284,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 135710
+    },
+    {
+      "epoch": 892.8947368421053,
+      "grad_norm": 0.8513756990432739,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 135720
+    },
+    {
+      "epoch": 892.9605263157895,
+      "grad_norm": 1.1040880680084229,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 135730
+    },
+    {
+      "epoch": 893.0263157894736,
+      "grad_norm": 1.076196551322937,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 135740
+    },
+    {
+      "epoch": 893.0921052631579,
+      "grad_norm": 1.1001917123794556,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 135750
+    },
+    {
+      "epoch": 893.1578947368421,
+      "grad_norm": 1.5378665924072266,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 135760
+    },
+    {
+      "epoch": 893.2236842105264,
+      "grad_norm": 1.166962742805481,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 135770
+    },
+    {
+      "epoch": 893.2894736842105,
+      "grad_norm": 0.90530925989151,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 135780
+    },
+    {
+      "epoch": 893.3552631578947,
+      "grad_norm": 1.342469334602356,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 135790
+    },
+    {
+      "epoch": 893.421052631579,
+      "grad_norm": 1.1305315494537354,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 135800
+    },
+    {
+      "epoch": 893.4868421052631,
+      "grad_norm": 0.8744671940803528,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 135810
+    },
+    {
+      "epoch": 893.5526315789474,
+      "grad_norm": 1.1060692071914673,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 135820
+    },
+    {
+      "epoch": 893.6184210526316,
+      "grad_norm": 1.1510794162750244,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 135830
+    },
+    {
+      "epoch": 893.6842105263158,
+      "grad_norm": 1.0159302949905396,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 135840
+    },
+    {
+      "epoch": 893.75,
+      "grad_norm": 1.0674599409103394,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 135850
+    },
+    {
+      "epoch": 893.8157894736842,
+      "grad_norm": 0.7770407199859619,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 135860
+    },
+    {
+      "epoch": 893.8815789473684,
+      "grad_norm": 1.1732828617095947,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 135870
+    },
+    {
+      "epoch": 893.9473684210526,
+      "grad_norm": 0.732951283454895,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 135880
+    },
+    {
+      "epoch": 894.0131578947369,
+      "grad_norm": 0.8434010148048401,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 135890
+    },
+    {
+      "epoch": 894.078947368421,
+      "grad_norm": 0.9561564922332764,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 135900
+    },
+    {
+      "epoch": 894.1447368421053,
+      "grad_norm": 1.377158761024475,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 135910
+    },
+    {
+      "epoch": 894.2105263157895,
+      "grad_norm": 1.1105989217758179,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 135920
+    },
+    {
+      "epoch": 894.2763157894736,
+      "grad_norm": 1.2619684934616089,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 135930
+    },
+    {
+      "epoch": 894.3421052631579,
+      "grad_norm": 1.026212453842163,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 135940
+    },
+    {
+      "epoch": 894.4078947368421,
+      "grad_norm": 0.7990017533302307,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 135950
+    },
+    {
+      "epoch": 894.4736842105264,
+      "grad_norm": 0.9422027468681335,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 135960
+    },
+    {
+      "epoch": 894.5394736842105,
+      "grad_norm": 0.796430230140686,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 135970
+    },
+    {
+      "epoch": 894.6052631578947,
+      "grad_norm": 1.2357088327407837,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 135980
+    },
+    {
+      "epoch": 894.671052631579,
+      "grad_norm": 1.2409306764602661,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 135990
+    },
+    {
+      "epoch": 894.7368421052631,
+      "grad_norm": 1.2816016674041748,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 136000
+    },
+    {
+      "epoch": 894.8026315789474,
+      "grad_norm": 0.6768025755882263,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 136010
+    },
+    {
+      "epoch": 894.8684210526316,
+      "grad_norm": 0.9706000685691833,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 136020
+    },
+    {
+      "epoch": 894.9342105263158,
+      "grad_norm": 0.9608473777770996,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 136030
+    },
+    {
+      "epoch": 895.0,
+      "grad_norm": 1.143330693244934,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 136040
+    },
+    {
+      "epoch": 895.0657894736842,
+      "grad_norm": 0.8923169374465942,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 136050
+    },
+    {
+      "epoch": 895.1315789473684,
+      "grad_norm": 0.7669290900230408,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 136060
+    },
+    {
+      "epoch": 895.1973684210526,
+      "grad_norm": 0.9958640933036804,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 136070
+    },
+    {
+      "epoch": 895.2631578947369,
+      "grad_norm": 1.1290862560272217,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 136080
+    },
+    {
+      "epoch": 895.328947368421,
+      "grad_norm": 0.9855304956436157,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 136090
+    },
+    {
+      "epoch": 895.3947368421053,
+      "grad_norm": 0.9755940437316895,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 136100
+    },
+    {
+      "epoch": 895.4605263157895,
+      "grad_norm": 0.8567885160446167,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 136110
+    },
+    {
+      "epoch": 895.5263157894736,
+      "grad_norm": 1.0460125207901,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 136120
+    },
+    {
+      "epoch": 895.5921052631579,
+      "grad_norm": 1.2655810117721558,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 136130
+    },
+    {
+      "epoch": 895.6578947368421,
+      "grad_norm": 0.951598048210144,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 136140
+    },
+    {
+      "epoch": 895.7236842105264,
+      "grad_norm": 1.1518741846084595,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 136150
+    },
+    {
+      "epoch": 895.7894736842105,
+      "grad_norm": 0.9149439334869385,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 136160
+    },
+    {
+      "epoch": 895.8552631578947,
+      "grad_norm": 1.0804107189178467,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 136170
+    },
+    {
+      "epoch": 895.921052631579,
+      "grad_norm": 1.2547401189804077,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 136180
+    },
+    {
+      "epoch": 895.9868421052631,
+      "grad_norm": 1.0589017868041992,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 136190
+    },
+    {
+      "epoch": 896.0526315789474,
+      "grad_norm": 1.040658950805664,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 136200
+    },
+    {
+      "epoch": 896.1184210526316,
+      "grad_norm": 1.1037824153900146,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 136210
+    },
+    {
+      "epoch": 896.1842105263158,
+      "grad_norm": 1.4946725368499756,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 136220
+    },
+    {
+      "epoch": 896.25,
+      "grad_norm": 1.519361972808838,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 136230
+    },
+    {
+      "epoch": 896.3157894736842,
+      "grad_norm": 1.1634271144866943,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 136240
+    },
+    {
+      "epoch": 896.3815789473684,
+      "grad_norm": 1.3984472751617432,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 136250
+    },
+    {
+      "epoch": 896.4473684210526,
+      "grad_norm": 1.7315332889556885,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 136260
+    },
+    {
+      "epoch": 896.5131578947369,
+      "grad_norm": 1.5922656059265137,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 136270
+    },
+    {
+      "epoch": 896.578947368421,
+      "grad_norm": 1.4069749116897583,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 136280
+    },
+    {
+      "epoch": 896.6447368421053,
+      "grad_norm": 1.244490146636963,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 136290
+    },
+    {
+      "epoch": 896.7105263157895,
+      "grad_norm": 1.1110283136367798,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 136300
+    },
+    {
+      "epoch": 896.7763157894736,
+      "grad_norm": 1.0167648792266846,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 136310
+    },
+    {
+      "epoch": 896.8421052631579,
+      "grad_norm": 1.137952208518982,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 136320
+    },
+    {
+      "epoch": 896.9078947368421,
+      "grad_norm": 1.0673314332962036,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 136330
+    },
+    {
+      "epoch": 896.9736842105264,
+      "grad_norm": 1.012425184249878,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 136340
+    },
+    {
+      "epoch": 897.0394736842105,
+      "grad_norm": 0.7594325542449951,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 136350
+    },
+    {
+      "epoch": 897.1052631578947,
+      "grad_norm": 1.1178792715072632,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 136360
+    },
+    {
+      "epoch": 897.171052631579,
+      "grad_norm": 1.0943617820739746,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 136370
+    },
+    {
+      "epoch": 897.2368421052631,
+      "grad_norm": 0.9920550584793091,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 136380
+    },
+    {
+      "epoch": 897.3026315789474,
+      "grad_norm": 1.0962131023406982,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 136390
+    },
+    {
+      "epoch": 897.3684210526316,
+      "grad_norm": 0.8344330787658691,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 136400
+    },
+    {
+      "epoch": 897.4342105263158,
+      "grad_norm": 1.1164144277572632,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 136410
+    },
+    {
+      "epoch": 897.5,
+      "grad_norm": 1.07270348072052,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 136420
+    },
+    {
+      "epoch": 897.5657894736842,
+      "grad_norm": 0.843373715877533,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 136430
+    },
+    {
+      "epoch": 897.6315789473684,
+      "grad_norm": 0.8556021451950073,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 136440
+    },
+    {
+      "epoch": 897.6973684210526,
+      "grad_norm": 0.8964880704879761,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 136450
+    },
+    {
+      "epoch": 897.7631578947369,
+      "grad_norm": 1.153839111328125,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 136460
+    },
+    {
+      "epoch": 897.828947368421,
+      "grad_norm": 1.0658332109451294,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 136470
+    },
+    {
+      "epoch": 897.8947368421053,
+      "grad_norm": 1.2250584363937378,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 136480
+    },
+    {
+      "epoch": 897.9605263157895,
+      "grad_norm": 1.1773885488510132,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 136490
+    },
+    {
+      "epoch": 898.0263157894736,
+      "grad_norm": 1.2498626708984375,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 136500
+    },
+    {
+      "epoch": 898.0921052631579,
+      "grad_norm": 0.980217695236206,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 136510
+    },
+    {
+      "epoch": 898.1578947368421,
+      "grad_norm": 0.8119854927062988,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 136520
+    },
+    {
+      "epoch": 898.2236842105264,
+      "grad_norm": 0.8097248077392578,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 136530
+    },
+    {
+      "epoch": 898.2894736842105,
+      "grad_norm": 1.080000638961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 136540
+    },
+    {
+      "epoch": 898.3552631578947,
+      "grad_norm": 1.0661876201629639,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 136550
+    },
+    {
+      "epoch": 898.421052631579,
+      "grad_norm": 1.1550641059875488,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 136560
+    },
+    {
+      "epoch": 898.4868421052631,
+      "grad_norm": 1.1951998472213745,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 136570
+    },
+    {
+      "epoch": 898.5526315789474,
+      "grad_norm": 0.6485775709152222,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 136580
+    },
+    {
+      "epoch": 898.6184210526316,
+      "grad_norm": 0.8973692059516907,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 136590
+    },
+    {
+      "epoch": 898.6842105263158,
+      "grad_norm": 0.7221608757972717,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 136600
+    },
+    {
+      "epoch": 898.75,
+      "grad_norm": 0.7160447835922241,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 136610
+    },
+    {
+      "epoch": 898.8157894736842,
+      "grad_norm": 0.9838132262229919,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 136620
+    },
+    {
+      "epoch": 898.8815789473684,
+      "grad_norm": 0.8849008679389954,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 136630
+    },
+    {
+      "epoch": 898.9473684210526,
+      "grad_norm": 1.1569770574569702,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 136640
+    },
+    {
+      "epoch": 899.0131578947369,
+      "grad_norm": 0.594029426574707,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 136650
+    },
+    {
+      "epoch": 899.078947368421,
+      "grad_norm": 0.8497085571289062,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 136660
+    },
+    {
+      "epoch": 899.1447368421053,
+      "grad_norm": 0.8578495383262634,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 136670
+    },
+    {
+      "epoch": 899.2105263157895,
+      "grad_norm": 1.1420329809188843,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 136680
+    },
+    {
+      "epoch": 899.2763157894736,
+      "grad_norm": 1.0156878232955933,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 136690
+    },
+    {
+      "epoch": 899.3421052631579,
+      "grad_norm": 0.8283445239067078,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 136700
+    },
+    {
+      "epoch": 899.4078947368421,
+      "grad_norm": 0.8237717747688293,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 136710
+    },
+    {
+      "epoch": 899.4736842105264,
+      "grad_norm": 0.9158399105072021,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 136720
+    },
+    {
+      "epoch": 899.5394736842105,
+      "grad_norm": 0.850222110748291,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 136730
+    },
+    {
+      "epoch": 899.6052631578947,
+      "grad_norm": 1.1023249626159668,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 136740
+    },
+    {
+      "epoch": 899.671052631579,
+      "grad_norm": 0.9188584685325623,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 136750
+    },
+    {
+      "epoch": 899.7368421052631,
+      "grad_norm": 1.0491944551467896,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 136760
+    },
+    {
+      "epoch": 899.8026315789474,
+      "grad_norm": 0.7473717927932739,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 136770
+    },
+    {
+      "epoch": 899.8684210526316,
+      "grad_norm": 0.8892393112182617,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 136780
+    },
+    {
+      "epoch": 899.9342105263158,
+      "grad_norm": 1.0245060920715332,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 136790
+    },
+    {
+      "epoch": 900.0,
+      "grad_norm": 1.3008759021759033,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 136800
+    },
+    {
+      "epoch": 900.0657894736842,
+      "grad_norm": 1.4540356397628784,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 136810
+    },
+    {
+      "epoch": 900.1315789473684,
+      "grad_norm": 1.342063069343567,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 136820
+    },
+    {
+      "epoch": 900.1973684210526,
+      "grad_norm": 1.1454083919525146,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 136830
+    },
+    {
+      "epoch": 900.2631578947369,
+      "grad_norm": 1.6022448539733887,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 136840
+    },
+    {
+      "epoch": 900.328947368421,
+      "grad_norm": 0.9221236109733582,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 136850
+    },
+    {
+      "epoch": 900.3947368421053,
+      "grad_norm": 1.3476216793060303,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 136860
+    },
+    {
+      "epoch": 900.4605263157895,
+      "grad_norm": 0.8298095464706421,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 136870
+    },
+    {
+      "epoch": 900.5263157894736,
+      "grad_norm": 0.7791982293128967,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 136880
+    },
+    {
+      "epoch": 900.5921052631579,
+      "grad_norm": 1.068069577217102,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 136890
+    },
+    {
+      "epoch": 900.6578947368421,
+      "grad_norm": 0.7247030138969421,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 136900
+    },
+    {
+      "epoch": 900.7236842105264,
+      "grad_norm": 0.8441376686096191,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 136910
+    },
+    {
+      "epoch": 900.7894736842105,
+      "grad_norm": 1.0622183084487915,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 136920
+    },
+    {
+      "epoch": 900.8552631578947,
+      "grad_norm": 0.9633458256721497,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 136930
+    },
+    {
+      "epoch": 900.921052631579,
+      "grad_norm": 1.0135974884033203,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 136940
+    },
+    {
+      "epoch": 900.9868421052631,
+      "grad_norm": 0.9505342841148376,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 136950
+    },
+    {
+      "epoch": 901.0526315789474,
+      "grad_norm": 0.7464510202407837,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 136960
+    },
+    {
+      "epoch": 901.1184210526316,
+      "grad_norm": 1.1256731748580933,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 136970
+    },
+    {
+      "epoch": 901.1842105263158,
+      "grad_norm": 1.1023762226104736,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 136980
+    },
+    {
+      "epoch": 901.25,
+      "grad_norm": 1.3419615030288696,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 136990
+    },
+    {
+      "epoch": 901.3157894736842,
+      "grad_norm": 1.1724894046783447,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 137000
+    },
+    {
+      "epoch": 901.3815789473684,
+      "grad_norm": 1.1471381187438965,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 137010
+    },
+    {
+      "epoch": 901.4473684210526,
+      "grad_norm": 1.5692983865737915,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 137020
+    },
+    {
+      "epoch": 901.5131578947369,
+      "grad_norm": 1.1495898962020874,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 137030
+    },
+    {
+      "epoch": 901.578947368421,
+      "grad_norm": 0.7561367154121399,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 137040
+    },
+    {
+      "epoch": 901.6447368421053,
+      "grad_norm": 0.8060752749443054,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 137050
+    },
+    {
+      "epoch": 901.7105263157895,
+      "grad_norm": 1.0309655666351318,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 137060
+    },
+    {
+      "epoch": 901.7763157894736,
+      "grad_norm": 1.3897480964660645,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 137070
+    },
+    {
+      "epoch": 901.8421052631579,
+      "grad_norm": 1.1561673879623413,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 137080
+    },
+    {
+      "epoch": 901.9078947368421,
+      "grad_norm": 1.1159383058547974,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 137090
+    },
+    {
+      "epoch": 901.9736842105264,
+      "grad_norm": 1.2140202522277832,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 137100
+    },
+    {
+      "epoch": 902.0394736842105,
+      "grad_norm": 1.0120538473129272,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 137110
+    },
+    {
+      "epoch": 902.1052631578947,
+      "grad_norm": 1.3522586822509766,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 137120
+    },
+    {
+      "epoch": 902.171052631579,
+      "grad_norm": 1.1707468032836914,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 137130
+    },
+    {
+      "epoch": 902.2368421052631,
+      "grad_norm": 1.209934949874878,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 137140
+    },
+    {
+      "epoch": 902.3026315789474,
+      "grad_norm": 0.8782662153244019,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 137150
+    },
+    {
+      "epoch": 902.3684210526316,
+      "grad_norm": 0.9009442329406738,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 137160
+    },
+    {
+      "epoch": 902.4342105263158,
+      "grad_norm": 1.024381399154663,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 137170
+    },
+    {
+      "epoch": 902.5,
+      "grad_norm": 1.0990251302719116,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 137180
+    },
+    {
+      "epoch": 902.5657894736842,
+      "grad_norm": 1.1666136980056763,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 137190
+    },
+    {
+      "epoch": 902.6315789473684,
+      "grad_norm": 1.1644891500473022,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 137200
+    },
+    {
+      "epoch": 902.6973684210526,
+      "grad_norm": 1.3089399337768555,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 137210
+    },
+    {
+      "epoch": 902.7631578947369,
+      "grad_norm": 1.2203880548477173,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 137220
+    },
+    {
+      "epoch": 902.828947368421,
+      "grad_norm": 1.0612380504608154,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 137230
+    },
+    {
+      "epoch": 902.8947368421053,
+      "grad_norm": 1.1838760375976562,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 137240
+    },
+    {
+      "epoch": 902.9605263157895,
+      "grad_norm": 1.0632514953613281,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 137250
+    },
+    {
+      "epoch": 903.0263157894736,
+      "grad_norm": 0.9610679745674133,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 137260
+    },
+    {
+      "epoch": 903.0921052631579,
+      "grad_norm": 0.670998215675354,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 137270
+    },
+    {
+      "epoch": 903.1578947368421,
+      "grad_norm": 0.9140186905860901,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 137280
+    },
+    {
+      "epoch": 903.2236842105264,
+      "grad_norm": 1.182522177696228,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 137290
+    },
+    {
+      "epoch": 903.2894736842105,
+      "grad_norm": 0.9858769774436951,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 137300
+    },
+    {
+      "epoch": 903.3552631578947,
+      "grad_norm": 1.1477785110473633,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 137310
+    },
+    {
+      "epoch": 903.421052631579,
+      "grad_norm": 1.2154635190963745,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 137320
+    },
+    {
+      "epoch": 903.4868421052631,
+      "grad_norm": 0.9853901863098145,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 137330
+    },
+    {
+      "epoch": 903.5526315789474,
+      "grad_norm": 1.2945221662521362,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 137340
+    },
+    {
+      "epoch": 903.6184210526316,
+      "grad_norm": 1.0276622772216797,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 137350
+    },
+    {
+      "epoch": 903.6842105263158,
+      "grad_norm": 0.8371639251708984,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 137360
+    },
+    {
+      "epoch": 903.75,
+      "grad_norm": 1.3738861083984375,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 137370
+    },
+    {
+      "epoch": 903.8157894736842,
+      "grad_norm": 0.9806889295578003,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 137380
+    },
+    {
+      "epoch": 903.8815789473684,
+      "grad_norm": 1.3452116250991821,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 137390
+    },
+    {
+      "epoch": 903.9473684210526,
+      "grad_norm": 1.2465362548828125,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 137400
+    },
+    {
+      "epoch": 904.0131578947369,
+      "grad_norm": 0.8776252269744873,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 137410
+    },
+    {
+      "epoch": 904.078947368421,
+      "grad_norm": 0.8679521083831787,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 137420
+    },
+    {
+      "epoch": 904.1447368421053,
+      "grad_norm": 1.0201027393341064,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 137430
+    },
+    {
+      "epoch": 904.2105263157895,
+      "grad_norm": 1.2104623317718506,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 137440
+    },
+    {
+      "epoch": 904.2763157894736,
+      "grad_norm": 0.9057734608650208,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 137450
+    },
+    {
+      "epoch": 904.3421052631579,
+      "grad_norm": 1.0167741775512695,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 137460
+    },
+    {
+      "epoch": 904.4078947368421,
+      "grad_norm": 0.5943556427955627,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 137470
+    },
+    {
+      "epoch": 904.4736842105264,
+      "grad_norm": 0.8920695781707764,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 137480
+    },
+    {
+      "epoch": 904.5394736842105,
+      "grad_norm": 0.9045788645744324,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 137490
+    },
+    {
+      "epoch": 904.6052631578947,
+      "grad_norm": 0.9521625638008118,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 137500
+    },
+    {
+      "epoch": 904.671052631579,
+      "grad_norm": 0.8317375779151917,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 137510
+    },
+    {
+      "epoch": 904.7368421052631,
+      "grad_norm": 0.6917269229888916,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 137520
+    },
+    {
+      "epoch": 904.8026315789474,
+      "grad_norm": 1.1361898183822632,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 137530
+    },
+    {
+      "epoch": 904.8684210526316,
+      "grad_norm": 1.2201398611068726,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 137540
+    },
+    {
+      "epoch": 904.9342105263158,
+      "grad_norm": 0.8375370502471924,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 137550
+    },
+    {
+      "epoch": 905.0,
+      "grad_norm": 1.1140497922897339,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 137560
+    },
+    {
+      "epoch": 905.0657894736842,
+      "grad_norm": 1.0850721597671509,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 137570
+    },
+    {
+      "epoch": 905.1315789473684,
+      "grad_norm": 0.9781196117401123,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 137580
+    },
+    {
+      "epoch": 905.1973684210526,
+      "grad_norm": 1.1050325632095337,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 137590
+    },
+    {
+      "epoch": 905.2631578947369,
+      "grad_norm": 0.8729236721992493,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 137600
+    },
+    {
+      "epoch": 905.328947368421,
+      "grad_norm": 1.088930368423462,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 137610
+    },
+    {
+      "epoch": 905.3947368421053,
+      "grad_norm": 1.292374849319458,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 137620
+    },
+    {
+      "epoch": 905.4605263157895,
+      "grad_norm": 0.9945433735847473,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 137630
+    },
+    {
+      "epoch": 905.5263157894736,
+      "grad_norm": 1.1098923683166504,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 137640
+    },
+    {
+      "epoch": 905.5921052631579,
+      "grad_norm": 1.15111243724823,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 137650
+    },
+    {
+      "epoch": 905.6578947368421,
+      "grad_norm": 0.8712033629417419,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 137660
+    },
+    {
+      "epoch": 905.7236842105264,
+      "grad_norm": 1.0952752828598022,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 137670
+    },
+    {
+      "epoch": 905.7894736842105,
+      "grad_norm": 1.2315001487731934,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 137680
+    },
+    {
+      "epoch": 905.8552631578947,
+      "grad_norm": 1.5495014190673828,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 137690
+    },
+    {
+      "epoch": 905.921052631579,
+      "grad_norm": 1.7285585403442383,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 137700
+    },
+    {
+      "epoch": 905.9868421052631,
+      "grad_norm": 1.7616963386535645,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 137710
+    },
+    {
+      "epoch": 906.0526315789474,
+      "grad_norm": 1.4196006059646606,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 137720
+    },
+    {
+      "epoch": 906.1184210526316,
+      "grad_norm": 1.3745293617248535,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 137730
+    },
+    {
+      "epoch": 906.1842105263158,
+      "grad_norm": 0.8986107707023621,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 137740
+    },
+    {
+      "epoch": 906.25,
+      "grad_norm": 0.9261276721954346,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 137750
+    },
+    {
+      "epoch": 906.3157894736842,
+      "grad_norm": 1.2320727109909058,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 137760
+    },
+    {
+      "epoch": 906.3815789473684,
+      "grad_norm": 1.3749173879623413,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 137770
+    },
+    {
+      "epoch": 906.4473684210526,
+      "grad_norm": 1.2523115873336792,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 137780
+    },
+    {
+      "epoch": 906.5131578947369,
+      "grad_norm": 0.9561618566513062,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 137790
+    },
+    {
+      "epoch": 906.578947368421,
+      "grad_norm": 1.0065858364105225,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 137800
+    },
+    {
+      "epoch": 906.6447368421053,
+      "grad_norm": 1.0665801763534546,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 137810
+    },
+    {
+      "epoch": 906.7105263157895,
+      "grad_norm": 1.0746973752975464,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 137820
+    },
+    {
+      "epoch": 906.7763157894736,
+      "grad_norm": 0.6593524217605591,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 137830
+    },
+    {
+      "epoch": 906.8421052631579,
+      "grad_norm": 0.9232041239738464,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 137840
+    },
+    {
+      "epoch": 906.9078947368421,
+      "grad_norm": 0.9710672497749329,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 137850
+    },
+    {
+      "epoch": 906.9736842105264,
+      "grad_norm": 1.034181833267212,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 137860
+    },
+    {
+      "epoch": 907.0394736842105,
+      "grad_norm": 0.5571140646934509,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 137870
+    },
+    {
+      "epoch": 907.1052631578947,
+      "grad_norm": 1.2709097862243652,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 137880
+    },
+    {
+      "epoch": 907.171052631579,
+      "grad_norm": 1.4115272760391235,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 137890
+    },
+    {
+      "epoch": 907.2368421052631,
+      "grad_norm": 1.275320291519165,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 137900
+    },
+    {
+      "epoch": 907.3026315789474,
+      "grad_norm": 0.980765163898468,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 137910
+    },
+    {
+      "epoch": 907.3684210526316,
+      "grad_norm": 0.9128990173339844,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 137920
+    },
+    {
+      "epoch": 907.4342105263158,
+      "grad_norm": 0.8334782719612122,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 137930
+    },
+    {
+      "epoch": 907.5,
+      "grad_norm": 0.7317301630973816,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 137940
+    },
+    {
+      "epoch": 907.5657894736842,
+      "grad_norm": 1.2653545141220093,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 137950
+    },
+    {
+      "epoch": 907.6315789473684,
+      "grad_norm": 1.2031755447387695,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 137960
+    },
+    {
+      "epoch": 907.6973684210526,
+      "grad_norm": 0.6481014490127563,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 137970
+    },
+    {
+      "epoch": 907.7631578947369,
+      "grad_norm": 1.1213470697402954,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 137980
+    },
+    {
+      "epoch": 907.828947368421,
+      "grad_norm": 1.4939583539962769,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 137990
+    },
+    {
+      "epoch": 907.8947368421053,
+      "grad_norm": 1.7831441164016724,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 138000
+    },
+    {
+      "epoch": 907.9605263157895,
+      "grad_norm": 1.7211501598358154,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 138010
+    },
+    {
+      "epoch": 908.0263157894736,
+      "grad_norm": 1.5099759101867676,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 138020
+    },
+    {
+      "epoch": 908.0921052631579,
+      "grad_norm": 1.137043833732605,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 138030
+    },
+    {
+      "epoch": 908.1578947368421,
+      "grad_norm": 1.2676162719726562,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 138040
+    },
+    {
+      "epoch": 908.2236842105264,
+      "grad_norm": 0.994552731513977,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 138050
+    },
+    {
+      "epoch": 908.2894736842105,
+      "grad_norm": 1.0138901472091675,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 138060
+    },
+    {
+      "epoch": 908.3552631578947,
+      "grad_norm": 0.9943720698356628,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 138070
+    },
+    {
+      "epoch": 908.421052631579,
+      "grad_norm": 1.1211268901824951,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 138080
+    },
+    {
+      "epoch": 908.4868421052631,
+      "grad_norm": 1.2332216501235962,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 138090
+    },
+    {
+      "epoch": 908.5526315789474,
+      "grad_norm": 1.1603976488113403,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 138100
+    },
+    {
+      "epoch": 908.6184210526316,
+      "grad_norm": 0.789871871471405,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 138110
+    },
+    {
+      "epoch": 908.6842105263158,
+      "grad_norm": 1.0970319509506226,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 138120
+    },
+    {
+      "epoch": 908.75,
+      "grad_norm": 1.0479586124420166,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 138130
+    },
+    {
+      "epoch": 908.8157894736842,
+      "grad_norm": 0.932201623916626,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 138140
+    },
+    {
+      "epoch": 908.8815789473684,
+      "grad_norm": 0.9041379690170288,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 138150
+    },
+    {
+      "epoch": 908.9473684210526,
+      "grad_norm": 0.9784705638885498,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 138160
+    },
+    {
+      "epoch": 909.0131578947369,
+      "grad_norm": 1.0539556741714478,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 138170
+    },
+    {
+      "epoch": 909.078947368421,
+      "grad_norm": 1.054840087890625,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 138180
+    },
+    {
+      "epoch": 909.1447368421053,
+      "grad_norm": 0.817751944065094,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 138190
+    },
+    {
+      "epoch": 909.2105263157895,
+      "grad_norm": 1.2767053842544556,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 138200
+    },
+    {
+      "epoch": 909.2763157894736,
+      "grad_norm": 0.9269699454307556,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 138210
+    },
+    {
+      "epoch": 909.3421052631579,
+      "grad_norm": 0.7192962765693665,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 138220
+    },
+    {
+      "epoch": 909.4078947368421,
+      "grad_norm": 1.0182446241378784,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 138230
+    },
+    {
+      "epoch": 909.4736842105264,
+      "grad_norm": 0.8939855694770813,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 138240
+    },
+    {
+      "epoch": 909.5394736842105,
+      "grad_norm": 1.3759862184524536,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 138250
+    },
+    {
+      "epoch": 909.6052631578947,
+      "grad_norm": 1.0938113927841187,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 138260
+    },
+    {
+      "epoch": 909.671052631579,
+      "grad_norm": 1.2169560194015503,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 138270
+    },
+    {
+      "epoch": 909.7368421052631,
+      "grad_norm": 0.8270349502563477,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 138280
+    },
+    {
+      "epoch": 909.8026315789474,
+      "grad_norm": 0.9773039221763611,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 138290
+    },
+    {
+      "epoch": 909.8684210526316,
+      "grad_norm": 0.9139556884765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 138300
+    },
+    {
+      "epoch": 909.9342105263158,
+      "grad_norm": 1.0602258443832397,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 138310
+    },
+    {
+      "epoch": 910.0,
+      "grad_norm": 1.1161372661590576,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 138320
+    },
+    {
+      "epoch": 910.0657894736842,
+      "grad_norm": 1.1189671754837036,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 138330
+    },
+    {
+      "epoch": 910.1315789473684,
+      "grad_norm": 0.8698909282684326,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 138340
+    },
+    {
+      "epoch": 910.1973684210526,
+      "grad_norm": 0.9428373575210571,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 138350
+    },
+    {
+      "epoch": 910.2631578947369,
+      "grad_norm": 0.6504638195037842,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 138360
+    },
+    {
+      "epoch": 910.328947368421,
+      "grad_norm": 0.8180758357048035,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 138370
+    },
+    {
+      "epoch": 910.3947368421053,
+      "grad_norm": 1.1417499780654907,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 138380
+    },
+    {
+      "epoch": 910.4605263157895,
+      "grad_norm": 0.8425560593605042,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 138390
+    },
+    {
+      "epoch": 910.5263157894736,
+      "grad_norm": 1.2811640501022339,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 138400
+    },
+    {
+      "epoch": 910.5921052631579,
+      "grad_norm": 1.0059995651245117,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 138410
+    },
+    {
+      "epoch": 910.6578947368421,
+      "grad_norm": 1.4118963479995728,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 138420
+    },
+    {
+      "epoch": 910.7236842105264,
+      "grad_norm": 1.055229663848877,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 138430
+    },
+    {
+      "epoch": 910.7894736842105,
+      "grad_norm": 1.1390751600265503,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 138440
+    },
+    {
+      "epoch": 910.8552631578947,
+      "grad_norm": 1.3545476198196411,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 138450
+    },
+    {
+      "epoch": 910.921052631579,
+      "grad_norm": 1.0266139507293701,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 138460
+    },
+    {
+      "epoch": 910.9868421052631,
+      "grad_norm": 1.0077507495880127,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 138470
+    },
+    {
+      "epoch": 911.0526315789474,
+      "grad_norm": 1.1615369319915771,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 138480
+    },
+    {
+      "epoch": 911.1184210526316,
+      "grad_norm": 1.2945265769958496,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 138490
+    },
+    {
+      "epoch": 911.1842105263158,
+      "grad_norm": 0.7563729286193848,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 138500
+    },
+    {
+      "epoch": 911.25,
+      "grad_norm": 0.941920280456543,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 138510
+    },
+    {
+      "epoch": 911.3157894736842,
+      "grad_norm": 0.8620559573173523,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 138520
+    },
+    {
+      "epoch": 911.3815789473684,
+      "grad_norm": 1.0454643964767456,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 138530
+    },
+    {
+      "epoch": 911.4473684210526,
+      "grad_norm": 1.2060420513153076,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 138540
+    },
+    {
+      "epoch": 911.5131578947369,
+      "grad_norm": 1.0942254066467285,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 138550
+    },
+    {
+      "epoch": 911.578947368421,
+      "grad_norm": 0.7879404425621033,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 138560
+    },
+    {
+      "epoch": 911.6447368421053,
+      "grad_norm": 1.4239997863769531,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 138570
+    },
+    {
+      "epoch": 911.7105263157895,
+      "grad_norm": 1.384261965751648,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 138580
+    },
+    {
+      "epoch": 911.7763157894736,
+      "grad_norm": 1.2264975309371948,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 138590
+    },
+    {
+      "epoch": 911.8421052631579,
+      "grad_norm": 1.1380722522735596,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 138600
+    },
+    {
+      "epoch": 911.9078947368421,
+      "grad_norm": 1.4093822240829468,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 138610
+    },
+    {
+      "epoch": 911.9736842105264,
+      "grad_norm": 0.9658661484718323,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 138620
+    },
+    {
+      "epoch": 912.0394736842105,
+      "grad_norm": 1.6494758129119873,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 138630
+    },
+    {
+      "epoch": 912.1052631578947,
+      "grad_norm": 1.223433494567871,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 138640
+    },
+    {
+      "epoch": 912.171052631579,
+      "grad_norm": 1.1293435096740723,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 138650
+    },
+    {
+      "epoch": 912.2368421052631,
+      "grad_norm": 1.3169997930526733,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 138660
+    },
+    {
+      "epoch": 912.3026315789474,
+      "grad_norm": 1.1258878707885742,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 138670
+    },
+    {
+      "epoch": 912.3684210526316,
+      "grad_norm": 1.0870180130004883,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 138680
+    },
+    {
+      "epoch": 912.4342105263158,
+      "grad_norm": 1.2303227186203003,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 138690
+    },
+    {
+      "epoch": 912.5,
+      "grad_norm": 1.3241380453109741,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 138700
+    },
+    {
+      "epoch": 912.5657894736842,
+      "grad_norm": 0.9058986306190491,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 138710
+    },
+    {
+      "epoch": 912.6315789473684,
+      "grad_norm": 1.2740260362625122,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 138720
+    },
+    {
+      "epoch": 912.6973684210526,
+      "grad_norm": 1.073053240776062,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 138730
+    },
+    {
+      "epoch": 912.7631578947369,
+      "grad_norm": 0.9886819124221802,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 138740
+    },
+    {
+      "epoch": 912.828947368421,
+      "grad_norm": 1.464430332183838,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 138750
+    },
+    {
+      "epoch": 912.8947368421053,
+      "grad_norm": 1.4598714113235474,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 138760
+    },
+    {
+      "epoch": 912.9605263157895,
+      "grad_norm": 1.233940839767456,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 138770
+    },
+    {
+      "epoch": 913.0263157894736,
+      "grad_norm": 0.9200678467750549,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 138780
+    },
+    {
+      "epoch": 913.0921052631579,
+      "grad_norm": 0.763102650642395,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 138790
+    },
+    {
+      "epoch": 913.1578947368421,
+      "grad_norm": 0.944960355758667,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 138800
+    },
+    {
+      "epoch": 913.2236842105264,
+      "grad_norm": 0.9596688747406006,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 138810
+    },
+    {
+      "epoch": 913.2894736842105,
+      "grad_norm": 0.8753271102905273,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 138820
+    },
+    {
+      "epoch": 913.3552631578947,
+      "grad_norm": 0.8596283197402954,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 138830
+    },
+    {
+      "epoch": 913.421052631579,
+      "grad_norm": 0.9290410280227661,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 138840
+    },
+    {
+      "epoch": 913.4868421052631,
+      "grad_norm": 0.703997015953064,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 138850
+    },
+    {
+      "epoch": 913.5526315789474,
+      "grad_norm": 0.7406630516052246,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 138860
+    },
+    {
+      "epoch": 913.6184210526316,
+      "grad_norm": 0.5769844055175781,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 138870
+    },
+    {
+      "epoch": 913.6842105263158,
+      "grad_norm": 0.8807429075241089,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 138880
+    },
+    {
+      "epoch": 913.75,
+      "grad_norm": 1.0065295696258545,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 138890
+    },
+    {
+      "epoch": 913.8157894736842,
+      "grad_norm": 0.9491748213768005,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 138900
+    },
+    {
+      "epoch": 913.8815789473684,
+      "grad_norm": 0.8654152750968933,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 138910
+    },
+    {
+      "epoch": 913.9473684210526,
+      "grad_norm": 0.9335588216781616,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 138920
+    },
+    {
+      "epoch": 914.0131578947369,
+      "grad_norm": 1.165829062461853,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 138930
+    },
+    {
+      "epoch": 914.078947368421,
+      "grad_norm": 0.8912755250930786,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 138940
+    },
+    {
+      "epoch": 914.1447368421053,
+      "grad_norm": 1.1820372343063354,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 138950
+    },
+    {
+      "epoch": 914.2105263157895,
+      "grad_norm": 1.3358891010284424,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 138960
+    },
+    {
+      "epoch": 914.2763157894736,
+      "grad_norm": 0.878984272480011,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 138970
+    },
+    {
+      "epoch": 914.3421052631579,
+      "grad_norm": 1.2799499034881592,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 138980
+    },
+    {
+      "epoch": 914.4078947368421,
+      "grad_norm": 0.9388821125030518,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 138990
+    },
+    {
+      "epoch": 914.4736842105264,
+      "grad_norm": 1.0505149364471436,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 139000
+    },
+    {
+      "epoch": 914.5394736842105,
+      "grad_norm": 0.869010329246521,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 139010
+    },
+    {
+      "epoch": 914.6052631578947,
+      "grad_norm": 0.8790243864059448,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 139020
+    },
+    {
+      "epoch": 914.671052631579,
+      "grad_norm": 0.7046595811843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 139030
+    },
+    {
+      "epoch": 914.7368421052631,
+      "grad_norm": 0.9645804166793823,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 139040
+    },
+    {
+      "epoch": 914.8026315789474,
+      "grad_norm": 1.3589966297149658,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 139050
+    },
+    {
+      "epoch": 914.8684210526316,
+      "grad_norm": 1.1540025472640991,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 139060
+    },
+    {
+      "epoch": 914.9342105263158,
+      "grad_norm": 0.794073224067688,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 139070
+    },
+    {
+      "epoch": 915.0,
+      "grad_norm": 0.9464465975761414,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 139080
+    },
+    {
+      "epoch": 915.0657894736842,
+      "grad_norm": 1.0088104009628296,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 139090
+    },
+    {
+      "epoch": 915.1315789473684,
+      "grad_norm": 1.040697693824768,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 139100
+    },
+    {
+      "epoch": 915.1973684210526,
+      "grad_norm": 1.2374768257141113,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 139110
+    },
+    {
+      "epoch": 915.2631578947369,
+      "grad_norm": 1.1254475116729736,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 139120
+    },
+    {
+      "epoch": 915.328947368421,
+      "grad_norm": 1.0068753957748413,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 139130
+    },
+    {
+      "epoch": 915.3947368421053,
+      "grad_norm": 1.0009926557540894,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 139140
+    },
+    {
+      "epoch": 915.4605263157895,
+      "grad_norm": 1.104837417602539,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 139150
+    },
+    {
+      "epoch": 915.5263157894736,
+      "grad_norm": 0.9055717587471008,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 139160
+    },
+    {
+      "epoch": 915.5921052631579,
+      "grad_norm": 1.2613590955734253,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 139170
+    },
+    {
+      "epoch": 915.6578947368421,
+      "grad_norm": 0.9277036786079407,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 139180
+    },
+    {
+      "epoch": 915.7236842105264,
+      "grad_norm": 0.8551871180534363,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 139190
+    },
+    {
+      "epoch": 915.7894736842105,
+      "grad_norm": 1.1076236963272095,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 139200
+    },
+    {
+      "epoch": 915.8552631578947,
+      "grad_norm": 1.120894432067871,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 139210
+    },
+    {
+      "epoch": 915.921052631579,
+      "grad_norm": 0.8614113926887512,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 139220
+    },
+    {
+      "epoch": 915.9868421052631,
+      "grad_norm": 0.8355217576026917,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 139230
+    },
+    {
+      "epoch": 916.0526315789474,
+      "grad_norm": 0.9857189655303955,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 139240
+    },
+    {
+      "epoch": 916.1184210526316,
+      "grad_norm": 0.8549261689186096,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 139250
+    },
+    {
+      "epoch": 916.1842105263158,
+      "grad_norm": 1.0027446746826172,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 139260
+    },
+    {
+      "epoch": 916.25,
+      "grad_norm": 0.9751135110855103,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 139270
+    },
+    {
+      "epoch": 916.3157894736842,
+      "grad_norm": 1.3326411247253418,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 139280
+    },
+    {
+      "epoch": 916.3815789473684,
+      "grad_norm": 0.8798542618751526,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 139290
+    },
+    {
+      "epoch": 916.4473684210526,
+      "grad_norm": 1.432790994644165,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 139300
+    },
+    {
+      "epoch": 916.5131578947369,
+      "grad_norm": 1.3682509660720825,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 139310
+    },
+    {
+      "epoch": 916.578947368421,
+      "grad_norm": 0.9973912835121155,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 139320
+    },
+    {
+      "epoch": 916.6447368421053,
+      "grad_norm": 1.2774707078933716,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 139330
+    },
+    {
+      "epoch": 916.7105263157895,
+      "grad_norm": 1.1853746175765991,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 139340
+    },
+    {
+      "epoch": 916.7763157894736,
+      "grad_norm": 1.0939549207687378,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 139350
+    },
+    {
+      "epoch": 916.8421052631579,
+      "grad_norm": 1.6797022819519043,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 139360
+    },
+    {
+      "epoch": 916.9078947368421,
+      "grad_norm": 0.9860899448394775,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 139370
+    },
+    {
+      "epoch": 916.9736842105264,
+      "grad_norm": 1.0874377489089966,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 139380
+    },
+    {
+      "epoch": 917.0394736842105,
+      "grad_norm": 0.8851874470710754,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 139390
+    },
+    {
+      "epoch": 917.1052631578947,
+      "grad_norm": 1.039806604385376,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 139400
+    },
+    {
+      "epoch": 917.171052631579,
+      "grad_norm": 1.1718597412109375,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 139410
+    },
+    {
+      "epoch": 917.2368421052631,
+      "grad_norm": 0.9298036098480225,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 139420
+    },
+    {
+      "epoch": 917.3026315789474,
+      "grad_norm": 1.1916438341140747,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 139430
+    },
+    {
+      "epoch": 917.3684210526316,
+      "grad_norm": 1.2922661304473877,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 139440
+    },
+    {
+      "epoch": 917.4342105263158,
+      "grad_norm": 1.0661784410476685,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 139450
+    },
+    {
+      "epoch": 917.5,
+      "grad_norm": 1.1906532049179077,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 139460
+    },
+    {
+      "epoch": 917.5657894736842,
+      "grad_norm": 1.1400047540664673,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 139470
+    },
+    {
+      "epoch": 917.6315789473684,
+      "grad_norm": 1.0732362270355225,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 139480
+    },
+    {
+      "epoch": 917.6973684210526,
+      "grad_norm": 0.8187721371650696,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 139490
+    },
+    {
+      "epoch": 917.7631578947369,
+      "grad_norm": 0.8629974722862244,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 139500
+    },
+    {
+      "epoch": 917.828947368421,
+      "grad_norm": 1.0016518831253052,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 139510
+    },
+    {
+      "epoch": 917.8947368421053,
+      "grad_norm": 0.9016565680503845,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 139520
+    },
+    {
+      "epoch": 917.9605263157895,
+      "grad_norm": 0.7531505227088928,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 139530
+    },
+    {
+      "epoch": 918.0263157894736,
+      "grad_norm": 0.846524178981781,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 139540
+    },
+    {
+      "epoch": 918.0921052631579,
+      "grad_norm": 1.01736581325531,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 139550
+    },
+    {
+      "epoch": 918.1578947368421,
+      "grad_norm": 0.9176644086837769,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 139560
+    },
+    {
+      "epoch": 918.2236842105264,
+      "grad_norm": 1.1797696352005005,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 139570
+    },
+    {
+      "epoch": 918.2894736842105,
+      "grad_norm": 0.9965087175369263,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 139580
+    },
+    {
+      "epoch": 918.3552631578947,
+      "grad_norm": 1.053809642791748,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 139590
+    },
+    {
+      "epoch": 918.421052631579,
+      "grad_norm": 0.8142527341842651,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 139600
+    },
+    {
+      "epoch": 918.4868421052631,
+      "grad_norm": 1.2485212087631226,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 139610
+    },
+    {
+      "epoch": 918.5526315789474,
+      "grad_norm": 0.9339953660964966,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 139620
+    },
+    {
+      "epoch": 918.6184210526316,
+      "grad_norm": 1.036667823791504,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 139630
+    },
+    {
+      "epoch": 918.6842105263158,
+      "grad_norm": 0.8027102947235107,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 139640
+    },
+    {
+      "epoch": 918.75,
+      "grad_norm": 0.7558383345603943,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 139650
+    },
+    {
+      "epoch": 918.8157894736842,
+      "grad_norm": 0.6426530480384827,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 139660
+    },
+    {
+      "epoch": 918.8815789473684,
+      "grad_norm": 0.6960042715072632,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 139670
+    },
+    {
+      "epoch": 918.9473684210526,
+      "grad_norm": 0.6835951209068298,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 139680
+    },
+    {
+      "epoch": 919.0131578947369,
+      "grad_norm": 1.270308017730713,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 139690
+    },
+    {
+      "epoch": 919.078947368421,
+      "grad_norm": 1.1782221794128418,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 139700
+    },
+    {
+      "epoch": 919.1447368421053,
+      "grad_norm": 1.0767455101013184,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 139710
+    },
+    {
+      "epoch": 919.2105263157895,
+      "grad_norm": 0.8092596530914307,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 139720
+    },
+    {
+      "epoch": 919.2763157894736,
+      "grad_norm": 0.9447590112686157,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 139730
+    },
+    {
+      "epoch": 919.3421052631579,
+      "grad_norm": 1.1488889455795288,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 139740
+    },
+    {
+      "epoch": 919.4078947368421,
+      "grad_norm": 0.6854469180107117,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 139750
+    },
+    {
+      "epoch": 919.4736842105264,
+      "grad_norm": 0.9892016649246216,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 139760
+    },
+    {
+      "epoch": 919.5394736842105,
+      "grad_norm": 1.0678046941757202,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 139770
+    },
+    {
+      "epoch": 919.6052631578947,
+      "grad_norm": 1.0261939764022827,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 139780
+    },
+    {
+      "epoch": 919.671052631579,
+      "grad_norm": 0.7115074396133423,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 139790
+    },
+    {
+      "epoch": 919.7368421052631,
+      "grad_norm": 0.8169745802879333,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 139800
+    },
+    {
+      "epoch": 919.8026315789474,
+      "grad_norm": 0.951320469379425,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 139810
+    },
+    {
+      "epoch": 919.8684210526316,
+      "grad_norm": 1.1289328336715698,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 139820
+    },
+    {
+      "epoch": 919.9342105263158,
+      "grad_norm": 0.8910213112831116,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 139830
+    },
+    {
+      "epoch": 920.0,
+      "grad_norm": 1.091223120689392,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 139840
+    },
+    {
+      "epoch": 920.0657894736842,
+      "grad_norm": 1.1098347902297974,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 139850
+    },
+    {
+      "epoch": 920.1315789473684,
+      "grad_norm": 0.8560648560523987,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 139860
+    },
+    {
+      "epoch": 920.1973684210526,
+      "grad_norm": 0.8434080481529236,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 139870
+    },
+    {
+      "epoch": 920.2631578947369,
+      "grad_norm": 1.10075044631958,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 139880
+    },
+    {
+      "epoch": 920.328947368421,
+      "grad_norm": 1.0007864236831665,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 139890
+    },
+    {
+      "epoch": 920.3947368421053,
+      "grad_norm": 0.9732034802436829,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 139900
+    },
+    {
+      "epoch": 920.4605263157895,
+      "grad_norm": 0.9646409749984741,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 139910
+    },
+    {
+      "epoch": 920.5263157894736,
+      "grad_norm": 0.9488770961761475,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 139920
+    },
+    {
+      "epoch": 920.5921052631579,
+      "grad_norm": 1.1824666261672974,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 139930
+    },
+    {
+      "epoch": 920.6578947368421,
+      "grad_norm": 0.9511621594429016,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 139940
+    },
+    {
+      "epoch": 920.7236842105264,
+      "grad_norm": 1.0742391347885132,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 139950
+    },
+    {
+      "epoch": 920.7894736842105,
+      "grad_norm": 1.4730793237686157,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 139960
+    },
+    {
+      "epoch": 920.8552631578947,
+      "grad_norm": 1.3852665424346924,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 139970
+    },
+    {
+      "epoch": 920.921052631579,
+      "grad_norm": 1.0065243244171143,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 139980
+    },
+    {
+      "epoch": 920.9868421052631,
+      "grad_norm": 1.2180964946746826,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 139990
+    },
+    {
+      "epoch": 921.0526315789474,
+      "grad_norm": 1.0379401445388794,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 140000
+    },
+    {
+      "epoch": 921.1184210526316,
+      "grad_norm": 1.255040168762207,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 140010
+    },
+    {
+      "epoch": 921.1842105263158,
+      "grad_norm": 1.2452787160873413,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 140020
+    },
+    {
+      "epoch": 921.25,
+      "grad_norm": 1.25413978099823,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 140030
+    },
+    {
+      "epoch": 921.3157894736842,
+      "grad_norm": 0.6751289963722229,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 140040
+    },
+    {
+      "epoch": 921.3815789473684,
+      "grad_norm": 1.1533355712890625,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 140050
+    },
+    {
+      "epoch": 921.4473684210526,
+      "grad_norm": 0.6121186017990112,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 140060
+    },
+    {
+      "epoch": 921.5131578947369,
+      "grad_norm": 1.2446112632751465,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 140070
+    },
+    {
+      "epoch": 921.578947368421,
+      "grad_norm": 1.057027816772461,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 140080
+    },
+    {
+      "epoch": 921.6447368421053,
+      "grad_norm": 0.7790706753730774,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 140090
+    },
+    {
+      "epoch": 921.7105263157895,
+      "grad_norm": 0.9158756732940674,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 140100
+    },
+    {
+      "epoch": 921.7763157894736,
+      "grad_norm": 0.7222694754600525,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 140110
+    },
+    {
+      "epoch": 921.8421052631579,
+      "grad_norm": 1.073899269104004,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 140120
+    },
+    {
+      "epoch": 921.9078947368421,
+      "grad_norm": 1.1622812747955322,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 140130
+    },
+    {
+      "epoch": 921.9736842105264,
+      "grad_norm": 1.201113224029541,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 140140
+    },
+    {
+      "epoch": 922.0394736842105,
+      "grad_norm": 1.0527993440628052,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 140150
+    },
+    {
+      "epoch": 922.1052631578947,
+      "grad_norm": 0.7997422814369202,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 140160
+    },
+    {
+      "epoch": 922.171052631579,
+      "grad_norm": 0.9327260851860046,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 140170
+    },
+    {
+      "epoch": 922.2368421052631,
+      "grad_norm": 0.844416618347168,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 140180
+    },
+    {
+      "epoch": 922.3026315789474,
+      "grad_norm": 0.9128046035766602,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 140190
+    },
+    {
+      "epoch": 922.3684210526316,
+      "grad_norm": 0.7824735045433044,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 140200
+    },
+    {
+      "epoch": 922.4342105263158,
+      "grad_norm": 0.961676836013794,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 140210
+    },
+    {
+      "epoch": 922.5,
+      "grad_norm": 0.7900063991546631,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 140220
+    },
+    {
+      "epoch": 922.5657894736842,
+      "grad_norm": 1.3658584356307983,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 140230
+    },
+    {
+      "epoch": 922.6315789473684,
+      "grad_norm": 1.0145326852798462,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 140240
+    },
+    {
+      "epoch": 922.6973684210526,
+      "grad_norm": 1.2029974460601807,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 140250
+    },
+    {
+      "epoch": 922.7631578947369,
+      "grad_norm": 1.2081828117370605,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 140260
+    },
+    {
+      "epoch": 922.828947368421,
+      "grad_norm": 1.177101492881775,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 140270
+    },
+    {
+      "epoch": 922.8947368421053,
+      "grad_norm": 1.2548049688339233,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 140280
+    },
+    {
+      "epoch": 922.9605263157895,
+      "grad_norm": 1.279402732849121,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 140290
+    },
+    {
+      "epoch": 923.0263157894736,
+      "grad_norm": 0.8556655645370483,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 140300
+    },
+    {
+      "epoch": 923.0921052631579,
+      "grad_norm": 0.7919434905052185,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 140310
+    },
+    {
+      "epoch": 923.1578947368421,
+      "grad_norm": 1.0582438707351685,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 140320
+    },
+    {
+      "epoch": 923.2236842105264,
+      "grad_norm": 0.8901892304420471,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 140330
+    },
+    {
+      "epoch": 923.2894736842105,
+      "grad_norm": 0.9587327837944031,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 140340
+    },
+    {
+      "epoch": 923.3552631578947,
+      "grad_norm": 0.9833970665931702,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 140350
+    },
+    {
+      "epoch": 923.421052631579,
+      "grad_norm": 0.9788154363632202,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 140360
+    },
+    {
+      "epoch": 923.4868421052631,
+      "grad_norm": 0.7878212332725525,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 140370
+    },
+    {
+      "epoch": 923.5526315789474,
+      "grad_norm": 0.7902630567550659,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 140380
+    },
+    {
+      "epoch": 923.6184210526316,
+      "grad_norm": 1.0521742105484009,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 140390
+    },
+    {
+      "epoch": 923.6842105263158,
+      "grad_norm": 1.1913330554962158,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 140400
+    },
+    {
+      "epoch": 923.75,
+      "grad_norm": 0.9974690079689026,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 140410
+    },
+    {
+      "epoch": 923.8157894736842,
+      "grad_norm": 0.7740110158920288,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 140420
+    },
+    {
+      "epoch": 923.8815789473684,
+      "grad_norm": 0.651350736618042,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 140430
+    },
+    {
+      "epoch": 923.9473684210526,
+      "grad_norm": 0.8533127903938293,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 140440
+    },
+    {
+      "epoch": 924.0131578947369,
+      "grad_norm": 1.2042335271835327,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 140450
+    },
+    {
+      "epoch": 924.078947368421,
+      "grad_norm": 1.0779392719268799,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 140460
+    },
+    {
+      "epoch": 924.1447368421053,
+      "grad_norm": 1.2157610654830933,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 140470
+    },
+    {
+      "epoch": 924.2105263157895,
+      "grad_norm": 1.7160738706588745,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 140480
+    },
+    {
+      "epoch": 924.2763157894736,
+      "grad_norm": 0.9967711567878723,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 140490
+    },
+    {
+      "epoch": 924.3421052631579,
+      "grad_norm": 0.9991487264633179,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 140500
+    },
+    {
+      "epoch": 924.4078947368421,
+      "grad_norm": 1.028993844985962,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 140510
+    },
+    {
+      "epoch": 924.4736842105264,
+      "grad_norm": 1.1722681522369385,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 140520
+    },
+    {
+      "epoch": 924.5394736842105,
+      "grad_norm": 0.9538052678108215,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 140530
+    },
+    {
+      "epoch": 924.6052631578947,
+      "grad_norm": 0.9765692353248596,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 140540
+    },
+    {
+      "epoch": 924.671052631579,
+      "grad_norm": 1.1148061752319336,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 140550
+    },
+    {
+      "epoch": 924.7368421052631,
+      "grad_norm": 1.2510484457015991,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 140560
+    },
+    {
+      "epoch": 924.8026315789474,
+      "grad_norm": 0.6285231709480286,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 140570
+    },
+    {
+      "epoch": 924.8684210526316,
+      "grad_norm": 0.8659400939941406,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 140580
+    },
+    {
+      "epoch": 924.9342105263158,
+      "grad_norm": 1.054602861404419,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 140590
+    },
+    {
+      "epoch": 925.0,
+      "grad_norm": 1.0485122203826904,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 140600
+    },
+    {
+      "epoch": 925.0657894736842,
+      "grad_norm": 1.1340656280517578,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 140610
+    },
+    {
+      "epoch": 925.1315789473684,
+      "grad_norm": 0.945837140083313,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 140620
+    },
+    {
+      "epoch": 925.1973684210526,
+      "grad_norm": 0.9775026440620422,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 140630
+    },
+    {
+      "epoch": 925.2631578947369,
+      "grad_norm": 0.979673445224762,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 140640
+    },
+    {
+      "epoch": 925.328947368421,
+      "grad_norm": 1.1637605428695679,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 140650
+    },
+    {
+      "epoch": 925.3947368421053,
+      "grad_norm": 1.587935209274292,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 140660
+    },
+    {
+      "epoch": 925.4605263157895,
+      "grad_norm": 1.3600460290908813,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 140670
+    },
+    {
+      "epoch": 925.5263157894736,
+      "grad_norm": 1.253381609916687,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 140680
+    },
+    {
+      "epoch": 925.5921052631579,
+      "grad_norm": 1.4726488590240479,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 140690
+    },
+    {
+      "epoch": 925.6578947368421,
+      "grad_norm": 1.187400221824646,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 140700
+    },
+    {
+      "epoch": 925.7236842105264,
+      "grad_norm": 1.0697247982025146,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 140710
+    },
+    {
+      "epoch": 925.7894736842105,
+      "grad_norm": 1.059090256690979,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 140720
+    },
+    {
+      "epoch": 925.8552631578947,
+      "grad_norm": 1.2741175889968872,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 140730
+    },
+    {
+      "epoch": 925.921052631579,
+      "grad_norm": 1.1128478050231934,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 140740
+    },
+    {
+      "epoch": 925.9868421052631,
+      "grad_norm": 0.7608397006988525,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 140750
+    },
+    {
+      "epoch": 926.0526315789474,
+      "grad_norm": 1.069777250289917,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 140760
+    },
+    {
+      "epoch": 926.1184210526316,
+      "grad_norm": 1.0431427955627441,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 140770
+    },
+    {
+      "epoch": 926.1842105263158,
+      "grad_norm": 1.0773409605026245,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 140780
+    },
+    {
+      "epoch": 926.25,
+      "grad_norm": 1.208251714706421,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 140790
+    },
+    {
+      "epoch": 926.3157894736842,
+      "grad_norm": 1.0173733234405518,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 140800
+    },
+    {
+      "epoch": 926.3815789473684,
+      "grad_norm": 0.7675154209136963,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 140810
+    },
+    {
+      "epoch": 926.4473684210526,
+      "grad_norm": 1.237313985824585,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 140820
+    },
+    {
+      "epoch": 926.5131578947369,
+      "grad_norm": 0.9644690752029419,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 140830
+    },
+    {
+      "epoch": 926.578947368421,
+      "grad_norm": 1.1485686302185059,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 140840
+    },
+    {
+      "epoch": 926.6447368421053,
+      "grad_norm": 1.82488214969635,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 140850
+    },
+    {
+      "epoch": 926.7105263157895,
+      "grad_norm": 1.2388635873794556,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 140860
+    },
+    {
+      "epoch": 926.7763157894736,
+      "grad_norm": 1.3352988958358765,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 140870
+    },
+    {
+      "epoch": 926.8421052631579,
+      "grad_norm": 1.1297030448913574,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 140880
+    },
+    {
+      "epoch": 926.9078947368421,
+      "grad_norm": 1.162842869758606,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 140890
+    },
+    {
+      "epoch": 926.9736842105264,
+      "grad_norm": 1.2064037322998047,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 140900
+    },
+    {
+      "epoch": 927.0394736842105,
+      "grad_norm": 1.08427894115448,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 140910
+    },
+    {
+      "epoch": 927.1052631578947,
+      "grad_norm": 1.1430708169937134,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 140920
+    },
+    {
+      "epoch": 927.171052631579,
+      "grad_norm": 1.0991872549057007,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 140930
+    },
+    {
+      "epoch": 927.2368421052631,
+      "grad_norm": 1.4145089387893677,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 140940
+    },
+    {
+      "epoch": 927.3026315789474,
+      "grad_norm": 0.8903805613517761,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 140950
+    },
+    {
+      "epoch": 927.3684210526316,
+      "grad_norm": 1.02116858959198,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 140960
+    },
+    {
+      "epoch": 927.4342105263158,
+      "grad_norm": 1.196993350982666,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 140970
+    },
+    {
+      "epoch": 927.5,
+      "grad_norm": 1.419411063194275,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 140980
+    },
+    {
+      "epoch": 927.5657894736842,
+      "grad_norm": 1.4672049283981323,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 140990
+    },
+    {
+      "epoch": 927.6315789473684,
+      "grad_norm": 0.8084446787834167,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 141000
+    },
+    {
+      "epoch": 927.6973684210526,
+      "grad_norm": 1.2789831161499023,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 141010
+    },
+    {
+      "epoch": 927.7631578947369,
+      "grad_norm": 0.6588245630264282,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 141020
+    },
+    {
+      "epoch": 927.828947368421,
+      "grad_norm": 0.9887574911117554,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 141030
+    },
+    {
+      "epoch": 927.8947368421053,
+      "grad_norm": 0.8936231136322021,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 141040
+    },
+    {
+      "epoch": 927.9605263157895,
+      "grad_norm": 0.9490195512771606,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 141050
+    },
+    {
+      "epoch": 928.0263157894736,
+      "grad_norm": 1.2485547065734863,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 141060
+    },
+    {
+      "epoch": 928.0921052631579,
+      "grad_norm": 1.1075419187545776,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 141070
+    },
+    {
+      "epoch": 928.1578947368421,
+      "grad_norm": 1.1165497303009033,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 141080
+    },
+    {
+      "epoch": 928.2236842105264,
+      "grad_norm": 1.352881669998169,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 141090
+    },
+    {
+      "epoch": 928.2894736842105,
+      "grad_norm": 1.4939100742340088,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 141100
+    },
+    {
+      "epoch": 928.3552631578947,
+      "grad_norm": 1.2259972095489502,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 141110
+    },
+    {
+      "epoch": 928.421052631579,
+      "grad_norm": 0.6991260647773743,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 141120
+    },
+    {
+      "epoch": 928.4868421052631,
+      "grad_norm": 0.9250067472457886,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 141130
+    },
+    {
+      "epoch": 928.5526315789474,
+      "grad_norm": 1.0126291513442993,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 141140
+    },
+    {
+      "epoch": 928.6184210526316,
+      "grad_norm": 1.092654824256897,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 141150
+    },
+    {
+      "epoch": 928.6842105263158,
+      "grad_norm": 1.143467903137207,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 141160
+    },
+    {
+      "epoch": 928.75,
+      "grad_norm": 1.1244598627090454,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 141170
+    },
+    {
+      "epoch": 928.8157894736842,
+      "grad_norm": 0.8869340419769287,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 141180
+    },
+    {
+      "epoch": 928.8815789473684,
+      "grad_norm": 1.0493642091751099,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 141190
+    },
+    {
+      "epoch": 928.9473684210526,
+      "grad_norm": 1.048553228378296,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 141200
+    },
+    {
+      "epoch": 929.0131578947369,
+      "grad_norm": 1.0063235759735107,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 141210
+    },
+    {
+      "epoch": 929.078947368421,
+      "grad_norm": 0.803489625453949,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 141220
+    },
+    {
+      "epoch": 929.1447368421053,
+      "grad_norm": 0.7754166126251221,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 141230
+    },
+    {
+      "epoch": 929.2105263157895,
+      "grad_norm": 1.123893141746521,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 141240
+    },
+    {
+      "epoch": 929.2763157894736,
+      "grad_norm": 0.8346624970436096,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 141250
+    },
+    {
+      "epoch": 929.3421052631579,
+      "grad_norm": 1.2501662969589233,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 141260
+    },
+    {
+      "epoch": 929.4078947368421,
+      "grad_norm": 0.9028252959251404,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 141270
+    },
+    {
+      "epoch": 929.4736842105264,
+      "grad_norm": 1.1409492492675781,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 141280
+    },
+    {
+      "epoch": 929.5394736842105,
+      "grad_norm": 0.795866847038269,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 141290
+    },
+    {
+      "epoch": 929.6052631578947,
+      "grad_norm": 1.0980581045150757,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 141300
+    },
+    {
+      "epoch": 929.671052631579,
+      "grad_norm": 0.9573304057121277,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 141310
+    },
+    {
+      "epoch": 929.7368421052631,
+      "grad_norm": 0.9308114051818848,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 141320
+    },
+    {
+      "epoch": 929.8026315789474,
+      "grad_norm": 1.13672935962677,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 141330
+    },
+    {
+      "epoch": 929.8684210526316,
+      "grad_norm": 1.0716716051101685,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 141340
+    },
+    {
+      "epoch": 929.9342105263158,
+      "grad_norm": 0.8969205617904663,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 141350
+    },
+    {
+      "epoch": 930.0,
+      "grad_norm": 0.7731922268867493,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 141360
+    },
+    {
+      "epoch": 930.0657894736842,
+      "grad_norm": 1.054961085319519,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 141370
+    },
+    {
+      "epoch": 930.1315789473684,
+      "grad_norm": 1.0062198638916016,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 141380
+    },
+    {
+      "epoch": 930.1973684210526,
+      "grad_norm": 0.8831532597541809,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 141390
+    },
+    {
+      "epoch": 930.2631578947369,
+      "grad_norm": 0.94657963514328,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 141400
+    },
+    {
+      "epoch": 930.328947368421,
+      "grad_norm": 0.7149097919464111,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 141410
+    },
+    {
+      "epoch": 930.3947368421053,
+      "grad_norm": 1.1722291707992554,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 141420
+    },
+    {
+      "epoch": 930.4605263157895,
+      "grad_norm": 1.2141605615615845,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 141430
+    },
+    {
+      "epoch": 930.5263157894736,
+      "grad_norm": 1.0134645700454712,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 141440
+    },
+    {
+      "epoch": 930.5921052631579,
+      "grad_norm": 1.4195690155029297,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 141450
+    },
+    {
+      "epoch": 930.6578947368421,
+      "grad_norm": 1.0043601989746094,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 141460
+    },
+    {
+      "epoch": 930.7236842105264,
+      "grad_norm": 0.9379559755325317,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 141470
+    },
+    {
+      "epoch": 930.7894736842105,
+      "grad_norm": 1.1322745084762573,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 141480
+    },
+    {
+      "epoch": 930.8552631578947,
+      "grad_norm": 1.1862339973449707,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 141490
+    },
+    {
+      "epoch": 930.921052631579,
+      "grad_norm": 1.0361219644546509,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 141500
+    },
+    {
+      "epoch": 930.9868421052631,
+      "grad_norm": 1.4550143480300903,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 141510
+    },
+    {
+      "epoch": 931.0526315789474,
+      "grad_norm": 1.0843679904937744,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 141520
+    },
+    {
+      "epoch": 931.1184210526316,
+      "grad_norm": 1.4889967441558838,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 141530
+    },
+    {
+      "epoch": 931.1842105263158,
+      "grad_norm": 1.130838394165039,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 141540
+    },
+    {
+      "epoch": 931.25,
+      "grad_norm": 1.1887781620025635,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 141550
+    },
+    {
+      "epoch": 931.3157894736842,
+      "grad_norm": 0.8891682028770447,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 141560
+    },
+    {
+      "epoch": 931.3815789473684,
+      "grad_norm": 1.069966197013855,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 141570
+    },
+    {
+      "epoch": 931.4473684210526,
+      "grad_norm": 1.2888097763061523,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 141580
+    },
+    {
+      "epoch": 931.5131578947369,
+      "grad_norm": 0.7364023327827454,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 141590
+    },
+    {
+      "epoch": 931.578947368421,
+      "grad_norm": 1.0230684280395508,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 141600
+    },
+    {
+      "epoch": 931.6447368421053,
+      "grad_norm": 1.1566507816314697,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 141610
+    },
+    {
+      "epoch": 931.7105263157895,
+      "grad_norm": 1.0186455249786377,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 141620
+    },
+    {
+      "epoch": 931.7763157894736,
+      "grad_norm": 1.209118127822876,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 141630
+    },
+    {
+      "epoch": 931.8421052631579,
+      "grad_norm": 0.766610324382782,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 141640
+    },
+    {
+      "epoch": 931.9078947368421,
+      "grad_norm": 1.0066529512405396,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 141650
+    },
+    {
+      "epoch": 931.9736842105264,
+      "grad_norm": 0.9950891137123108,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 141660
+    },
+    {
+      "epoch": 932.0394736842105,
+      "grad_norm": 1.1890411376953125,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 141670
+    },
+    {
+      "epoch": 932.1052631578947,
+      "grad_norm": 0.9757727384567261,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 141680
+    },
+    {
+      "epoch": 932.171052631579,
+      "grad_norm": 0.7399367690086365,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 141690
+    },
+    {
+      "epoch": 932.2368421052631,
+      "grad_norm": 0.8191750645637512,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 141700
+    },
+    {
+      "epoch": 932.3026315789474,
+      "grad_norm": 1.0755620002746582,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 141710
+    },
+    {
+      "epoch": 932.3684210526316,
+      "grad_norm": 1.099184513092041,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 141720
+    },
+    {
+      "epoch": 932.4342105263158,
+      "grad_norm": 0.948433518409729,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 141730
+    },
+    {
+      "epoch": 932.5,
+      "grad_norm": 1.2848048210144043,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 141740
+    },
+    {
+      "epoch": 932.5657894736842,
+      "grad_norm": 1.0524227619171143,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 141750
+    },
+    {
+      "epoch": 932.6315789473684,
+      "grad_norm": 1.1864573955535889,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 141760
+    },
+    {
+      "epoch": 932.6973684210526,
+      "grad_norm": 1.1344093084335327,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 141770
+    },
+    {
+      "epoch": 932.7631578947369,
+      "grad_norm": 1.0128337144851685,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 141780
+    },
+    {
+      "epoch": 932.828947368421,
+      "grad_norm": 0.9922648668289185,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 141790
+    },
+    {
+      "epoch": 932.8947368421053,
+      "grad_norm": 1.1825958490371704,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 141800
+    },
+    {
+      "epoch": 932.9605263157895,
+      "grad_norm": 1.0141911506652832,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 141810
+    },
+    {
+      "epoch": 933.0263157894736,
+      "grad_norm": 1.0173567533493042,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 141820
+    },
+    {
+      "epoch": 933.0921052631579,
+      "grad_norm": 1.0617711544036865,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 141830
+    },
+    {
+      "epoch": 933.1578947368421,
+      "grad_norm": 1.0690877437591553,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 141840
+    },
+    {
+      "epoch": 933.2236842105264,
+      "grad_norm": 0.7836157083511353,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 141850
+    },
+    {
+      "epoch": 933.2894736842105,
+      "grad_norm": 0.8588806986808777,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 141860
+    },
+    {
+      "epoch": 933.3552631578947,
+      "grad_norm": 0.8288969993591309,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 141870
+    },
+    {
+      "epoch": 933.421052631579,
+      "grad_norm": 0.8952768445014954,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 141880
+    },
+    {
+      "epoch": 933.4868421052631,
+      "grad_norm": 0.7412847280502319,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 141890
+    },
+    {
+      "epoch": 933.5526315789474,
+      "grad_norm": 0.5983102321624756,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 141900
+    },
+    {
+      "epoch": 933.6184210526316,
+      "grad_norm": 0.8310821652412415,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 141910
+    },
+    {
+      "epoch": 933.6842105263158,
+      "grad_norm": 1.0020363330841064,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 141920
+    },
+    {
+      "epoch": 933.75,
+      "grad_norm": 0.9288676381111145,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 141930
+    },
+    {
+      "epoch": 933.8157894736842,
+      "grad_norm": 0.8493418097496033,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 141940
+    },
+    {
+      "epoch": 933.8815789473684,
+      "grad_norm": 1.2082266807556152,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 141950
+    },
+    {
+      "epoch": 933.9473684210526,
+      "grad_norm": 1.1100561618804932,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 141960
+    },
+    {
+      "epoch": 934.0131578947369,
+      "grad_norm": 1.3064969778060913,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 141970
+    },
+    {
+      "epoch": 934.078947368421,
+      "grad_norm": 1.045897126197815,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 141980
+    },
+    {
+      "epoch": 934.1447368421053,
+      "grad_norm": 1.0338056087493896,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 141990
+    },
+    {
+      "epoch": 934.2105263157895,
+      "grad_norm": 0.882927656173706,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 142000
+    },
+    {
+      "epoch": 934.2763157894736,
+      "grad_norm": 0.8924429416656494,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 142010
+    },
+    {
+      "epoch": 934.3421052631579,
+      "grad_norm": 1.2137278318405151,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 142020
+    },
+    {
+      "epoch": 934.4078947368421,
+      "grad_norm": 1.2545146942138672,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 142030
+    },
+    {
+      "epoch": 934.4736842105264,
+      "grad_norm": 1.1060088872909546,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 142040
+    },
+    {
+      "epoch": 934.5394736842105,
+      "grad_norm": 0.8980741500854492,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 142050
+    },
+    {
+      "epoch": 934.6052631578947,
+      "grad_norm": 1.3434455394744873,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 142060
+    },
+    {
+      "epoch": 934.671052631579,
+      "grad_norm": 1.541412591934204,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 142070
+    },
+    {
+      "epoch": 934.7368421052631,
+      "grad_norm": 0.7517480850219727,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 142080
+    },
+    {
+      "epoch": 934.8026315789474,
+      "grad_norm": 0.9728532433509827,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 142090
+    },
+    {
+      "epoch": 934.8684210526316,
+      "grad_norm": 0.956839382648468,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 142100
+    },
+    {
+      "epoch": 934.9342105263158,
+      "grad_norm": 0.6664096117019653,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 142110
+    },
+    {
+      "epoch": 935.0,
+      "grad_norm": 0.6346357464790344,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 142120
+    },
+    {
+      "epoch": 935.0657894736842,
+      "grad_norm": 0.8186964392662048,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 142130
+    },
+    {
+      "epoch": 935.1315789473684,
+      "grad_norm": 1.007627248764038,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 142140
+    },
+    {
+      "epoch": 935.1973684210526,
+      "grad_norm": 0.8972962498664856,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 142150
+    },
+    {
+      "epoch": 935.2631578947369,
+      "grad_norm": 0.863339364528656,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 142160
+    },
+    {
+      "epoch": 935.328947368421,
+      "grad_norm": 1.014875054359436,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 142170
+    },
+    {
+      "epoch": 935.3947368421053,
+      "grad_norm": 1.0720738172531128,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 142180
+    },
+    {
+      "epoch": 935.4605263157895,
+      "grad_norm": 0.8214248418807983,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 142190
+    },
+    {
+      "epoch": 935.5263157894736,
+      "grad_norm": 0.9831026792526245,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 142200
+    },
+    {
+      "epoch": 935.5921052631579,
+      "grad_norm": 1.0691466331481934,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 142210
+    },
+    {
+      "epoch": 935.6578947368421,
+      "grad_norm": 1.3758347034454346,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 142220
+    },
+    {
+      "epoch": 935.7236842105264,
+      "grad_norm": 1.7172006368637085,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 142230
+    },
+    {
+      "epoch": 935.7894736842105,
+      "grad_norm": 1.1777650117874146,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 142240
+    },
+    {
+      "epoch": 935.8552631578947,
+      "grad_norm": 1.176995038986206,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 142250
+    },
+    {
+      "epoch": 935.921052631579,
+      "grad_norm": 1.0358529090881348,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 142260
+    },
+    {
+      "epoch": 935.9868421052631,
+      "grad_norm": 1.0105960369110107,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 142270
+    },
+    {
+      "epoch": 936.0526315789474,
+      "grad_norm": 1.2849767208099365,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 142280
+    },
+    {
+      "epoch": 936.1184210526316,
+      "grad_norm": 1.315524935722351,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 142290
+    },
+    {
+      "epoch": 936.1842105263158,
+      "grad_norm": 1.1446727514266968,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 142300
+    },
+    {
+      "epoch": 936.25,
+      "grad_norm": 1.1680877208709717,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 142310
+    },
+    {
+      "epoch": 936.3157894736842,
+      "grad_norm": 1.1997969150543213,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 142320
+    },
+    {
+      "epoch": 936.3815789473684,
+      "grad_norm": 1.043163537979126,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 142330
+    },
+    {
+      "epoch": 936.4473684210526,
+      "grad_norm": 1.2314527034759521,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 142340
+    },
+    {
+      "epoch": 936.5131578947369,
+      "grad_norm": 0.8051645755767822,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 142350
+    },
+    {
+      "epoch": 936.578947368421,
+      "grad_norm": 0.8518595099449158,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 142360
+    },
+    {
+      "epoch": 936.6447368421053,
+      "grad_norm": 1.2750550508499146,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 142370
+    },
+    {
+      "epoch": 936.7105263157895,
+      "grad_norm": 0.7354443073272705,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 142380
+    },
+    {
+      "epoch": 936.7763157894736,
+      "grad_norm": 0.8145555853843689,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 142390
+    },
+    {
+      "epoch": 936.8421052631579,
+      "grad_norm": 1.1294350624084473,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 142400
+    },
+    {
+      "epoch": 936.9078947368421,
+      "grad_norm": 0.8376908302307129,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 142410
+    },
+    {
+      "epoch": 936.9736842105264,
+      "grad_norm": 0.9719381332397461,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 142420
+    },
+    {
+      "epoch": 937.0394736842105,
+      "grad_norm": 0.9003921747207642,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 142430
+    },
+    {
+      "epoch": 937.1052631578947,
+      "grad_norm": 1.2315677404403687,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 142440
+    },
+    {
+      "epoch": 937.171052631579,
+      "grad_norm": 1.2337232828140259,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 142450
+    },
+    {
+      "epoch": 937.2368421052631,
+      "grad_norm": 0.866338312625885,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 142460
+    },
+    {
+      "epoch": 937.3026315789474,
+      "grad_norm": 1.071451187133789,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 142470
+    },
+    {
+      "epoch": 937.3684210526316,
+      "grad_norm": 1.0966578722000122,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 142480
+    },
+    {
+      "epoch": 937.4342105263158,
+      "grad_norm": 0.882290780544281,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 142490
+    },
+    {
+      "epoch": 937.5,
+      "grad_norm": 0.9930370450019836,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 142500
+    },
+    {
+      "epoch": 937.5657894736842,
+      "grad_norm": 0.729529857635498,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 142510
+    },
+    {
+      "epoch": 937.6315789473684,
+      "grad_norm": 0.8702946305274963,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 142520
+    },
+    {
+      "epoch": 937.6973684210526,
+      "grad_norm": 0.8152609467506409,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 142530
+    },
+    {
+      "epoch": 937.7631578947369,
+      "grad_norm": 1.1005128622055054,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 142540
+    },
+    {
+      "epoch": 937.828947368421,
+      "grad_norm": 1.302595615386963,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 142550
+    },
+    {
+      "epoch": 937.8947368421053,
+      "grad_norm": 1.0444657802581787,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 142560
+    },
+    {
+      "epoch": 937.9605263157895,
+      "grad_norm": 0.947675883769989,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 142570
+    },
+    {
+      "epoch": 938.0263157894736,
+      "grad_norm": 0.942642867565155,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 142580
+    },
+    {
+      "epoch": 938.0921052631579,
+      "grad_norm": 1.1774204969406128,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 142590
+    },
+    {
+      "epoch": 938.1578947368421,
+      "grad_norm": 1.1025643348693848,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 142600
+    },
+    {
+      "epoch": 938.2236842105264,
+      "grad_norm": 1.1018234491348267,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 142610
+    },
+    {
+      "epoch": 938.2894736842105,
+      "grad_norm": 1.5480759143829346,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 142620
+    },
+    {
+      "epoch": 938.3552631578947,
+      "grad_norm": 1.1027711629867554,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 142630
+    },
+    {
+      "epoch": 938.421052631579,
+      "grad_norm": 0.9052314162254333,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 142640
+    },
+    {
+      "epoch": 938.4868421052631,
+      "grad_norm": 1.1163532733917236,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 142650
+    },
+    {
+      "epoch": 938.5526315789474,
+      "grad_norm": 1.5282114744186401,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 142660
+    },
+    {
+      "epoch": 938.6184210526316,
+      "grad_norm": 1.0808197259902954,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 142670
+    },
+    {
+      "epoch": 938.6842105263158,
+      "grad_norm": 0.9166187644004822,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 142680
+    },
+    {
+      "epoch": 938.75,
+      "grad_norm": 1.2333557605743408,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 142690
+    },
+    {
+      "epoch": 938.8157894736842,
+      "grad_norm": 1.2331442832946777,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 142700
+    },
+    {
+      "epoch": 938.8815789473684,
+      "grad_norm": 0.7584700584411621,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 142710
+    },
+    {
+      "epoch": 938.9473684210526,
+      "grad_norm": 0.8076496720314026,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 142720
+    },
+    {
+      "epoch": 939.0131578947369,
+      "grad_norm": 1.1252182722091675,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 142730
+    },
+    {
+      "epoch": 939.078947368421,
+      "grad_norm": 0.7238128185272217,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 142740
+    },
+    {
+      "epoch": 939.1447368421053,
+      "grad_norm": 0.9610776305198669,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 142750
+    },
+    {
+      "epoch": 939.2105263157895,
+      "grad_norm": 1.0641119480133057,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 142760
+    },
+    {
+      "epoch": 939.2763157894736,
+      "grad_norm": 1.4816539287567139,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 142770
+    },
+    {
+      "epoch": 939.3421052631579,
+      "grad_norm": 1.748676061630249,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 142780
+    },
+    {
+      "epoch": 939.4078947368421,
+      "grad_norm": 1.8980330228805542,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 142790
+    },
+    {
+      "epoch": 939.4736842105264,
+      "grad_norm": 1.2307404279708862,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 142800
+    },
+    {
+      "epoch": 939.5394736842105,
+      "grad_norm": 1.277494192123413,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 142810
+    },
+    {
+      "epoch": 939.6052631578947,
+      "grad_norm": 1.3489617109298706,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 142820
+    },
+    {
+      "epoch": 939.671052631579,
+      "grad_norm": 1.0544668436050415,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 142830
+    },
+    {
+      "epoch": 939.7368421052631,
+      "grad_norm": 1.1810656785964966,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 142840
+    },
+    {
+      "epoch": 939.8026315789474,
+      "grad_norm": 0.905704140663147,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 142850
+    },
+    {
+      "epoch": 939.8684210526316,
+      "grad_norm": 1.0323855876922607,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 142860
+    },
+    {
+      "epoch": 939.9342105263158,
+      "grad_norm": 1.2508949041366577,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 142870
+    },
+    {
+      "epoch": 940.0,
+      "grad_norm": 1.1149414777755737,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 142880
+    },
+    {
+      "epoch": 940.0657894736842,
+      "grad_norm": 1.280163288116455,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 142890
+    },
+    {
+      "epoch": 940.1315789473684,
+      "grad_norm": 1.270648717880249,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 142900
+    },
+    {
+      "epoch": 940.1973684210526,
+      "grad_norm": 0.8688375949859619,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 142910
+    },
+    {
+      "epoch": 940.2631578947369,
+      "grad_norm": 1.2751940488815308,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 142920
+    },
+    {
+      "epoch": 940.328947368421,
+      "grad_norm": 1.1398617029190063,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 142930
+    },
+    {
+      "epoch": 940.3947368421053,
+      "grad_norm": 0.7770438194274902,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 142940
+    },
+    {
+      "epoch": 940.4605263157895,
+      "grad_norm": 1.470767855644226,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 142950
+    },
+    {
+      "epoch": 940.5263157894736,
+      "grad_norm": 1.3184136152267456,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 142960
+    },
+    {
+      "epoch": 940.5921052631579,
+      "grad_norm": 0.9164167642593384,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 142970
+    },
+    {
+      "epoch": 940.6578947368421,
+      "grad_norm": 1.149084210395813,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 142980
+    },
+    {
+      "epoch": 940.7236842105264,
+      "grad_norm": 1.2743107080459595,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 142990
+    },
+    {
+      "epoch": 940.7894736842105,
+      "grad_norm": 0.6566851139068604,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 143000
+    },
+    {
+      "epoch": 940.8552631578947,
+      "grad_norm": 0.6921648979187012,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 143010
+    },
+    {
+      "epoch": 940.921052631579,
+      "grad_norm": 0.9658145904541016,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 143020
+    },
+    {
+      "epoch": 940.9868421052631,
+      "grad_norm": 1.0340406894683838,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 143030
+    },
+    {
+      "epoch": 941.0526315789474,
+      "grad_norm": 1.061692237854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 143040
+    },
+    {
+      "epoch": 941.1184210526316,
+      "grad_norm": 0.8725217580795288,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 143050
+    },
+    {
+      "epoch": 941.1842105263158,
+      "grad_norm": 1.1090641021728516,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 143060
+    },
+    {
+      "epoch": 941.25,
+      "grad_norm": 1.313765287399292,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 143070
+    },
+    {
+      "epoch": 941.3157894736842,
+      "grad_norm": 0.8687812089920044,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 143080
+    },
+    {
+      "epoch": 941.3815789473684,
+      "grad_norm": 1.066633701324463,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 143090
+    },
+    {
+      "epoch": 941.4473684210526,
+      "grad_norm": 1.1810579299926758,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 143100
+    },
+    {
+      "epoch": 941.5131578947369,
+      "grad_norm": 1.057855486869812,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 143110
+    },
+    {
+      "epoch": 941.578947368421,
+      "grad_norm": 0.7218169569969177,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 143120
+    },
+    {
+      "epoch": 941.6447368421053,
+      "grad_norm": 1.1620811223983765,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 143130
+    },
+    {
+      "epoch": 941.7105263157895,
+      "grad_norm": 1.2973016500473022,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 143140
+    },
+    {
+      "epoch": 941.7763157894736,
+      "grad_norm": 1.3135859966278076,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 143150
+    },
+    {
+      "epoch": 941.8421052631579,
+      "grad_norm": 0.8703665733337402,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 143160
+    },
+    {
+      "epoch": 941.9078947368421,
+      "grad_norm": 0.9815681576728821,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 143170
+    },
+    {
+      "epoch": 941.9736842105264,
+      "grad_norm": 1.3823065757751465,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 143180
+    },
+    {
+      "epoch": 942.0394736842105,
+      "grad_norm": 0.974730372428894,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 143190
+    },
+    {
+      "epoch": 942.1052631578947,
+      "grad_norm": 0.7570364475250244,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 143200
+    },
+    {
+      "epoch": 942.171052631579,
+      "grad_norm": 1.0894103050231934,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 143210
+    },
+    {
+      "epoch": 942.2368421052631,
+      "grad_norm": 1.1041042804718018,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 143220
+    },
+    {
+      "epoch": 942.3026315789474,
+      "grad_norm": 1.2231684923171997,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 143230
+    },
+    {
+      "epoch": 942.3684210526316,
+      "grad_norm": 1.2694424390792847,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 143240
+    },
+    {
+      "epoch": 942.4342105263158,
+      "grad_norm": 0.8678814172744751,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 143250
+    },
+    {
+      "epoch": 942.5,
+      "grad_norm": 0.9810147285461426,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 143260
+    },
+    {
+      "epoch": 942.5657894736842,
+      "grad_norm": 0.6947391629219055,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 143270
+    },
+    {
+      "epoch": 942.6315789473684,
+      "grad_norm": 1.0292843580245972,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 143280
+    },
+    {
+      "epoch": 942.6973684210526,
+      "grad_norm": 0.93792724609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 143290
+    },
+    {
+      "epoch": 942.7631578947369,
+      "grad_norm": 0.827596127986908,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 143300
+    },
+    {
+      "epoch": 942.828947368421,
+      "grad_norm": 1.1035057306289673,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 143310
+    },
+    {
+      "epoch": 942.8947368421053,
+      "grad_norm": 1.2476816177368164,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 143320
+    },
+    {
+      "epoch": 942.9605263157895,
+      "grad_norm": 1.1506506204605103,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 143330
+    },
+    {
+      "epoch": 943.0263157894736,
+      "grad_norm": 0.9531042575836182,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 143340
+    },
+    {
+      "epoch": 943.0921052631579,
+      "grad_norm": 1.2359414100646973,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 143350
+    },
+    {
+      "epoch": 943.1578947368421,
+      "grad_norm": 1.264983057975769,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 143360
+    },
+    {
+      "epoch": 943.2236842105264,
+      "grad_norm": 1.0347520112991333,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 143370
+    },
+    {
+      "epoch": 943.2894736842105,
+      "grad_norm": 1.0723800659179688,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 143380
+    },
+    {
+      "epoch": 943.3552631578947,
+      "grad_norm": 1.2043391466140747,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 143390
+    },
+    {
+      "epoch": 943.421052631579,
+      "grad_norm": 0.9921755790710449,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 143400
+    },
+    {
+      "epoch": 943.4868421052631,
+      "grad_norm": 1.0856060981750488,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 143410
+    },
+    {
+      "epoch": 943.5526315789474,
+      "grad_norm": 1.3713430166244507,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 143420
+    },
+    {
+      "epoch": 943.6184210526316,
+      "grad_norm": 1.2866997718811035,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 143430
+    },
+    {
+      "epoch": 943.6842105263158,
+      "grad_norm": 1.1391348838806152,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 143440
+    },
+    {
+      "epoch": 943.75,
+      "grad_norm": 0.8373425006866455,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 143450
+    },
+    {
+      "epoch": 943.8157894736842,
+      "grad_norm": 0.9597638845443726,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 143460
+    },
+    {
+      "epoch": 943.8815789473684,
+      "grad_norm": 0.904534637928009,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 143470
+    },
+    {
+      "epoch": 943.9473684210526,
+      "grad_norm": 0.8183276057243347,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 143480
+    },
+    {
+      "epoch": 944.0131578947369,
+      "grad_norm": 0.7242514491081238,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 143490
+    },
+    {
+      "epoch": 944.078947368421,
+      "grad_norm": 0.8846292495727539,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 143500
+    },
+    {
+      "epoch": 944.1447368421053,
+      "grad_norm": 0.9731491804122925,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 143510
+    },
+    {
+      "epoch": 944.2105263157895,
+      "grad_norm": 0.6665592193603516,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 143520
+    },
+    {
+      "epoch": 944.2763157894736,
+      "grad_norm": 0.75457763671875,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 143530
+    },
+    {
+      "epoch": 944.3421052631579,
+      "grad_norm": 0.9339364767074585,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 143540
+    },
+    {
+      "epoch": 944.4078947368421,
+      "grad_norm": 0.8604110479354858,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 143550
+    },
+    {
+      "epoch": 944.4736842105264,
+      "grad_norm": 1.0775866508483887,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 143560
+    },
+    {
+      "epoch": 944.5394736842105,
+      "grad_norm": 1.1625889539718628,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 143570
+    },
+    {
+      "epoch": 944.6052631578947,
+      "grad_norm": 1.0786641836166382,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 143580
+    },
+    {
+      "epoch": 944.671052631579,
+      "grad_norm": 0.6444934010505676,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 143590
+    },
+    {
+      "epoch": 944.7368421052631,
+      "grad_norm": 0.9392868876457214,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 143600
+    },
+    {
+      "epoch": 944.8026315789474,
+      "grad_norm": 1.0685558319091797,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 143610
+    },
+    {
+      "epoch": 944.8684210526316,
+      "grad_norm": 1.4054182767868042,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 143620
+    },
+    {
+      "epoch": 944.9342105263158,
+      "grad_norm": 1.2101002931594849,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 143630
+    },
+    {
+      "epoch": 945.0,
+      "grad_norm": 1.5603113174438477,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 143640
+    },
+    {
+      "epoch": 945.0657894736842,
+      "grad_norm": 1.0184569358825684,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 143650
+    },
+    {
+      "epoch": 945.1315789473684,
+      "grad_norm": 0.7555046081542969,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 143660
+    },
+    {
+      "epoch": 945.1973684210526,
+      "grad_norm": 0.8718606233596802,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 143670
+    },
+    {
+      "epoch": 945.2631578947369,
+      "grad_norm": 1.0357540845870972,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 143680
+    },
+    {
+      "epoch": 945.328947368421,
+      "grad_norm": 0.9740762710571289,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 143690
+    },
+    {
+      "epoch": 945.3947368421053,
+      "grad_norm": 0.9810712337493896,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 143700
+    },
+    {
+      "epoch": 945.4605263157895,
+      "grad_norm": 0.9435747265815735,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 143710
+    },
+    {
+      "epoch": 945.5263157894736,
+      "grad_norm": 1.1019375324249268,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 143720
+    },
+    {
+      "epoch": 945.5921052631579,
+      "grad_norm": 0.8351660370826721,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 143730
+    },
+    {
+      "epoch": 945.6578947368421,
+      "grad_norm": 0.8502940535545349,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 143740
+    },
+    {
+      "epoch": 945.7236842105264,
+      "grad_norm": 0.7618118524551392,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 143750
+    },
+    {
+      "epoch": 945.7894736842105,
+      "grad_norm": 1.3105641603469849,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 143760
+    },
+    {
+      "epoch": 945.8552631578947,
+      "grad_norm": 0.9641187787055969,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 143770
+    },
+    {
+      "epoch": 945.921052631579,
+      "grad_norm": 1.303466796875,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 143780
+    },
+    {
+      "epoch": 945.9868421052631,
+      "grad_norm": 1.175897479057312,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 143790
+    },
+    {
+      "epoch": 946.0526315789474,
+      "grad_norm": 1.1815438270568848,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 143800
+    },
+    {
+      "epoch": 946.1184210526316,
+      "grad_norm": 1.078926920890808,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 143810
+    },
+    {
+      "epoch": 946.1842105263158,
+      "grad_norm": 0.9007603526115417,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 143820
+    },
+    {
+      "epoch": 946.25,
+      "grad_norm": 1.0532934665679932,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 143830
+    },
+    {
+      "epoch": 946.3157894736842,
+      "grad_norm": 1.0850658416748047,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 143840
+    },
+    {
+      "epoch": 946.3815789473684,
+      "grad_norm": 0.9188476800918579,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 143850
+    },
+    {
+      "epoch": 946.4473684210526,
+      "grad_norm": 0.854782223701477,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 143860
+    },
+    {
+      "epoch": 946.5131578947369,
+      "grad_norm": 0.5295913815498352,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 143870
+    },
+    {
+      "epoch": 946.578947368421,
+      "grad_norm": 0.8378633856773376,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 143880
+    },
+    {
+      "epoch": 946.6447368421053,
+      "grad_norm": 1.1583908796310425,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 143890
+    },
+    {
+      "epoch": 946.7105263157895,
+      "grad_norm": 0.8956261873245239,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 143900
+    },
+    {
+      "epoch": 946.7763157894736,
+      "grad_norm": 1.1853934526443481,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 143910
+    },
+    {
+      "epoch": 946.8421052631579,
+      "grad_norm": 0.9609845876693726,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 143920
+    },
+    {
+      "epoch": 946.9078947368421,
+      "grad_norm": 0.7970736622810364,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 143930
+    },
+    {
+      "epoch": 946.9736842105264,
+      "grad_norm": 1.046696424484253,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 143940
+    },
+    {
+      "epoch": 947.0394736842105,
+      "grad_norm": 1.5037174224853516,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 143950
+    },
+    {
+      "epoch": 947.1052631578947,
+      "grad_norm": 1.2524503469467163,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 143960
+    },
+    {
+      "epoch": 947.171052631579,
+      "grad_norm": 0.7371581792831421,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 143970
+    },
+    {
+      "epoch": 947.2368421052631,
+      "grad_norm": 0.883450984954834,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 143980
+    },
+    {
+      "epoch": 947.3026315789474,
+      "grad_norm": 1.1274125576019287,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 143990
+    },
+    {
+      "epoch": 947.3684210526316,
+      "grad_norm": 1.2218645811080933,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 144000
+    },
+    {
+      "epoch": 947.4342105263158,
+      "grad_norm": 0.9843568205833435,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 144010
+    },
+    {
+      "epoch": 947.5,
+      "grad_norm": 0.8719574213027954,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 144020
+    },
+    {
+      "epoch": 947.5657894736842,
+      "grad_norm": 0.8319658637046814,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 144030
+    },
+    {
+      "epoch": 947.6315789473684,
+      "grad_norm": 0.8840770721435547,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 144040
+    },
+    {
+      "epoch": 947.6973684210526,
+      "grad_norm": 1.1723517179489136,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 144050
+    },
+    {
+      "epoch": 947.7631578947369,
+      "grad_norm": 0.925780177116394,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 144060
+    },
+    {
+      "epoch": 947.828947368421,
+      "grad_norm": 0.7539234161376953,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 144070
+    },
+    {
+      "epoch": 947.8947368421053,
+      "grad_norm": 1.0115832090377808,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 144080
+    },
+    {
+      "epoch": 947.9605263157895,
+      "grad_norm": 1.076154112815857,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 144090
+    },
+    {
+      "epoch": 948.0263157894736,
+      "grad_norm": 1.1594144105911255,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 144100
+    },
+    {
+      "epoch": 948.0921052631579,
+      "grad_norm": 0.8752143383026123,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 144110
+    },
+    {
+      "epoch": 948.1578947368421,
+      "grad_norm": 0.816047191619873,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 144120
+    },
+    {
+      "epoch": 948.2236842105264,
+      "grad_norm": 0.9532915949821472,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 144130
+    },
+    {
+      "epoch": 948.2894736842105,
+      "grad_norm": 1.082932710647583,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 144140
+    },
+    {
+      "epoch": 948.3552631578947,
+      "grad_norm": 0.8170779347419739,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 144150
+    },
+    {
+      "epoch": 948.421052631579,
+      "grad_norm": 1.189034104347229,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 144160
+    },
+    {
+      "epoch": 948.4868421052631,
+      "grad_norm": 1.0919978618621826,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 144170
+    },
+    {
+      "epoch": 948.5526315789474,
+      "grad_norm": 1.05934476852417,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 144180
+    },
+    {
+      "epoch": 948.6184210526316,
+      "grad_norm": 0.8420997858047485,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 144190
+    },
+    {
+      "epoch": 948.6842105263158,
+      "grad_norm": 1.1011571884155273,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 144200
+    },
+    {
+      "epoch": 948.75,
+      "grad_norm": 0.7747015953063965,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 144210
+    },
+    {
+      "epoch": 948.8157894736842,
+      "grad_norm": 1.1915805339813232,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 144220
+    },
+    {
+      "epoch": 948.8815789473684,
+      "grad_norm": 1.0432209968566895,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 144230
+    },
+    {
+      "epoch": 948.9473684210526,
+      "grad_norm": 0.9377793669700623,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 144240
+    },
+    {
+      "epoch": 949.0131578947369,
+      "grad_norm": 0.8904802203178406,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 144250
+    },
+    {
+      "epoch": 949.078947368421,
+      "grad_norm": 1.0895966291427612,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 144260
+    },
+    {
+      "epoch": 949.1447368421053,
+      "grad_norm": 1.2289841175079346,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 144270
+    },
+    {
+      "epoch": 949.2105263157895,
+      "grad_norm": 0.9211072325706482,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 144280
+    },
+    {
+      "epoch": 949.2763157894736,
+      "grad_norm": 1.0456809997558594,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 144290
+    },
+    {
+      "epoch": 949.3421052631579,
+      "grad_norm": 0.9694221019744873,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 144300
+    },
+    {
+      "epoch": 949.4078947368421,
+      "grad_norm": 0.9555386304855347,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 144310
+    },
+    {
+      "epoch": 949.4736842105264,
+      "grad_norm": 0.9614490866661072,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 144320
+    },
+    {
+      "epoch": 949.5394736842105,
+      "grad_norm": 1.557354211807251,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 144330
+    },
+    {
+      "epoch": 949.6052631578947,
+      "grad_norm": 1.0724836587905884,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 144340
+    },
+    {
+      "epoch": 949.671052631579,
+      "grad_norm": 1.0988471508026123,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 144350
+    },
+    {
+      "epoch": 949.7368421052631,
+      "grad_norm": 1.2191790342330933,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 144360
+    },
+    {
+      "epoch": 949.8026315789474,
+      "grad_norm": 1.3018871545791626,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 144370
+    },
+    {
+      "epoch": 949.8684210526316,
+      "grad_norm": 0.7598888874053955,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 144380
+    },
+    {
+      "epoch": 949.9342105263158,
+      "grad_norm": 0.9216148257255554,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 144390
+    },
+    {
+      "epoch": 950.0,
+      "grad_norm": 1.0104155540466309,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 144400
+    },
+    {
+      "epoch": 950.0657894736842,
+      "grad_norm": 1.0783321857452393,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 144410
+    },
+    {
+      "epoch": 950.1315789473684,
+      "grad_norm": 1.3584924936294556,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 144420
+    },
+    {
+      "epoch": 950.1973684210526,
+      "grad_norm": 1.443320393562317,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 144430
+    },
+    {
+      "epoch": 950.2631578947369,
+      "grad_norm": 1.016124963760376,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 144440
+    },
+    {
+      "epoch": 950.328947368421,
+      "grad_norm": 1.2364863157272339,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 144450
+    },
+    {
+      "epoch": 950.3947368421053,
+      "grad_norm": 1.195246696472168,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 144460
+    },
+    {
+      "epoch": 950.4605263157895,
+      "grad_norm": 1.5904250144958496,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 144470
+    },
+    {
+      "epoch": 950.5263157894736,
+      "grad_norm": 1.104093074798584,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 144480
+    },
+    {
+      "epoch": 950.5921052631579,
+      "grad_norm": 1.0603477954864502,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 144490
+    },
+    {
+      "epoch": 950.6578947368421,
+      "grad_norm": 1.1990110874176025,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 144500
+    },
+    {
+      "epoch": 950.7236842105264,
+      "grad_norm": 1.1858569383621216,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 144510
+    },
+    {
+      "epoch": 950.7894736842105,
+      "grad_norm": 0.9741637706756592,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 144520
+    },
+    {
+      "epoch": 950.8552631578947,
+      "grad_norm": 0.9094181656837463,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 144530
+    },
+    {
+      "epoch": 950.921052631579,
+      "grad_norm": 1.022029995918274,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 144540
+    },
+    {
+      "epoch": 950.9868421052631,
+      "grad_norm": 0.9624213576316833,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 144550
+    },
+    {
+      "epoch": 951.0526315789474,
+      "grad_norm": 1.2031329870224,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 144560
+    },
+    {
+      "epoch": 951.1184210526316,
+      "grad_norm": 1.1659698486328125,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 144570
+    },
+    {
+      "epoch": 951.1842105263158,
+      "grad_norm": 1.0349775552749634,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 144580
+    },
+    {
+      "epoch": 951.25,
+      "grad_norm": 1.165041208267212,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 144590
+    },
+    {
+      "epoch": 951.3157894736842,
+      "grad_norm": 1.2488925457000732,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 144600
+    },
+    {
+      "epoch": 951.3815789473684,
+      "grad_norm": 1.0464668273925781,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 144610
+    },
+    {
+      "epoch": 951.4473684210526,
+      "grad_norm": 0.7799762487411499,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 144620
+    },
+    {
+      "epoch": 951.5131578947369,
+      "grad_norm": 0.6983115077018738,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 144630
+    },
+    {
+      "epoch": 951.578947368421,
+      "grad_norm": 1.1071887016296387,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 144640
+    },
+    {
+      "epoch": 951.6447368421053,
+      "grad_norm": 0.8498964309692383,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 144650
+    },
+    {
+      "epoch": 951.7105263157895,
+      "grad_norm": 0.8821257948875427,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 144660
+    },
+    {
+      "epoch": 951.7763157894736,
+      "grad_norm": 0.9334793090820312,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 144670
+    },
+    {
+      "epoch": 951.8421052631579,
+      "grad_norm": 1.2155643701553345,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 144680
+    },
+    {
+      "epoch": 951.9078947368421,
+      "grad_norm": 0.8632481694221497,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 144690
+    },
+    {
+      "epoch": 951.9736842105264,
+      "grad_norm": 1.1241223812103271,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 144700
+    },
+    {
+      "epoch": 952.0394736842105,
+      "grad_norm": 1.0715571641921997,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 144710
+    },
+    {
+      "epoch": 952.1052631578947,
+      "grad_norm": 1.1770998239517212,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 144720
+    },
+    {
+      "epoch": 952.171052631579,
+      "grad_norm": 1.2517147064208984,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 144730
+    },
+    {
+      "epoch": 952.2368421052631,
+      "grad_norm": 1.0367016792297363,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 144740
+    },
+    {
+      "epoch": 952.3026315789474,
+      "grad_norm": 0.8607826232910156,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 144750
+    },
+    {
+      "epoch": 952.3684210526316,
+      "grad_norm": 1.0406628847122192,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 144760
+    },
+    {
+      "epoch": 952.4342105263158,
+      "grad_norm": 0.97654128074646,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 144770
+    },
+    {
+      "epoch": 952.5,
+      "grad_norm": 0.7601302266120911,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 144780
+    },
+    {
+      "epoch": 952.5657894736842,
+      "grad_norm": 0.7820169925689697,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 144790
+    },
+    {
+      "epoch": 952.6315789473684,
+      "grad_norm": 0.8376019597053528,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 144800
+    },
+    {
+      "epoch": 952.6973684210526,
+      "grad_norm": 1.023985505104065,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 144810
+    },
+    {
+      "epoch": 952.7631578947369,
+      "grad_norm": 0.743279218673706,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 144820
+    },
+    {
+      "epoch": 952.828947368421,
+      "grad_norm": 1.3411636352539062,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 144830
+    },
+    {
+      "epoch": 952.8947368421053,
+      "grad_norm": 1.0974597930908203,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 144840
+    },
+    {
+      "epoch": 952.9605263157895,
+      "grad_norm": 1.0889559984207153,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 144850
+    },
+    {
+      "epoch": 953.0263157894736,
+      "grad_norm": 1.1542203426361084,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 144860
+    },
+    {
+      "epoch": 953.0921052631579,
+      "grad_norm": 1.394636631011963,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 144870
+    },
+    {
+      "epoch": 953.1578947368421,
+      "grad_norm": 1.16937255859375,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 144880
+    },
+    {
+      "epoch": 953.2236842105264,
+      "grad_norm": 1.2651804685592651,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 144890
+    },
+    {
+      "epoch": 953.2894736842105,
+      "grad_norm": 1.2077585458755493,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 144900
+    },
+    {
+      "epoch": 953.3552631578947,
+      "grad_norm": 1.1569517850875854,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 144910
+    },
+    {
+      "epoch": 953.421052631579,
+      "grad_norm": 1.34952712059021,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 144920
+    },
+    {
+      "epoch": 953.4868421052631,
+      "grad_norm": 1.0990208387374878,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 144930
+    },
+    {
+      "epoch": 953.5526315789474,
+      "grad_norm": 0.9614241719245911,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 144940
+    },
+    {
+      "epoch": 953.6184210526316,
+      "grad_norm": 1.349439024925232,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 144950
+    },
+    {
+      "epoch": 953.6842105263158,
+      "grad_norm": 1.501696228981018,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 144960
+    },
+    {
+      "epoch": 953.75,
+      "grad_norm": 1.1203855276107788,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 144970
+    },
+    {
+      "epoch": 953.8157894736842,
+      "grad_norm": 1.2413620948791504,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 144980
+    },
+    {
+      "epoch": 953.8815789473684,
+      "grad_norm": 1.3755046129226685,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 144990
+    },
+    {
+      "epoch": 953.9473684210526,
+      "grad_norm": 1.0295451879501343,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 145000
+    },
+    {
+      "epoch": 954.0131578947369,
+      "grad_norm": 1.3562036752700806,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 145010
+    },
+    {
+      "epoch": 954.078947368421,
+      "grad_norm": 1.4135730266571045,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 145020
+    },
+    {
+      "epoch": 954.1447368421053,
+      "grad_norm": 1.433528184890747,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 145030
+    },
+    {
+      "epoch": 954.2105263157895,
+      "grad_norm": 1.3277394771575928,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 145040
+    },
+    {
+      "epoch": 954.2763157894736,
+      "grad_norm": 0.9119917154312134,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 145050
+    },
+    {
+      "epoch": 954.3421052631579,
+      "grad_norm": 1.1164143085479736,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 145060
+    },
+    {
+      "epoch": 954.4078947368421,
+      "grad_norm": 1.1159082651138306,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 145070
+    },
+    {
+      "epoch": 954.4736842105264,
+      "grad_norm": 1.0839544534683228,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 145080
+    },
+    {
+      "epoch": 954.5394736842105,
+      "grad_norm": 0.9123933911323547,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 145090
+    },
+    {
+      "epoch": 954.6052631578947,
+      "grad_norm": 0.9606594443321228,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 145100
+    },
+    {
+      "epoch": 954.671052631579,
+      "grad_norm": 0.8391530513763428,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 145110
+    },
+    {
+      "epoch": 954.7368421052631,
+      "grad_norm": 0.8009461164474487,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 145120
+    },
+    {
+      "epoch": 954.8026315789474,
+      "grad_norm": 1.1401244401931763,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 145130
+    },
+    {
+      "epoch": 954.8684210526316,
+      "grad_norm": 1.1111904382705688,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 145140
+    },
+    {
+      "epoch": 954.9342105263158,
+      "grad_norm": 0.8661221265792847,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 145150
+    },
+    {
+      "epoch": 955.0,
+      "grad_norm": 1.0058561563491821,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 145160
+    },
+    {
+      "epoch": 955.0657894736842,
+      "grad_norm": 1.0411880016326904,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 145170
+    },
+    {
+      "epoch": 955.1315789473684,
+      "grad_norm": 1.146001935005188,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 145180
+    },
+    {
+      "epoch": 955.1973684210526,
+      "grad_norm": 1.115849256515503,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 145190
+    },
+    {
+      "epoch": 955.2631578947369,
+      "grad_norm": 0.991651713848114,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 145200
+    },
+    {
+      "epoch": 955.328947368421,
+      "grad_norm": 1.1527934074401855,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 145210
+    },
+    {
+      "epoch": 955.3947368421053,
+      "grad_norm": 1.2311185598373413,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 145220
+    },
+    {
+      "epoch": 955.4605263157895,
+      "grad_norm": 1.1631335020065308,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 145230
+    },
+    {
+      "epoch": 955.5263157894736,
+      "grad_norm": 0.657287061214447,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 145240
+    },
+    {
+      "epoch": 955.5921052631579,
+      "grad_norm": 0.9889928698539734,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 145250
+    },
+    {
+      "epoch": 955.6578947368421,
+      "grad_norm": 0.9157257676124573,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 145260
+    },
+    {
+      "epoch": 955.7236842105264,
+      "grad_norm": 0.983443558216095,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 145270
+    },
+    {
+      "epoch": 955.7894736842105,
+      "grad_norm": 1.1161870956420898,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 145280
+    },
+    {
+      "epoch": 955.8552631578947,
+      "grad_norm": 0.9558767080307007,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 145290
+    },
+    {
+      "epoch": 955.921052631579,
+      "grad_norm": 1.0487935543060303,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 145300
+    },
+    {
+      "epoch": 955.9868421052631,
+      "grad_norm": 0.7724936604499817,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 145310
+    },
+    {
+      "epoch": 956.0526315789474,
+      "grad_norm": 1.0620529651641846,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 145320
+    },
+    {
+      "epoch": 956.1184210526316,
+      "grad_norm": 1.1085588932037354,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 145330
+    },
+    {
+      "epoch": 956.1842105263158,
+      "grad_norm": 1.2441157102584839,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 145340
+    },
+    {
+      "epoch": 956.25,
+      "grad_norm": 1.0353556871414185,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 145350
+    },
+    {
+      "epoch": 956.3157894736842,
+      "grad_norm": 1.4679001569747925,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 145360
+    },
+    {
+      "epoch": 956.3815789473684,
+      "grad_norm": 1.2972333431243896,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 145370
+    },
+    {
+      "epoch": 956.4473684210526,
+      "grad_norm": 0.9384067058563232,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 145380
+    },
+    {
+      "epoch": 956.5131578947369,
+      "grad_norm": 1.0650990009307861,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 145390
+    },
+    {
+      "epoch": 956.578947368421,
+      "grad_norm": 1.5110204219818115,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 145400
+    },
+    {
+      "epoch": 956.6447368421053,
+      "grad_norm": 0.7420325875282288,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 145410
+    },
+    {
+      "epoch": 956.7105263157895,
+      "grad_norm": 1.3069480657577515,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 145420
+    },
+    {
+      "epoch": 956.7763157894736,
+      "grad_norm": 1.2609871625900269,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 145430
+    },
+    {
+      "epoch": 956.8421052631579,
+      "grad_norm": 1.0048247575759888,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 145440
+    },
+    {
+      "epoch": 956.9078947368421,
+      "grad_norm": 1.0545076131820679,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 145450
+    },
+    {
+      "epoch": 956.9736842105264,
+      "grad_norm": 0.8621100187301636,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 145460
+    },
+    {
+      "epoch": 957.0394736842105,
+      "grad_norm": 1.0694085359573364,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 145470
+    },
+    {
+      "epoch": 957.1052631578947,
+      "grad_norm": 1.2032498121261597,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 145480
+    },
+    {
+      "epoch": 957.171052631579,
+      "grad_norm": 0.7613986134529114,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 145490
+    },
+    {
+      "epoch": 957.2368421052631,
+      "grad_norm": 0.8016976714134216,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 145500
+    },
+    {
+      "epoch": 957.3026315789474,
+      "grad_norm": 1.1267141103744507,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 145510
+    },
+    {
+      "epoch": 957.3684210526316,
+      "grad_norm": 1.0526936054229736,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 145520
+    },
+    {
+      "epoch": 957.4342105263158,
+      "grad_norm": 1.1312087774276733,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 145530
+    },
+    {
+      "epoch": 957.5,
+      "grad_norm": 0.8225216865539551,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 145540
+    },
+    {
+      "epoch": 957.5657894736842,
+      "grad_norm": 1.3235195875167847,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 145550
+    },
+    {
+      "epoch": 957.6315789473684,
+      "grad_norm": 0.9246519804000854,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 145560
+    },
+    {
+      "epoch": 957.6973684210526,
+      "grad_norm": 1.227251648902893,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 145570
+    },
+    {
+      "epoch": 957.7631578947369,
+      "grad_norm": 1.1541143655776978,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 145580
+    },
+    {
+      "epoch": 957.828947368421,
+      "grad_norm": 1.001869797706604,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 145590
+    },
+    {
+      "epoch": 957.8947368421053,
+      "grad_norm": 1.1103819608688354,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 145600
+    },
+    {
+      "epoch": 957.9605263157895,
+      "grad_norm": 1.1512665748596191,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 145610
+    },
+    {
+      "epoch": 958.0263157894736,
+      "grad_norm": 1.135569453239441,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 145620
+    },
+    {
+      "epoch": 958.0921052631579,
+      "grad_norm": 0.9894860982894897,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 145630
+    },
+    {
+      "epoch": 958.1578947368421,
+      "grad_norm": 1.2264050245285034,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 145640
+    },
+    {
+      "epoch": 958.2236842105264,
+      "grad_norm": 0.9081289768218994,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 145650
+    },
+    {
+      "epoch": 958.2894736842105,
+      "grad_norm": 1.1225621700286865,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 145660
+    },
+    {
+      "epoch": 958.3552631578947,
+      "grad_norm": 1.4110243320465088,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 145670
+    },
+    {
+      "epoch": 958.421052631579,
+      "grad_norm": 1.2238539457321167,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 145680
+    },
+    {
+      "epoch": 958.4868421052631,
+      "grad_norm": 0.7691843509674072,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 145690
+    },
+    {
+      "epoch": 958.5526315789474,
+      "grad_norm": 1.0191400051116943,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 145700
+    },
+    {
+      "epoch": 958.6184210526316,
+      "grad_norm": 1.1331918239593506,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 145710
+    },
+    {
+      "epoch": 958.6842105263158,
+      "grad_norm": 1.3987983465194702,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 145720
+    },
+    {
+      "epoch": 958.75,
+      "grad_norm": 1.1458479166030884,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 145730
+    },
+    {
+      "epoch": 958.8157894736842,
+      "grad_norm": 0.9019910097122192,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 145740
+    },
+    {
+      "epoch": 958.8815789473684,
+      "grad_norm": 1.2355693578720093,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 145750
+    },
+    {
+      "epoch": 958.9473684210526,
+      "grad_norm": 1.0789332389831543,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 145760
+    },
+    {
+      "epoch": 959.0131578947369,
+      "grad_norm": 1.0482940673828125,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 145770
+    },
+    {
+      "epoch": 959.078947368421,
+      "grad_norm": 1.026578426361084,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 145780
+    },
+    {
+      "epoch": 959.1447368421053,
+      "grad_norm": 1.144579529762268,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 145790
+    },
+    {
+      "epoch": 959.2105263157895,
+      "grad_norm": 1.0373647212982178,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 145800
+    },
+    {
+      "epoch": 959.2763157894736,
+      "grad_norm": 1.0401736497879028,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 145810
+    },
+    {
+      "epoch": 959.3421052631579,
+      "grad_norm": 1.3450522422790527,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 145820
+    },
+    {
+      "epoch": 959.4078947368421,
+      "grad_norm": 0.786882221698761,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 145830
+    },
+    {
+      "epoch": 959.4736842105264,
+      "grad_norm": 1.122706413269043,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 145840
+    },
+    {
+      "epoch": 959.5394736842105,
+      "grad_norm": 1.224389672279358,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 145850
+    },
+    {
+      "epoch": 959.6052631578947,
+      "grad_norm": 0.8540865778923035,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 145860
+    },
+    {
+      "epoch": 959.671052631579,
+      "grad_norm": 1.0747767686843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 145870
+    },
+    {
+      "epoch": 959.7368421052631,
+      "grad_norm": 0.7939700484275818,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 145880
+    },
+    {
+      "epoch": 959.8026315789474,
+      "grad_norm": 0.817993700504303,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 145890
+    },
+    {
+      "epoch": 959.8684210526316,
+      "grad_norm": 0.7488849759101868,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 145900
+    },
+    {
+      "epoch": 959.9342105263158,
+      "grad_norm": 1.1702375411987305,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 145910
+    },
+    {
+      "epoch": 960.0,
+      "grad_norm": 1.0733966827392578,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 145920
+    },
+    {
+      "epoch": 960.0657894736842,
+      "grad_norm": 0.836497962474823,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 145930
+    },
+    {
+      "epoch": 960.1315789473684,
+      "grad_norm": 1.1901671886444092,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 145940
+    },
+    {
+      "epoch": 960.1973684210526,
+      "grad_norm": 1.657729148864746,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 145950
+    },
+    {
+      "epoch": 960.2631578947369,
+      "grad_norm": 1.0714272260665894,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 145960
+    },
+    {
+      "epoch": 960.328947368421,
+      "grad_norm": 1.1627241373062134,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 145970
+    },
+    {
+      "epoch": 960.3947368421053,
+      "grad_norm": 0.7160118222236633,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 145980
+    },
+    {
+      "epoch": 960.4605263157895,
+      "grad_norm": 1.2324270009994507,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 145990
+    },
+    {
+      "epoch": 960.5263157894736,
+      "grad_norm": 0.9793133735656738,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 146000
+    },
+    {
+      "epoch": 960.5921052631579,
+      "grad_norm": 1.2766015529632568,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 146010
+    },
+    {
+      "epoch": 960.6578947368421,
+      "grad_norm": 0.7441828846931458,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 146020
+    },
+    {
+      "epoch": 960.7236842105264,
+      "grad_norm": 1.006022572517395,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 146030
+    },
+    {
+      "epoch": 960.7894736842105,
+      "grad_norm": 0.781061053276062,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 146040
+    },
+    {
+      "epoch": 960.8552631578947,
+      "grad_norm": 1.0156453847885132,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 146050
+    },
+    {
+      "epoch": 960.921052631579,
+      "grad_norm": 1.5238510370254517,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 146060
+    },
+    {
+      "epoch": 960.9868421052631,
+      "grad_norm": 1.1294466257095337,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 146070
+    },
+    {
+      "epoch": 961.0526315789474,
+      "grad_norm": 1.103088140487671,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 146080
+    },
+    {
+      "epoch": 961.1184210526316,
+      "grad_norm": 1.1292765140533447,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 146090
+    },
+    {
+      "epoch": 961.1842105263158,
+      "grad_norm": 1.060682773590088,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 146100
+    },
+    {
+      "epoch": 961.25,
+      "grad_norm": 0.584579348564148,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 146110
+    },
+    {
+      "epoch": 961.3157894736842,
+      "grad_norm": 0.8006397485733032,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 146120
+    },
+    {
+      "epoch": 961.3815789473684,
+      "grad_norm": 0.8636497855186462,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 146130
+    },
+    {
+      "epoch": 961.4473684210526,
+      "grad_norm": 0.9201158881187439,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 146140
+    },
+    {
+      "epoch": 961.5131578947369,
+      "grad_norm": 1.3234913349151611,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 146150
+    },
+    {
+      "epoch": 961.578947368421,
+      "grad_norm": 0.9483774900436401,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 146160
+    },
+    {
+      "epoch": 961.6447368421053,
+      "grad_norm": 0.9402949213981628,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 146170
+    },
+    {
+      "epoch": 961.7105263157895,
+      "grad_norm": 0.809597909450531,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 146180
+    },
+    {
+      "epoch": 961.7763157894736,
+      "grad_norm": 1.229201078414917,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 146190
+    },
+    {
+      "epoch": 961.8421052631579,
+      "grad_norm": 1.1061270236968994,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 146200
+    },
+    {
+      "epoch": 961.9078947368421,
+      "grad_norm": 0.8046864867210388,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 146210
+    },
+    {
+      "epoch": 961.9736842105264,
+      "grad_norm": 0.7022920846939087,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 146220
+    },
+    {
+      "epoch": 962.0394736842105,
+      "grad_norm": 1.0460927486419678,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 146230
+    },
+    {
+      "epoch": 962.1052631578947,
+      "grad_norm": 1.061414361000061,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 146240
+    },
+    {
+      "epoch": 962.171052631579,
+      "grad_norm": 0.8802205324172974,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 146250
+    },
+    {
+      "epoch": 962.2368421052631,
+      "grad_norm": 1.0647673606872559,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 146260
+    },
+    {
+      "epoch": 962.3026315789474,
+      "grad_norm": 1.1186468601226807,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 146270
+    },
+    {
+      "epoch": 962.3684210526316,
+      "grad_norm": 1.0472915172576904,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 146280
+    },
+    {
+      "epoch": 962.4342105263158,
+      "grad_norm": 0.8644057512283325,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 146290
+    },
+    {
+      "epoch": 962.5,
+      "grad_norm": 0.8172399997711182,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 146300
+    },
+    {
+      "epoch": 962.5657894736842,
+      "grad_norm": 0.9596365094184875,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 146310
+    },
+    {
+      "epoch": 962.6315789473684,
+      "grad_norm": 0.8653599619865417,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 146320
+    },
+    {
+      "epoch": 962.6973684210526,
+      "grad_norm": 0.9772462844848633,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 146330
+    },
+    {
+      "epoch": 962.7631578947369,
+      "grad_norm": 1.0098869800567627,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 146340
+    },
+    {
+      "epoch": 962.828947368421,
+      "grad_norm": 1.2209604978561401,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 146350
+    },
+    {
+      "epoch": 962.8947368421053,
+      "grad_norm": 0.9694502949714661,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 146360
+    },
+    {
+      "epoch": 962.9605263157895,
+      "grad_norm": 0.981524646282196,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 146370
+    },
+    {
+      "epoch": 963.0263157894736,
+      "grad_norm": 1.0773963928222656,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 146380
+    },
+    {
+      "epoch": 963.0921052631579,
+      "grad_norm": 1.1981137990951538,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 146390
+    },
+    {
+      "epoch": 963.1578947368421,
+      "grad_norm": 1.1366533041000366,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 146400
+    },
+    {
+      "epoch": 963.2236842105264,
+      "grad_norm": 1.1924341917037964,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 146410
+    },
+    {
+      "epoch": 963.2894736842105,
+      "grad_norm": 0.9857580065727234,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 146420
+    },
+    {
+      "epoch": 963.3552631578947,
+      "grad_norm": 1.2033591270446777,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 146430
+    },
+    {
+      "epoch": 963.421052631579,
+      "grad_norm": 1.2146192789077759,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 146440
+    },
+    {
+      "epoch": 963.4868421052631,
+      "grad_norm": 1.1512936353683472,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 146450
+    },
+    {
+      "epoch": 963.5526315789474,
+      "grad_norm": 0.6414240598678589,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 146460
+    },
+    {
+      "epoch": 963.6184210526316,
+      "grad_norm": 0.7209314107894897,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 146470
+    },
+    {
+      "epoch": 963.6842105263158,
+      "grad_norm": 0.8891003727912903,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 146480
+    },
+    {
+      "epoch": 963.75,
+      "grad_norm": 0.7356894016265869,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 146490
+    },
+    {
+      "epoch": 963.8157894736842,
+      "grad_norm": 0.9845222234725952,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 146500
+    },
+    {
+      "epoch": 963.8815789473684,
+      "grad_norm": 1.4602724313735962,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 146510
+    },
+    {
+      "epoch": 963.9473684210526,
+      "grad_norm": 0.9599493741989136,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 146520
+    },
+    {
+      "epoch": 964.0131578947369,
+      "grad_norm": 1.1864409446716309,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 146530
+    },
+    {
+      "epoch": 964.078947368421,
+      "grad_norm": 1.6686468124389648,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 146540
+    },
+    {
+      "epoch": 964.1447368421053,
+      "grad_norm": 1.6375855207443237,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 146550
+    },
+    {
+      "epoch": 964.2105263157895,
+      "grad_norm": 1.250057339668274,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 146560
+    },
+    {
+      "epoch": 964.2763157894736,
+      "grad_norm": 1.3468494415283203,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 146570
+    },
+    {
+      "epoch": 964.3421052631579,
+      "grad_norm": 1.369868516921997,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 146580
+    },
+    {
+      "epoch": 964.4078947368421,
+      "grad_norm": 0.8721538186073303,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 146590
+    },
+    {
+      "epoch": 964.4736842105264,
+      "grad_norm": 1.0958600044250488,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 146600
+    },
+    {
+      "epoch": 964.5394736842105,
+      "grad_norm": 1.1461358070373535,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 146610
+    },
+    {
+      "epoch": 964.6052631578947,
+      "grad_norm": 1.3515912294387817,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 146620
+    },
+    {
+      "epoch": 964.671052631579,
+      "grad_norm": 1.3492871522903442,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 146630
+    },
+    {
+      "epoch": 964.7368421052631,
+      "grad_norm": 0.7146100997924805,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 146640
+    },
+    {
+      "epoch": 964.8026315789474,
+      "grad_norm": 1.150172233581543,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 146650
+    },
+    {
+      "epoch": 964.8684210526316,
+      "grad_norm": 1.4243590831756592,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 146660
+    },
+    {
+      "epoch": 964.9342105263158,
+      "grad_norm": 1.0917390584945679,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 146670
+    },
+    {
+      "epoch": 965.0,
+      "grad_norm": 0.888741135597229,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 146680
+    },
+    {
+      "epoch": 965.0657894736842,
+      "grad_norm": 0.8921936750411987,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 146690
+    },
+    {
+      "epoch": 965.1315789473684,
+      "grad_norm": 1.1336225271224976,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 146700
+    },
+    {
+      "epoch": 965.1973684210526,
+      "grad_norm": 1.2168223857879639,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 146710
+    },
+    {
+      "epoch": 965.2631578947369,
+      "grad_norm": 0.9628434181213379,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 146720
+    },
+    {
+      "epoch": 965.328947368421,
+      "grad_norm": 0.9987488389015198,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 146730
+    },
+    {
+      "epoch": 965.3947368421053,
+      "grad_norm": 1.300894856452942,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 146740
+    },
+    {
+      "epoch": 965.4605263157895,
+      "grad_norm": 3.1058502197265625,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 146750
+    },
+    {
+      "epoch": 965.5263157894736,
+      "grad_norm": 1.6994332075119019,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 146760
+    },
+    {
+      "epoch": 965.5921052631579,
+      "grad_norm": 0.7340602874755859,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 146770
+    },
+    {
+      "epoch": 965.6578947368421,
+      "grad_norm": 0.9732682108879089,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 146780
+    },
+    {
+      "epoch": 965.7236842105264,
+      "grad_norm": 1.0283761024475098,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 146790
+    },
+    {
+      "epoch": 965.7894736842105,
+      "grad_norm": 0.8493642807006836,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 146800
+    },
+    {
+      "epoch": 965.8552631578947,
+      "grad_norm": 0.8938899636268616,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 146810
+    },
+    {
+      "epoch": 965.921052631579,
+      "grad_norm": 0.8142083287239075,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 146820
+    },
+    {
+      "epoch": 965.9868421052631,
+      "grad_norm": 0.9816213250160217,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 146830
+    },
+    {
+      "epoch": 966.0526315789474,
+      "grad_norm": 0.7372590899467468,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 146840
+    },
+    {
+      "epoch": 966.1184210526316,
+      "grad_norm": 1.1608000993728638,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 146850
+    },
+    {
+      "epoch": 966.1842105263158,
+      "grad_norm": 1.0333927869796753,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 146860
+    },
+    {
+      "epoch": 966.25,
+      "grad_norm": 1.392578125,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 146870
+    },
+    {
+      "epoch": 966.3157894736842,
+      "grad_norm": 1.003764271736145,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 146880
+    },
+    {
+      "epoch": 966.3815789473684,
+      "grad_norm": 1.026443362236023,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 146890
+    },
+    {
+      "epoch": 966.4473684210526,
+      "grad_norm": 0.8039220571517944,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 146900
+    },
+    {
+      "epoch": 966.5131578947369,
+      "grad_norm": 1.3538212776184082,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 146910
+    },
+    {
+      "epoch": 966.578947368421,
+      "grad_norm": 1.2053329944610596,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 146920
+    },
+    {
+      "epoch": 966.6447368421053,
+      "grad_norm": 1.1715484857559204,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 146930
+    },
+    {
+      "epoch": 966.7105263157895,
+      "grad_norm": 1.387887716293335,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 146940
+    },
+    {
+      "epoch": 966.7763157894736,
+      "grad_norm": 0.7918421626091003,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 146950
+    },
+    {
+      "epoch": 966.8421052631579,
+      "grad_norm": 1.122507929801941,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 146960
+    },
+    {
+      "epoch": 966.9078947368421,
+      "grad_norm": 1.20033597946167,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 146970
+    },
+    {
+      "epoch": 966.9736842105264,
+      "grad_norm": 1.156383991241455,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 146980
+    },
+    {
+      "epoch": 967.0394736842105,
+      "grad_norm": 0.760665774345398,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 146990
+    },
+    {
+      "epoch": 967.1052631578947,
+      "grad_norm": 0.9186726808547974,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 147000
+    },
+    {
+      "epoch": 967.171052631579,
+      "grad_norm": 1.0184344053268433,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 147010
+    },
+    {
+      "epoch": 967.2368421052631,
+      "grad_norm": 1.3089451789855957,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 147020
+    },
+    {
+      "epoch": 967.3026315789474,
+      "grad_norm": 0.8527283072471619,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 147030
+    },
+    {
+      "epoch": 967.3684210526316,
+      "grad_norm": 1.0763221979141235,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 147040
+    },
+    {
+      "epoch": 967.4342105263158,
+      "grad_norm": 1.0858137607574463,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 147050
+    },
+    {
+      "epoch": 967.5,
+      "grad_norm": 0.9243344068527222,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 147060
+    },
+    {
+      "epoch": 967.5657894736842,
+      "grad_norm": 0.9228748083114624,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 147070
+    },
+    {
+      "epoch": 967.6315789473684,
+      "grad_norm": 1.0944608449935913,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 147080
+    },
+    {
+      "epoch": 967.6973684210526,
+      "grad_norm": 0.9257031679153442,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 147090
+    },
+    {
+      "epoch": 967.7631578947369,
+      "grad_norm": 1.181267261505127,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 147100
+    },
+    {
+      "epoch": 967.828947368421,
+      "grad_norm": 1.0577452182769775,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 147110
+    },
+    {
+      "epoch": 967.8947368421053,
+      "grad_norm": 0.8401322960853577,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 147120
+    },
+    {
+      "epoch": 967.9605263157895,
+      "grad_norm": 0.9169796705245972,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 147130
+    },
+    {
+      "epoch": 968.0263157894736,
+      "grad_norm": 0.8156391978263855,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 147140
+    },
+    {
+      "epoch": 968.0921052631579,
+      "grad_norm": 1.2525750398635864,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 147150
+    },
+    {
+      "epoch": 968.1578947368421,
+      "grad_norm": 1.0658366680145264,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 147160
+    },
+    {
+      "epoch": 968.2236842105264,
+      "grad_norm": 0.9549158215522766,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 147170
+    },
+    {
+      "epoch": 968.2894736842105,
+      "grad_norm": 1.1841901540756226,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 147180
+    },
+    {
+      "epoch": 968.3552631578947,
+      "grad_norm": 0.855679452419281,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 147190
+    },
+    {
+      "epoch": 968.421052631579,
+      "grad_norm": 1.03905189037323,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 147200
+    },
+    {
+      "epoch": 968.4868421052631,
+      "grad_norm": 0.7143714427947998,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 147210
+    },
+    {
+      "epoch": 968.5526315789474,
+      "grad_norm": 0.8746243715286255,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 147220
+    },
+    {
+      "epoch": 968.6184210526316,
+      "grad_norm": 0.8814296126365662,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 147230
+    },
+    {
+      "epoch": 968.6842105263158,
+      "grad_norm": 0.7824357151985168,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 147240
+    },
+    {
+      "epoch": 968.75,
+      "grad_norm": 0.8747170567512512,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 147250
+    },
+    {
+      "epoch": 968.8157894736842,
+      "grad_norm": 1.2082691192626953,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 147260
+    },
+    {
+      "epoch": 968.8815789473684,
+      "grad_norm": 1.2725180387496948,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 147270
+    },
+    {
+      "epoch": 968.9473684210526,
+      "grad_norm": 1.4602564573287964,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 147280
+    },
+    {
+      "epoch": 969.0131578947369,
+      "grad_norm": 0.9811068773269653,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 147290
+    },
+    {
+      "epoch": 969.078947368421,
+      "grad_norm": 1.359781265258789,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 147300
+    },
+    {
+      "epoch": 969.1447368421053,
+      "grad_norm": 1.2391654253005981,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 147310
+    },
+    {
+      "epoch": 969.2105263157895,
+      "grad_norm": 1.193811297416687,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 147320
+    },
+    {
+      "epoch": 969.2763157894736,
+      "grad_norm": 0.9369627237319946,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 147330
+    },
+    {
+      "epoch": 969.3421052631579,
+      "grad_norm": 1.2277706861495972,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 147340
+    },
+    {
+      "epoch": 969.4078947368421,
+      "grad_norm": 0.6797551512718201,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 147350
+    },
+    {
+      "epoch": 969.4736842105264,
+      "grad_norm": 0.8890407085418701,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 147360
+    },
+    {
+      "epoch": 969.5394736842105,
+      "grad_norm": 0.6497656106948853,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 147370
+    },
+    {
+      "epoch": 969.6052631578947,
+      "grad_norm": 1.0025545358657837,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 147380
+    },
+    {
+      "epoch": 969.671052631579,
+      "grad_norm": 1.0839303731918335,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 147390
+    },
+    {
+      "epoch": 969.7368421052631,
+      "grad_norm": 0.9364657402038574,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 147400
+    },
+    {
+      "epoch": 969.8026315789474,
+      "grad_norm": 0.725238561630249,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 147410
+    },
+    {
+      "epoch": 969.8684210526316,
+      "grad_norm": 0.7093734741210938,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 147420
+    },
+    {
+      "epoch": 969.9342105263158,
+      "grad_norm": 1.0567338466644287,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 147430
+    },
+    {
+      "epoch": 970.0,
+      "grad_norm": 1.1926231384277344,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 147440
+    },
+    {
+      "epoch": 970.0657894736842,
+      "grad_norm": 0.7869142889976501,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 147450
+    },
+    {
+      "epoch": 970.1315789473684,
+      "grad_norm": 1.23822820186615,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 147460
+    },
+    {
+      "epoch": 970.1973684210526,
+      "grad_norm": 1.007323145866394,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 147470
+    },
+    {
+      "epoch": 970.2631578947369,
+      "grad_norm": 1.408695101737976,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 147480
+    },
+    {
+      "epoch": 970.328947368421,
+      "grad_norm": 1.1631524562835693,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 147490
+    },
+    {
+      "epoch": 970.3947368421053,
+      "grad_norm": 0.9714431166648865,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 147500
+    },
+    {
+      "epoch": 970.4605263157895,
+      "grad_norm": 0.8139010071754456,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 147510
+    },
+    {
+      "epoch": 970.5263157894736,
+      "grad_norm": 1.2274115085601807,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 147520
+    },
+    {
+      "epoch": 970.5921052631579,
+      "grad_norm": 1.0801491737365723,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 147530
+    },
+    {
+      "epoch": 970.6578947368421,
+      "grad_norm": 0.7565146088600159,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 147540
+    },
+    {
+      "epoch": 970.7236842105264,
+      "grad_norm": 0.9171380400657654,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 147550
+    },
+    {
+      "epoch": 970.7894736842105,
+      "grad_norm": 0.9315175414085388,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 147560
+    },
+    {
+      "epoch": 970.8552631578947,
+      "grad_norm": 0.9675357341766357,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 147570
+    },
+    {
+      "epoch": 970.921052631579,
+      "grad_norm": 0.8446849584579468,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 147580
+    },
+    {
+      "epoch": 970.9868421052631,
+      "grad_norm": 1.2151329517364502,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 147590
+    },
+    {
+      "epoch": 971.0526315789474,
+      "grad_norm": 0.8625166416168213,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 147600
+    },
+    {
+      "epoch": 971.1184210526316,
+      "grad_norm": 0.904151439666748,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 147610
+    },
+    {
+      "epoch": 971.1842105263158,
+      "grad_norm": 1.3609187602996826,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 147620
+    },
+    {
+      "epoch": 971.25,
+      "grad_norm": 2.187743663787842,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 147630
+    },
+    {
+      "epoch": 971.3157894736842,
+      "grad_norm": 2.0440964698791504,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 147640
+    },
+    {
+      "epoch": 971.3815789473684,
+      "grad_norm": 1.9944370985031128,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 147650
+    },
+    {
+      "epoch": 971.4473684210526,
+      "grad_norm": 1.8147251605987549,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 147660
+    },
+    {
+      "epoch": 971.5131578947369,
+      "grad_norm": 1.0225766897201538,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 147670
+    },
+    {
+      "epoch": 971.578947368421,
+      "grad_norm": 1.2411375045776367,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 147680
+    },
+    {
+      "epoch": 971.6447368421053,
+      "grad_norm": 1.3855613470077515,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 147690
+    },
+    {
+      "epoch": 971.7105263157895,
+      "grad_norm": 1.0319336652755737,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 147700
+    },
+    {
+      "epoch": 971.7763157894736,
+      "grad_norm": 1.0877528190612793,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 147710
+    },
+    {
+      "epoch": 971.8421052631579,
+      "grad_norm": 1.5878164768218994,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 147720
+    },
+    {
+      "epoch": 971.9078947368421,
+      "grad_norm": 1.735569715499878,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 147730
+    },
+    {
+      "epoch": 971.9736842105264,
+      "grad_norm": 1.2337061166763306,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 147740
+    },
+    {
+      "epoch": 972.0394736842105,
+      "grad_norm": 0.926618754863739,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 147750
+    },
+    {
+      "epoch": 972.1052631578947,
+      "grad_norm": 0.951611340045929,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 147760
+    },
+    {
+      "epoch": 972.171052631579,
+      "grad_norm": 1.1412699222564697,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 147770
+    },
+    {
+      "epoch": 972.2368421052631,
+      "grad_norm": 1.0385075807571411,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 147780
+    },
+    {
+      "epoch": 972.3026315789474,
+      "grad_norm": 0.8071399331092834,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 147790
+    },
+    {
+      "epoch": 972.3684210526316,
+      "grad_norm": 1.131365418434143,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 147800
+    },
+    {
+      "epoch": 972.4342105263158,
+      "grad_norm": 1.10077702999115,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 147810
+    },
+    {
+      "epoch": 972.5,
+      "grad_norm": 1.1625159978866577,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 147820
+    },
+    {
+      "epoch": 972.5657894736842,
+      "grad_norm": 0.8516972064971924,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 147830
+    },
+    {
+      "epoch": 972.6315789473684,
+      "grad_norm": 0.9543007612228394,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 147840
+    },
+    {
+      "epoch": 972.6973684210526,
+      "grad_norm": 0.7978515028953552,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 147850
+    },
+    {
+      "epoch": 972.7631578947369,
+      "grad_norm": 1.0908616781234741,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 147860
+    },
+    {
+      "epoch": 972.828947368421,
+      "grad_norm": 1.0295151472091675,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 147870
+    },
+    {
+      "epoch": 972.8947368421053,
+      "grad_norm": 0.67076176404953,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 147880
+    },
+    {
+      "epoch": 972.9605263157895,
+      "grad_norm": 1.1055384874343872,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 147890
+    },
+    {
+      "epoch": 973.0263157894736,
+      "grad_norm": 0.9399993419647217,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 147900
+    },
+    {
+      "epoch": 973.0921052631579,
+      "grad_norm": 0.9942353367805481,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 147910
+    },
+    {
+      "epoch": 973.1578947368421,
+      "grad_norm": 0.7977069020271301,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 147920
+    },
+    {
+      "epoch": 973.2236842105264,
+      "grad_norm": 0.940716028213501,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 147930
+    },
+    {
+      "epoch": 973.2894736842105,
+      "grad_norm": 0.5818641185760498,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 147940
+    },
+    {
+      "epoch": 973.3552631578947,
+      "grad_norm": 0.9071598052978516,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 147950
+    },
+    {
+      "epoch": 973.421052631579,
+      "grad_norm": 0.9769167900085449,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 147960
+    },
+    {
+      "epoch": 973.4868421052631,
+      "grad_norm": 0.7918068766593933,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 147970
+    },
+    {
+      "epoch": 973.5526315789474,
+      "grad_norm": 1.44162917137146,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 147980
+    },
+    {
+      "epoch": 973.6184210526316,
+      "grad_norm": 1.3190975189208984,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 147990
+    },
+    {
+      "epoch": 973.6842105263158,
+      "grad_norm": 0.8876311182975769,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 148000
+    },
+    {
+      "epoch": 973.75,
+      "grad_norm": 0.6711481213569641,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 148010
+    },
+    {
+      "epoch": 973.8157894736842,
+      "grad_norm": 0.9575647115707397,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 148020
+    },
+    {
+      "epoch": 973.8815789473684,
+      "grad_norm": 1.1577619314193726,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 148030
+    },
+    {
+      "epoch": 973.9473684210526,
+      "grad_norm": 1.1602065563201904,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 148040
+    },
+    {
+      "epoch": 974.0131578947369,
+      "grad_norm": 1.1503347158432007,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 148050
+    },
+    {
+      "epoch": 974.078947368421,
+      "grad_norm": 1.1186825037002563,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 148060
+    },
+    {
+      "epoch": 974.1447368421053,
+      "grad_norm": 1.2692493200302124,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 148070
+    },
+    {
+      "epoch": 974.2105263157895,
+      "grad_norm": 1.3095154762268066,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 148080
+    },
+    {
+      "epoch": 974.2763157894736,
+      "grad_norm": 1.1788098812103271,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 148090
+    },
+    {
+      "epoch": 974.3421052631579,
+      "grad_norm": 0.8337370753288269,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 148100
+    },
+    {
+      "epoch": 974.4078947368421,
+      "grad_norm": 0.8174479007720947,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 148110
+    },
+    {
+      "epoch": 974.4736842105264,
+      "grad_norm": 1.0070465803146362,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 148120
+    },
+    {
+      "epoch": 974.5394736842105,
+      "grad_norm": 1.2746952772140503,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 148130
+    },
+    {
+      "epoch": 974.6052631578947,
+      "grad_norm": 1.2314714193344116,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 148140
+    },
+    {
+      "epoch": 974.671052631579,
+      "grad_norm": 0.9068085551261902,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 148150
+    },
+    {
+      "epoch": 974.7368421052631,
+      "grad_norm": 1.3437057733535767,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 148160
+    },
+    {
+      "epoch": 974.8026315789474,
+      "grad_norm": 1.2839545011520386,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 148170
+    },
+    {
+      "epoch": 974.8684210526316,
+      "grad_norm": 1.1615701913833618,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 148180
+    },
+    {
+      "epoch": 974.9342105263158,
+      "grad_norm": 1.104874849319458,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 148190
+    },
+    {
+      "epoch": 975.0,
+      "grad_norm": 1.0236458778381348,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 148200
+    },
+    {
+      "epoch": 975.0657894736842,
+      "grad_norm": 0.8276492953300476,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 148210
+    },
+    {
+      "epoch": 975.1315789473684,
+      "grad_norm": 0.7698920369148254,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 148220
+    },
+    {
+      "epoch": 975.1973684210526,
+      "grad_norm": 1.0697041749954224,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 148230
+    },
+    {
+      "epoch": 975.2631578947369,
+      "grad_norm": 0.6203532218933105,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 148240
+    },
+    {
+      "epoch": 975.328947368421,
+      "grad_norm": 1.2870906591415405,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 148250
+    },
+    {
+      "epoch": 975.3947368421053,
+      "grad_norm": 1.1638849973678589,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 148260
+    },
+    {
+      "epoch": 975.4605263157895,
+      "grad_norm": 0.8770232200622559,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 148270
+    },
+    {
+      "epoch": 975.5263157894736,
+      "grad_norm": 0.8042388558387756,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 148280
+    },
+    {
+      "epoch": 975.5921052631579,
+      "grad_norm": 1.1777013540267944,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 148290
+    },
+    {
+      "epoch": 975.6578947368421,
+      "grad_norm": 1.2170054912567139,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 148300
+    },
+    {
+      "epoch": 975.7236842105264,
+      "grad_norm": 1.1343997716903687,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 148310
+    },
+    {
+      "epoch": 975.7894736842105,
+      "grad_norm": 0.9868319034576416,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 148320
+    },
+    {
+      "epoch": 975.8552631578947,
+      "grad_norm": 1.1468241214752197,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 148330
+    },
+    {
+      "epoch": 975.921052631579,
+      "grad_norm": 0.9679098129272461,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 148340
+    },
+    {
+      "epoch": 975.9868421052631,
+      "grad_norm": 1.3126945495605469,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 148350
+    },
+    {
+      "epoch": 976.0526315789474,
+      "grad_norm": 1.1734507083892822,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 148360
+    },
+    {
+      "epoch": 976.1184210526316,
+      "grad_norm": 0.9262369871139526,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 148370
+    },
+    {
+      "epoch": 976.1842105263158,
+      "grad_norm": 1.1285349130630493,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 148380
+    },
+    {
+      "epoch": 976.25,
+      "grad_norm": 0.9451377987861633,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 148390
+    },
+    {
+      "epoch": 976.3157894736842,
+      "grad_norm": 1.0291858911514282,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 148400
+    },
+    {
+      "epoch": 976.3815789473684,
+      "grad_norm": 1.141808271408081,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 148410
+    },
+    {
+      "epoch": 976.4473684210526,
+      "grad_norm": 0.9214851260185242,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 148420
+    },
+    {
+      "epoch": 976.5131578947369,
+      "grad_norm": 1.1481177806854248,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 148430
+    },
+    {
+      "epoch": 976.578947368421,
+      "grad_norm": 1.2398234605789185,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 148440
+    },
+    {
+      "epoch": 976.6447368421053,
+      "grad_norm": 0.9766506552696228,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 148450
+    },
+    {
+      "epoch": 976.7105263157895,
+      "grad_norm": 1.284214735031128,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 148460
+    },
+    {
+      "epoch": 976.7763157894736,
+      "grad_norm": 1.0955685377120972,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 148470
+    },
+    {
+      "epoch": 976.8421052631579,
+      "grad_norm": 1.0362792015075684,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 148480
+    },
+    {
+      "epoch": 976.9078947368421,
+      "grad_norm": 1.067434310913086,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 148490
+    },
+    {
+      "epoch": 976.9736842105264,
+      "grad_norm": 0.7627929449081421,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 148500
+    },
+    {
+      "epoch": 977.0394736842105,
+      "grad_norm": 0.9590473175048828,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 148510
+    },
+    {
+      "epoch": 977.1052631578947,
+      "grad_norm": 0.8360632061958313,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 148520
+    },
+    {
+      "epoch": 977.171052631579,
+      "grad_norm": 0.7430837154388428,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 148530
+    },
+    {
+      "epoch": 977.2368421052631,
+      "grad_norm": 0.9567493796348572,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 148540
+    },
+    {
+      "epoch": 977.3026315789474,
+      "grad_norm": 1.1349761486053467,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 148550
+    },
+    {
+      "epoch": 977.3684210526316,
+      "grad_norm": 1.2582765817642212,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 148560
+    },
+    {
+      "epoch": 977.4342105263158,
+      "grad_norm": 1.0692235231399536,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 148570
+    },
+    {
+      "epoch": 977.5,
+      "grad_norm": 1.2210439443588257,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 148580
+    },
+    {
+      "epoch": 977.5657894736842,
+      "grad_norm": 0.9458500742912292,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 148590
+    },
+    {
+      "epoch": 977.6315789473684,
+      "grad_norm": 1.099303960800171,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 148600
+    },
+    {
+      "epoch": 977.6973684210526,
+      "grad_norm": 0.9207699298858643,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 148610
+    },
+    {
+      "epoch": 977.7631578947369,
+      "grad_norm": 1.0685609579086304,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 148620
+    },
+    {
+      "epoch": 977.828947368421,
+      "grad_norm": 1.1274516582489014,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 148630
+    },
+    {
+      "epoch": 977.8947368421053,
+      "grad_norm": 1.022341012954712,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 148640
+    },
+    {
+      "epoch": 977.9605263157895,
+      "grad_norm": 1.0944082736968994,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 148650
+    },
+    {
+      "epoch": 978.0263157894736,
+      "grad_norm": 1.1251676082611084,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 148660
+    },
+    {
+      "epoch": 978.0921052631579,
+      "grad_norm": 0.7965459823608398,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 148670
+    },
+    {
+      "epoch": 978.1578947368421,
+      "grad_norm": 0.8056169152259827,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 148680
+    },
+    {
+      "epoch": 978.2236842105264,
+      "grad_norm": 0.9947016835212708,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 148690
+    },
+    {
+      "epoch": 978.2894736842105,
+      "grad_norm": 1.521198034286499,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 148700
+    },
+    {
+      "epoch": 978.3552631578947,
+      "grad_norm": 1.4726166725158691,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 148710
+    },
+    {
+      "epoch": 978.421052631579,
+      "grad_norm": 1.2374329566955566,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 148720
+    },
+    {
+      "epoch": 978.4868421052631,
+      "grad_norm": 1.090682029724121,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 148730
+    },
+    {
+      "epoch": 978.5526315789474,
+      "grad_norm": 1.163394570350647,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 148740
+    },
+    {
+      "epoch": 978.6184210526316,
+      "grad_norm": 1.0703188180923462,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 148750
+    },
+    {
+      "epoch": 978.6842105263158,
+      "grad_norm": 1.2367560863494873,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 148760
+    },
+    {
+      "epoch": 978.75,
+      "grad_norm": 1.329300880432129,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 148770
+    },
+    {
+      "epoch": 978.8157894736842,
+      "grad_norm": 1.083126425743103,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 148780
+    },
+    {
+      "epoch": 978.8815789473684,
+      "grad_norm": 0.7448536157608032,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 148790
+    },
+    {
+      "epoch": 978.9473684210526,
+      "grad_norm": 1.0757248401641846,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 148800
+    },
+    {
+      "epoch": 979.0131578947369,
+      "grad_norm": 0.8935508131980896,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 148810
+    },
+    {
+      "epoch": 979.078947368421,
+      "grad_norm": 1.3141487836837769,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 148820
+    },
+    {
+      "epoch": 979.1447368421053,
+      "grad_norm": 0.9629881978034973,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 148830
+    },
+    {
+      "epoch": 979.2105263157895,
+      "grad_norm": 0.9605873823165894,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 148840
+    },
+    {
+      "epoch": 979.2763157894736,
+      "grad_norm": 1.030029058456421,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 148850
+    },
+    {
+      "epoch": 979.3421052631579,
+      "grad_norm": 1.113380789756775,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 148860
+    },
+    {
+      "epoch": 979.4078947368421,
+      "grad_norm": 1.1046963930130005,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 148870
+    },
+    {
+      "epoch": 979.4736842105264,
+      "grad_norm": 1.0526591539382935,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 148880
+    },
+    {
+      "epoch": 979.5394736842105,
+      "grad_norm": 1.6122370958328247,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 148890
+    },
+    {
+      "epoch": 979.6052631578947,
+      "grad_norm": 1.042912244796753,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 148900
+    },
+    {
+      "epoch": 979.671052631579,
+      "grad_norm": 1.0555074214935303,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 148910
+    },
+    {
+      "epoch": 979.7368421052631,
+      "grad_norm": 0.8293302655220032,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 148920
+    },
+    {
+      "epoch": 979.8026315789474,
+      "grad_norm": 0.7137147188186646,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 148930
+    },
+    {
+      "epoch": 979.8684210526316,
+      "grad_norm": 1.0557599067687988,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 148940
+    },
+    {
+      "epoch": 979.9342105263158,
+      "grad_norm": 1.147873044013977,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 148950
+    },
+    {
+      "epoch": 980.0,
+      "grad_norm": 0.8311463594436646,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 148960
+    },
+    {
+      "epoch": 980.0657894736842,
+      "grad_norm": 1.0311176776885986,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 148970
+    },
+    {
+      "epoch": 980.1315789473684,
+      "grad_norm": 0.8780924677848816,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 148980
+    },
+    {
+      "epoch": 980.1973684210526,
+      "grad_norm": 1.235633373260498,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 148990
+    },
+    {
+      "epoch": 980.2631578947369,
+      "grad_norm": 1.298965573310852,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 149000
+    },
+    {
+      "epoch": 980.328947368421,
+      "grad_norm": 0.9419639706611633,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 149010
+    },
+    {
+      "epoch": 980.3947368421053,
+      "grad_norm": 1.0413622856140137,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 149020
+    },
+    {
+      "epoch": 980.4605263157895,
+      "grad_norm": 0.7547554969787598,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 149030
+    },
+    {
+      "epoch": 980.5263157894736,
+      "grad_norm": 0.7510955333709717,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 149040
+    },
+    {
+      "epoch": 980.5921052631579,
+      "grad_norm": 0.9087187647819519,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 149050
+    },
+    {
+      "epoch": 980.6578947368421,
+      "grad_norm": 1.227208137512207,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 149060
+    },
+    {
+      "epoch": 980.7236842105264,
+      "grad_norm": 0.9901775121688843,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 149070
+    },
+    {
+      "epoch": 980.7894736842105,
+      "grad_norm": 1.1645238399505615,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 149080
+    },
+    {
+      "epoch": 980.8552631578947,
+      "grad_norm": 0.9713094234466553,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 149090
+    },
+    {
+      "epoch": 980.921052631579,
+      "grad_norm": 1.0313467979431152,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 149100
+    },
+    {
+      "epoch": 980.9868421052631,
+      "grad_norm": 0.6963071227073669,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 149110
+    },
+    {
+      "epoch": 981.0526315789474,
+      "grad_norm": 1.0932525396347046,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 149120
+    },
+    {
+      "epoch": 981.1184210526316,
+      "grad_norm": 1.1071008443832397,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 149130
+    },
+    {
+      "epoch": 981.1842105263158,
+      "grad_norm": 1.3636506795883179,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 149140
+    },
+    {
+      "epoch": 981.25,
+      "grad_norm": 0.9255510568618774,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 149150
+    },
+    {
+      "epoch": 981.3157894736842,
+      "grad_norm": 0.9791567921638489,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 149160
+    },
+    {
+      "epoch": 981.3815789473684,
+      "grad_norm": 0.8821341395378113,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 149170
+    },
+    {
+      "epoch": 981.4473684210526,
+      "grad_norm": 0.9280576705932617,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 149180
+    },
+    {
+      "epoch": 981.5131578947369,
+      "grad_norm": 0.7593532800674438,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 149190
+    },
+    {
+      "epoch": 981.578947368421,
+      "grad_norm": 1.2766715288162231,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 149200
+    },
+    {
+      "epoch": 981.6447368421053,
+      "grad_norm": 0.7514734864234924,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 149210
+    },
+    {
+      "epoch": 981.7105263157895,
+      "grad_norm": 0.6935755014419556,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 149220
+    },
+    {
+      "epoch": 981.7763157894736,
+      "grad_norm": 1.0693104267120361,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 149230
+    },
+    {
+      "epoch": 981.8421052631579,
+      "grad_norm": 1.1052137613296509,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 149240
+    },
+    {
+      "epoch": 981.9078947368421,
+      "grad_norm": 1.1656912565231323,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 149250
+    },
+    {
+      "epoch": 981.9736842105264,
+      "grad_norm": 1.1336760520935059,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 149260
+    },
+    {
+      "epoch": 982.0394736842105,
+      "grad_norm": 1.118245244026184,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 149270
+    },
+    {
+      "epoch": 982.1052631578947,
+      "grad_norm": 1.0137847661972046,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 149280
+    },
+    {
+      "epoch": 982.171052631579,
+      "grad_norm": 1.1577128171920776,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 149290
+    },
+    {
+      "epoch": 982.2368421052631,
+      "grad_norm": 0.8299029469490051,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 149300
+    },
+    {
+      "epoch": 982.3026315789474,
+      "grad_norm": 1.0582658052444458,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 149310
+    },
+    {
+      "epoch": 982.3684210526316,
+      "grad_norm": 1.0629546642303467,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 149320
+    },
+    {
+      "epoch": 982.4342105263158,
+      "grad_norm": 1.2285642623901367,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 149330
+    },
+    {
+      "epoch": 982.5,
+      "grad_norm": 1.1044747829437256,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 149340
+    },
+    {
+      "epoch": 982.5657894736842,
+      "grad_norm": 1.2919867038726807,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 149350
+    },
+    {
+      "epoch": 982.6315789473684,
+      "grad_norm": 0.8287044763565063,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 149360
+    },
+    {
+      "epoch": 982.6973684210526,
+      "grad_norm": 0.9002633094787598,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 149370
+    },
+    {
+      "epoch": 982.7631578947369,
+      "grad_norm": 1.065321683883667,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 149380
+    },
+    {
+      "epoch": 982.828947368421,
+      "grad_norm": 1.0330926179885864,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 149390
+    },
+    {
+      "epoch": 982.8947368421053,
+      "grad_norm": 0.989842414855957,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 149400
+    },
+    {
+      "epoch": 982.9605263157895,
+      "grad_norm": 0.905006468296051,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 149410
+    },
+    {
+      "epoch": 983.0263157894736,
+      "grad_norm": 1.150158405303955,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 149420
+    },
+    {
+      "epoch": 983.0921052631579,
+      "grad_norm": 0.8461920619010925,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 149430
+    },
+    {
+      "epoch": 983.1578947368421,
+      "grad_norm": 0.8330326676368713,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 149440
+    },
+    {
+      "epoch": 983.2236842105264,
+      "grad_norm": 1.2453408241271973,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 149450
+    },
+    {
+      "epoch": 983.2894736842105,
+      "grad_norm": 1.0621312856674194,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 149460
+    },
+    {
+      "epoch": 983.3552631578947,
+      "grad_norm": 1.0289655923843384,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 149470
+    },
+    {
+      "epoch": 983.421052631579,
+      "grad_norm": 0.8648130297660828,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 149480
+    },
+    {
+      "epoch": 983.4868421052631,
+      "grad_norm": 1.0741007328033447,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 149490
+    },
+    {
+      "epoch": 983.5526315789474,
+      "grad_norm": 0.8858330845832825,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 149500
+    },
+    {
+      "epoch": 983.6184210526316,
+      "grad_norm": 1.25303053855896,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 149510
+    },
+    {
+      "epoch": 983.6842105263158,
+      "grad_norm": 1.2206391096115112,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 149520
+    },
+    {
+      "epoch": 983.75,
+      "grad_norm": 0.9137042164802551,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 149530
+    },
+    {
+      "epoch": 983.8157894736842,
+      "grad_norm": 1.025648593902588,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 149540
+    },
+    {
+      "epoch": 983.8815789473684,
+      "grad_norm": 1.0423038005828857,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 149550
+    },
+    {
+      "epoch": 983.9473684210526,
+      "grad_norm": 1.0618716478347778,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 149560
+    },
+    {
+      "epoch": 984.0131578947369,
+      "grad_norm": 1.030285120010376,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 149570
+    },
+    {
+      "epoch": 984.078947368421,
+      "grad_norm": 0.8739547729492188,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 149580
+    },
+    {
+      "epoch": 984.1447368421053,
+      "grad_norm": 1.1206870079040527,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 149590
+    },
+    {
+      "epoch": 984.2105263157895,
+      "grad_norm": 1.278941035270691,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 149600
+    },
+    {
+      "epoch": 984.2763157894736,
+      "grad_norm": 1.7599936723709106,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 149610
+    },
+    {
+      "epoch": 984.3421052631579,
+      "grad_norm": 1.6677428483963013,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 149620
+    },
+    {
+      "epoch": 984.4078947368421,
+      "grad_norm": 3.871687412261963,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 149630
+    },
+    {
+      "epoch": 984.4736842105264,
+      "grad_norm": 2.421388864517212,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 149640
+    },
+    {
+      "epoch": 984.5394736842105,
+      "grad_norm": 1.2500579357147217,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 149650
+    },
+    {
+      "epoch": 984.6052631578947,
+      "grad_norm": 1.5379196405410767,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 149660
+    },
+    {
+      "epoch": 984.671052631579,
+      "grad_norm": 1.7158979177474976,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 149670
+    },
+    {
+      "epoch": 984.7368421052631,
+      "grad_norm": 1.4845489263534546,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 149680
+    },
+    {
+      "epoch": 984.8026315789474,
+      "grad_norm": 1.1629608869552612,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 149690
+    },
+    {
+      "epoch": 984.8684210526316,
+      "grad_norm": 1.415179967880249,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 149700
+    },
+    {
+      "epoch": 984.9342105263158,
+      "grad_norm": 1.1638578176498413,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 149710
+    },
+    {
+      "epoch": 985.0,
+      "grad_norm": 0.9588046669960022,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 149720
+    },
+    {
+      "epoch": 985.0657894736842,
+      "grad_norm": 1.2830522060394287,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 149730
+    },
+    {
+      "epoch": 985.1315789473684,
+      "grad_norm": 0.9208570122718811,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 149740
+    },
+    {
+      "epoch": 985.1973684210526,
+      "grad_norm": 0.5856902003288269,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 149750
+    },
+    {
+      "epoch": 985.2631578947369,
+      "grad_norm": 0.9705900549888611,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 149760
+    },
+    {
+      "epoch": 985.328947368421,
+      "grad_norm": 1.1862601041793823,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 149770
+    },
+    {
+      "epoch": 985.3947368421053,
+      "grad_norm": 0.9023541808128357,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 149780
+    },
+    {
+      "epoch": 985.4605263157895,
+      "grad_norm": 1.044999361038208,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 149790
+    },
+    {
+      "epoch": 985.5263157894736,
+      "grad_norm": 0.9709221124649048,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 149800
+    },
+    {
+      "epoch": 985.5921052631579,
+      "grad_norm": 1.0954318046569824,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 149810
+    },
+    {
+      "epoch": 985.6578947368421,
+      "grad_norm": 1.156111478805542,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 149820
+    },
+    {
+      "epoch": 985.7236842105264,
+      "grad_norm": 0.8454775810241699,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 149830
+    },
+    {
+      "epoch": 985.7894736842105,
+      "grad_norm": 0.880204439163208,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 149840
+    },
+    {
+      "epoch": 985.8552631578947,
+      "grad_norm": 1.2712900638580322,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 149850
+    },
+    {
+      "epoch": 985.921052631579,
+      "grad_norm": 1.0623518228530884,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 149860
+    },
+    {
+      "epoch": 985.9868421052631,
+      "grad_norm": 0.7809160947799683,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 149870
+    },
+    {
+      "epoch": 986.0526315789474,
+      "grad_norm": 0.9675153493881226,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 149880
+    },
+    {
+      "epoch": 986.1184210526316,
+      "grad_norm": 0.8535943031311035,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 149890
+    },
+    {
+      "epoch": 986.1842105263158,
+      "grad_norm": 0.9090612530708313,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 149900
+    },
+    {
+      "epoch": 986.25,
+      "grad_norm": 0.7848944067955017,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 149910
+    },
+    {
+      "epoch": 986.3157894736842,
+      "grad_norm": 1.056888461112976,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 149920
+    },
+    {
+      "epoch": 986.3815789473684,
+      "grad_norm": 0.9026320576667786,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 149930
+    },
+    {
+      "epoch": 986.4473684210526,
+      "grad_norm": 1.167649745941162,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 149940
+    },
+    {
+      "epoch": 986.5131578947369,
+      "grad_norm": 1.0454814434051514,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 149950
+    },
+    {
+      "epoch": 986.578947368421,
+      "grad_norm": 1.3439925909042358,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 149960
+    },
+    {
+      "epoch": 986.6447368421053,
+      "grad_norm": 1.020798921585083,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 149970
+    },
+    {
+      "epoch": 986.7105263157895,
+      "grad_norm": 1.1719021797180176,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 149980
+    },
+    {
+      "epoch": 986.7763157894736,
+      "grad_norm": 1.0493474006652832,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 149990
+    },
+    {
+      "epoch": 986.8421052631579,
+      "grad_norm": 0.8416689038276672,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 150000
+    },
+    {
+      "epoch": 986.9078947368421,
+      "grad_norm": 0.9856203198432922,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 150010
+    },
+    {
+      "epoch": 986.9736842105264,
+      "grad_norm": 0.9143205881118774,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 150020
+    },
+    {
+      "epoch": 987.0394736842105,
+      "grad_norm": 0.9702136516571045,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 150030
+    },
+    {
+      "epoch": 987.1052631578947,
+      "grad_norm": 0.7779248356819153,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 150040
+    },
+    {
+      "epoch": 987.171052631579,
+      "grad_norm": 1.1739939451217651,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 150050
+    },
+    {
+      "epoch": 987.2368421052631,
+      "grad_norm": 1.0114635229110718,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 150060
+    },
+    {
+      "epoch": 987.3026315789474,
+      "grad_norm": 1.0713098049163818,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 150070
+    },
+    {
+      "epoch": 987.3684210526316,
+      "grad_norm": 0.9097891449928284,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 150080
+    },
+    {
+      "epoch": 987.4342105263158,
+      "grad_norm": 1.1985975503921509,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 150090
+    },
+    {
+      "epoch": 987.5,
+      "grad_norm": 1.0965287685394287,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 150100
+    },
+    {
+      "epoch": 987.5657894736842,
+      "grad_norm": 0.873046338558197,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 150110
+    },
+    {
+      "epoch": 987.6315789473684,
+      "grad_norm": 1.016603708267212,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 150120
+    },
+    {
+      "epoch": 987.6973684210526,
+      "grad_norm": 0.9413048028945923,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 150130
+    },
+    {
+      "epoch": 987.7631578947369,
+      "grad_norm": 1.0264689922332764,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 150140
+    },
+    {
+      "epoch": 987.828947368421,
+      "grad_norm": 0.916080117225647,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 150150
+    },
+    {
+      "epoch": 987.8947368421053,
+      "grad_norm": 1.110188364982605,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 150160
+    },
+    {
+      "epoch": 987.9605263157895,
+      "grad_norm": 1.0848978757858276,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 150170
+    },
+    {
+      "epoch": 988.0263157894736,
+      "grad_norm": 0.9753615856170654,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 150180
+    },
+    {
+      "epoch": 988.0921052631579,
+      "grad_norm": 1.2816380262374878,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 150190
+    },
+    {
+      "epoch": 988.1578947368421,
+      "grad_norm": 1.2264199256896973,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 150200
+    },
+    {
+      "epoch": 988.2236842105264,
+      "grad_norm": 1.3432106971740723,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 150210
+    },
+    {
+      "epoch": 988.2894736842105,
+      "grad_norm": 1.5210658311843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 150220
+    },
+    {
+      "epoch": 988.3552631578947,
+      "grad_norm": 1.3455909490585327,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 150230
+    },
+    {
+      "epoch": 988.421052631579,
+      "grad_norm": 1.2642890214920044,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 150240
+    },
+    {
+      "epoch": 988.4868421052631,
+      "grad_norm": 1.0704741477966309,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 150250
+    },
+    {
+      "epoch": 988.5526315789474,
+      "grad_norm": 1.0484412908554077,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 150260
+    },
+    {
+      "epoch": 988.6184210526316,
+      "grad_norm": 0.8334908485412598,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 150270
+    },
+    {
+      "epoch": 988.6842105263158,
+      "grad_norm": 0.8684769868850708,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 150280
+    },
+    {
+      "epoch": 988.75,
+      "grad_norm": 0.8539243936538696,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 150290
+    },
+    {
+      "epoch": 988.8157894736842,
+      "grad_norm": 0.8212605118751526,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 150300
+    },
+    {
+      "epoch": 988.8815789473684,
+      "grad_norm": 1.1426969766616821,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 150310
+    },
+    {
+      "epoch": 988.9473684210526,
+      "grad_norm": 1.2686213254928589,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 150320
+    },
+    {
+      "epoch": 989.0131578947369,
+      "grad_norm": 1.0295464992523193,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 150330
+    },
+    {
+      "epoch": 989.078947368421,
+      "grad_norm": 0.9937509298324585,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 150340
+    },
+    {
+      "epoch": 989.1447368421053,
+      "grad_norm": 0.8517789840698242,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 150350
+    },
+    {
+      "epoch": 989.2105263157895,
+      "grad_norm": 1.2996900081634521,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 150360
+    },
+    {
+      "epoch": 989.2763157894736,
+      "grad_norm": 0.7057400345802307,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 150370
+    },
+    {
+      "epoch": 989.3421052631579,
+      "grad_norm": 0.8088485598564148,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 150380
+    },
+    {
+      "epoch": 989.4078947368421,
+      "grad_norm": 0.9495823979377747,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 150390
+    },
+    {
+      "epoch": 989.4736842105264,
+      "grad_norm": 1.176836609840393,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 150400
+    },
+    {
+      "epoch": 989.5394736842105,
+      "grad_norm": 1.438713788986206,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 150410
+    },
+    {
+      "epoch": 989.6052631578947,
+      "grad_norm": 1.0466101169586182,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 150420
+    },
+    {
+      "epoch": 989.671052631579,
+      "grad_norm": 0.9159618020057678,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 150430
+    },
+    {
+      "epoch": 989.7368421052631,
+      "grad_norm": 1.128126859664917,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 150440
+    },
+    {
+      "epoch": 989.8026315789474,
+      "grad_norm": 0.6916261911392212,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 150450
+    },
+    {
+      "epoch": 989.8684210526316,
+      "grad_norm": 1.2188067436218262,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 150460
+    },
+    {
+      "epoch": 989.9342105263158,
+      "grad_norm": 1.330909013748169,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 150470
+    },
+    {
+      "epoch": 990.0,
+      "grad_norm": 1.1757625341415405,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 150480
+    },
+    {
+      "epoch": 990.0657894736842,
+      "grad_norm": 1.4033007621765137,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 150490
+    },
+    {
+      "epoch": 990.1315789473684,
+      "grad_norm": 1.0067665576934814,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 150500
+    },
+    {
+      "epoch": 990.1973684210526,
+      "grad_norm": 1.1867026090621948,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 150510
+    },
+    {
+      "epoch": 990.2631578947369,
+      "grad_norm": 1.306778073310852,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 150520
+    },
+    {
+      "epoch": 990.328947368421,
+      "grad_norm": 1.033635139465332,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 150530
+    },
+    {
+      "epoch": 990.3947368421053,
+      "grad_norm": 1.0581187009811401,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 150540
+    },
+    {
+      "epoch": 990.4605263157895,
+      "grad_norm": 1.396301507949829,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 150550
+    },
+    {
+      "epoch": 990.5263157894736,
+      "grad_norm": 1.2779043912887573,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 150560
+    },
+    {
+      "epoch": 990.5921052631579,
+      "grad_norm": 1.0077518224716187,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 150570
+    },
+    {
+      "epoch": 990.6578947368421,
+      "grad_norm": 1.1723346710205078,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 150580
+    },
+    {
+      "epoch": 990.7236842105264,
+      "grad_norm": 1.0024058818817139,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 150590
+    },
+    {
+      "epoch": 990.7894736842105,
+      "grad_norm": 1.120664358139038,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 150600
+    },
+    {
+      "epoch": 990.8552631578947,
+      "grad_norm": 1.2388149499893188,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 150610
+    },
+    {
+      "epoch": 990.921052631579,
+      "grad_norm": 1.3285748958587646,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 150620
+    },
+    {
+      "epoch": 990.9868421052631,
+      "grad_norm": 0.933190643787384,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 150630
+    },
+    {
+      "epoch": 991.0526315789474,
+      "grad_norm": 1.2380552291870117,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 150640
+    },
+    {
+      "epoch": 991.1184210526316,
+      "grad_norm": 1.042554497718811,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 150650
+    },
+    {
+      "epoch": 991.1842105263158,
+      "grad_norm": 1.0768581628799438,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 150660
+    },
+    {
+      "epoch": 991.25,
+      "grad_norm": 0.9357948899269104,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 150670
+    },
+    {
+      "epoch": 991.3157894736842,
+      "grad_norm": 1.2259186506271362,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 150680
+    },
+    {
+      "epoch": 991.3815789473684,
+      "grad_norm": 0.9076092839241028,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 150690
+    },
+    {
+      "epoch": 991.4473684210526,
+      "grad_norm": 1.0756639242172241,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 150700
+    },
+    {
+      "epoch": 991.5131578947369,
+      "grad_norm": 0.8654159307479858,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 150710
+    },
+    {
+      "epoch": 991.578947368421,
+      "grad_norm": 0.5353500843048096,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 150720
+    },
+    {
+      "epoch": 991.6447368421053,
+      "grad_norm": 1.0881078243255615,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 150730
+    },
+    {
+      "epoch": 991.7105263157895,
+      "grad_norm": 0.8417906165122986,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 150740
+    },
+    {
+      "epoch": 991.7763157894736,
+      "grad_norm": 1.3755824565887451,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 150750
+    },
+    {
+      "epoch": 991.8421052631579,
+      "grad_norm": 1.0867851972579956,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 150760
+    },
+    {
+      "epoch": 991.9078947368421,
+      "grad_norm": 1.0686115026474,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 150770
+    },
+    {
+      "epoch": 991.9736842105264,
+      "grad_norm": 0.9699664115905762,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 150780
+    },
+    {
+      "epoch": 992.0394736842105,
+      "grad_norm": 1.1987810134887695,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 150790
+    },
+    {
+      "epoch": 992.1052631578947,
+      "grad_norm": 0.8176727294921875,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 150800
+    },
+    {
+      "epoch": 992.171052631579,
+      "grad_norm": 0.8529284596443176,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 150810
+    },
+    {
+      "epoch": 992.2368421052631,
+      "grad_norm": 1.179165005683899,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 150820
+    },
+    {
+      "epoch": 992.3026315789474,
+      "grad_norm": 0.9331127405166626,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 150830
+    },
+    {
+      "epoch": 992.3684210526316,
+      "grad_norm": 1.0699684619903564,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 150840
+    },
+    {
+      "epoch": 992.4342105263158,
+      "grad_norm": 1.0798276662826538,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 150850
+    },
+    {
+      "epoch": 992.5,
+      "grad_norm": 1.126177430152893,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 150860
+    },
+    {
+      "epoch": 992.5657894736842,
+      "grad_norm": 1.269126057624817,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 150870
+    },
+    {
+      "epoch": 992.6315789473684,
+      "grad_norm": 1.2540189027786255,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 150880
+    },
+    {
+      "epoch": 992.6973684210526,
+      "grad_norm": 0.9607136845588684,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 150890
+    },
+    {
+      "epoch": 992.7631578947369,
+      "grad_norm": 1.192214846611023,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 150900
+    },
+    {
+      "epoch": 992.828947368421,
+      "grad_norm": 0.9914960265159607,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 150910
+    },
+    {
+      "epoch": 992.8947368421053,
+      "grad_norm": 0.728797435760498,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 150920
+    },
+    {
+      "epoch": 992.9605263157895,
+      "grad_norm": 0.8233456611633301,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 150930
+    },
+    {
+      "epoch": 993.0263157894736,
+      "grad_norm": 0.8725035190582275,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 150940
+    },
+    {
+      "epoch": 993.0921052631579,
+      "grad_norm": 1.1844197511672974,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 150950
+    },
+    {
+      "epoch": 993.1578947368421,
+      "grad_norm": 0.7963541150093079,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 150960
+    },
+    {
+      "epoch": 993.2236842105264,
+      "grad_norm": 1.0642098188400269,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 150970
+    },
+    {
+      "epoch": 993.2894736842105,
+      "grad_norm": 0.7806707620620728,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 150980
+    },
+    {
+      "epoch": 993.3552631578947,
+      "grad_norm": 0.9475646615028381,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 150990
+    },
+    {
+      "epoch": 993.421052631579,
+      "grad_norm": 0.715223491191864,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 151000
+    },
+    {
+      "epoch": 993.4868421052631,
+      "grad_norm": 1.004929542541504,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 151010
+    },
+    {
+      "epoch": 993.5526315789474,
+      "grad_norm": 0.9862182140350342,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 151020
+    },
+    {
+      "epoch": 993.6184210526316,
+      "grad_norm": 0.9281184673309326,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 151030
+    },
+    {
+      "epoch": 993.6842105263158,
+      "grad_norm": 1.0093213319778442,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 151040
+    },
+    {
+      "epoch": 993.75,
+      "grad_norm": 1.3052515983581543,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 151050
+    },
+    {
+      "epoch": 993.8157894736842,
+      "grad_norm": 1.137257695198059,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 151060
+    },
+    {
+      "epoch": 993.8815789473684,
+      "grad_norm": 1.3269919157028198,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 151070
+    },
+    {
+      "epoch": 993.9473684210526,
+      "grad_norm": 1.2152782678604126,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 151080
+    },
+    {
+      "epoch": 994.0131578947369,
+      "grad_norm": 1.2013344764709473,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 151090
+    },
+    {
+      "epoch": 994.078947368421,
+      "grad_norm": 1.0185185670852661,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 151100
+    },
+    {
+      "epoch": 994.1447368421053,
+      "grad_norm": 1.1128599643707275,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 151110
+    },
+    {
+      "epoch": 994.2105263157895,
+      "grad_norm": 1.1425211429595947,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 151120
+    },
+    {
+      "epoch": 994.2763157894736,
+      "grad_norm": 1.0568586587905884,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 151130
+    },
+    {
+      "epoch": 994.3421052631579,
+      "grad_norm": 1.080623984336853,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 151140
+    },
+    {
+      "epoch": 994.4078947368421,
+      "grad_norm": 1.4741694927215576,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 151150
+    },
+    {
+      "epoch": 994.4736842105264,
+      "grad_norm": 2.169677972793579,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 151160
+    },
+    {
+      "epoch": 994.5394736842105,
+      "grad_norm": 2.1970841884613037,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 151170
+    },
+    {
+      "epoch": 994.6052631578947,
+      "grad_norm": 1.8922151327133179,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 151180
+    },
+    {
+      "epoch": 994.671052631579,
+      "grad_norm": 1.2851279973983765,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 151190
+    },
+    {
+      "epoch": 994.7368421052631,
+      "grad_norm": 1.3797351121902466,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 151200
+    },
+    {
+      "epoch": 994.8026315789474,
+      "grad_norm": 1.318306565284729,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 151210
+    },
+    {
+      "epoch": 994.8684210526316,
+      "grad_norm": 1.115478754043579,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 151220
+    },
+    {
+      "epoch": 994.9342105263158,
+      "grad_norm": 1.0066384077072144,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 151230
+    },
+    {
+      "epoch": 995.0,
+      "grad_norm": 1.1105729341506958,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 151240
+    },
+    {
+      "epoch": 995.0657894736842,
+      "grad_norm": 1.399383544921875,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 151250
+    },
+    {
+      "epoch": 995.1315789473684,
+      "grad_norm": 1.160544991493225,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 151260
+    },
+    {
+      "epoch": 995.1973684210526,
+      "grad_norm": 0.9492695331573486,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 151270
+    },
+    {
+      "epoch": 995.2631578947369,
+      "grad_norm": 1.0131347179412842,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 151280
+    },
+    {
+      "epoch": 995.328947368421,
+      "grad_norm": 0.9944673776626587,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 151290
+    },
+    {
+      "epoch": 995.3947368421053,
+      "grad_norm": 1.1447815895080566,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 151300
+    },
+    {
+      "epoch": 995.4605263157895,
+      "grad_norm": 1.2639286518096924,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 151310
+    },
+    {
+      "epoch": 995.5263157894736,
+      "grad_norm": 0.9355853199958801,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 151320
+    },
+    {
+      "epoch": 995.5921052631579,
+      "grad_norm": 1.3493982553482056,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 151330
+    },
+    {
+      "epoch": 995.6578947368421,
+      "grad_norm": 1.0639145374298096,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 151340
+    },
+    {
+      "epoch": 995.7236842105264,
+      "grad_norm": 0.9887959957122803,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 151350
+    },
+    {
+      "epoch": 995.7894736842105,
+      "grad_norm": 1.3102302551269531,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 151360
+    },
+    {
+      "epoch": 995.8552631578947,
+      "grad_norm": 1.0294629335403442,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 151370
+    },
+    {
+      "epoch": 995.921052631579,
+      "grad_norm": 1.3485372066497803,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 151380
+    },
+    {
+      "epoch": 995.9868421052631,
+      "grad_norm": 1.2144447565078735,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 151390
+    },
+    {
+      "epoch": 996.0526315789474,
+      "grad_norm": 0.7425600290298462,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 151400
+    },
+    {
+      "epoch": 996.1184210526316,
+      "grad_norm": 1.1900805234909058,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 151410
+    },
+    {
+      "epoch": 996.1842105263158,
+      "grad_norm": 0.8845036029815674,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 151420
+    },
+    {
+      "epoch": 996.25,
+      "grad_norm": 0.965152382850647,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 151430
+    },
+    {
+      "epoch": 996.3157894736842,
+      "grad_norm": 1.1030590534210205,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 151440
+    },
+    {
+      "epoch": 996.3815789473684,
+      "grad_norm": 0.915751576423645,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 151450
+    },
+    {
+      "epoch": 996.4473684210526,
+      "grad_norm": 0.9805639982223511,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 151460
+    },
+    {
+      "epoch": 996.5131578947369,
+      "grad_norm": 0.8894463777542114,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 151470
+    },
+    {
+      "epoch": 996.578947368421,
+      "grad_norm": 1.0241661071777344,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 151480
+    },
+    {
+      "epoch": 996.6447368421053,
+      "grad_norm": 1.265334129333496,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 151490
+    },
+    {
+      "epoch": 996.7105263157895,
+      "grad_norm": 0.9865074753761292,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 151500
+    },
+    {
+      "epoch": 996.7763157894736,
+      "grad_norm": 0.9851489067077637,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 151510
+    },
+    {
+      "epoch": 996.8421052631579,
+      "grad_norm": 0.9616159200668335,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 151520
+    },
+    {
+      "epoch": 996.9078947368421,
+      "grad_norm": 1.051690936088562,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 151530
+    },
+    {
+      "epoch": 996.9736842105264,
+      "grad_norm": 0.7457857131958008,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 151540
+    },
+    {
+      "epoch": 997.0394736842105,
+      "grad_norm": 1.3598742485046387,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 151550
+    },
+    {
+      "epoch": 997.1052631578947,
+      "grad_norm": 1.0818504095077515,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 151560
+    },
+    {
+      "epoch": 997.171052631579,
+      "grad_norm": 1.1341614723205566,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 151570
+    },
+    {
+      "epoch": 997.2368421052631,
+      "grad_norm": 1.3437490463256836,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 151580
+    },
+    {
+      "epoch": 997.3026315789474,
+      "grad_norm": 0.998374879360199,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 151590
+    },
+    {
+      "epoch": 997.3684210526316,
+      "grad_norm": 1.2805155515670776,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 151600
+    },
+    {
+      "epoch": 997.4342105263158,
+      "grad_norm": 1.1499323844909668,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 151610
+    },
+    {
+      "epoch": 997.5,
+      "grad_norm": 1.3379521369934082,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 151620
+    },
+    {
+      "epoch": 997.5657894736842,
+      "grad_norm": 1.0792102813720703,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 151630
+    },
+    {
+      "epoch": 997.6315789473684,
+      "grad_norm": 1.2518432140350342,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 151640
+    },
+    {
+      "epoch": 997.6973684210526,
+      "grad_norm": 1.0179204940795898,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 151650
+    },
+    {
+      "epoch": 997.7631578947369,
+      "grad_norm": 0.97569340467453,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 151660
+    },
+    {
+      "epoch": 997.828947368421,
+      "grad_norm": 1.1537328958511353,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 151670
+    },
+    {
+      "epoch": 997.8947368421053,
+      "grad_norm": 1.0750672817230225,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 151680
+    },
+    {
+      "epoch": 997.9605263157895,
+      "grad_norm": 0.9960630536079407,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 151690
+    },
+    {
+      "epoch": 998.0263157894736,
+      "grad_norm": 0.7261005640029907,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 151700
+    },
+    {
+      "epoch": 998.0921052631579,
+      "grad_norm": 1.1404473781585693,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 151710
+    },
+    {
+      "epoch": 998.1578947368421,
+      "grad_norm": 1.0211684703826904,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 151720
+    },
+    {
+      "epoch": 998.2236842105264,
+      "grad_norm": 0.8978713750839233,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 151730
+    },
+    {
+      "epoch": 998.2894736842105,
+      "grad_norm": 1.1412054300308228,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 151740
+    },
+    {
+      "epoch": 998.3552631578947,
+      "grad_norm": 0.9569369554519653,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 151750
+    },
+    {
+      "epoch": 998.421052631579,
+      "grad_norm": 1.211241602897644,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 151760
+    },
+    {
+      "epoch": 998.4868421052631,
+      "grad_norm": 1.1356269121170044,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 151770
+    },
+    {
+      "epoch": 998.5526315789474,
+      "grad_norm": 1.2875263690948486,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 151780
+    },
+    {
+      "epoch": 998.6184210526316,
+      "grad_norm": 0.9279645085334778,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 151790
+    },
+    {
+      "epoch": 998.6842105263158,
+      "grad_norm": 1.1666724681854248,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 151800
+    },
+    {
+      "epoch": 998.75,
+      "grad_norm": 1.2689951658248901,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 151810
+    },
+    {
+      "epoch": 998.8157894736842,
+      "grad_norm": 0.8863130807876587,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 151820
+    },
+    {
+      "epoch": 998.8815789473684,
+      "grad_norm": 1.002847671508789,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 151830
+    },
+    {
+      "epoch": 998.9473684210526,
+      "grad_norm": 1.0156797170639038,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 151840
+    },
+    {
+      "epoch": 999.0131578947369,
+      "grad_norm": 1.1730620861053467,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 151850
+    },
+    {
+      "epoch": 999.078947368421,
+      "grad_norm": 1.2348865270614624,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 151860
+    },
+    {
+      "epoch": 999.1447368421053,
+      "grad_norm": 1.2273298501968384,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 151870
+    },
+    {
+      "epoch": 999.2105263157895,
+      "grad_norm": 1.1343059539794922,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 151880
+    },
+    {
+      "epoch": 999.2763157894736,
+      "grad_norm": 1.2365325689315796,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 151890
+    },
+    {
+      "epoch": 999.3421052631579,
+      "grad_norm": 0.8025848269462585,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 151900
+    },
+    {
+      "epoch": 999.4078947368421,
+      "grad_norm": 1.318703055381775,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 151910
+    },
+    {
+      "epoch": 999.4736842105264,
+      "grad_norm": 0.9021982550621033,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 151920
+    },
+    {
+      "epoch": 999.5394736842105,
+      "grad_norm": 1.2123674154281616,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 151930
+    },
+    {
+      "epoch": 999.6052631578947,
+      "grad_norm": 1.0211516618728638,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 151940
+    },
+    {
+      "epoch": 999.671052631579,
+      "grad_norm": 0.8161410689353943,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 151950
+    },
+    {
+      "epoch": 999.7368421052631,
+      "grad_norm": 0.7967602610588074,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 151960
+    },
+    {
+      "epoch": 999.8026315789474,
+      "grad_norm": 1.032511591911316,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 151970
+    },
+    {
+      "epoch": 999.8684210526316,
+      "grad_norm": 0.9636939764022827,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 151980
+    },
+    {
+      "epoch": 999.9342105263158,
+      "grad_norm": 0.7413888573646545,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 151990
+    },
+    {
+      "epoch": 1000.0,
+      "grad_norm": 1.27151358127594,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 152000
+    },
+    {
+      "epoch": 1000.0657894736842,
+      "grad_norm": 1.314128041267395,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 152010
+    },
+    {
+      "epoch": 1000.1315789473684,
+      "grad_norm": 1.3646230697631836,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 152020
+    },
+    {
+      "epoch": 1000.1973684210526,
+      "grad_norm": 1.1015452146530151,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 152030
+    },
+    {
+      "epoch": 1000.2631578947369,
+      "grad_norm": 1.1657509803771973,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 152040
+    },
+    {
+      "epoch": 1000.328947368421,
+      "grad_norm": 1.5875916481018066,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 152050
+    },
+    {
+      "epoch": 1000.3947368421053,
+      "grad_norm": 0.9831330180168152,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 152060
+    },
+    {
+      "epoch": 1000.4605263157895,
+      "grad_norm": 1.056758999824524,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 152070
+    },
+    {
+      "epoch": 1000.5263157894736,
+      "grad_norm": 0.9852983355522156,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 152080
+    },
+    {
+      "epoch": 1000.5921052631579,
+      "grad_norm": 1.0023329257965088,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 152090
+    },
+    {
+      "epoch": 1000.6578947368421,
+      "grad_norm": 0.8812468647956848,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 152100
+    },
+    {
+      "epoch": 1000.7236842105264,
+      "grad_norm": 1.064321756362915,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 152110
+    },
+    {
+      "epoch": 1000.7894736842105,
+      "grad_norm": 0.9033085107803345,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 152120
+    },
+    {
+      "epoch": 1000.8552631578947,
+      "grad_norm": 1.2103757858276367,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 152130
+    },
+    {
+      "epoch": 1000.921052631579,
+      "grad_norm": 0.9150286316871643,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 152140
+    },
+    {
+      "epoch": 1000.9868421052631,
+      "grad_norm": 0.9393565058708191,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 152150
+    },
+    {
+      "epoch": 1001.0526315789474,
+      "grad_norm": 1.163619875907898,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 152160
+    },
+    {
+      "epoch": 1001.1184210526316,
+      "grad_norm": 0.695565938949585,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 152170
+    },
+    {
+      "epoch": 1001.1842105263158,
+      "grad_norm": 0.837020754814148,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 152180
+    },
+    {
+      "epoch": 1001.25,
+      "grad_norm": 0.9212221503257751,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 152190
+    },
+    {
+      "epoch": 1001.3157894736842,
+      "grad_norm": 1.2509287595748901,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 152200
+    },
+    {
+      "epoch": 1001.3815789473684,
+      "grad_norm": 1.2080440521240234,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 152210
+    },
+    {
+      "epoch": 1001.4473684210526,
+      "grad_norm": 0.846920907497406,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 152220
+    },
+    {
+      "epoch": 1001.5131578947369,
+      "grad_norm": 1.3782720565795898,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 152230
+    },
+    {
+      "epoch": 1001.578947368421,
+      "grad_norm": 1.222617506980896,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 152240
+    },
+    {
+      "epoch": 1001.6447368421053,
+      "grad_norm": 0.953065812587738,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 152250
+    },
+    {
+      "epoch": 1001.7105263157895,
+      "grad_norm": 0.9601914882659912,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 152260
+    },
+    {
+      "epoch": 1001.7763157894736,
+      "grad_norm": 0.7010998725891113,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 152270
+    },
+    {
+      "epoch": 1001.8421052631579,
+      "grad_norm": 1.2162070274353027,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 152280
+    },
+    {
+      "epoch": 1001.9078947368421,
+      "grad_norm": 0.6597650647163391,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 152290
+    },
+    {
+      "epoch": 1001.9736842105264,
+      "grad_norm": 1.0291619300842285,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 152300
+    },
+    {
+      "epoch": 1002.0394736842105,
+      "grad_norm": 1.1354762315750122,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 152310
+    },
+    {
+      "epoch": 1002.1052631578947,
+      "grad_norm": 0.8112242221832275,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 152320
+    },
+    {
+      "epoch": 1002.171052631579,
+      "grad_norm": 1.0882664918899536,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 152330
+    },
+    {
+      "epoch": 1002.2368421052631,
+      "grad_norm": 0.919173538684845,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 152340
+    },
+    {
+      "epoch": 1002.3026315789474,
+      "grad_norm": 1.0165963172912598,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 152350
+    },
+    {
+      "epoch": 1002.3684210526316,
+      "grad_norm": 1.0748934745788574,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 152360
+    },
+    {
+      "epoch": 1002.4342105263158,
+      "grad_norm": 1.1250693798065186,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 152370
+    },
+    {
+      "epoch": 1002.5,
+      "grad_norm": 1.0457911491394043,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 152380
+    },
+    {
+      "epoch": 1002.5657894736842,
+      "grad_norm": 1.0000548362731934,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 152390
+    },
+    {
+      "epoch": 1002.6315789473684,
+      "grad_norm": 1.0795695781707764,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 152400
+    },
+    {
+      "epoch": 1002.6973684210526,
+      "grad_norm": 0.7151395082473755,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 152410
+    },
+    {
+      "epoch": 1002.7631578947369,
+      "grad_norm": 0.9494915008544922,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 152420
+    },
+    {
+      "epoch": 1002.828947368421,
+      "grad_norm": 0.8552742004394531,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 152430
+    },
+    {
+      "epoch": 1002.8947368421053,
+      "grad_norm": 0.8879970908164978,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 152440
+    },
+    {
+      "epoch": 1002.9605263157895,
+      "grad_norm": 1.2099488973617554,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 152450
+    },
+    {
+      "epoch": 1003.0263157894736,
+      "grad_norm": 0.8229483366012573,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 152460
+    },
+    {
+      "epoch": 1003.0921052631579,
+      "grad_norm": 0.604310154914856,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 152470
+    },
+    {
+      "epoch": 1003.1578947368421,
+      "grad_norm": 1.2384097576141357,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 152480
+    },
+    {
+      "epoch": 1003.2236842105264,
+      "grad_norm": 0.9572075009346008,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 152490
+    },
+    {
+      "epoch": 1003.2894736842105,
+      "grad_norm": 1.0197408199310303,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 152500
+    },
+    {
+      "epoch": 1003.3552631578947,
+      "grad_norm": 0.9657206535339355,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 152510
+    },
+    {
+      "epoch": 1003.421052631579,
+      "grad_norm": 0.9685333967208862,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 152520
+    },
+    {
+      "epoch": 1003.4868421052631,
+      "grad_norm": 1.1198996305465698,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 152530
+    },
+    {
+      "epoch": 1003.5526315789474,
+      "grad_norm": 1.1644282341003418,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 152540
+    },
+    {
+      "epoch": 1003.6184210526316,
+      "grad_norm": 0.9588817954063416,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 152550
+    },
+    {
+      "epoch": 1003.6842105263158,
+      "grad_norm": 0.7863638997077942,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 152560
+    },
+    {
+      "epoch": 1003.75,
+      "grad_norm": 1.0634634494781494,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 152570
+    },
+    {
+      "epoch": 1003.8157894736842,
+      "grad_norm": 1.312638759613037,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 152580
+    },
+    {
+      "epoch": 1003.8815789473684,
+      "grad_norm": 1.0929923057556152,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 152590
+    },
+    {
+      "epoch": 1003.9473684210526,
+      "grad_norm": 1.4399696588516235,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 152600
+    },
+    {
+      "epoch": 1004.0131578947369,
+      "grad_norm": 1.2716460227966309,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 152610
+    },
+    {
+      "epoch": 1004.078947368421,
+      "grad_norm": 1.0641130208969116,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 152620
+    },
+    {
+      "epoch": 1004.1447368421053,
+      "grad_norm": 0.9772552847862244,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 152630
+    },
+    {
+      "epoch": 1004.2105263157895,
+      "grad_norm": 0.9149281978607178,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 152640
+    },
+    {
+      "epoch": 1004.2763157894736,
+      "grad_norm": 1.0702228546142578,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 152650
+    },
+    {
+      "epoch": 1004.3421052631579,
+      "grad_norm": 1.0155905485153198,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 152660
+    },
+    {
+      "epoch": 1004.4078947368421,
+      "grad_norm": 1.0387847423553467,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 152670
+    },
+    {
+      "epoch": 1004.4736842105264,
+      "grad_norm": 1.1399638652801514,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 152680
+    },
+    {
+      "epoch": 1004.5394736842105,
+      "grad_norm": 1.349084734916687,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 152690
+    },
+    {
+      "epoch": 1004.6052631578947,
+      "grad_norm": 1.1907349824905396,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 152700
+    },
+    {
+      "epoch": 1004.671052631579,
+      "grad_norm": 1.4231032133102417,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 152710
+    },
+    {
+      "epoch": 1004.7368421052631,
+      "grad_norm": 1.0504357814788818,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 152720
+    },
+    {
+      "epoch": 1004.8026315789474,
+      "grad_norm": 0.8143443465232849,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 152730
+    },
+    {
+      "epoch": 1004.8684210526316,
+      "grad_norm": 1.1146156787872314,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 152740
+    },
+    {
+      "epoch": 1004.9342105263158,
+      "grad_norm": 1.0395196676254272,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 152750
+    },
+    {
+      "epoch": 1005.0,
+      "grad_norm": 1.0455608367919922,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 152760
+    },
+    {
+      "epoch": 1005.0657894736842,
+      "grad_norm": 0.8449133038520813,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 152770
+    },
+    {
+      "epoch": 1005.1315789473684,
+      "grad_norm": 0.9601693749427795,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 152780
+    },
+    {
+      "epoch": 1005.1973684210526,
+      "grad_norm": 0.9671313762664795,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 152790
+    },
+    {
+      "epoch": 1005.2631578947369,
+      "grad_norm": 0.8322471976280212,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 152800
+    },
+    {
+      "epoch": 1005.328947368421,
+      "grad_norm": 0.9465487003326416,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 152810
+    },
+    {
+      "epoch": 1005.3947368421053,
+      "grad_norm": 0.8888710141181946,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 152820
+    },
+    {
+      "epoch": 1005.4605263157895,
+      "grad_norm": 1.081684947013855,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 152830
+    },
+    {
+      "epoch": 1005.5263157894736,
+      "grad_norm": 0.9783938527107239,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 152840
+    },
+    {
+      "epoch": 1005.5921052631579,
+      "grad_norm": 1.3762357234954834,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 152850
+    },
+    {
+      "epoch": 1005.6578947368421,
+      "grad_norm": 1.3803174495697021,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 152860
+    },
+    {
+      "epoch": 1005.7236842105264,
+      "grad_norm": 1.0648322105407715,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 152870
+    },
+    {
+      "epoch": 1005.7894736842105,
+      "grad_norm": 1.019877314567566,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 152880
+    },
+    {
+      "epoch": 1005.8552631578947,
+      "grad_norm": 1.0689724683761597,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 152890
+    },
+    {
+      "epoch": 1005.921052631579,
+      "grad_norm": 1.0110927820205688,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 152900
+    },
+    {
+      "epoch": 1005.9868421052631,
+      "grad_norm": 0.8026441335678101,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 152910
+    },
+    {
+      "epoch": 1006.0526315789474,
+      "grad_norm": 0.9327909350395203,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 152920
+    },
+    {
+      "epoch": 1006.1184210526316,
+      "grad_norm": 0.9593319296836853,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 152930
+    },
+    {
+      "epoch": 1006.1842105263158,
+      "grad_norm": 0.9734617471694946,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 152940
+    },
+    {
+      "epoch": 1006.25,
+      "grad_norm": 0.8981104493141174,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 152950
+    },
+    {
+      "epoch": 1006.3157894736842,
+      "grad_norm": 0.6825907826423645,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 152960
+    },
+    {
+      "epoch": 1006.3815789473684,
+      "grad_norm": 1.2038465738296509,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 152970
+    },
+    {
+      "epoch": 1006.4473684210526,
+      "grad_norm": 1.119794487953186,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 152980
+    },
+    {
+      "epoch": 1006.5131578947369,
+      "grad_norm": 1.064179539680481,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 152990
+    },
+    {
+      "epoch": 1006.578947368421,
+      "grad_norm": 1.1156647205352783,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 153000
+    },
+    {
+      "epoch": 1006.6447368421053,
+      "grad_norm": 1.3254365921020508,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 153010
+    },
+    {
+      "epoch": 1006.7105263157895,
+      "grad_norm": 1.4035636186599731,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 153020
+    },
+    {
+      "epoch": 1006.7763157894736,
+      "grad_norm": 1.1695518493652344,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153030
+    },
+    {
+      "epoch": 1006.8421052631579,
+      "grad_norm": 1.270020842552185,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 153040
+    },
+    {
+      "epoch": 1006.9078947368421,
+      "grad_norm": 1.3155039548873901,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 153050
+    },
+    {
+      "epoch": 1006.9736842105264,
+      "grad_norm": 0.8515135645866394,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 153060
+    },
+    {
+      "epoch": 1007.0394736842105,
+      "grad_norm": 0.9089174866676331,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 153070
+    },
+    {
+      "epoch": 1007.1052631578947,
+      "grad_norm": 1.1836307048797607,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 153080
+    },
+    {
+      "epoch": 1007.171052631579,
+      "grad_norm": 1.1035048961639404,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 153090
+    },
+    {
+      "epoch": 1007.2368421052631,
+      "grad_norm": 1.2062311172485352,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 153100
+    },
+    {
+      "epoch": 1007.3026315789474,
+      "grad_norm": 1.3278850317001343,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 153110
+    },
+    {
+      "epoch": 1007.3684210526316,
+      "grad_norm": 1.200394630432129,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153120
+    },
+    {
+      "epoch": 1007.4342105263158,
+      "grad_norm": 1.2503161430358887,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 153130
+    },
+    {
+      "epoch": 1007.5,
+      "grad_norm": 1.327488899230957,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 153140
+    },
+    {
+      "epoch": 1007.5657894736842,
+      "grad_norm": 0.6946982145309448,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 153150
+    },
+    {
+      "epoch": 1007.6315789473684,
+      "grad_norm": 0.8254473805427551,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 153160
+    },
+    {
+      "epoch": 1007.6973684210526,
+      "grad_norm": 0.8691716194152832,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 153170
+    },
+    {
+      "epoch": 1007.7631578947369,
+      "grad_norm": 1.0709201097488403,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 153180
+    },
+    {
+      "epoch": 1007.828947368421,
+      "grad_norm": 1.361719012260437,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 153190
+    },
+    {
+      "epoch": 1007.8947368421053,
+      "grad_norm": 1.2241108417510986,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 153200
+    },
+    {
+      "epoch": 1007.9605263157895,
+      "grad_norm": 1.4689909219741821,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 153210
+    },
+    {
+      "epoch": 1008.0263157894736,
+      "grad_norm": 1.4543124437332153,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 153220
+    },
+    {
+      "epoch": 1008.0921052631579,
+      "grad_norm": 1.4135468006134033,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 153230
+    },
+    {
+      "epoch": 1008.1578947368421,
+      "grad_norm": 1.1939821243286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 153240
+    },
+    {
+      "epoch": 1008.2236842105264,
+      "grad_norm": 1.4653980731964111,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 153250
+    },
+    {
+      "epoch": 1008.2894736842105,
+      "grad_norm": 1.4187058210372925,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153260
+    },
+    {
+      "epoch": 1008.3552631578947,
+      "grad_norm": 1.3027122020721436,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 153270
+    },
+    {
+      "epoch": 1008.421052631579,
+      "grad_norm": 1.2403491735458374,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153280
+    },
+    {
+      "epoch": 1008.4868421052631,
+      "grad_norm": 0.8869355916976929,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153290
+    },
+    {
+      "epoch": 1008.5526315789474,
+      "grad_norm": 1.038567066192627,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 153300
+    },
+    {
+      "epoch": 1008.6184210526316,
+      "grad_norm": 1.1018482446670532,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 153310
+    },
+    {
+      "epoch": 1008.6842105263158,
+      "grad_norm": 0.9891707301139832,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 153320
+    },
+    {
+      "epoch": 1008.75,
+      "grad_norm": 1.3126126527786255,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 153330
+    },
+    {
+      "epoch": 1008.8157894736842,
+      "grad_norm": 1.0261906385421753,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 153340
+    },
+    {
+      "epoch": 1008.8815789473684,
+      "grad_norm": 1.0585100650787354,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 153350
+    },
+    {
+      "epoch": 1008.9473684210526,
+      "grad_norm": 0.7990329265594482,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 153360
+    },
+    {
+      "epoch": 1009.0131578947369,
+      "grad_norm": 0.703535795211792,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 153370
+    },
+    {
+      "epoch": 1009.078947368421,
+      "grad_norm": 0.7562394142150879,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 153380
+    },
+    {
+      "epoch": 1009.1447368421053,
+      "grad_norm": 0.6611744165420532,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 153390
+    },
+    {
+      "epoch": 1009.2105263157895,
+      "grad_norm": 0.9061070084571838,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 153400
+    },
+    {
+      "epoch": 1009.2763157894736,
+      "grad_norm": 1.1760482788085938,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 153410
+    },
+    {
+      "epoch": 1009.3421052631579,
+      "grad_norm": 1.6927746534347534,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 153420
+    },
+    {
+      "epoch": 1009.4078947368421,
+      "grad_norm": 1.375290870666504,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 153430
+    },
+    {
+      "epoch": 1009.4736842105264,
+      "grad_norm": 1.3933255672454834,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153440
+    },
+    {
+      "epoch": 1009.5394736842105,
+      "grad_norm": 1.4885764122009277,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153450
+    },
+    {
+      "epoch": 1009.6052631578947,
+      "grad_norm": 0.8054714798927307,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 153460
+    },
+    {
+      "epoch": 1009.671052631579,
+      "grad_norm": 0.9713701605796814,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 153470
+    },
+    {
+      "epoch": 1009.7368421052631,
+      "grad_norm": 1.0270944833755493,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 153480
+    },
+    {
+      "epoch": 1009.8026315789474,
+      "grad_norm": 1.0508967638015747,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 153490
+    },
+    {
+      "epoch": 1009.8684210526316,
+      "grad_norm": 1.0318694114685059,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 153500
+    },
+    {
+      "epoch": 1009.9342105263158,
+      "grad_norm": 1.081775426864624,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 153510
+    },
+    {
+      "epoch": 1010.0,
+      "grad_norm": 0.6069177389144897,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 153520
+    },
+    {
+      "epoch": 1010.0657894736842,
+      "grad_norm": 0.6230463981628418,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 153530
+    },
+    {
+      "epoch": 1010.1315789473684,
+      "grad_norm": 0.8272234201431274,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 153540
+    },
+    {
+      "epoch": 1010.1973684210526,
+      "grad_norm": 1.0551385879516602,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 153550
+    },
+    {
+      "epoch": 1010.2631578947369,
+      "grad_norm": 1.039683222770691,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 153560
+    },
+    {
+      "epoch": 1010.328947368421,
+      "grad_norm": 0.7899230122566223,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 153570
+    },
+    {
+      "epoch": 1010.3947368421053,
+      "grad_norm": 0.8216849565505981,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153580
+    },
+    {
+      "epoch": 1010.4605263157895,
+      "grad_norm": 0.9571535587310791,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 153590
+    },
+    {
+      "epoch": 1010.5263157894736,
+      "grad_norm": 1.1962392330169678,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 153600
+    },
+    {
+      "epoch": 1010.5921052631579,
+      "grad_norm": 0.8668336868286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 153610
+    },
+    {
+      "epoch": 1010.6578947368421,
+      "grad_norm": 1.1946544647216797,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 153620
+    },
+    {
+      "epoch": 1010.7236842105264,
+      "grad_norm": 0.8684710264205933,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 153630
+    },
+    {
+      "epoch": 1010.7894736842105,
+      "grad_norm": 1.0615448951721191,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 153640
+    },
+    {
+      "epoch": 1010.8552631578947,
+      "grad_norm": 1.3180712461471558,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 153650
+    },
+    {
+      "epoch": 1010.921052631579,
+      "grad_norm": 1.4111500978469849,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 153660
+    },
+    {
+      "epoch": 1010.9868421052631,
+      "grad_norm": 1.3890069723129272,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 153670
+    },
+    {
+      "epoch": 1011.0526315789474,
+      "grad_norm": 0.9825431108474731,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 153680
+    },
+    {
+      "epoch": 1011.1184210526316,
+      "grad_norm": 0.8479264974594116,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 153690
+    },
+    {
+      "epoch": 1011.1842105263158,
+      "grad_norm": 1.2472234964370728,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 153700
+    },
+    {
+      "epoch": 1011.25,
+      "grad_norm": 1.0906670093536377,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 153710
+    },
+    {
+      "epoch": 1011.3157894736842,
+      "grad_norm": 1.144405722618103,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 153720
+    },
+    {
+      "epoch": 1011.3815789473684,
+      "grad_norm": 0.8326566219329834,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 153730
+    },
+    {
+      "epoch": 1011.4473684210526,
+      "grad_norm": 1.1683098077774048,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 153740
+    },
+    {
+      "epoch": 1011.5131578947369,
+      "grad_norm": 0.7336865663528442,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 153750
+    },
+    {
+      "epoch": 1011.578947368421,
+      "grad_norm": 1.1023454666137695,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 153760
+    },
+    {
+      "epoch": 1011.6447368421053,
+      "grad_norm": 1.045922040939331,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 153770
+    },
+    {
+      "epoch": 1011.7105263157895,
+      "grad_norm": 0.9625352621078491,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 153780
+    },
+    {
+      "epoch": 1011.7763157894736,
+      "grad_norm": 1.1642197370529175,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 153790
+    },
+    {
+      "epoch": 1011.8421052631579,
+      "grad_norm": 1.1206856966018677,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 153800
+    },
+    {
+      "epoch": 1011.9078947368421,
+      "grad_norm": 1.251189947128296,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153810
+    },
+    {
+      "epoch": 1011.9736842105264,
+      "grad_norm": 0.9257858991622925,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 153820
+    },
+    {
+      "epoch": 1012.0394736842105,
+      "grad_norm": 0.9416394233703613,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 153830
+    },
+    {
+      "epoch": 1012.1052631578947,
+      "grad_norm": 0.7011066675186157,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153840
+    },
+    {
+      "epoch": 1012.171052631579,
+      "grad_norm": 0.9953075051307678,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 153850
+    },
+    {
+      "epoch": 1012.2368421052631,
+      "grad_norm": 1.0580404996871948,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 153860
+    },
+    {
+      "epoch": 1012.3026315789474,
+      "grad_norm": 0.9159753918647766,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 153870
+    },
+    {
+      "epoch": 1012.3684210526316,
+      "grad_norm": 1.2467912435531616,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 153880
+    },
+    {
+      "epoch": 1012.4342105263158,
+      "grad_norm": 0.8988513350486755,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 153890
+    },
+    {
+      "epoch": 1012.5,
+      "grad_norm": 0.8975652456283569,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 153900
+    },
+    {
+      "epoch": 1012.5657894736842,
+      "grad_norm": 0.9276600480079651,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 153910
+    },
+    {
+      "epoch": 1012.6315789473684,
+      "grad_norm": 1.1802819967269897,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 153920
+    },
+    {
+      "epoch": 1012.6973684210526,
+      "grad_norm": 1.5919638872146606,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 153930
+    },
+    {
+      "epoch": 1012.7631578947369,
+      "grad_norm": 0.8894348740577698,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 153940
+    },
+    {
+      "epoch": 1012.828947368421,
+      "grad_norm": 0.9059845209121704,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 153950
+    },
+    {
+      "epoch": 1012.8947368421053,
+      "grad_norm": 0.7619372010231018,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 153960
+    },
+    {
+      "epoch": 1012.9605263157895,
+      "grad_norm": 1.105398416519165,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 153970
+    },
+    {
+      "epoch": 1013.0263157894736,
+      "grad_norm": 0.6441566944122314,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 153980
+    },
+    {
+      "epoch": 1013.0921052631579,
+      "grad_norm": 1.1921231746673584,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 153990
+    },
+    {
+      "epoch": 1013.1578947368421,
+      "grad_norm": 1.3228693008422852,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 154000
+    },
+    {
+      "epoch": 1013.2236842105264,
+      "grad_norm": 1.2285857200622559,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 154010
+    },
+    {
+      "epoch": 1013.2894736842105,
+      "grad_norm": 0.5511453747749329,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 154020
+    },
+    {
+      "epoch": 1013.3552631578947,
+      "grad_norm": 0.8343952894210815,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 154030
+    },
+    {
+      "epoch": 1013.421052631579,
+      "grad_norm": 0.9927038550376892,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 154040
+    },
+    {
+      "epoch": 1013.4868421052631,
+      "grad_norm": 0.9461319446563721,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 154050
+    },
+    {
+      "epoch": 1013.5526315789474,
+      "grad_norm": 1.170270562171936,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 154060
+    },
+    {
+      "epoch": 1013.6184210526316,
+      "grad_norm": 1.1801016330718994,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 154070
+    },
+    {
+      "epoch": 1013.6842105263158,
+      "grad_norm": 1.064570665359497,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 154080
+    },
+    {
+      "epoch": 1013.75,
+      "grad_norm": 0.9943475127220154,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 154090
+    },
+    {
+      "epoch": 1013.8157894736842,
+      "grad_norm": 1.1326581239700317,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 154100
+    },
+    {
+      "epoch": 1013.8815789473684,
+      "grad_norm": 1.2644318342208862,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 154110
+    },
+    {
+      "epoch": 1013.9473684210526,
+      "grad_norm": 0.9471270442008972,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 154120
+    },
+    {
+      "epoch": 1014.0131578947369,
+      "grad_norm": 1.1460727453231812,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 154130
+    },
+    {
+      "epoch": 1014.078947368421,
+      "grad_norm": 1.2603262662887573,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 154140
+    },
+    {
+      "epoch": 1014.1447368421053,
+      "grad_norm": 0.8152709007263184,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 154150
+    },
+    {
+      "epoch": 1014.2105263157895,
+      "grad_norm": 0.7039467692375183,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 154160
+    },
+    {
+      "epoch": 1014.2763157894736,
+      "grad_norm": 1.0728949308395386,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 154170
+    },
+    {
+      "epoch": 1014.3421052631579,
+      "grad_norm": 0.7830701470375061,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 154180
+    },
+    {
+      "epoch": 1014.4078947368421,
+      "grad_norm": 0.8367193937301636,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 154190
+    },
+    {
+      "epoch": 1014.4736842105264,
+      "grad_norm": 1.973212480545044,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 154200
+    },
+    {
+      "epoch": 1014.5394736842105,
+      "grad_norm": 1.815258502960205,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 154210
+    },
+    {
+      "epoch": 1014.6052631578947,
+      "grad_norm": 2.393010377883911,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 154220
+    },
+    {
+      "epoch": 1014.671052631579,
+      "grad_norm": 1.425652027130127,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 154230
+    },
+    {
+      "epoch": 1014.7368421052631,
+      "grad_norm": 1.2232511043548584,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 154240
+    },
+    {
+      "epoch": 1014.8026315789474,
+      "grad_norm": 1.2349510192871094,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 154250
+    },
+    {
+      "epoch": 1014.8684210526316,
+      "grad_norm": 1.3472286462783813,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 154260
+    },
+    {
+      "epoch": 1014.9342105263158,
+      "grad_norm": 1.1970043182373047,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 154270
+    },
+    {
+      "epoch": 1015.0,
+      "grad_norm": 0.9629507660865784,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 154280
+    },
+    {
+      "epoch": 1015.0657894736842,
+      "grad_norm": 0.9816302061080933,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 154290
+    },
+    {
+      "epoch": 1015.1315789473684,
+      "grad_norm": 1.125002384185791,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 154300
+    },
+    {
+      "epoch": 1015.1973684210526,
+      "grad_norm": 1.284298062324524,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 154310
+    },
+    {
+      "epoch": 1015.2631578947369,
+      "grad_norm": 1.4155656099319458,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 154320
+    },
+    {
+      "epoch": 1015.328947368421,
+      "grad_norm": 1.0492305755615234,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 154330
+    },
+    {
+      "epoch": 1015.3947368421053,
+      "grad_norm": 1.006813406944275,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 154340
+    },
+    {
+      "epoch": 1015.4605263157895,
+      "grad_norm": 1.409071683883667,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 154350
+    },
+    {
+      "epoch": 1015.5263157894736,
+      "grad_norm": 1.6593658924102783,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 154360
+    },
+    {
+      "epoch": 1015.5921052631579,
+      "grad_norm": 1.4697237014770508,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 154370
+    },
+    {
+      "epoch": 1015.6578947368421,
+      "grad_norm": 1.9840725660324097,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 154380
+    },
+    {
+      "epoch": 1015.7236842105264,
+      "grad_norm": 1.4490352869033813,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 154390
+    },
+    {
+      "epoch": 1015.7894736842105,
+      "grad_norm": 1.1679290533065796,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 154400
+    },
+    {
+      "epoch": 1015.8552631578947,
+      "grad_norm": 1.219805121421814,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 154410
+    },
+    {
+      "epoch": 1015.921052631579,
+      "grad_norm": 1.1907403469085693,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 154420
+    },
+    {
+      "epoch": 1015.9868421052631,
+      "grad_norm": 1.4506357908248901,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 154430
+    },
+    {
+      "epoch": 1016.0526315789474,
+      "grad_norm": 1.450147032737732,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 154440
+    },
+    {
+      "epoch": 1016.1184210526316,
+      "grad_norm": 1.0505656003952026,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 154450
+    },
+    {
+      "epoch": 1016.1842105263158,
+      "grad_norm": 1.4850653409957886,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 154460
+    },
+    {
+      "epoch": 1016.25,
+      "grad_norm": 1.165228247642517,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 154470
+    },
+    {
+      "epoch": 1016.3157894736842,
+      "grad_norm": 0.9636247754096985,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 154480
+    },
+    {
+      "epoch": 1016.3815789473684,
+      "grad_norm": 1.0024343729019165,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 154490
+    },
+    {
+      "epoch": 1016.4473684210526,
+      "grad_norm": 0.8500187397003174,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 154500
+    },
+    {
+      "epoch": 1016.5131578947369,
+      "grad_norm": 1.01841139793396,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 154510
+    },
+    {
+      "epoch": 1016.578947368421,
+      "grad_norm": 0.774018406867981,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 154520
+    },
+    {
+      "epoch": 1016.6447368421053,
+      "grad_norm": 1.0345419645309448,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 154530
+    },
+    {
+      "epoch": 1016.7105263157895,
+      "grad_norm": 1.3381010293960571,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 154540
+    },
+    {
+      "epoch": 1016.7763157894736,
+      "grad_norm": 1.0903173685073853,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 154550
+    },
+    {
+      "epoch": 1016.8421052631579,
+      "grad_norm": 0.7370033860206604,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 154560
+    },
+    {
+      "epoch": 1016.9078947368421,
+      "grad_norm": 0.9964326620101929,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 154570
+    },
+    {
+      "epoch": 1016.9736842105264,
+      "grad_norm": 1.0719693899154663,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 154580
+    },
+    {
+      "epoch": 1017.0394736842105,
+      "grad_norm": 1.0311167240142822,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 154590
+    },
+    {
+      "epoch": 1017.1052631578947,
+      "grad_norm": 1.1289986371994019,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 154600
+    },
+    {
+      "epoch": 1017.171052631579,
+      "grad_norm": 0.8267912864685059,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 154610
+    },
+    {
+      "epoch": 1017.2368421052631,
+      "grad_norm": 1.1450597047805786,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 154620
+    },
+    {
+      "epoch": 1017.3026315789474,
+      "grad_norm": 0.8567523956298828,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 154630
+    },
+    {
+      "epoch": 1017.3684210526316,
+      "grad_norm": 0.6963806748390198,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 154640
+    },
+    {
+      "epoch": 1017.4342105263158,
+      "grad_norm": 0.7073983550071716,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 154650
+    },
+    {
+      "epoch": 1017.5,
+      "grad_norm": 0.6711755394935608,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 154660
+    },
+    {
+      "epoch": 1017.5657894736842,
+      "grad_norm": 0.7169049382209778,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 154670
+    },
+    {
+      "epoch": 1017.6315789473684,
+      "grad_norm": 0.6535640954971313,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 154680
+    },
+    {
+      "epoch": 1017.6973684210526,
+      "grad_norm": 0.5502938032150269,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 154690
+    },
+    {
+      "epoch": 1017.7631578947369,
+      "grad_norm": 0.9529855847358704,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 154700
+    },
+    {
+      "epoch": 1017.828947368421,
+      "grad_norm": 1.1591817140579224,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 154710
+    },
+    {
+      "epoch": 1017.8947368421053,
+      "grad_norm": 1.1533323526382446,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 154720
+    },
+    {
+      "epoch": 1017.9605263157895,
+      "grad_norm": 1.2087000608444214,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 154730
+    },
+    {
+      "epoch": 1018.0263157894736,
+      "grad_norm": 0.936263918876648,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 154740
+    },
+    {
+      "epoch": 1018.0921052631579,
+      "grad_norm": 1.1925992965698242,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 154750
+    },
+    {
+      "epoch": 1018.1578947368421,
+      "grad_norm": 1.5353156328201294,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 154760
+    },
+    {
+      "epoch": 1018.2236842105264,
+      "grad_norm": 1.3622137308120728,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 154770
+    },
+    {
+      "epoch": 1018.2894736842105,
+      "grad_norm": 1.236350417137146,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 154780
+    },
+    {
+      "epoch": 1018.3552631578947,
+      "grad_norm": 1.0077983140945435,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 154790
+    },
+    {
+      "epoch": 1018.421052631579,
+      "grad_norm": 1.119370460510254,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 154800
+    },
+    {
+      "epoch": 1018.4868421052631,
+      "grad_norm": 1.009660005569458,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 154810
+    },
+    {
+      "epoch": 1018.5526315789474,
+      "grad_norm": 1.031388759613037,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 154820
+    },
+    {
+      "epoch": 1018.6184210526316,
+      "grad_norm": 0.6950995326042175,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 154830
+    },
+    {
+      "epoch": 1018.6842105263158,
+      "grad_norm": 0.9287242293357849,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 154840
+    },
+    {
+      "epoch": 1018.75,
+      "grad_norm": 0.8696134090423584,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 154850
+    },
+    {
+      "epoch": 1018.8157894736842,
+      "grad_norm": 0.8959367275238037,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 154860
+    },
+    {
+      "epoch": 1018.8815789473684,
+      "grad_norm": 1.239129900932312,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 154870
+    },
+    {
+      "epoch": 1018.9473684210526,
+      "grad_norm": 0.8386613726615906,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 154880
+    },
+    {
+      "epoch": 1019.0131578947369,
+      "grad_norm": 1.1517994403839111,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 154890
+    },
+    {
+      "epoch": 1019.078947368421,
+      "grad_norm": 1.2050758600234985,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 154900
+    },
+    {
+      "epoch": 1019.1447368421053,
+      "grad_norm": 1.15315842628479,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 154910
+    },
+    {
+      "epoch": 1019.2105263157895,
+      "grad_norm": 0.8363189697265625,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 154920
+    },
+    {
+      "epoch": 1019.2763157894736,
+      "grad_norm": 0.8743520379066467,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 154930
+    },
+    {
+      "epoch": 1019.3421052631579,
+      "grad_norm": 1.1668791770935059,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 154940
+    },
+    {
+      "epoch": 1019.4078947368421,
+      "grad_norm": 1.0123094320297241,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 154950
+    },
+    {
+      "epoch": 1019.4736842105264,
+      "grad_norm": 0.5643197894096375,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 154960
+    },
+    {
+      "epoch": 1019.5394736842105,
+      "grad_norm": 0.9855898022651672,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 154970
+    },
+    {
+      "epoch": 1019.6052631578947,
+      "grad_norm": 0.9356517791748047,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 154980
+    },
+    {
+      "epoch": 1019.671052631579,
+      "grad_norm": 0.853830099105835,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 154990
+    },
+    {
+      "epoch": 1019.7368421052631,
+      "grad_norm": 1.0371325016021729,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 155000
+    },
+    {
+      "epoch": 1019.8026315789474,
+      "grad_norm": 0.9705318212509155,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 155010
+    },
+    {
+      "epoch": 1019.8684210526316,
+      "grad_norm": 0.9544471502304077,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 155020
+    },
+    {
+      "epoch": 1019.9342105263158,
+      "grad_norm": 1.049113392829895,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 155030
+    },
+    {
+      "epoch": 1020.0,
+      "grad_norm": 1.123701572418213,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 155040
+    },
+    {
+      "epoch": 1020.0657894736842,
+      "grad_norm": 1.035402536392212,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 155050
+    },
+    {
+      "epoch": 1020.1315789473684,
+      "grad_norm": 1.1114774942398071,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 155060
+    },
+    {
+      "epoch": 1020.1973684210526,
+      "grad_norm": 0.7469269037246704,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 155070
+    },
+    {
+      "epoch": 1020.2631578947369,
+      "grad_norm": 1.2120532989501953,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 155080
+    },
+    {
+      "epoch": 1020.328947368421,
+      "grad_norm": 0.9209999442100525,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 155090
+    },
+    {
+      "epoch": 1020.3947368421053,
+      "grad_norm": 1.0287814140319824,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 155100
+    },
+    {
+      "epoch": 1020.4605263157895,
+      "grad_norm": 1.212156057357788,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 155110
+    },
+    {
+      "epoch": 1020.5263157894736,
+      "grad_norm": 1.101295828819275,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 155120
+    },
+    {
+      "epoch": 1020.5921052631579,
+      "grad_norm": 1.2472134828567505,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 155130
+    },
+    {
+      "epoch": 1020.6578947368421,
+      "grad_norm": 1.4082173109054565,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 155140
+    },
+    {
+      "epoch": 1020.7236842105264,
+      "grad_norm": 1.0719764232635498,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 155150
+    },
+    {
+      "epoch": 1020.7894736842105,
+      "grad_norm": 1.4822707176208496,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 155160
+    },
+    {
+      "epoch": 1020.8552631578947,
+      "grad_norm": 1.1437565088272095,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 155170
+    },
+    {
+      "epoch": 1020.921052631579,
+      "grad_norm": 1.258529543876648,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 155180
+    },
+    {
+      "epoch": 1020.9868421052631,
+      "grad_norm": 1.0920777320861816,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 155190
+    },
+    {
+      "epoch": 1021.0526315789474,
+      "grad_norm": 1.4850847721099854,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 155200
+    },
+    {
+      "epoch": 1021.1184210526316,
+      "grad_norm": 0.7385463118553162,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 155210
+    },
+    {
+      "epoch": 1021.1842105263158,
+      "grad_norm": 1.168807864189148,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 155220
+    },
+    {
+      "epoch": 1021.25,
+      "grad_norm": 1.1659116744995117,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 155230
+    },
+    {
+      "epoch": 1021.3157894736842,
+      "grad_norm": 0.993746817111969,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 155240
+    },
+    {
+      "epoch": 1021.3815789473684,
+      "grad_norm": 1.1375447511672974,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 155250
+    },
+    {
+      "epoch": 1021.4473684210526,
+      "grad_norm": 0.8850059509277344,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 155260
+    },
+    {
+      "epoch": 1021.5131578947369,
+      "grad_norm": 0.8182586431503296,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 155270
+    },
+    {
+      "epoch": 1021.578947368421,
+      "grad_norm": 0.7995529770851135,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 155280
+    },
+    {
+      "epoch": 1021.6447368421053,
+      "grad_norm": 1.0963636636734009,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 155290
+    },
+    {
+      "epoch": 1021.7105263157895,
+      "grad_norm": 1.1766555309295654,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 155300
+    },
+    {
+      "epoch": 1021.7763157894736,
+      "grad_norm": 1.4533417224884033,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 155310
+    },
+    {
+      "epoch": 1021.8421052631579,
+      "grad_norm": 0.9066959023475647,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 155320
+    },
+    {
+      "epoch": 1021.9078947368421,
+      "grad_norm": 0.9514365196228027,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 155330
+    },
+    {
+      "epoch": 1021.9736842105264,
+      "grad_norm": 0.9749160408973694,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 155340
+    },
+    {
+      "epoch": 1022.0394736842105,
+      "grad_norm": 1.3155367374420166,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 155350
+    },
+    {
+      "epoch": 1022.1052631578947,
+      "grad_norm": 0.8676218390464783,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 155360
+    },
+    {
+      "epoch": 1022.171052631579,
+      "grad_norm": 1.047354817390442,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 155370
+    },
+    {
+      "epoch": 1022.2368421052631,
+      "grad_norm": 1.1073765754699707,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 155380
+    },
+    {
+      "epoch": 1022.3026315789474,
+      "grad_norm": 1.0393955707550049,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 155390
+    },
+    {
+      "epoch": 1022.3684210526316,
+      "grad_norm": 0.886330783367157,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 155400
+    },
+    {
+      "epoch": 1022.4342105263158,
+      "grad_norm": 0.9492865800857544,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 155410
+    },
+    {
+      "epoch": 1022.5,
+      "grad_norm": 1.4241212606430054,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 155420
+    },
+    {
+      "epoch": 1022.5657894736842,
+      "grad_norm": 1.7502048015594482,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 155430
+    },
+    {
+      "epoch": 1022.6315789473684,
+      "grad_norm": 1.311353087425232,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 155440
+    },
+    {
+      "epoch": 1022.6973684210526,
+      "grad_norm": 1.1434524059295654,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 155450
+    },
+    {
+      "epoch": 1022.7631578947369,
+      "grad_norm": 1.5268412828445435,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 155460
+    },
+    {
+      "epoch": 1022.828947368421,
+      "grad_norm": 1.0521647930145264,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 155470
+    },
+    {
+      "epoch": 1022.8947368421053,
+      "grad_norm": 1.2657438516616821,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 155480
+    },
+    {
+      "epoch": 1022.9605263157895,
+      "grad_norm": 1.5762451887130737,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 155490
+    },
+    {
+      "epoch": 1023.0263157894736,
+      "grad_norm": 0.9546888470649719,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 155500
+    },
+    {
+      "epoch": 1023.0921052631579,
+      "grad_norm": 0.966810941696167,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 155510
+    },
+    {
+      "epoch": 1023.1578947368421,
+      "grad_norm": 1.2217552661895752,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 155520
+    },
+    {
+      "epoch": 1023.2236842105264,
+      "grad_norm": 1.2993433475494385,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 155530
+    },
+    {
+      "epoch": 1023.2894736842105,
+      "grad_norm": 1.1217420101165771,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 155540
+    },
+    {
+      "epoch": 1023.3552631578947,
+      "grad_norm": 1.1864551305770874,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 155550
+    },
+    {
+      "epoch": 1023.421052631579,
+      "grad_norm": 1.1899709701538086,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 155560
+    },
+    {
+      "epoch": 1023.4868421052631,
+      "grad_norm": 1.223408818244934,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 155570
+    },
+    {
+      "epoch": 1023.5526315789474,
+      "grad_norm": 1.22601318359375,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 155580
+    },
+    {
+      "epoch": 1023.6184210526316,
+      "grad_norm": 0.761501133441925,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 155590
+    },
+    {
+      "epoch": 1023.6842105263158,
+      "grad_norm": 1.0578773021697998,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 155600
+    },
+    {
+      "epoch": 1023.75,
+      "grad_norm": 1.3183938264846802,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 155610
+    },
+    {
+      "epoch": 1023.8157894736842,
+      "grad_norm": 1.4918771982192993,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 155620
+    },
+    {
+      "epoch": 1023.8815789473684,
+      "grad_norm": 1.0870895385742188,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 155630
+    },
+    {
+      "epoch": 1023.9473684210526,
+      "grad_norm": 1.3157042264938354,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 155640
+    },
+    {
+      "epoch": 1024.0131578947369,
+      "grad_norm": 1.3045361042022705,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 155650
+    },
+    {
+      "epoch": 1024.078947368421,
+      "grad_norm": 1.0080130100250244,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 155660
+    },
+    {
+      "epoch": 1024.1447368421052,
+      "grad_norm": 1.2834631204605103,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 155670
+    },
+    {
+      "epoch": 1024.2105263157894,
+      "grad_norm": 1.2468620538711548,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 155680
+    },
+    {
+      "epoch": 1024.2763157894738,
+      "grad_norm": 0.9517583250999451,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 155690
+    },
+    {
+      "epoch": 1024.342105263158,
+      "grad_norm": 1.0547385215759277,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 155700
+    },
+    {
+      "epoch": 1024.407894736842,
+      "grad_norm": 1.4880155324935913,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 155710
+    },
+    {
+      "epoch": 1024.4736842105262,
+      "grad_norm": 1.133327603340149,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 155720
+    },
+    {
+      "epoch": 1024.5394736842106,
+      "grad_norm": 1.1422955989837646,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 155730
+    },
+    {
+      "epoch": 1024.6052631578948,
+      "grad_norm": 1.3525456190109253,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 155740
+    },
+    {
+      "epoch": 1024.671052631579,
+      "grad_norm": 1.063536524772644,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 155750
+    },
+    {
+      "epoch": 1024.7368421052631,
+      "grad_norm": 1.0734652280807495,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 155760
+    },
+    {
+      "epoch": 1024.8026315789473,
+      "grad_norm": 1.1811069250106812,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 155770
+    },
+    {
+      "epoch": 1024.8684210526317,
+      "grad_norm": 0.923505425453186,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 155780
+    },
+    {
+      "epoch": 1024.9342105263158,
+      "grad_norm": 0.7430877685546875,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 155790
+    },
+    {
+      "epoch": 1025.0,
+      "grad_norm": 0.5086251497268677,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 155800
+    },
+    {
+      "epoch": 1025.0657894736842,
+      "grad_norm": 0.8926265239715576,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 155810
+    },
+    {
+      "epoch": 1025.1315789473683,
+      "grad_norm": 1.6458979845046997,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 155820
+    },
+    {
+      "epoch": 1025.1973684210527,
+      "grad_norm": 0.9550696611404419,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 155830
+    },
+    {
+      "epoch": 1025.2631578947369,
+      "grad_norm": 1.3163942098617554,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 155840
+    },
+    {
+      "epoch": 1025.328947368421,
+      "grad_norm": 1.4571993350982666,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 155850
+    },
+    {
+      "epoch": 1025.3947368421052,
+      "grad_norm": 1.3019025325775146,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 155860
+    },
+    {
+      "epoch": 1025.4605263157894,
+      "grad_norm": 1.154705286026001,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 155870
+    },
+    {
+      "epoch": 1025.5263157894738,
+      "grad_norm": 1.1474753618240356,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 155880
+    },
+    {
+      "epoch": 1025.592105263158,
+      "grad_norm": 0.7636915445327759,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 155890
+    },
+    {
+      "epoch": 1025.657894736842,
+      "grad_norm": 1.04159414768219,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 155900
+    },
+    {
+      "epoch": 1025.7236842105262,
+      "grad_norm": 1.030527114868164,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 155910
+    },
+    {
+      "epoch": 1025.7894736842106,
+      "grad_norm": 1.1383177042007446,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 155920
+    },
+    {
+      "epoch": 1025.8552631578948,
+      "grad_norm": 1.3946573734283447,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 155930
+    },
+    {
+      "epoch": 1025.921052631579,
+      "grad_norm": 1.2071908712387085,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 155940
+    },
+    {
+      "epoch": 1025.9868421052631,
+      "grad_norm": 1.3407089710235596,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 155950
+    },
+    {
+      "epoch": 1026.0526315789473,
+      "grad_norm": 0.9338118433952332,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 155960
+    },
+    {
+      "epoch": 1026.1184210526317,
+      "grad_norm": 1.0861817598342896,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 155970
+    },
+    {
+      "epoch": 1026.1842105263158,
+      "grad_norm": 1.192749261856079,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 155980
+    },
+    {
+      "epoch": 1026.25,
+      "grad_norm": 1.2024480104446411,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 155990
+    },
+    {
+      "epoch": 1026.3157894736842,
+      "grad_norm": 1.0778899192810059,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 156000
+    },
+    {
+      "epoch": 1026.3815789473683,
+      "grad_norm": 1.0766499042510986,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 156010
+    },
+    {
+      "epoch": 1026.4473684210527,
+      "grad_norm": 1.1861462593078613,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 156020
+    },
+    {
+      "epoch": 1026.5131578947369,
+      "grad_norm": 1.0940054655075073,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 156030
+    },
+    {
+      "epoch": 1026.578947368421,
+      "grad_norm": 1.3079572916030884,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 156040
+    },
+    {
+      "epoch": 1026.6447368421052,
+      "grad_norm": 1.1062140464782715,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 156050
+    },
+    {
+      "epoch": 1026.7105263157894,
+      "grad_norm": 1.2502998113632202,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 156060
+    },
+    {
+      "epoch": 1026.7763157894738,
+      "grad_norm": 0.586124837398529,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 156070
+    },
+    {
+      "epoch": 1026.842105263158,
+      "grad_norm": 0.6738919019699097,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 156080
+    },
+    {
+      "epoch": 1026.907894736842,
+      "grad_norm": 0.9908142685890198,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 156090
+    },
+    {
+      "epoch": 1026.9736842105262,
+      "grad_norm": 1.2288180589675903,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 156100
+    },
+    {
+      "epoch": 1027.0394736842106,
+      "grad_norm": 1.325110673904419,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 156110
+    },
+    {
+      "epoch": 1027.1052631578948,
+      "grad_norm": 1.1032160520553589,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 156120
+    },
+    {
+      "epoch": 1027.171052631579,
+      "grad_norm": 1.2170844078063965,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 156130
+    },
+    {
+      "epoch": 1027.2368421052631,
+      "grad_norm": 0.6858425140380859,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 156140
+    },
+    {
+      "epoch": 1027.3026315789473,
+      "grad_norm": 1.3081910610198975,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 156150
+    },
+    {
+      "epoch": 1027.3684210526317,
+      "grad_norm": 1.1171272993087769,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 156160
+    },
+    {
+      "epoch": 1027.4342105263158,
+      "grad_norm": 1.35430109500885,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 156170
+    },
+    {
+      "epoch": 1027.5,
+      "grad_norm": 1.3601711988449097,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 156180
+    },
+    {
+      "epoch": 1027.5657894736842,
+      "grad_norm": 1.4882049560546875,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 156190
+    },
+    {
+      "epoch": 1027.6315789473683,
+      "grad_norm": 0.9452913999557495,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 156200
+    },
+    {
+      "epoch": 1027.6973684210527,
+      "grad_norm": 1.0488688945770264,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 156210
+    },
+    {
+      "epoch": 1027.7631578947369,
+      "grad_norm": 1.0508980751037598,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 156220
+    },
+    {
+      "epoch": 1027.828947368421,
+      "grad_norm": 1.4618175029754639,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 156230
+    },
+    {
+      "epoch": 1027.8947368421052,
+      "grad_norm": 0.8647827506065369,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 156240
+    },
+    {
+      "epoch": 1027.9605263157894,
+      "grad_norm": 1.1985787153244019,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 156250
+    },
+    {
+      "epoch": 1028.0263157894738,
+      "grad_norm": 1.6828774213790894,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 156260
+    },
+    {
+      "epoch": 1028.092105263158,
+      "grad_norm": 1.015224814414978,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 156270
+    },
+    {
+      "epoch": 1028.157894736842,
+      "grad_norm": 1.148722767829895,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 156280
+    },
+    {
+      "epoch": 1028.2236842105262,
+      "grad_norm": 0.7410078048706055,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 156290
+    },
+    {
+      "epoch": 1028.2894736842106,
+      "grad_norm": 0.7929697036743164,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 156300
+    },
+    {
+      "epoch": 1028.3552631578948,
+      "grad_norm": 1.1712801456451416,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 156310
+    },
+    {
+      "epoch": 1028.421052631579,
+      "grad_norm": 0.8382496237754822,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 156320
+    },
+    {
+      "epoch": 1028.4868421052631,
+      "grad_norm": 0.9845030307769775,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 156330
+    },
+    {
+      "epoch": 1028.5526315789473,
+      "grad_norm": 0.8377711176872253,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 156340
+    },
+    {
+      "epoch": 1028.6184210526317,
+      "grad_norm": 0.874224066734314,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 156350
+    },
+    {
+      "epoch": 1028.6842105263158,
+      "grad_norm": 1.0614688396453857,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 156360
+    },
+    {
+      "epoch": 1028.75,
+      "grad_norm": 0.947669267654419,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 156370
+    },
+    {
+      "epoch": 1028.8157894736842,
+      "grad_norm": 0.8915820121765137,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 156380
+    },
+    {
+      "epoch": 1028.8815789473683,
+      "grad_norm": 1.1027073860168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 156390
+    },
+    {
+      "epoch": 1028.9473684210527,
+      "grad_norm": 0.9126038551330566,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 156400
+    },
+    {
+      "epoch": 1029.0131578947369,
+      "grad_norm": 1.0201152563095093,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 156410
+    },
+    {
+      "epoch": 1029.078947368421,
+      "grad_norm": 0.7695257067680359,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 156420
+    },
+    {
+      "epoch": 1029.1447368421052,
+      "grad_norm": 1.1396857500076294,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 156430
+    },
+    {
+      "epoch": 1029.2105263157894,
+      "grad_norm": 1.5026952028274536,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 156440
+    },
+    {
+      "epoch": 1029.2763157894738,
+      "grad_norm": 1.048147201538086,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 156450
+    },
+    {
+      "epoch": 1029.342105263158,
+      "grad_norm": 1.0914058685302734,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 156460
+    },
+    {
+      "epoch": 1029.407894736842,
+      "grad_norm": 0.9121741056442261,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 156470
+    },
+    {
+      "epoch": 1029.4736842105262,
+      "grad_norm": 1.072133183479309,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 156480
+    },
+    {
+      "epoch": 1029.5394736842106,
+      "grad_norm": 1.3796448707580566,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 156490
+    },
+    {
+      "epoch": 1029.6052631578948,
+      "grad_norm": 1.3151845932006836,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 156500
+    },
+    {
+      "epoch": 1029.671052631579,
+      "grad_norm": 0.7077886462211609,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 156510
+    },
+    {
+      "epoch": 1029.7368421052631,
+      "grad_norm": 0.8222764134407043,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 156520
+    },
+    {
+      "epoch": 1029.8026315789473,
+      "grad_norm": 1.199191689491272,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 156530
+    },
+    {
+      "epoch": 1029.8684210526317,
+      "grad_norm": 1.1930209398269653,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 156540
+    },
+    {
+      "epoch": 1029.9342105263158,
+      "grad_norm": 1.1827071905136108,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 156550
+    },
+    {
+      "epoch": 1030.0,
+      "grad_norm": 1.1457871198654175,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 156560
+    },
+    {
+      "epoch": 1030.0657894736842,
+      "grad_norm": 1.4931516647338867,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 156570
+    },
+    {
+      "epoch": 1030.1315789473683,
+      "grad_norm": 1.4106398820877075,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 156580
+    },
+    {
+      "epoch": 1030.1973684210527,
+      "grad_norm": 1.3719513416290283,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 156590
+    },
+    {
+      "epoch": 1030.2631578947369,
+      "grad_norm": 0.9716209173202515,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 156600
+    },
+    {
+      "epoch": 1030.328947368421,
+      "grad_norm": 1.2926265001296997,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 156610
+    },
+    {
+      "epoch": 1030.3947368421052,
+      "grad_norm": 0.9236207604408264,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 156620
+    },
+    {
+      "epoch": 1030.4605263157894,
+      "grad_norm": 0.8633676767349243,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 156630
+    },
+    {
+      "epoch": 1030.5263157894738,
+      "grad_norm": 0.5557491183280945,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 156640
+    },
+    {
+      "epoch": 1030.592105263158,
+      "grad_norm": 1.302513837814331,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 156650
+    },
+    {
+      "epoch": 1030.657894736842,
+      "grad_norm": 0.9874489307403564,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 156660
+    },
+    {
+      "epoch": 1030.7236842105262,
+      "grad_norm": 1.4245355129241943,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 156670
+    },
+    {
+      "epoch": 1030.7894736842106,
+      "grad_norm": 1.2239059209823608,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 156680
+    },
+    {
+      "epoch": 1030.8552631578948,
+      "grad_norm": 1.1711642742156982,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 156690
+    },
+    {
+      "epoch": 1030.921052631579,
+      "grad_norm": 1.1151902675628662,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 156700
+    },
+    {
+      "epoch": 1030.9868421052631,
+      "grad_norm": 1.4098039865493774,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 156710
+    },
+    {
+      "epoch": 1031.0526315789473,
+      "grad_norm": 1.140479326248169,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 156720
+    },
+    {
+      "epoch": 1031.1184210526317,
+      "grad_norm": 1.3065927028656006,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 156730
+    },
+    {
+      "epoch": 1031.1842105263158,
+      "grad_norm": 0.9672998785972595,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 156740
+    },
+    {
+      "epoch": 1031.25,
+      "grad_norm": 1.2625794410705566,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 156750
+    },
+    {
+      "epoch": 1031.3157894736842,
+      "grad_norm": 1.0192890167236328,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 156760
+    },
+    {
+      "epoch": 1031.3815789473683,
+      "grad_norm": 1.1616865396499634,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 156770
+    },
+    {
+      "epoch": 1031.4473684210527,
+      "grad_norm": 0.9945463538169861,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 156780
+    },
+    {
+      "epoch": 1031.5131578947369,
+      "grad_norm": 0.9444164037704468,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 156790
+    },
+    {
+      "epoch": 1031.578947368421,
+      "grad_norm": 0.7794418931007385,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 156800
+    },
+    {
+      "epoch": 1031.6447368421052,
+      "grad_norm": 0.908078670501709,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 156810
+    },
+    {
+      "epoch": 1031.7105263157894,
+      "grad_norm": 0.7587025165557861,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 156820
+    },
+    {
+      "epoch": 1031.7763157894738,
+      "grad_norm": 1.2462350130081177,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 156830
+    },
+    {
+      "epoch": 1031.842105263158,
+      "grad_norm": 1.1004960536956787,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 156840
+    },
+    {
+      "epoch": 1031.907894736842,
+      "grad_norm": 0.6818149089813232,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 156850
+    },
+    {
+      "epoch": 1031.9736842105262,
+      "grad_norm": 1.0181989669799805,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 156860
+    },
+    {
+      "epoch": 1032.0394736842106,
+      "grad_norm": 1.3769375085830688,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 156870
+    },
+    {
+      "epoch": 1032.1052631578948,
+      "grad_norm": 1.301367998123169,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 156880
+    },
+    {
+      "epoch": 1032.171052631579,
+      "grad_norm": 1.4614466428756714,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 156890
+    },
+    {
+      "epoch": 1032.2368421052631,
+      "grad_norm": 1.141822099685669,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 156900
+    },
+    {
+      "epoch": 1032.3026315789473,
+      "grad_norm": 0.8680024147033691,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 156910
+    },
+    {
+      "epoch": 1032.3684210526317,
+      "grad_norm": 1.0369664430618286,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 156920
+    },
+    {
+      "epoch": 1032.4342105263158,
+      "grad_norm": 1.025201439857483,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 156930
+    },
+    {
+      "epoch": 1032.5,
+      "grad_norm": 0.8427085876464844,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 156940
+    },
+    {
+      "epoch": 1032.5657894736842,
+      "grad_norm": 1.090246558189392,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 156950
+    },
+    {
+      "epoch": 1032.6315789473683,
+      "grad_norm": 0.9469527006149292,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 156960
+    },
+    {
+      "epoch": 1032.6973684210527,
+      "grad_norm": 0.9149708151817322,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 156970
+    },
+    {
+      "epoch": 1032.7631578947369,
+      "grad_norm": 0.9780547618865967,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 156980
+    },
+    {
+      "epoch": 1032.828947368421,
+      "grad_norm": 1.3159875869750977,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 156990
+    },
+    {
+      "epoch": 1032.8947368421052,
+      "grad_norm": 0.9465406537055969,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 157000
+    },
+    {
+      "epoch": 1032.9605263157894,
+      "grad_norm": 1.0383718013763428,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 157010
+    },
+    {
+      "epoch": 1033.0263157894738,
+      "grad_norm": 0.9563059210777283,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 157020
+    },
+    {
+      "epoch": 1033.092105263158,
+      "grad_norm": 0.8377731442451477,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 157030
+    },
+    {
+      "epoch": 1033.157894736842,
+      "grad_norm": 0.7005869150161743,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 157040
+    },
+    {
+      "epoch": 1033.2236842105262,
+      "grad_norm": 0.8357473611831665,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 157050
+    },
+    {
+      "epoch": 1033.2894736842106,
+      "grad_norm": 0.877869188785553,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 157060
+    },
+    {
+      "epoch": 1033.3552631578948,
+      "grad_norm": 1.0630418062210083,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 157070
+    },
+    {
+      "epoch": 1033.421052631579,
+      "grad_norm": 1.6160914897918701,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 157080
+    },
+    {
+      "epoch": 1033.4868421052631,
+      "grad_norm": 1.2976188659667969,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 157090
+    },
+    {
+      "epoch": 1033.5526315789473,
+      "grad_norm": 1.338534951210022,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 157100
+    },
+    {
+      "epoch": 1033.6184210526317,
+      "grad_norm": 1.0486924648284912,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 157110
+    },
+    {
+      "epoch": 1033.6842105263158,
+      "grad_norm": 0.957435667514801,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 157120
+    },
+    {
+      "epoch": 1033.75,
+      "grad_norm": 1.0611435174942017,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 157130
+    },
+    {
+      "epoch": 1033.8157894736842,
+      "grad_norm": 1.471934199333191,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 157140
+    },
+    {
+      "epoch": 1033.8815789473683,
+      "grad_norm": 1.0346896648406982,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 157150
+    },
+    {
+      "epoch": 1033.9473684210527,
+      "grad_norm": 1.5586978197097778,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 157160
+    },
+    {
+      "epoch": 1034.0131578947369,
+      "grad_norm": 1.5314991474151611,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 157170
+    },
+    {
+      "epoch": 1034.078947368421,
+      "grad_norm": 1.3219138383865356,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 157180
+    },
+    {
+      "epoch": 1034.1447368421052,
+      "grad_norm": 0.7580617070198059,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 157190
+    },
+    {
+      "epoch": 1034.2105263157894,
+      "grad_norm": 1.2841469049453735,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 157200
+    },
+    {
+      "epoch": 1034.2763157894738,
+      "grad_norm": 0.8952348828315735,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 157210
+    },
+    {
+      "epoch": 1034.342105263158,
+      "grad_norm": 1.0828073024749756,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 157220
+    },
+    {
+      "epoch": 1034.407894736842,
+      "grad_norm": 1.3251006603240967,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 157230
+    },
+    {
+      "epoch": 1034.4736842105262,
+      "grad_norm": 0.8319105505943298,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 157240
+    },
+    {
+      "epoch": 1034.5394736842106,
+      "grad_norm": 0.6340340971946716,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 157250
+    },
+    {
+      "epoch": 1034.6052631578948,
+      "grad_norm": 1.0439341068267822,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 157260
+    },
+    {
+      "epoch": 1034.671052631579,
+      "grad_norm": 0.9129518270492554,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 157270
+    },
+    {
+      "epoch": 1034.7368421052631,
+      "grad_norm": 1.1103804111480713,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 157280
+    },
+    {
+      "epoch": 1034.8026315789473,
+      "grad_norm": 1.1742290258407593,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 157290
+    },
+    {
+      "epoch": 1034.8684210526317,
+      "grad_norm": 1.2057838439941406,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 157300
+    },
+    {
+      "epoch": 1034.9342105263158,
+      "grad_norm": 0.9219057559967041,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 157310
+    },
+    {
+      "epoch": 1035.0,
+      "grad_norm": 0.6480765342712402,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 157320
+    },
+    {
+      "epoch": 1035.0657894736842,
+      "grad_norm": 0.9799567461013794,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 157330
+    },
+    {
+      "epoch": 1035.1315789473683,
+      "grad_norm": 1.2882983684539795,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 157340
+    },
+    {
+      "epoch": 1035.1973684210527,
+      "grad_norm": 0.887322187423706,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 157350
+    },
+    {
+      "epoch": 1035.2631578947369,
+      "grad_norm": 0.8471446633338928,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 157360
+    },
+    {
+      "epoch": 1035.328947368421,
+      "grad_norm": 0.8873788118362427,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 157370
+    },
+    {
+      "epoch": 1035.3947368421052,
+      "grad_norm": 1.0279638767242432,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 157380
+    },
+    {
+      "epoch": 1035.4605263157894,
+      "grad_norm": 0.9138223528862,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 157390
+    },
+    {
+      "epoch": 1035.5263157894738,
+      "grad_norm": 1.492469072341919,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 157400
+    },
+    {
+      "epoch": 1035.592105263158,
+      "grad_norm": 1.4782291650772095,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 157410
+    },
+    {
+      "epoch": 1035.657894736842,
+      "grad_norm": 1.0031368732452393,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 157420
+    },
+    {
+      "epoch": 1035.7236842105262,
+      "grad_norm": 1.0063573122024536,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 157430
+    },
+    {
+      "epoch": 1035.7894736842106,
+      "grad_norm": 1.1823172569274902,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 157440
+    },
+    {
+      "epoch": 1035.8552631578948,
+      "grad_norm": 0.9905596971511841,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 157450
+    },
+    {
+      "epoch": 1035.921052631579,
+      "grad_norm": 0.9706788063049316,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 157460
+    },
+    {
+      "epoch": 1035.9868421052631,
+      "grad_norm": 1.1120086908340454,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 157470
+    },
+    {
+      "epoch": 1036.0526315789473,
+      "grad_norm": 1.1422266960144043,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 157480
+    },
+    {
+      "epoch": 1036.1184210526317,
+      "grad_norm": 1.3374935388565063,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 157490
+    },
+    {
+      "epoch": 1036.1842105263158,
+      "grad_norm": 1.3589179515838623,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 157500
+    },
+    {
+      "epoch": 1036.25,
+      "grad_norm": 1.1102532148361206,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 157510
+    },
+    {
+      "epoch": 1036.3157894736842,
+      "grad_norm": 1.357604742050171,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 157520
+    },
+    {
+      "epoch": 1036.3815789473683,
+      "grad_norm": 1.054775595664978,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 157530
+    },
+    {
+      "epoch": 1036.4473684210527,
+      "grad_norm": 1.0131275653839111,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 157540
+    },
+    {
+      "epoch": 1036.5131578947369,
+      "grad_norm": 0.73055100440979,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 157550
+    },
+    {
+      "epoch": 1036.578947368421,
+      "grad_norm": 0.8774827122688293,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 157560
+    },
+    {
+      "epoch": 1036.6447368421052,
+      "grad_norm": 1.4485974311828613,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 157570
+    },
+    {
+      "epoch": 1036.7105263157894,
+      "grad_norm": 1.3411765098571777,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 157580
+    },
+    {
+      "epoch": 1036.7763157894738,
+      "grad_norm": 1.4272361993789673,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 157590
+    },
+    {
+      "epoch": 1036.842105263158,
+      "grad_norm": 1.3204584121704102,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 157600
+    },
+    {
+      "epoch": 1036.907894736842,
+      "grad_norm": 1.5034716129302979,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 157610
+    },
+    {
+      "epoch": 1036.9736842105262,
+      "grad_norm": 1.6326338052749634,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 157620
+    },
+    {
+      "epoch": 1037.0394736842106,
+      "grad_norm": 0.9135128259658813,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 157630
+    },
+    {
+      "epoch": 1037.1052631578948,
+      "grad_norm": 1.148511290550232,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 157640
+    },
+    {
+      "epoch": 1037.171052631579,
+      "grad_norm": 1.2435686588287354,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 157650
+    },
+    {
+      "epoch": 1037.2368421052631,
+      "grad_norm": 0.9658685326576233,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 157660
+    },
+    {
+      "epoch": 1037.3026315789473,
+      "grad_norm": 1.3001043796539307,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 157670
+    },
+    {
+      "epoch": 1037.3684210526317,
+      "grad_norm": 1.0400381088256836,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 157680
+    },
+    {
+      "epoch": 1037.4342105263158,
+      "grad_norm": 0.8855817914009094,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 157690
+    },
+    {
+      "epoch": 1037.5,
+      "grad_norm": 1.0237966775894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 157700
+    },
+    {
+      "epoch": 1037.5657894736842,
+      "grad_norm": 1.0766881704330444,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 157710
+    },
+    {
+      "epoch": 1037.6315789473683,
+      "grad_norm": 1.1923819780349731,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 157720
+    },
+    {
+      "epoch": 1037.6973684210527,
+      "grad_norm": 1.0676945447921753,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 157730
+    },
+    {
+      "epoch": 1037.7631578947369,
+      "grad_norm": 0.5870857238769531,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 157740
+    },
+    {
+      "epoch": 1037.828947368421,
+      "grad_norm": 0.9011507034301758,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 157750
+    },
+    {
+      "epoch": 1037.8947368421052,
+      "grad_norm": 1.1840976476669312,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 157760
+    },
+    {
+      "epoch": 1037.9605263157894,
+      "grad_norm": 1.0965074300765991,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 157770
+    },
+    {
+      "epoch": 1038.0263157894738,
+      "grad_norm": 0.8631574511528015,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 157780
+    },
+    {
+      "epoch": 1038.092105263158,
+      "grad_norm": 1.3120415210723877,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 157790
+    },
+    {
+      "epoch": 1038.157894736842,
+      "grad_norm": 1.1779505014419556,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 157800
+    },
+    {
+      "epoch": 1038.2236842105262,
+      "grad_norm": 1.042224645614624,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 157810
+    },
+    {
+      "epoch": 1038.2894736842106,
+      "grad_norm": 0.9169209599494934,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 157820
+    },
+    {
+      "epoch": 1038.3552631578948,
+      "grad_norm": 1.129440188407898,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 157830
+    },
+    {
+      "epoch": 1038.421052631579,
+      "grad_norm": 0.9302949905395508,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 157840
+    },
+    {
+      "epoch": 1038.4868421052631,
+      "grad_norm": 1.3047235012054443,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 157850
+    },
+    {
+      "epoch": 1038.5526315789473,
+      "grad_norm": 1.4257992506027222,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 157860
+    },
+    {
+      "epoch": 1038.6184210526317,
+      "grad_norm": 1.1425153017044067,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 157870
+    },
+    {
+      "epoch": 1038.6842105263158,
+      "grad_norm": 1.1358925104141235,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 157880
+    },
+    {
+      "epoch": 1038.75,
+      "grad_norm": 0.9900014400482178,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 157890
+    },
+    {
+      "epoch": 1038.8157894736842,
+      "grad_norm": 0.9139213562011719,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 157900
+    },
+    {
+      "epoch": 1038.8815789473683,
+      "grad_norm": 1.2955433130264282,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 157910
+    },
+    {
+      "epoch": 1038.9473684210527,
+      "grad_norm": 1.2196770906448364,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 157920
+    },
+    {
+      "epoch": 1039.0131578947369,
+      "grad_norm": 1.0511000156402588,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 157930
+    },
+    {
+      "epoch": 1039.078947368421,
+      "grad_norm": 1.1892273426055908,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 157940
+    },
+    {
+      "epoch": 1039.1447368421052,
+      "grad_norm": 0.8552981019020081,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 157950
+    },
+    {
+      "epoch": 1039.2105263157894,
+      "grad_norm": 0.9567652940750122,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 157960
+    },
+    {
+      "epoch": 1039.2763157894738,
+      "grad_norm": 0.9324297904968262,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 157970
+    },
+    {
+      "epoch": 1039.342105263158,
+      "grad_norm": 0.8563692569732666,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 157980
+    },
+    {
+      "epoch": 1039.407894736842,
+      "grad_norm": 1.0775697231292725,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 157990
+    },
+    {
+      "epoch": 1039.4736842105262,
+      "grad_norm": 1.0669533014297485,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 158000
+    },
+    {
+      "epoch": 1039.5394736842106,
+      "grad_norm": 1.2241381406784058,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 158010
+    },
+    {
+      "epoch": 1039.6052631578948,
+      "grad_norm": 1.0516544580459595,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 158020
+    },
+    {
+      "epoch": 1039.671052631579,
+      "grad_norm": 1.1746127605438232,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 158030
+    },
+    {
+      "epoch": 1039.7368421052631,
+      "grad_norm": 0.8839475512504578,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 158040
+    },
+    {
+      "epoch": 1039.8026315789473,
+      "grad_norm": 0.8313326835632324,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 158050
+    },
+    {
+      "epoch": 1039.8684210526317,
+      "grad_norm": 1.1891549825668335,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 158060
+    },
+    {
+      "epoch": 1039.9342105263158,
+      "grad_norm": 0.9349226355552673,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 158070
+    },
+    {
+      "epoch": 1040.0,
+      "grad_norm": 1.3668376207351685,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 158080
+    },
+    {
+      "epoch": 1040.0657894736842,
+      "grad_norm": 1.0891430377960205,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 158090
+    },
+    {
+      "epoch": 1040.1315789473683,
+      "grad_norm": 1.2689098119735718,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 158100
+    },
+    {
+      "epoch": 1040.1973684210527,
+      "grad_norm": 1.3867899179458618,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 158110
+    },
+    {
+      "epoch": 1040.2631578947369,
+      "grad_norm": 1.1549094915390015,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 158120
+    },
+    {
+      "epoch": 1040.328947368421,
+      "grad_norm": 1.0771297216415405,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 158130
+    },
+    {
+      "epoch": 1040.3947368421052,
+      "grad_norm": 0.6854061484336853,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 158140
+    },
+    {
+      "epoch": 1040.4605263157894,
+      "grad_norm": 0.7592263221740723,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 158150
+    },
+    {
+      "epoch": 1040.5263157894738,
+      "grad_norm": 0.7394665479660034,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 158160
+    },
+    {
+      "epoch": 1040.592105263158,
+      "grad_norm": 1.2182101011276245,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 158170
+    },
+    {
+      "epoch": 1040.657894736842,
+      "grad_norm": 1.0940667390823364,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 158180
+    },
+    {
+      "epoch": 1040.7236842105262,
+      "grad_norm": 1.154238224029541,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 158190
+    },
+    {
+      "epoch": 1040.7894736842106,
+      "grad_norm": 1.0772144794464111,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 158200
+    },
+    {
+      "epoch": 1040.8552631578948,
+      "grad_norm": 0.7135392427444458,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 158210
+    },
+    {
+      "epoch": 1040.921052631579,
+      "grad_norm": 0.6920236945152283,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 158220
+    },
+    {
+      "epoch": 1040.9868421052631,
+      "grad_norm": 0.9873380661010742,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 158230
+    },
+    {
+      "epoch": 1041.0526315789473,
+      "grad_norm": 1.1470561027526855,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 158240
+    },
+    {
+      "epoch": 1041.1184210526317,
+      "grad_norm": 1.1026843786239624,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 158250
+    },
+    {
+      "epoch": 1041.1842105263158,
+      "grad_norm": 0.9850237369537354,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 158260
+    },
+    {
+      "epoch": 1041.25,
+      "grad_norm": 0.8078471422195435,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 158270
+    },
+    {
+      "epoch": 1041.3157894736842,
+      "grad_norm": 0.9304453134536743,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 158280
+    },
+    {
+      "epoch": 1041.3815789473683,
+      "grad_norm": 1.4547866582870483,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 158290
+    },
+    {
+      "epoch": 1041.4473684210527,
+      "grad_norm": 1.2319010496139526,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 158300
+    },
+    {
+      "epoch": 1041.5131578947369,
+      "grad_norm": 0.9949256181716919,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 158310
+    },
+    {
+      "epoch": 1041.578947368421,
+      "grad_norm": 0.8701847195625305,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 158320
+    },
+    {
+      "epoch": 1041.6447368421052,
+      "grad_norm": 1.0713963508605957,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 158330
+    },
+    {
+      "epoch": 1041.7105263157894,
+      "grad_norm": 1.043212652206421,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 158340
+    },
+    {
+      "epoch": 1041.7763157894738,
+      "grad_norm": 1.2368879318237305,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 158350
+    },
+    {
+      "epoch": 1041.842105263158,
+      "grad_norm": 1.1877309083938599,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 158360
+    },
+    {
+      "epoch": 1041.907894736842,
+      "grad_norm": 1.265149712562561,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 158370
+    },
+    {
+      "epoch": 1041.9736842105262,
+      "grad_norm": 1.0216879844665527,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 158380
+    },
+    {
+      "epoch": 1042.0394736842106,
+      "grad_norm": 0.9168482422828674,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 158390
+    },
+    {
+      "epoch": 1042.1052631578948,
+      "grad_norm": 1.1165847778320312,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 158400
+    },
+    {
+      "epoch": 1042.171052631579,
+      "grad_norm": 1.5731300115585327,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 158410
+    },
+    {
+      "epoch": 1042.2368421052631,
+      "grad_norm": 1.1611987352371216,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 158420
+    },
+    {
+      "epoch": 1042.3026315789473,
+      "grad_norm": 1.191115140914917,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 158430
+    },
+    {
+      "epoch": 1042.3684210526317,
+      "grad_norm": 0.9672328233718872,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 158440
+    },
+    {
+      "epoch": 1042.4342105263158,
+      "grad_norm": 1.1341195106506348,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 158450
+    },
+    {
+      "epoch": 1042.5,
+      "grad_norm": 1.0045264959335327,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 158460
+    },
+    {
+      "epoch": 1042.5657894736842,
+      "grad_norm": 1.13510000705719,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 158470
+    },
+    {
+      "epoch": 1042.6315789473683,
+      "grad_norm": 0.9069583415985107,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 158480
+    },
+    {
+      "epoch": 1042.6973684210527,
+      "grad_norm": 1.0426357984542847,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 158490
+    },
+    {
+      "epoch": 1042.7631578947369,
+      "grad_norm": 1.1042143106460571,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 158500
+    },
+    {
+      "epoch": 1042.828947368421,
+      "grad_norm": 1.1533740758895874,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 158510
+    },
+    {
+      "epoch": 1042.8947368421052,
+      "grad_norm": 1.0955190658569336,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 158520
+    },
+    {
+      "epoch": 1042.9605263157894,
+      "grad_norm": 1.552657961845398,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 158530
+    },
+    {
+      "epoch": 1043.0263157894738,
+      "grad_norm": 0.9286198019981384,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 158540
+    },
+    {
+      "epoch": 1043.092105263158,
+      "grad_norm": 0.8410740494728088,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 158550
+    },
+    {
+      "epoch": 1043.157894736842,
+      "grad_norm": 1.1539233922958374,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 158560
+    },
+    {
+      "epoch": 1043.2236842105262,
+      "grad_norm": 1.0042409896850586,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 158570
+    },
+    {
+      "epoch": 1043.2894736842106,
+      "grad_norm": 1.2449952363967896,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 158580
+    },
+    {
+      "epoch": 1043.3552631578948,
+      "grad_norm": 1.4503757953643799,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 158590
+    },
+    {
+      "epoch": 1043.421052631579,
+      "grad_norm": 1.995700478553772,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 158600
+    },
+    {
+      "epoch": 1043.4868421052631,
+      "grad_norm": 1.310492992401123,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 158610
+    },
+    {
+      "epoch": 1043.5526315789473,
+      "grad_norm": 1.1206649541854858,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 158620
+    },
+    {
+      "epoch": 1043.6184210526317,
+      "grad_norm": 1.135241985321045,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 158630
+    },
+    {
+      "epoch": 1043.6842105263158,
+      "grad_norm": 1.2304860353469849,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 158640
+    },
+    {
+      "epoch": 1043.75,
+      "grad_norm": 1.1069141626358032,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 158650
+    },
+    {
+      "epoch": 1043.8157894736842,
+      "grad_norm": 1.013854742050171,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 158660
+    },
+    {
+      "epoch": 1043.8815789473683,
+      "grad_norm": 0.8795697093009949,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 158670
+    },
+    {
+      "epoch": 1043.9473684210527,
+      "grad_norm": 1.4216660261154175,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 158680
+    },
+    {
+      "epoch": 1044.0131578947369,
+      "grad_norm": 1.174999475479126,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 158690
+    },
+    {
+      "epoch": 1044.078947368421,
+      "grad_norm": 1.3855254650115967,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 158700
+    },
+    {
+      "epoch": 1044.1447368421052,
+      "grad_norm": 0.9125514626502991,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 158710
+    },
+    {
+      "epoch": 1044.2105263157894,
+      "grad_norm": 0.9567851424217224,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 158720
+    },
+    {
+      "epoch": 1044.2763157894738,
+      "grad_norm": 0.9960494041442871,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 158730
+    },
+    {
+      "epoch": 1044.342105263158,
+      "grad_norm": 0.8722960948944092,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 158740
+    },
+    {
+      "epoch": 1044.407894736842,
+      "grad_norm": 0.9810813665390015,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 158750
+    },
+    {
+      "epoch": 1044.4736842105262,
+      "grad_norm": 1.2386258840560913,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 158760
+    },
+    {
+      "epoch": 1044.5394736842106,
+      "grad_norm": 1.000181794166565,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 158770
+    },
+    {
+      "epoch": 1044.6052631578948,
+      "grad_norm": 0.927920937538147,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 158780
+    },
+    {
+      "epoch": 1044.671052631579,
+      "grad_norm": 0.8654837608337402,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 158790
+    },
+    {
+      "epoch": 1044.7368421052631,
+      "grad_norm": 0.9293181896209717,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 158800
+    },
+    {
+      "epoch": 1044.8026315789473,
+      "grad_norm": 1.0467267036437988,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 158810
+    },
+    {
+      "epoch": 1044.8684210526317,
+      "grad_norm": 0.9053267240524292,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 158820
+    },
+    {
+      "epoch": 1044.9342105263158,
+      "grad_norm": 1.1089938879013062,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 158830
+    },
+    {
+      "epoch": 1045.0,
+      "grad_norm": 1.0741143226623535,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 158840
+    },
+    {
+      "epoch": 1045.0657894736842,
+      "grad_norm": 0.8008183836936951,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 158850
+    },
+    {
+      "epoch": 1045.1315789473683,
+      "grad_norm": 0.9834315776824951,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 158860
+    },
+    {
+      "epoch": 1045.1973684210527,
+      "grad_norm": 0.8978926539421082,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 158870
+    },
+    {
+      "epoch": 1045.2631578947369,
+      "grad_norm": 0.7383962273597717,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 158880
+    },
+    {
+      "epoch": 1045.328947368421,
+      "grad_norm": 0.7250087261199951,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 158890
+    },
+    {
+      "epoch": 1045.3947368421052,
+      "grad_norm": 1.158569097518921,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 158900
+    },
+    {
+      "epoch": 1045.4605263157894,
+      "grad_norm": 1.0793335437774658,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 158910
+    },
+    {
+      "epoch": 1045.5263157894738,
+      "grad_norm": 1.2114217281341553,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 158920
+    },
+    {
+      "epoch": 1045.592105263158,
+      "grad_norm": 1.464314579963684,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 158930
+    },
+    {
+      "epoch": 1045.657894736842,
+      "grad_norm": 1.0885869264602661,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 158940
+    },
+    {
+      "epoch": 1045.7236842105262,
+      "grad_norm": 1.1773313283920288,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 158950
+    },
+    {
+      "epoch": 1045.7894736842106,
+      "grad_norm": 1.345479965209961,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 158960
+    },
+    {
+      "epoch": 1045.8552631578948,
+      "grad_norm": 1.1355724334716797,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 158970
+    },
+    {
+      "epoch": 1045.921052631579,
+      "grad_norm": 1.21616792678833,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 158980
+    },
+    {
+      "epoch": 1045.9868421052631,
+      "grad_norm": 0.895064651966095,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 158990
+    },
+    {
+      "epoch": 1046.0526315789473,
+      "grad_norm": 1.1562937498092651,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 159000
+    },
+    {
+      "epoch": 1046.1184210526317,
+      "grad_norm": 1.1668152809143066,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 159010
+    },
+    {
+      "epoch": 1046.1842105263158,
+      "grad_norm": 1.03926682472229,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 159020
+    },
+    {
+      "epoch": 1046.25,
+      "grad_norm": 1.0261049270629883,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 159030
+    },
+    {
+      "epoch": 1046.3157894736842,
+      "grad_norm": 1.0766314268112183,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 159040
+    },
+    {
+      "epoch": 1046.3815789473683,
+      "grad_norm": 0.6256817579269409,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 159050
+    },
+    {
+      "epoch": 1046.4473684210527,
+      "grad_norm": 0.874019980430603,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 159060
+    },
+    {
+      "epoch": 1046.5131578947369,
+      "grad_norm": 1.0828053951263428,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 159070
+    },
+    {
+      "epoch": 1046.578947368421,
+      "grad_norm": 1.11966872215271,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 159080
+    },
+    {
+      "epoch": 1046.6447368421052,
+      "grad_norm": 0.9557555913925171,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 159090
+    },
+    {
+      "epoch": 1046.7105263157894,
+      "grad_norm": 1.2064489126205444,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 159100
+    },
+    {
+      "epoch": 1046.7763157894738,
+      "grad_norm": 1.3529369831085205,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 159110
+    },
+    {
+      "epoch": 1046.842105263158,
+      "grad_norm": 0.934829592704773,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 159120
+    },
+    {
+      "epoch": 1046.907894736842,
+      "grad_norm": 1.274330735206604,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 159130
+    },
+    {
+      "epoch": 1046.9736842105262,
+      "grad_norm": 1.0497575998306274,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 159140
+    },
+    {
+      "epoch": 1047.0394736842106,
+      "grad_norm": 1.1160510778427124,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 159150
+    },
+    {
+      "epoch": 1047.1052631578948,
+      "grad_norm": 1.325922966003418,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 159160
+    },
+    {
+      "epoch": 1047.171052631579,
+      "grad_norm": 1.4824130535125732,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 159170
+    },
+    {
+      "epoch": 1047.2368421052631,
+      "grad_norm": 1.2637231349945068,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 159180
+    },
+    {
+      "epoch": 1047.3026315789473,
+      "grad_norm": 1.0419176816940308,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 159190
+    },
+    {
+      "epoch": 1047.3684210526317,
+      "grad_norm": 1.0075671672821045,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 159200
+    },
+    {
+      "epoch": 1047.4342105263158,
+      "grad_norm": 0.9810393452644348,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 159210
+    },
+    {
+      "epoch": 1047.5,
+      "grad_norm": 0.9505169987678528,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 159220
+    },
+    {
+      "epoch": 1047.5657894736842,
+      "grad_norm": 0.9999101758003235,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 159230
+    },
+    {
+      "epoch": 1047.6315789473683,
+      "grad_norm": 0.7792953848838806,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 159240
+    },
+    {
+      "epoch": 1047.6973684210527,
+      "grad_norm": 1.1231037378311157,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 159250
+    },
+    {
+      "epoch": 1047.7631578947369,
+      "grad_norm": 1.1759856939315796,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 159260
+    },
+    {
+      "epoch": 1047.828947368421,
+      "grad_norm": 0.9894112348556519,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 159270
+    },
+    {
+      "epoch": 1047.8947368421052,
+      "grad_norm": 0.9776474833488464,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 159280
+    },
+    {
+      "epoch": 1047.9605263157894,
+      "grad_norm": 1.2519046068191528,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 159290
+    },
+    {
+      "epoch": 1048.0263157894738,
+      "grad_norm": 1.459942102432251,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 159300
+    },
+    {
+      "epoch": 1048.092105263158,
+      "grad_norm": 1.0710073709487915,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 159310
+    },
+    {
+      "epoch": 1048.157894736842,
+      "grad_norm": 0.962834358215332,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 159320
+    },
+    {
+      "epoch": 1048.2236842105262,
+      "grad_norm": 0.8130425810813904,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 159330
+    },
+    {
+      "epoch": 1048.2894736842106,
+      "grad_norm": 0.9603180289268494,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 159340
+    },
+    {
+      "epoch": 1048.3552631578948,
+      "grad_norm": 0.8210400342941284,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 159350
+    },
+    {
+      "epoch": 1048.421052631579,
+      "grad_norm": 1.1121681928634644,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 159360
+    },
+    {
+      "epoch": 1048.4868421052631,
+      "grad_norm": 1.0714201927185059,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 159370
+    },
+    {
+      "epoch": 1048.5526315789473,
+      "grad_norm": 0.9843019247055054,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 159380
+    },
+    {
+      "epoch": 1048.6184210526317,
+      "grad_norm": 1.1511754989624023,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 159390
+    },
+    {
+      "epoch": 1048.6842105263158,
+      "grad_norm": 0.6616480946540833,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 159400
+    },
+    {
+      "epoch": 1048.75,
+      "grad_norm": 1.1609398126602173,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 159410
+    },
+    {
+      "epoch": 1048.8157894736842,
+      "grad_norm": 0.9391231536865234,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 159420
+    },
+    {
+      "epoch": 1048.8815789473683,
+      "grad_norm": 0.9525076150894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 159430
+    },
+    {
+      "epoch": 1048.9473684210527,
+      "grad_norm": 0.9699547290802002,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 159440
+    },
+    {
+      "epoch": 1049.0131578947369,
+      "grad_norm": 0.7067571878433228,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 159450
+    },
+    {
+      "epoch": 1049.078947368421,
+      "grad_norm": 0.7832255959510803,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 159460
+    },
+    {
+      "epoch": 1049.1447368421052,
+      "grad_norm": 0.945345938205719,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 159470
+    },
+    {
+      "epoch": 1049.2105263157894,
+      "grad_norm": 1.0841176509857178,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 159480
+    },
+    {
+      "epoch": 1049.2763157894738,
+      "grad_norm": 1.3195167779922485,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 159490
+    },
+    {
+      "epoch": 1049.342105263158,
+      "grad_norm": 1.5180623531341553,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 159500
+    },
+    {
+      "epoch": 1049.407894736842,
+      "grad_norm": 0.7986946105957031,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 159510
+    },
+    {
+      "epoch": 1049.4736842105262,
+      "grad_norm": 1.2574127912521362,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 159520
+    },
+    {
+      "epoch": 1049.5394736842106,
+      "grad_norm": 1.320098876953125,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 159530
+    },
+    {
+      "epoch": 1049.6052631578948,
+      "grad_norm": 0.890644371509552,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 159540
+    },
+    {
+      "epoch": 1049.671052631579,
+      "grad_norm": 1.4320951700210571,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 159550
+    },
+    {
+      "epoch": 1049.7368421052631,
+      "grad_norm": 1.1257306337356567,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 159560
+    },
+    {
+      "epoch": 1049.8026315789473,
+      "grad_norm": 1.2194310426712036,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 159570
+    },
+    {
+      "epoch": 1049.8684210526317,
+      "grad_norm": 0.7975109219551086,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 159580
+    },
+    {
+      "epoch": 1049.9342105263158,
+      "grad_norm": 1.1322660446166992,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 159590
+    },
+    {
+      "epoch": 1050.0,
+      "grad_norm": 1.1089693307876587,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 159600
+    },
+    {
+      "epoch": 1050.0657894736842,
+      "grad_norm": 1.135398507118225,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 159610
+    },
+    {
+      "epoch": 1050.1315789473683,
+      "grad_norm": 1.2235279083251953,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 159620
+    },
+    {
+      "epoch": 1050.1973684210527,
+      "grad_norm": 1.2901211977005005,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 159630
+    },
+    {
+      "epoch": 1050.2631578947369,
+      "grad_norm": 1.0014219284057617,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 159640
+    },
+    {
+      "epoch": 1050.328947368421,
+      "grad_norm": 1.5262455940246582,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 159650
+    },
+    {
+      "epoch": 1050.3947368421052,
+      "grad_norm": 0.9812429547309875,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 159660
+    },
+    {
+      "epoch": 1050.4605263157894,
+      "grad_norm": 0.8760225176811218,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 159670
+    },
+    {
+      "epoch": 1050.5263157894738,
+      "grad_norm": 0.8203020691871643,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 159680
+    },
+    {
+      "epoch": 1050.592105263158,
+      "grad_norm": 1.3625808954238892,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 159690
+    },
+    {
+      "epoch": 1050.657894736842,
+      "grad_norm": 0.927275538444519,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 159700
+    },
+    {
+      "epoch": 1050.7236842105262,
+      "grad_norm": 1.043666124343872,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 159710
+    },
+    {
+      "epoch": 1050.7894736842106,
+      "grad_norm": 0.8062131404876709,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 159720
+    },
+    {
+      "epoch": 1050.8552631578948,
+      "grad_norm": 0.9691109657287598,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 159730
+    },
+    {
+      "epoch": 1050.921052631579,
+      "grad_norm": 1.3453434705734253,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 159740
+    },
+    {
+      "epoch": 1050.9868421052631,
+      "grad_norm": 1.1522691249847412,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 159750
+    },
+    {
+      "epoch": 1051.0526315789473,
+      "grad_norm": 1.2776633501052856,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 159760
+    },
+    {
+      "epoch": 1051.1184210526317,
+      "grad_norm": 1.5653990507125854,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 159770
+    },
+    {
+      "epoch": 1051.1842105263158,
+      "grad_norm": 1.3178924322128296,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 159780
+    },
+    {
+      "epoch": 1051.25,
+      "grad_norm": 1.3187295198440552,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 159790
+    },
+    {
+      "epoch": 1051.3157894736842,
+      "grad_norm": 0.9678728580474854,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 159800
+    },
+    {
+      "epoch": 1051.3815789473683,
+      "grad_norm": 0.9071379899978638,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 159810
+    },
+    {
+      "epoch": 1051.4473684210527,
+      "grad_norm": 1.175668478012085,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 159820
+    },
+    {
+      "epoch": 1051.5131578947369,
+      "grad_norm": 1.0487818717956543,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 159830
+    },
+    {
+      "epoch": 1051.578947368421,
+      "grad_norm": 0.8464121222496033,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 159840
+    },
+    {
+      "epoch": 1051.6447368421052,
+      "grad_norm": 1.1520172357559204,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 159850
+    },
+    {
+      "epoch": 1051.7105263157894,
+      "grad_norm": 0.6816383600234985,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 159860
+    },
+    {
+      "epoch": 1051.7763157894738,
+      "grad_norm": 1.1880346536636353,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 159870
+    },
+    {
+      "epoch": 1051.842105263158,
+      "grad_norm": 1.3333251476287842,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 159880
+    },
+    {
+      "epoch": 1051.907894736842,
+      "grad_norm": 1.1161757707595825,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 159890
+    },
+    {
+      "epoch": 1051.9736842105262,
+      "grad_norm": 1.5317634344100952,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 159900
+    },
+    {
+      "epoch": 1052.0394736842106,
+      "grad_norm": 1.2690035104751587,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 159910
+    },
+    {
+      "epoch": 1052.1052631578948,
+      "grad_norm": 1.0578007698059082,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 159920
+    },
+    {
+      "epoch": 1052.171052631579,
+      "grad_norm": 1.1168135404586792,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 159930
+    },
+    {
+      "epoch": 1052.2368421052631,
+      "grad_norm": 1.1406240463256836,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 159940
+    },
+    {
+      "epoch": 1052.3026315789473,
+      "grad_norm": 0.972987949848175,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 159950
+    },
+    {
+      "epoch": 1052.3684210526317,
+      "grad_norm": 0.7099289894104004,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 159960
+    },
+    {
+      "epoch": 1052.4342105263158,
+      "grad_norm": 0.7676605582237244,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 159970
+    },
+    {
+      "epoch": 1052.5,
+      "grad_norm": 0.9532294273376465,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 159980
+    },
+    {
+      "epoch": 1052.5657894736842,
+      "grad_norm": 0.8012232780456543,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 159990
+    },
+    {
+      "epoch": 1052.6315789473683,
+      "grad_norm": 1.0869078636169434,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 160000
+    },
+    {
+      "epoch": 1052.6973684210527,
+      "grad_norm": 0.9576671123504639,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 160010
+    },
+    {
+      "epoch": 1052.7631578947369,
+      "grad_norm": 0.9968484044075012,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 160020
+    },
+    {
+      "epoch": 1052.828947368421,
+      "grad_norm": 1.0012370347976685,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 160030
+    },
+    {
+      "epoch": 1052.8947368421052,
+      "grad_norm": 0.6246411800384521,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 160040
+    },
+    {
+      "epoch": 1052.9605263157894,
+      "grad_norm": 0.8564648032188416,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 160050
+    },
+    {
+      "epoch": 1053.0263157894738,
+      "grad_norm": 1.1350836753845215,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 160060
+    },
+    {
+      "epoch": 1053.092105263158,
+      "grad_norm": 0.8760911822319031,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 160070
+    },
+    {
+      "epoch": 1053.157894736842,
+      "grad_norm": 1.1660544872283936,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 160080
+    },
+    {
+      "epoch": 1053.2236842105262,
+      "grad_norm": 0.8360348343849182,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 160090
+    },
+    {
+      "epoch": 1053.2894736842106,
+      "grad_norm": 0.754542887210846,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 160100
+    },
+    {
+      "epoch": 1053.3552631578948,
+      "grad_norm": 1.2186027765274048,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 160110
+    },
+    {
+      "epoch": 1053.421052631579,
+      "grad_norm": 0.9988258481025696,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 160120
+    },
+    {
+      "epoch": 1053.4868421052631,
+      "grad_norm": 1.007232904434204,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 160130
+    },
+    {
+      "epoch": 1053.5526315789473,
+      "grad_norm": 1.2770189046859741,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 160140
+    },
+    {
+      "epoch": 1053.6184210526317,
+      "grad_norm": 1.163527488708496,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 160150
+    },
+    {
+      "epoch": 1053.6842105263158,
+      "grad_norm": 0.8490967154502869,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 160160
+    },
+    {
+      "epoch": 1053.75,
+      "grad_norm": 1.0161746740341187,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 160170
+    },
+    {
+      "epoch": 1053.8157894736842,
+      "grad_norm": 1.1093944311141968,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 160180
+    },
+    {
+      "epoch": 1053.8815789473683,
+      "grad_norm": 0.9495304822921753,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 160190
+    },
+    {
+      "epoch": 1053.9473684210527,
+      "grad_norm": 0.9384229779243469,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 160200
+    },
+    {
+      "epoch": 1054.0131578947369,
+      "grad_norm": 1.0374761819839478,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 160210
+    },
+    {
+      "epoch": 1054.078947368421,
+      "grad_norm": 1.3228940963745117,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 160220
+    },
+    {
+      "epoch": 1054.1447368421052,
+      "grad_norm": 1.1004372835159302,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 160230
+    },
+    {
+      "epoch": 1054.2105263157894,
+      "grad_norm": 1.346169352531433,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 160240
+    },
+    {
+      "epoch": 1054.2763157894738,
+      "grad_norm": 1.0477455854415894,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 160250
+    },
+    {
+      "epoch": 1054.342105263158,
+      "grad_norm": 1.119246244430542,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 160260
+    },
+    {
+      "epoch": 1054.407894736842,
+      "grad_norm": 1.2886263132095337,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 160270
+    },
+    {
+      "epoch": 1054.4736842105262,
+      "grad_norm": 0.7763897180557251,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 160280
+    },
+    {
+      "epoch": 1054.5394736842106,
+      "grad_norm": 0.8887063264846802,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 160290
+    },
+    {
+      "epoch": 1054.6052631578948,
+      "grad_norm": 1.2421995401382446,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 160300
+    },
+    {
+      "epoch": 1054.671052631579,
+      "grad_norm": 1.0714391469955444,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 160310
+    },
+    {
+      "epoch": 1054.7368421052631,
+      "grad_norm": 1.002956509590149,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 160320
+    },
+    {
+      "epoch": 1054.8026315789473,
+      "grad_norm": 1.0658334493637085,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 160330
+    },
+    {
+      "epoch": 1054.8684210526317,
+      "grad_norm": 0.7606706619262695,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 160340
+    },
+    {
+      "epoch": 1054.9342105263158,
+      "grad_norm": 0.995364785194397,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 160350
+    },
+    {
+      "epoch": 1055.0,
+      "grad_norm": 0.8916927576065063,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 160360
+    },
+    {
+      "epoch": 1055.0657894736842,
+      "grad_norm": 1.03330659866333,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 160370
+    },
+    {
+      "epoch": 1055.1315789473683,
+      "grad_norm": 0.8269760012626648,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 160380
+    },
+    {
+      "epoch": 1055.1973684210527,
+      "grad_norm": 0.9928635358810425,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 160390
+    },
+    {
+      "epoch": 1055.2631578947369,
+      "grad_norm": 1.2529430389404297,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 160400
+    },
+    {
+      "epoch": 1055.328947368421,
+      "grad_norm": 0.6669118404388428,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 160410
+    },
+    {
+      "epoch": 1055.3947368421052,
+      "grad_norm": 1.3909178972244263,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 160420
+    },
+    {
+      "epoch": 1055.4605263157894,
+      "grad_norm": 1.6014965772628784,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 160430
+    },
+    {
+      "epoch": 1055.5263157894738,
+      "grad_norm": 1.5059226751327515,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 160440
+    },
+    {
+      "epoch": 1055.592105263158,
+      "grad_norm": 0.8995516896247864,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 160450
+    },
+    {
+      "epoch": 1055.657894736842,
+      "grad_norm": 0.8544006943702698,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 160460
+    },
+    {
+      "epoch": 1055.7236842105262,
+      "grad_norm": 1.2715330123901367,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 160470
+    },
+    {
+      "epoch": 1055.7894736842106,
+      "grad_norm": 0.8547405004501343,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 160480
+    },
+    {
+      "epoch": 1055.8552631578948,
+      "grad_norm": 0.8670430183410645,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 160490
+    },
+    {
+      "epoch": 1055.921052631579,
+      "grad_norm": 0.853753387928009,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 160500
+    },
+    {
+      "epoch": 1055.9868421052631,
+      "grad_norm": 0.8636345267295837,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 160510
+    },
+    {
+      "epoch": 1056.0526315789473,
+      "grad_norm": 0.9810386896133423,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 160520
+    },
+    {
+      "epoch": 1056.1184210526317,
+      "grad_norm": 1.062951683998108,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 160530
+    },
+    {
+      "epoch": 1056.1842105263158,
+      "grad_norm": 1.1501662731170654,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 160540
+    },
+    {
+      "epoch": 1056.25,
+      "grad_norm": 1.5023175477981567,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 160550
+    },
+    {
+      "epoch": 1056.3157894736842,
+      "grad_norm": 1.048349380493164,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 160560
+    },
+    {
+      "epoch": 1056.3815789473683,
+      "grad_norm": 1.1809251308441162,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 160570
+    },
+    {
+      "epoch": 1056.4473684210527,
+      "grad_norm": 1.038750171661377,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 160580
+    },
+    {
+      "epoch": 1056.5131578947369,
+      "grad_norm": 1.030869483947754,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 160590
+    },
+    {
+      "epoch": 1056.578947368421,
+      "grad_norm": 1.0279252529144287,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 160600
+    },
+    {
+      "epoch": 1056.6447368421052,
+      "grad_norm": 0.8072715997695923,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 160610
+    },
+    {
+      "epoch": 1056.7105263157894,
+      "grad_norm": 1.0479940176010132,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 160620
+    },
+    {
+      "epoch": 1056.7763157894738,
+      "grad_norm": 0.9599220156669617,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 160630
+    },
+    {
+      "epoch": 1056.842105263158,
+      "grad_norm": 0.6841326355934143,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 160640
+    },
+    {
+      "epoch": 1056.907894736842,
+      "grad_norm": 1.1303530931472778,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 160650
+    },
+    {
+      "epoch": 1056.9736842105262,
+      "grad_norm": 1.0210925340652466,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 160660
+    },
+    {
+      "epoch": 1057.0394736842106,
+      "grad_norm": 0.7038403153419495,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 160670
+    },
+    {
+      "epoch": 1057.1052631578948,
+      "grad_norm": 1.3282930850982666,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 160680
+    },
+    {
+      "epoch": 1057.171052631579,
+      "grad_norm": 1.007394790649414,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 160690
+    },
+    {
+      "epoch": 1057.2368421052631,
+      "grad_norm": 0.7837812304496765,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 160700
+    },
+    {
+      "epoch": 1057.3026315789473,
+      "grad_norm": 0.9380816221237183,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 160710
+    },
+    {
+      "epoch": 1057.3684210526317,
+      "grad_norm": 1.3225820064544678,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 160720
+    },
+    {
+      "epoch": 1057.4342105263158,
+      "grad_norm": 0.6032253503799438,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 160730
+    },
+    {
+      "epoch": 1057.5,
+      "grad_norm": 1.0029069185256958,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 160740
+    },
+    {
+      "epoch": 1057.5657894736842,
+      "grad_norm": 0.8361908197402954,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 160750
+    },
+    {
+      "epoch": 1057.6315789473683,
+      "grad_norm": 1.2243342399597168,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 160760
+    },
+    {
+      "epoch": 1057.6973684210527,
+      "grad_norm": 0.9138869643211365,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 160770
+    },
+    {
+      "epoch": 1057.7631578947369,
+      "grad_norm": 1.0361911058425903,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 160780
+    },
+    {
+      "epoch": 1057.828947368421,
+      "grad_norm": 0.9030035138130188,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 160790
+    },
+    {
+      "epoch": 1057.8947368421052,
+      "grad_norm": 0.996807873249054,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 160800
+    },
+    {
+      "epoch": 1057.9605263157894,
+      "grad_norm": 0.8536729216575623,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 160810
+    },
+    {
+      "epoch": 1058.0263157894738,
+      "grad_norm": 1.0060828924179077,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 160820
+    },
+    {
+      "epoch": 1058.092105263158,
+      "grad_norm": 0.9179419875144958,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 160830
+    },
+    {
+      "epoch": 1058.157894736842,
+      "grad_norm": 0.8911098837852478,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 160840
+    },
+    {
+      "epoch": 1058.2236842105262,
+      "grad_norm": 1.0588878393173218,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 160850
+    },
+    {
+      "epoch": 1058.2894736842106,
+      "grad_norm": 0.6525148153305054,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 160860
+    },
+    {
+      "epoch": 1058.3552631578948,
+      "grad_norm": 0.6334079504013062,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 160870
+    },
+    {
+      "epoch": 1058.421052631579,
+      "grad_norm": 0.8859434127807617,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 160880
+    },
+    {
+      "epoch": 1058.4868421052631,
+      "grad_norm": 0.7293907999992371,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 160890
+    },
+    {
+      "epoch": 1058.5526315789473,
+      "grad_norm": 0.8205122351646423,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 160900
+    },
+    {
+      "epoch": 1058.6184210526317,
+      "grad_norm": 0.8700542449951172,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 160910
+    },
+    {
+      "epoch": 1058.6842105263158,
+      "grad_norm": 1.1697430610656738,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 160920
+    },
+    {
+      "epoch": 1058.75,
+      "grad_norm": 1.5094969272613525,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 160930
+    },
+    {
+      "epoch": 1058.8157894736842,
+      "grad_norm": 1.085866928100586,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 160940
+    },
+    {
+      "epoch": 1058.8815789473683,
+      "grad_norm": 0.935643196105957,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 160950
+    },
+    {
+      "epoch": 1058.9473684210527,
+      "grad_norm": 0.8698868751525879,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 160960
+    },
+    {
+      "epoch": 1059.0131578947369,
+      "grad_norm": 1.2459019422531128,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 160970
+    },
+    {
+      "epoch": 1059.078947368421,
+      "grad_norm": 1.4872862100601196,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 160980
+    },
+    {
+      "epoch": 1059.1447368421052,
+      "grad_norm": 1.7374746799468994,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 160990
+    },
+    {
+      "epoch": 1059.2105263157894,
+      "grad_norm": 1.4759396314620972,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 161000
+    },
+    {
+      "epoch": 1059.2763157894738,
+      "grad_norm": 1.3807487487792969,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 161010
+    },
+    {
+      "epoch": 1059.342105263158,
+      "grad_norm": 1.0829836130142212,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 161020
+    },
+    {
+      "epoch": 1059.407894736842,
+      "grad_norm": 1.3213567733764648,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 161030
+    },
+    {
+      "epoch": 1059.4736842105262,
+      "grad_norm": 1.1159719228744507,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 161040
+    },
+    {
+      "epoch": 1059.5394736842106,
+      "grad_norm": 1.0388392210006714,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 161050
+    },
+    {
+      "epoch": 1059.6052631578948,
+      "grad_norm": 1.4326999187469482,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 161060
+    },
+    {
+      "epoch": 1059.671052631579,
+      "grad_norm": 0.9729658365249634,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 161070
+    },
+    {
+      "epoch": 1059.7368421052631,
+      "grad_norm": 0.9425380229949951,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 161080
+    },
+    {
+      "epoch": 1059.8026315789473,
+      "grad_norm": 0.8282036185264587,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 161090
+    },
+    {
+      "epoch": 1059.8684210526317,
+      "grad_norm": 0.782004714012146,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 161100
+    },
+    {
+      "epoch": 1059.9342105263158,
+      "grad_norm": 0.8745496273040771,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 161110
+    },
+    {
+      "epoch": 1060.0,
+      "grad_norm": 1.0883170366287231,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 161120
+    },
+    {
+      "epoch": 1060.0657894736842,
+      "grad_norm": 0.7825089693069458,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 161130
+    },
+    {
+      "epoch": 1060.1315789473683,
+      "grad_norm": 0.7291836738586426,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 161140
+    },
+    {
+      "epoch": 1060.1973684210527,
+      "grad_norm": 0.9234983325004578,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 161150
+    },
+    {
+      "epoch": 1060.2631578947369,
+      "grad_norm": 1.0557537078857422,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 161160
+    },
+    {
+      "epoch": 1060.328947368421,
+      "grad_norm": 1.190638542175293,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 161170
+    },
+    {
+      "epoch": 1060.3947368421052,
+      "grad_norm": 1.2052501440048218,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 161180
+    },
+    {
+      "epoch": 1060.4605263157894,
+      "grad_norm": 0.8857777714729309,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 161190
+    },
+    {
+      "epoch": 1060.5263157894738,
+      "grad_norm": 0.9795964956283569,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 161200
+    },
+    {
+      "epoch": 1060.592105263158,
+      "grad_norm": 1.1422735452651978,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 161210
+    },
+    {
+      "epoch": 1060.657894736842,
+      "grad_norm": 0.8671726584434509,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 161220
+    },
+    {
+      "epoch": 1060.7236842105262,
+      "grad_norm": 1.1022329330444336,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 161230
+    },
+    {
+      "epoch": 1060.7894736842106,
+      "grad_norm": 1.2287969589233398,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 161240
+    },
+    {
+      "epoch": 1060.8552631578948,
+      "grad_norm": 1.1893408298492432,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 161250
+    },
+    {
+      "epoch": 1060.921052631579,
+      "grad_norm": 1.085235595703125,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 161260
+    },
+    {
+      "epoch": 1060.9868421052631,
+      "grad_norm": 1.2641836404800415,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 161270
+    },
+    {
+      "epoch": 1061.0526315789473,
+      "grad_norm": 0.9672636985778809,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 161280
+    },
+    {
+      "epoch": 1061.1184210526317,
+      "grad_norm": 0.8965618014335632,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 161290
+    },
+    {
+      "epoch": 1061.1842105263158,
+      "grad_norm": 0.9255189299583435,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 161300
+    },
+    {
+      "epoch": 1061.25,
+      "grad_norm": 0.9481072425842285,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 161310
+    },
+    {
+      "epoch": 1061.3157894736842,
+      "grad_norm": 1.216856837272644,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 161320
+    },
+    {
+      "epoch": 1061.3815789473683,
+      "grad_norm": 1.3531123399734497,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 161330
+    },
+    {
+      "epoch": 1061.4473684210527,
+      "grad_norm": 0.8785354495048523,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 161340
+    },
+    {
+      "epoch": 1061.5131578947369,
+      "grad_norm": 1.1641191244125366,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 161350
+    },
+    {
+      "epoch": 1061.578947368421,
+      "grad_norm": 1.3598567247390747,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 161360
+    },
+    {
+      "epoch": 1061.6447368421052,
+      "grad_norm": 0.9496684670448303,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 161370
+    },
+    {
+      "epoch": 1061.7105263157894,
+      "grad_norm": 1.231241226196289,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 161380
+    },
+    {
+      "epoch": 1061.7763157894738,
+      "grad_norm": 1.052296757698059,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 161390
+    },
+    {
+      "epoch": 1061.842105263158,
+      "grad_norm": 0.9695555567741394,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 161400
+    },
+    {
+      "epoch": 1061.907894736842,
+      "grad_norm": 1.0984798669815063,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 161410
+    },
+    {
+      "epoch": 1061.9736842105262,
+      "grad_norm": 0.8881370425224304,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 161420
+    },
+    {
+      "epoch": 1062.0394736842106,
+      "grad_norm": 0.9678500294685364,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 161430
+    },
+    {
+      "epoch": 1062.1052631578948,
+      "grad_norm": 1.1650303602218628,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 161440
+    },
+    {
+      "epoch": 1062.171052631579,
+      "grad_norm": 1.0884886980056763,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 161450
+    },
+    {
+      "epoch": 1062.2368421052631,
+      "grad_norm": 0.9503740668296814,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 161460
+    },
+    {
+      "epoch": 1062.3026315789473,
+      "grad_norm": 0.8754977583885193,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 161470
+    },
+    {
+      "epoch": 1062.3684210526317,
+      "grad_norm": 1.0203204154968262,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 161480
+    },
+    {
+      "epoch": 1062.4342105263158,
+      "grad_norm": 0.9040229320526123,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 161490
+    },
+    {
+      "epoch": 1062.5,
+      "grad_norm": 1.0845030546188354,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 161500
+    },
+    {
+      "epoch": 1062.5657894736842,
+      "grad_norm": 0.950481116771698,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 161510
+    },
+    {
+      "epoch": 1062.6315789473683,
+      "grad_norm": 1.1162766218185425,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 161520
+    },
+    {
+      "epoch": 1062.6973684210527,
+      "grad_norm": 1.0046415328979492,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 161530
+    },
+    {
+      "epoch": 1062.7631578947369,
+      "grad_norm": 0.8391357064247131,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 161540
+    },
+    {
+      "epoch": 1062.828947368421,
+      "grad_norm": 0.9628133773803711,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 161550
+    },
+    {
+      "epoch": 1062.8947368421052,
+      "grad_norm": 0.9832002520561218,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 161560
+    },
+    {
+      "epoch": 1062.9605263157894,
+      "grad_norm": 1.2245110273361206,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 161570
+    },
+    {
+      "epoch": 1063.0263157894738,
+      "grad_norm": 1.3383867740631104,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 161580
+    },
+    {
+      "epoch": 1063.092105263158,
+      "grad_norm": 1.1894196271896362,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 161590
+    },
+    {
+      "epoch": 1063.157894736842,
+      "grad_norm": 1.3893300294876099,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 161600
+    },
+    {
+      "epoch": 1063.2236842105262,
+      "grad_norm": 0.8097324371337891,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 161610
+    },
+    {
+      "epoch": 1063.2894736842106,
+      "grad_norm": 0.9299864172935486,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 161620
+    },
+    {
+      "epoch": 1063.3552631578948,
+      "grad_norm": 1.031350016593933,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 161630
+    },
+    {
+      "epoch": 1063.421052631579,
+      "grad_norm": 1.079654574394226,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 161640
+    },
+    {
+      "epoch": 1063.4868421052631,
+      "grad_norm": 1.1577768325805664,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 161650
+    },
+    {
+      "epoch": 1063.5526315789473,
+      "grad_norm": 1.0904746055603027,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 161660
+    },
+    {
+      "epoch": 1063.6184210526317,
+      "grad_norm": 1.2395423650741577,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 161670
+    },
+    {
+      "epoch": 1063.6842105263158,
+      "grad_norm": 1.0951871871948242,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 161680
+    },
+    {
+      "epoch": 1063.75,
+      "grad_norm": 1.186826467514038,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 161690
+    },
+    {
+      "epoch": 1063.8157894736842,
+      "grad_norm": 0.7495224475860596,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 161700
+    },
+    {
+      "epoch": 1063.8815789473683,
+      "grad_norm": 1.0695480108261108,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 161710
+    },
+    {
+      "epoch": 1063.9473684210527,
+      "grad_norm": 0.743630051612854,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 161720
+    },
+    {
+      "epoch": 1064.0131578947369,
+      "grad_norm": 0.8561877608299255,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 161730
+    },
+    {
+      "epoch": 1064.078947368421,
+      "grad_norm": 0.8216092586517334,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 161740
+    },
+    {
+      "epoch": 1064.1447368421052,
+      "grad_norm": 0.8354560136795044,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 161750
+    },
+    {
+      "epoch": 1064.2105263157894,
+      "grad_norm": 1.2202868461608887,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 161760
+    },
+    {
+      "epoch": 1064.2763157894738,
+      "grad_norm": 0.7838713526725769,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 161770
+    },
+    {
+      "epoch": 1064.342105263158,
+      "grad_norm": 1.285569190979004,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 161780
+    },
+    {
+      "epoch": 1064.407894736842,
+      "grad_norm": 0.9280838370323181,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 161790
+    },
+    {
+      "epoch": 1064.4736842105262,
+      "grad_norm": 0.8232817649841309,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 161800
+    },
+    {
+      "epoch": 1064.5394736842106,
+      "grad_norm": 1.015890121459961,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 161810
+    },
+    {
+      "epoch": 1064.6052631578948,
+      "grad_norm": 0.8009679317474365,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 161820
+    },
+    {
+      "epoch": 1064.671052631579,
+      "grad_norm": 0.6812798976898193,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 161830
+    },
+    {
+      "epoch": 1064.7368421052631,
+      "grad_norm": 1.3212003707885742,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 161840
+    },
+    {
+      "epoch": 1064.8026315789473,
+      "grad_norm": 0.8784815669059753,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 161850
+    },
+    {
+      "epoch": 1064.8684210526317,
+      "grad_norm": 1.0952194929122925,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 161860
+    },
+    {
+      "epoch": 1064.9342105263158,
+      "grad_norm": 1.038833498954773,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 161870
+    },
+    {
+      "epoch": 1065.0,
+      "grad_norm": 1.1288541555404663,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 161880
+    },
+    {
+      "epoch": 1065.0657894736842,
+      "grad_norm": 1.0623427629470825,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 161890
+    },
+    {
+      "epoch": 1065.1315789473683,
+      "grad_norm": 1.2442352771759033,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 161900
+    },
+    {
+      "epoch": 1065.1973684210527,
+      "grad_norm": 1.1538206338882446,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 161910
+    },
+    {
+      "epoch": 1065.2631578947369,
+      "grad_norm": 1.1074141263961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 161920
+    },
+    {
+      "epoch": 1065.328947368421,
+      "grad_norm": 1.0948708057403564,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 161930
+    },
+    {
+      "epoch": 1065.3947368421052,
+      "grad_norm": 0.8019455671310425,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 161940
+    },
+    {
+      "epoch": 1065.4605263157894,
+      "grad_norm": 0.6959496736526489,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 161950
+    },
+    {
+      "epoch": 1065.5263157894738,
+      "grad_norm": 0.9164180755615234,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 161960
+    },
+    {
+      "epoch": 1065.592105263158,
+      "grad_norm": 1.390481948852539,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 161970
+    },
+    {
+      "epoch": 1065.657894736842,
+      "grad_norm": 1.1339566707611084,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 161980
+    },
+    {
+      "epoch": 1065.7236842105262,
+      "grad_norm": 0.8999402523040771,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 161990
+    },
+    {
+      "epoch": 1065.7894736842106,
+      "grad_norm": 1.1724854707717896,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 162000
+    },
+    {
+      "epoch": 1065.8552631578948,
+      "grad_norm": 1.1292765140533447,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 162010
+    },
+    {
+      "epoch": 1065.921052631579,
+      "grad_norm": 1.3481417894363403,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 162020
+    },
+    {
+      "epoch": 1065.9868421052631,
+      "grad_norm": 1.1352733373641968,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 162030
+    },
+    {
+      "epoch": 1066.0526315789473,
+      "grad_norm": 1.0196239948272705,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 162040
+    },
+    {
+      "epoch": 1066.1184210526317,
+      "grad_norm": 1.2338160276412964,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 162050
+    },
+    {
+      "epoch": 1066.1842105263158,
+      "grad_norm": 0.9089396595954895,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 162060
+    },
+    {
+      "epoch": 1066.25,
+      "grad_norm": 0.8388129472732544,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 162070
+    },
+    {
+      "epoch": 1066.3157894736842,
+      "grad_norm": 1.036230206489563,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 162080
+    },
+    {
+      "epoch": 1066.3815789473683,
+      "grad_norm": 0.9070473909378052,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 162090
+    },
+    {
+      "epoch": 1066.4473684210527,
+      "grad_norm": 1.3986698389053345,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 162100
+    },
+    {
+      "epoch": 1066.5131578947369,
+      "grad_norm": 1.2524380683898926,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 162110
+    },
+    {
+      "epoch": 1066.578947368421,
+      "grad_norm": 1.3897756338119507,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 162120
+    },
+    {
+      "epoch": 1066.6447368421052,
+      "grad_norm": 1.2259730100631714,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 162130
+    },
+    {
+      "epoch": 1066.7105263157894,
+      "grad_norm": 1.0453299283981323,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 162140
+    },
+    {
+      "epoch": 1066.7763157894738,
+      "grad_norm": 1.3548429012298584,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 162150
+    },
+    {
+      "epoch": 1066.842105263158,
+      "grad_norm": 0.7889332175254822,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 162160
+    },
+    {
+      "epoch": 1066.907894736842,
+      "grad_norm": 0.773144006729126,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 162170
+    },
+    {
+      "epoch": 1066.9736842105262,
+      "grad_norm": 0.8035649657249451,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 162180
+    },
+    {
+      "epoch": 1067.0394736842106,
+      "grad_norm": 0.7761321067810059,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 162190
+    },
+    {
+      "epoch": 1067.1052631578948,
+      "grad_norm": 0.8864235877990723,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 162200
+    },
+    {
+      "epoch": 1067.171052631579,
+      "grad_norm": 1.0513126850128174,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 162210
+    },
+    {
+      "epoch": 1067.2368421052631,
+      "grad_norm": 0.9010508060455322,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 162220
+    },
+    {
+      "epoch": 1067.3026315789473,
+      "grad_norm": 1.1405491828918457,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 162230
+    },
+    {
+      "epoch": 1067.3684210526317,
+      "grad_norm": 0.7161151170730591,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 162240
+    },
+    {
+      "epoch": 1067.4342105263158,
+      "grad_norm": 0.7078927755355835,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 162250
+    },
+    {
+      "epoch": 1067.5,
+      "grad_norm": 0.9383971095085144,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 162260
+    },
+    {
+      "epoch": 1067.5657894736842,
+      "grad_norm": 0.7436659336090088,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 162270
+    },
+    {
+      "epoch": 1067.6315789473683,
+      "grad_norm": 0.8315485715866089,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 162280
+    },
+    {
+      "epoch": 1067.6973684210527,
+      "grad_norm": 1.013370156288147,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 162290
+    },
+    {
+      "epoch": 1067.7631578947369,
+      "grad_norm": 1.2187751531600952,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 162300
+    },
+    {
+      "epoch": 1067.828947368421,
+      "grad_norm": 1.2132238149642944,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 162310
+    },
+    {
+      "epoch": 1067.8947368421052,
+      "grad_norm": 1.1550824642181396,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 162320
+    },
+    {
+      "epoch": 1067.9605263157894,
+      "grad_norm": 1.193505883216858,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 162330
+    },
+    {
+      "epoch": 1068.0263157894738,
+      "grad_norm": 0.9410684108734131,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 162340
+    },
+    {
+      "epoch": 1068.092105263158,
+      "grad_norm": 1.1005580425262451,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 162350
+    },
+    {
+      "epoch": 1068.157894736842,
+      "grad_norm": 1.2267907857894897,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 162360
+    },
+    {
+      "epoch": 1068.2236842105262,
+      "grad_norm": 1.1919841766357422,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 162370
+    },
+    {
+      "epoch": 1068.2894736842106,
+      "grad_norm": 1.14139986038208,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 162380
+    },
+    {
+      "epoch": 1068.3552631578948,
+      "grad_norm": 0.8994879722595215,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 162390
+    },
+    {
+      "epoch": 1068.421052631579,
+      "grad_norm": 1.3189268112182617,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 162400
+    },
+    {
+      "epoch": 1068.4868421052631,
+      "grad_norm": 1.0653045177459717,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 162410
+    },
+    {
+      "epoch": 1068.5526315789473,
+      "grad_norm": 1.248468041419983,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 162420
+    },
+    {
+      "epoch": 1068.6184210526317,
+      "grad_norm": 1.4759578704833984,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 162430
+    },
+    {
+      "epoch": 1068.6842105263158,
+      "grad_norm": 1.3343039751052856,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 162440
+    },
+    {
+      "epoch": 1068.75,
+      "grad_norm": 0.91098952293396,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 162450
+    },
+    {
+      "epoch": 1068.8157894736842,
+      "grad_norm": 1.612722635269165,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 162460
+    },
+    {
+      "epoch": 1068.8815789473683,
+      "grad_norm": 1.5344117879867554,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 162470
+    },
+    {
+      "epoch": 1068.9473684210527,
+      "grad_norm": 1.2199701070785522,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 162480
+    },
+    {
+      "epoch": 1069.0131578947369,
+      "grad_norm": 1.03998601436615,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 162490
+    },
+    {
+      "epoch": 1069.078947368421,
+      "grad_norm": 1.0132770538330078,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 162500
+    },
+    {
+      "epoch": 1069.1447368421052,
+      "grad_norm": 1.1793491840362549,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 162510
+    },
+    {
+      "epoch": 1069.2105263157894,
+      "grad_norm": 1.0435994863510132,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 162520
+    },
+    {
+      "epoch": 1069.2763157894738,
+      "grad_norm": 0.7933632135391235,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 162530
+    },
+    {
+      "epoch": 1069.342105263158,
+      "grad_norm": 1.094799280166626,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 162540
+    },
+    {
+      "epoch": 1069.407894736842,
+      "grad_norm": 1.0293395519256592,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 162550
+    },
+    {
+      "epoch": 1069.4736842105262,
+      "grad_norm": 0.8514150381088257,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 162560
+    },
+    {
+      "epoch": 1069.5394736842106,
+      "grad_norm": 1.075880765914917,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 162570
+    },
+    {
+      "epoch": 1069.6052631578948,
+      "grad_norm": 0.8023967742919922,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 162580
+    },
+    {
+      "epoch": 1069.671052631579,
+      "grad_norm": 0.9331603050231934,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 162590
+    },
+    {
+      "epoch": 1069.7368421052631,
+      "grad_norm": 0.8573759198188782,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 162600
+    },
+    {
+      "epoch": 1069.8026315789473,
+      "grad_norm": 1.105576515197754,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 162610
+    },
+    {
+      "epoch": 1069.8684210526317,
+      "grad_norm": 1.3647798299789429,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 162620
+    },
+    {
+      "epoch": 1069.9342105263158,
+      "grad_norm": 0.9643642902374268,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 162630
+    },
+    {
+      "epoch": 1070.0,
+      "grad_norm": 1.3293102979660034,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 162640
+    },
+    {
+      "epoch": 1070.0657894736842,
+      "grad_norm": 1.421432375907898,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 162650
+    },
+    {
+      "epoch": 1070.1315789473683,
+      "grad_norm": 1.0988101959228516,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 162660
+    },
+    {
+      "epoch": 1070.1973684210527,
+      "grad_norm": 1.185173511505127,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 162670
+    },
+    {
+      "epoch": 1070.2631578947369,
+      "grad_norm": 0.96378093957901,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 162680
+    },
+    {
+      "epoch": 1070.328947368421,
+      "grad_norm": 1.0372625589370728,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 162690
+    },
+    {
+      "epoch": 1070.3947368421052,
+      "grad_norm": 0.8378893136978149,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 162700
+    },
+    {
+      "epoch": 1070.4605263157894,
+      "grad_norm": 1.0191593170166016,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 162710
+    },
+    {
+      "epoch": 1070.5263157894738,
+      "grad_norm": 1.2982977628707886,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 162720
+    },
+    {
+      "epoch": 1070.592105263158,
+      "grad_norm": 1.0704700946807861,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 162730
+    },
+    {
+      "epoch": 1070.657894736842,
+      "grad_norm": 1.0267466306686401,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 162740
+    },
+    {
+      "epoch": 1070.7236842105262,
+      "grad_norm": 1.131195068359375,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 162750
+    },
+    {
+      "epoch": 1070.7894736842106,
+      "grad_norm": 0.9964534044265747,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 162760
+    },
+    {
+      "epoch": 1070.8552631578948,
+      "grad_norm": 1.135125994682312,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 162770
+    },
+    {
+      "epoch": 1070.921052631579,
+      "grad_norm": 0.5625234842300415,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 162780
+    },
+    {
+      "epoch": 1070.9868421052631,
+      "grad_norm": 1.0314369201660156,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 162790
+    },
+    {
+      "epoch": 1071.0526315789473,
+      "grad_norm": 1.1041908264160156,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 162800
+    },
+    {
+      "epoch": 1071.1184210526317,
+      "grad_norm": 1.4432227611541748,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 162810
+    },
+    {
+      "epoch": 1071.1842105263158,
+      "grad_norm": 1.1122026443481445,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 162820
+    },
+    {
+      "epoch": 1071.25,
+      "grad_norm": 0.8845628499984741,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 162830
+    },
+    {
+      "epoch": 1071.3157894736842,
+      "grad_norm": 0.9934724569320679,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 162840
+    },
+    {
+      "epoch": 1071.3815789473683,
+      "grad_norm": 1.2053556442260742,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 162850
+    },
+    {
+      "epoch": 1071.4473684210527,
+      "grad_norm": 1.0750372409820557,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 162860
+    },
+    {
+      "epoch": 1071.5131578947369,
+      "grad_norm": 1.1821308135986328,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 162870
+    },
+    {
+      "epoch": 1071.578947368421,
+      "grad_norm": 0.8854652047157288,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 162880
+    },
+    {
+      "epoch": 1071.6447368421052,
+      "grad_norm": 0.9452785849571228,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 162890
+    },
+    {
+      "epoch": 1071.7105263157894,
+      "grad_norm": 0.4714410901069641,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 162900
+    },
+    {
+      "epoch": 1071.7763157894738,
+      "grad_norm": 0.9950037598609924,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 162910
+    },
+    {
+      "epoch": 1071.842105263158,
+      "grad_norm": 1.0008114576339722,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 162920
+    },
+    {
+      "epoch": 1071.907894736842,
+      "grad_norm": 0.9882568120956421,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 162930
+    },
+    {
+      "epoch": 1071.9736842105262,
+      "grad_norm": 0.9975512623786926,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 162940
+    },
+    {
+      "epoch": 1072.0394736842106,
+      "grad_norm": 0.9865131974220276,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 162950
+    },
+    {
+      "epoch": 1072.1052631578948,
+      "grad_norm": 0.8073837757110596,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 162960
+    },
+    {
+      "epoch": 1072.171052631579,
+      "grad_norm": 1.311487078666687,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 162970
+    },
+    {
+      "epoch": 1072.2368421052631,
+      "grad_norm": 1.1827518939971924,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 162980
+    },
+    {
+      "epoch": 1072.3026315789473,
+      "grad_norm": 1.183228611946106,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 162990
+    },
+    {
+      "epoch": 1072.3684210526317,
+      "grad_norm": 1.4253993034362793,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 163000
+    },
+    {
+      "epoch": 1072.4342105263158,
+      "grad_norm": 1.0491241216659546,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 163010
+    },
+    {
+      "epoch": 1072.5,
+      "grad_norm": 1.536728858947754,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 163020
+    },
+    {
+      "epoch": 1072.5657894736842,
+      "grad_norm": 0.9010704159736633,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 163030
+    },
+    {
+      "epoch": 1072.6315789473683,
+      "grad_norm": 0.9637726545333862,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 163040
+    },
+    {
+      "epoch": 1072.6973684210527,
+      "grad_norm": 1.410753846168518,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 163050
+    },
+    {
+      "epoch": 1072.7631578947369,
+      "grad_norm": 1.4014109373092651,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 163060
+    },
+    {
+      "epoch": 1072.828947368421,
+      "grad_norm": 1.3508261442184448,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 163070
+    },
+    {
+      "epoch": 1072.8947368421052,
+      "grad_norm": 1.2983468770980835,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 163080
+    },
+    {
+      "epoch": 1072.9605263157894,
+      "grad_norm": 1.4878889322280884,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 163090
+    },
+    {
+      "epoch": 1073.0263157894738,
+      "grad_norm": 1.155154824256897,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 163100
+    },
+    {
+      "epoch": 1073.092105263158,
+      "grad_norm": 1.2900110483169556,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 163110
+    },
+    {
+      "epoch": 1073.157894736842,
+      "grad_norm": 0.711246907711029,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 163120
+    },
+    {
+      "epoch": 1073.2236842105262,
+      "grad_norm": 1.4569929838180542,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 163130
+    },
+    {
+      "epoch": 1073.2894736842106,
+      "grad_norm": 1.12445068359375,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 163140
+    },
+    {
+      "epoch": 1073.3552631578948,
+      "grad_norm": 0.9855101704597473,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 163150
+    },
+    {
+      "epoch": 1073.421052631579,
+      "grad_norm": 0.9957090020179749,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 163160
+    },
+    {
+      "epoch": 1073.4868421052631,
+      "grad_norm": 0.8900773525238037,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 163170
+    },
+    {
+      "epoch": 1073.5526315789473,
+      "grad_norm": 0.8382232785224915,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 163180
+    },
+    {
+      "epoch": 1073.6184210526317,
+      "grad_norm": 1.113542914390564,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 163190
+    },
+    {
+      "epoch": 1073.6842105263158,
+      "grad_norm": 1.017556071281433,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 163200
+    },
+    {
+      "epoch": 1073.75,
+      "grad_norm": 1.1955180168151855,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 163210
+    },
+    {
+      "epoch": 1073.8157894736842,
+      "grad_norm": 1.223476529121399,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 163220
+    },
+    {
+      "epoch": 1073.8815789473683,
+      "grad_norm": 0.9942741394042969,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 163230
+    },
+    {
+      "epoch": 1073.9473684210527,
+      "grad_norm": 1.0802150964736938,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 163240
+    },
+    {
+      "epoch": 1074.0131578947369,
+      "grad_norm": 0.9715554714202881,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 163250
+    },
+    {
+      "epoch": 1074.078947368421,
+      "grad_norm": 1.1486918926239014,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 163260
+    },
+    {
+      "epoch": 1074.1447368421052,
+      "grad_norm": 1.1075643301010132,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 163270
+    },
+    {
+      "epoch": 1074.2105263157894,
+      "grad_norm": 1.1682664155960083,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 163280
+    },
+    {
+      "epoch": 1074.2763157894738,
+      "grad_norm": 1.3140562772750854,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 163290
+    },
+    {
+      "epoch": 1074.342105263158,
+      "grad_norm": 1.2166996002197266,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 163300
+    },
+    {
+      "epoch": 1074.407894736842,
+      "grad_norm": 1.1994695663452148,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 163310
+    },
+    {
+      "epoch": 1074.4736842105262,
+      "grad_norm": 0.7714230418205261,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 163320
+    },
+    {
+      "epoch": 1074.5394736842106,
+      "grad_norm": 0.611962080001831,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 163330
+    },
+    {
+      "epoch": 1074.6052631578948,
+      "grad_norm": 0.8996819853782654,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 163340
+    },
+    {
+      "epoch": 1074.671052631579,
+      "grad_norm": 0.9892165064811707,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 163350
+    },
+    {
+      "epoch": 1074.7368421052631,
+      "grad_norm": 1.245927333831787,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 163360
+    },
+    {
+      "epoch": 1074.8026315789473,
+      "grad_norm": 1.5583741664886475,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 163370
+    },
+    {
+      "epoch": 1074.8684210526317,
+      "grad_norm": 1.4896721839904785,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 163380
+    },
+    {
+      "epoch": 1074.9342105263158,
+      "grad_norm": 1.5544887781143188,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 163390
+    },
+    {
+      "epoch": 1075.0,
+      "grad_norm": 1.562259554862976,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 163400
+    },
+    {
+      "epoch": 1075.0657894736842,
+      "grad_norm": 1.1588703393936157,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 163410
+    },
+    {
+      "epoch": 1075.1315789473683,
+      "grad_norm": 1.3577895164489746,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 163420
+    },
+    {
+      "epoch": 1075.1973684210527,
+      "grad_norm": 1.0716686248779297,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 163430
+    },
+    {
+      "epoch": 1075.2631578947369,
+      "grad_norm": 1.1159995794296265,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 163440
+    },
+    {
+      "epoch": 1075.328947368421,
+      "grad_norm": 1.1047301292419434,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 163450
+    },
+    {
+      "epoch": 1075.3947368421052,
+      "grad_norm": 1.001532793045044,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 163460
+    },
+    {
+      "epoch": 1075.4605263157894,
+      "grad_norm": 1.4024605751037598,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 163470
+    },
+    {
+      "epoch": 1075.5263157894738,
+      "grad_norm": 1.3084982633590698,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 163480
+    },
+    {
+      "epoch": 1075.592105263158,
+      "grad_norm": 1.0299196243286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 163490
+    },
+    {
+      "epoch": 1075.657894736842,
+      "grad_norm": 0.941424548625946,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 163500
+    },
+    {
+      "epoch": 1075.7236842105262,
+      "grad_norm": 1.2184175252914429,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 163510
+    },
+    {
+      "epoch": 1075.7894736842106,
+      "grad_norm": 1.1736630201339722,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 163520
+    },
+    {
+      "epoch": 1075.8552631578948,
+      "grad_norm": 0.897025465965271,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 163530
+    },
+    {
+      "epoch": 1075.921052631579,
+      "grad_norm": 0.8720511198043823,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 163540
+    },
+    {
+      "epoch": 1075.9868421052631,
+      "grad_norm": 1.0315616130828857,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 163550
+    },
+    {
+      "epoch": 1076.0526315789473,
+      "grad_norm": 1.194000005722046,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 163560
+    },
+    {
+      "epoch": 1076.1184210526317,
+      "grad_norm": 1.108793020248413,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 163570
+    },
+    {
+      "epoch": 1076.1842105263158,
+      "grad_norm": 1.059388518333435,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 163580
+    },
+    {
+      "epoch": 1076.25,
+      "grad_norm": 1.1141692399978638,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 163590
+    },
+    {
+      "epoch": 1076.3157894736842,
+      "grad_norm": 0.8344478011131287,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 163600
+    },
+    {
+      "epoch": 1076.3815789473683,
+      "grad_norm": 1.0367026329040527,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 163610
+    },
+    {
+      "epoch": 1076.4473684210527,
+      "grad_norm": 1.339621663093567,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 163620
+    },
+    {
+      "epoch": 1076.5131578947369,
+      "grad_norm": 1.3752906322479248,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 163630
+    },
+    {
+      "epoch": 1076.578947368421,
+      "grad_norm": 0.8494938015937805,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 163640
+    },
+    {
+      "epoch": 1076.6447368421052,
+      "grad_norm": 0.6599777936935425,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 163650
+    },
+    {
+      "epoch": 1076.7105263157894,
+      "grad_norm": 1.3162485361099243,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 163660
+    },
+    {
+      "epoch": 1076.7763157894738,
+      "grad_norm": 0.9044468998908997,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 163670
+    },
+    {
+      "epoch": 1076.842105263158,
+      "grad_norm": 0.8075153231620789,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 163680
+    },
+    {
+      "epoch": 1076.907894736842,
+      "grad_norm": 1.406198501586914,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 163690
+    },
+    {
+      "epoch": 1076.9736842105262,
+      "grad_norm": 1.080063819885254,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 163700
+    },
+    {
+      "epoch": 1077.0394736842106,
+      "grad_norm": 1.1820083856582642,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 163710
+    },
+    {
+      "epoch": 1077.1052631578948,
+      "grad_norm": 1.0260475873947144,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 163720
+    },
+    {
+      "epoch": 1077.171052631579,
+      "grad_norm": 1.063698410987854,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 163730
+    },
+    {
+      "epoch": 1077.2368421052631,
+      "grad_norm": 0.7983181476593018,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 163740
+    },
+    {
+      "epoch": 1077.3026315789473,
+      "grad_norm": 0.8934413194656372,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 163750
+    },
+    {
+      "epoch": 1077.3684210526317,
+      "grad_norm": 0.701677680015564,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 163760
+    },
+    {
+      "epoch": 1077.4342105263158,
+      "grad_norm": 0.9841487407684326,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 163770
+    },
+    {
+      "epoch": 1077.5,
+      "grad_norm": 0.8910136222839355,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 163780
+    },
+    {
+      "epoch": 1077.5657894736842,
+      "grad_norm": 1.0197700262069702,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 163790
+    },
+    {
+      "epoch": 1077.6315789473683,
+      "grad_norm": 1.3116474151611328,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 163800
+    },
+    {
+      "epoch": 1077.6973684210527,
+      "grad_norm": 0.8362675905227661,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 163810
+    },
+    {
+      "epoch": 1077.7631578947369,
+      "grad_norm": 0.6043428182601929,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 163820
+    },
+    {
+      "epoch": 1077.828947368421,
+      "grad_norm": 1.1942695379257202,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 163830
+    },
+    {
+      "epoch": 1077.8947368421052,
+      "grad_norm": 0.9171984195709229,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 163840
+    },
+    {
+      "epoch": 1077.9605263157894,
+      "grad_norm": 1.100325584411621,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 163850
+    },
+    {
+      "epoch": 1078.0263157894738,
+      "grad_norm": 1.150246024131775,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 163860
+    },
+    {
+      "epoch": 1078.092105263158,
+      "grad_norm": 0.9796252250671387,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 163870
+    },
+    {
+      "epoch": 1078.157894736842,
+      "grad_norm": 0.7340827584266663,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 163880
+    },
+    {
+      "epoch": 1078.2236842105262,
+      "grad_norm": 0.9303230047225952,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 163890
+    },
+    {
+      "epoch": 1078.2894736842106,
+      "grad_norm": 0.9361478090286255,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 163900
+    },
+    {
+      "epoch": 1078.3552631578948,
+      "grad_norm": 1.0189319849014282,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 163910
+    },
+    {
+      "epoch": 1078.421052631579,
+      "grad_norm": 1.059217929840088,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 163920
+    },
+    {
+      "epoch": 1078.4868421052631,
+      "grad_norm": 1.3725210428237915,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 163930
+    },
+    {
+      "epoch": 1078.5526315789473,
+      "grad_norm": 1.200869083404541,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 163940
+    },
+    {
+      "epoch": 1078.6184210526317,
+      "grad_norm": 1.561780571937561,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 163950
+    },
+    {
+      "epoch": 1078.6842105263158,
+      "grad_norm": 1.174270510673523,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 163960
+    },
+    {
+      "epoch": 1078.75,
+      "grad_norm": 1.1414722204208374,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 163970
+    },
+    {
+      "epoch": 1078.8157894736842,
+      "grad_norm": 1.0247485637664795,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 163980
+    },
+    {
+      "epoch": 1078.8815789473683,
+      "grad_norm": 0.9565194249153137,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 163990
+    },
+    {
+      "epoch": 1078.9473684210527,
+      "grad_norm": 0.9359107613563538,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 164000
+    },
+    {
+      "epoch": 1079.0131578947369,
+      "grad_norm": 0.8053498268127441,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 164010
+    },
+    {
+      "epoch": 1079.078947368421,
+      "grad_norm": 1.111828088760376,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 164020
+    },
+    {
+      "epoch": 1079.1447368421052,
+      "grad_norm": 0.9666885137557983,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 164030
+    },
+    {
+      "epoch": 1079.2105263157894,
+      "grad_norm": 1.089706301689148,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 164040
+    },
+    {
+      "epoch": 1079.2763157894738,
+      "grad_norm": 1.2634003162384033,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 164050
+    },
+    {
+      "epoch": 1079.342105263158,
+      "grad_norm": 1.2256748676300049,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 164060
+    },
+    {
+      "epoch": 1079.407894736842,
+      "grad_norm": 1.0512341260910034,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 164070
+    },
+    {
+      "epoch": 1079.4736842105262,
+      "grad_norm": 1.1967103481292725,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 164080
+    },
+    {
+      "epoch": 1079.5394736842106,
+      "grad_norm": 0.9088690876960754,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 164090
+    },
+    {
+      "epoch": 1079.6052631578948,
+      "grad_norm": 0.8924160599708557,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 164100
+    },
+    {
+      "epoch": 1079.671052631579,
+      "grad_norm": 0.8244693875312805,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 164110
+    },
+    {
+      "epoch": 1079.7368421052631,
+      "grad_norm": 1.0724085569381714,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 164120
+    },
+    {
+      "epoch": 1079.8026315789473,
+      "grad_norm": 0.8203881978988647,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 164130
+    },
+    {
+      "epoch": 1079.8684210526317,
+      "grad_norm": 0.8975365161895752,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 164140
+    },
+    {
+      "epoch": 1079.9342105263158,
+      "grad_norm": 1.132312297821045,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 164150
+    },
+    {
+      "epoch": 1080.0,
+      "grad_norm": 1.0990959405899048,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 164160
+    },
+    {
+      "epoch": 1080.0657894736842,
+      "grad_norm": 0.9807849526405334,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 164170
+    },
+    {
+      "epoch": 1080.1315789473683,
+      "grad_norm": 1.4156996011734009,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 164180
+    },
+    {
+      "epoch": 1080.1973684210527,
+      "grad_norm": 1.217441439628601,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 164190
+    },
+    {
+      "epoch": 1080.2631578947369,
+      "grad_norm": 1.1044880151748657,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 164200
+    },
+    {
+      "epoch": 1080.328947368421,
+      "grad_norm": 1.1214570999145508,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 164210
+    },
+    {
+      "epoch": 1080.3947368421052,
+      "grad_norm": 0.7764980792999268,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 164220
+    },
+    {
+      "epoch": 1080.4605263157894,
+      "grad_norm": 1.1083272695541382,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 164230
+    },
+    {
+      "epoch": 1080.5263157894738,
+      "grad_norm": 0.908808171749115,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 164240
+    },
+    {
+      "epoch": 1080.592105263158,
+      "grad_norm": 0.9142073392868042,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 164250
+    },
+    {
+      "epoch": 1080.657894736842,
+      "grad_norm": 0.9220647215843201,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 164260
+    },
+    {
+      "epoch": 1080.7236842105262,
+      "grad_norm": 0.9497620463371277,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 164270
+    },
+    {
+      "epoch": 1080.7894736842106,
+      "grad_norm": 0.9928162693977356,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 164280
+    },
+    {
+      "epoch": 1080.8552631578948,
+      "grad_norm": 1.177316427230835,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 164290
+    },
+    {
+      "epoch": 1080.921052631579,
+      "grad_norm": 1.1457185745239258,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 164300
+    },
+    {
+      "epoch": 1080.9868421052631,
+      "grad_norm": 0.5666669011116028,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 164310
+    },
+    {
+      "epoch": 1081.0526315789473,
+      "grad_norm": 0.9142729640007019,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 164320
+    },
+    {
+      "epoch": 1081.1184210526317,
+      "grad_norm": 0.9657456874847412,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 164330
+    },
+    {
+      "epoch": 1081.1842105263158,
+      "grad_norm": 0.9326075315475464,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 164340
+    },
+    {
+      "epoch": 1081.25,
+      "grad_norm": 0.8688045740127563,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 164350
+    },
+    {
+      "epoch": 1081.3157894736842,
+      "grad_norm": 1.195733666419983,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 164360
+    },
+    {
+      "epoch": 1081.3815789473683,
+      "grad_norm": 1.1410993337631226,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 164370
+    },
+    {
+      "epoch": 1081.4473684210527,
+      "grad_norm": 0.9133111238479614,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 164380
+    },
+    {
+      "epoch": 1081.5131578947369,
+      "grad_norm": 0.9332903027534485,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 164390
+    },
+    {
+      "epoch": 1081.578947368421,
+      "grad_norm": 1.2633785009384155,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 164400
+    },
+    {
+      "epoch": 1081.6447368421052,
+      "grad_norm": 1.1889288425445557,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 164410
+    },
+    {
+      "epoch": 1081.7105263157894,
+      "grad_norm": 0.9122311472892761,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 164420
+    },
+    {
+      "epoch": 1081.7763157894738,
+      "grad_norm": 0.9126380085945129,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 164430
+    },
+    {
+      "epoch": 1081.842105263158,
+      "grad_norm": 0.9555377960205078,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 164440
+    },
+    {
+      "epoch": 1081.907894736842,
+      "grad_norm": 1.0733895301818848,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 164450
+    },
+    {
+      "epoch": 1081.9736842105262,
+      "grad_norm": 1.0887807607650757,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 164460
+    },
+    {
+      "epoch": 1082.0394736842106,
+      "grad_norm": 0.8500633835792542,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 164470
+    },
+    {
+      "epoch": 1082.1052631578948,
+      "grad_norm": 0.6595375537872314,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 164480
+    },
+    {
+      "epoch": 1082.171052631579,
+      "grad_norm": 1.3263654708862305,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 164490
+    },
+    {
+      "epoch": 1082.2368421052631,
+      "grad_norm": 0.9079999327659607,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 164500
+    },
+    {
+      "epoch": 1082.3026315789473,
+      "grad_norm": 0.9589836001396179,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 164510
+    },
+    {
+      "epoch": 1082.3684210526317,
+      "grad_norm": 1.1960152387619019,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 164520
+    },
+    {
+      "epoch": 1082.4342105263158,
+      "grad_norm": 0.9973219037055969,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 164530
+    },
+    {
+      "epoch": 1082.5,
+      "grad_norm": 1.1780554056167603,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 164540
+    },
+    {
+      "epoch": 1082.5657894736842,
+      "grad_norm": 1.2360888719558716,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 164550
+    },
+    {
+      "epoch": 1082.6315789473683,
+      "grad_norm": 1.0589853525161743,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 164560
+    },
+    {
+      "epoch": 1082.6973684210527,
+      "grad_norm": 1.2129298448562622,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 164570
+    },
+    {
+      "epoch": 1082.7631578947369,
+      "grad_norm": 0.8589812517166138,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 164580
+    },
+    {
+      "epoch": 1082.828947368421,
+      "grad_norm": 1.226012110710144,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 164590
+    },
+    {
+      "epoch": 1082.8947368421052,
+      "grad_norm": 0.921205461025238,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 164600
+    },
+    {
+      "epoch": 1082.9605263157894,
+      "grad_norm": 0.9451828598976135,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 164610
+    },
+    {
+      "epoch": 1083.0263157894738,
+      "grad_norm": 0.8518921732902527,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 164620
+    },
+    {
+      "epoch": 1083.092105263158,
+      "grad_norm": 0.7957374453544617,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 164630
+    },
+    {
+      "epoch": 1083.157894736842,
+      "grad_norm": 0.6601923108100891,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 164640
+    },
+    {
+      "epoch": 1083.2236842105262,
+      "grad_norm": 0.8923590779304504,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 164650
+    },
+    {
+      "epoch": 1083.2894736842106,
+      "grad_norm": 1.051577091217041,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 164660
+    },
+    {
+      "epoch": 1083.3552631578948,
+      "grad_norm": 1.3525593280792236,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 164670
+    },
+    {
+      "epoch": 1083.421052631579,
+      "grad_norm": 1.278694987297058,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 164680
+    },
+    {
+      "epoch": 1083.4868421052631,
+      "grad_norm": 1.5232925415039062,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 164690
+    },
+    {
+      "epoch": 1083.5526315789473,
+      "grad_norm": 1.3283989429473877,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 164700
+    },
+    {
+      "epoch": 1083.6184210526317,
+      "grad_norm": 1.3318917751312256,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 164710
+    },
+    {
+      "epoch": 1083.6842105263158,
+      "grad_norm": 1.2124128341674805,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 164720
+    },
+    {
+      "epoch": 1083.75,
+      "grad_norm": 1.336235761642456,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 164730
+    },
+    {
+      "epoch": 1083.8157894736842,
+      "grad_norm": 0.9546015858650208,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 164740
+    },
+    {
+      "epoch": 1083.8815789473683,
+      "grad_norm": 0.8885062336921692,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 164750
+    },
+    {
+      "epoch": 1083.9473684210527,
+      "grad_norm": 1.1417690515518188,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 164760
+    },
+    {
+      "epoch": 1084.0131578947369,
+      "grad_norm": 0.9051141738891602,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 164770
+    },
+    {
+      "epoch": 1084.078947368421,
+      "grad_norm": 1.195184588432312,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 164780
+    },
+    {
+      "epoch": 1084.1447368421052,
+      "grad_norm": 1.2856130599975586,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 164790
+    },
+    {
+      "epoch": 1084.2105263157894,
+      "grad_norm": 0.9908321499824524,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 164800
+    },
+    {
+      "epoch": 1084.2763157894738,
+      "grad_norm": 1.1323515176773071,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 164810
+    },
+    {
+      "epoch": 1084.342105263158,
+      "grad_norm": 1.0925931930541992,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 164820
+    },
+    {
+      "epoch": 1084.407894736842,
+      "grad_norm": 0.8730666637420654,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 164830
+    },
+    {
+      "epoch": 1084.4736842105262,
+      "grad_norm": 1.1487864255905151,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 164840
+    },
+    {
+      "epoch": 1084.5394736842106,
+      "grad_norm": 0.6327576041221619,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 164850
+    },
+    {
+      "epoch": 1084.6052631578948,
+      "grad_norm": 1.2673507928848267,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 164860
+    },
+    {
+      "epoch": 1084.671052631579,
+      "grad_norm": 1.0878818035125732,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 164870
+    },
+    {
+      "epoch": 1084.7368421052631,
+      "grad_norm": 1.1536747217178345,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 164880
+    },
+    {
+      "epoch": 1084.8026315789473,
+      "grad_norm": 1.274584174156189,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 164890
+    },
+    {
+      "epoch": 1084.8684210526317,
+      "grad_norm": 1.102174162864685,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 164900
+    },
+    {
+      "epoch": 1084.9342105263158,
+      "grad_norm": 0.8976937532424927,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 164910
+    },
+    {
+      "epoch": 1085.0,
+      "grad_norm": 1.2239928245544434,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 164920
+    },
+    {
+      "epoch": 1085.0657894736842,
+      "grad_norm": 0.8770923614501953,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 164930
+    },
+    {
+      "epoch": 1085.1315789473683,
+      "grad_norm": 0.9570636749267578,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 164940
+    },
+    {
+      "epoch": 1085.1973684210527,
+      "grad_norm": 0.9466798901557922,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 164950
+    },
+    {
+      "epoch": 1085.2631578947369,
+      "grad_norm": 0.897685170173645,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 164960
+    },
+    {
+      "epoch": 1085.328947368421,
+      "grad_norm": 0.710432767868042,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 164970
+    },
+    {
+      "epoch": 1085.3947368421052,
+      "grad_norm": 0.8062248826026917,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 164980
+    },
+    {
+      "epoch": 1085.4605263157894,
+      "grad_norm": 0.7032548785209656,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 164990
+    },
+    {
+      "epoch": 1085.5263157894738,
+      "grad_norm": 1.0174980163574219,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 165000
+    },
+    {
+      "epoch": 1085.592105263158,
+      "grad_norm": 0.7526353597640991,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 165010
+    },
+    {
+      "epoch": 1085.657894736842,
+      "grad_norm": 0.815987229347229,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 165020
+    },
+    {
+      "epoch": 1085.7236842105262,
+      "grad_norm": 0.6938906908035278,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 165030
+    },
+    {
+      "epoch": 1085.7894736842106,
+      "grad_norm": 1.025226354598999,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 165040
+    },
+    {
+      "epoch": 1085.8552631578948,
+      "grad_norm": 1.1357896327972412,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 165050
+    },
+    {
+      "epoch": 1085.921052631579,
+      "grad_norm": 1.0124887228012085,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 165060
+    },
+    {
+      "epoch": 1085.9868421052631,
+      "grad_norm": 0.9484216570854187,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 165070
+    },
+    {
+      "epoch": 1086.0526315789473,
+      "grad_norm": 1.2238960266113281,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 165080
+    },
+    {
+      "epoch": 1086.1184210526317,
+      "grad_norm": 1.4172849655151367,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 165090
+    },
+    {
+      "epoch": 1086.1842105263158,
+      "grad_norm": 1.292891263961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 165100
+    },
+    {
+      "epoch": 1086.25,
+      "grad_norm": 1.20212984085083,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 165110
+    },
+    {
+      "epoch": 1086.3157894736842,
+      "grad_norm": 1.3747832775115967,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 165120
+    },
+    {
+      "epoch": 1086.3815789473683,
+      "grad_norm": 1.1830034255981445,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 165130
+    },
+    {
+      "epoch": 1086.4473684210527,
+      "grad_norm": 1.2273931503295898,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 165140
+    },
+    {
+      "epoch": 1086.5131578947369,
+      "grad_norm": 0.9580655097961426,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 165150
+    },
+    {
+      "epoch": 1086.578947368421,
+      "grad_norm": 1.0733976364135742,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 165160
+    },
+    {
+      "epoch": 1086.6447368421052,
+      "grad_norm": 1.256765604019165,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 165170
+    },
+    {
+      "epoch": 1086.7105263157894,
+      "grad_norm": 0.8285326361656189,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 165180
+    },
+    {
+      "epoch": 1086.7763157894738,
+      "grad_norm": 0.8502132296562195,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 165190
+    },
+    {
+      "epoch": 1086.842105263158,
+      "grad_norm": 0.5578171610832214,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 165200
+    },
+    {
+      "epoch": 1086.907894736842,
+      "grad_norm": 0.8829189538955688,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 165210
+    },
+    {
+      "epoch": 1086.9736842105262,
+      "grad_norm": 0.8555329442024231,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 165220
+    },
+    {
+      "epoch": 1087.0394736842106,
+      "grad_norm": 1.1692776679992676,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 165230
+    },
+    {
+      "epoch": 1087.1052631578948,
+      "grad_norm": 0.903022825717926,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 165240
+    },
+    {
+      "epoch": 1087.171052631579,
+      "grad_norm": 0.9635505676269531,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 165250
+    },
+    {
+      "epoch": 1087.2368421052631,
+      "grad_norm": 1.0092943906784058,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 165260
+    },
+    {
+      "epoch": 1087.3026315789473,
+      "grad_norm": 0.9053812026977539,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 165270
+    },
+    {
+      "epoch": 1087.3684210526317,
+      "grad_norm": 1.0822906494140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 165280
+    },
+    {
+      "epoch": 1087.4342105263158,
+      "grad_norm": 0.9517032504081726,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 165290
+    },
+    {
+      "epoch": 1087.5,
+      "grad_norm": 0.995897114276886,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 165300
+    },
+    {
+      "epoch": 1087.5657894736842,
+      "grad_norm": 1.121558427810669,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 165310
+    },
+    {
+      "epoch": 1087.6315789473683,
+      "grad_norm": 0.905768632888794,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 165320
+    },
+    {
+      "epoch": 1087.6973684210527,
+      "grad_norm": 0.9619491100311279,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 165330
+    },
+    {
+      "epoch": 1087.7631578947369,
+      "grad_norm": 0.804125964641571,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 165340
+    },
+    {
+      "epoch": 1087.828947368421,
+      "grad_norm": 0.9737436175346375,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 165350
+    },
+    {
+      "epoch": 1087.8947368421052,
+      "grad_norm": 1.1739447116851807,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 165360
+    },
+    {
+      "epoch": 1087.9605263157894,
+      "grad_norm": 1.081647515296936,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 165370
+    },
+    {
+      "epoch": 1088.0263157894738,
+      "grad_norm": 1.38816499710083,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 165380
+    },
+    {
+      "epoch": 1088.092105263158,
+      "grad_norm": 1.0949757099151611,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 165390
+    },
+    {
+      "epoch": 1088.157894736842,
+      "grad_norm": 1.0412479639053345,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 165400
+    },
+    {
+      "epoch": 1088.2236842105262,
+      "grad_norm": 1.2765264511108398,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 165410
+    },
+    {
+      "epoch": 1088.2894736842106,
+      "grad_norm": 1.1357839107513428,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 165420
+    },
+    {
+      "epoch": 1088.3552631578948,
+      "grad_norm": 1.2164040803909302,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 165430
+    },
+    {
+      "epoch": 1088.421052631579,
+      "grad_norm": 1.2040680646896362,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 165440
+    },
+    {
+      "epoch": 1088.4868421052631,
+      "grad_norm": 1.342111349105835,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 165450
+    },
+    {
+      "epoch": 1088.5526315789473,
+      "grad_norm": 1.203928828239441,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 165460
+    },
+    {
+      "epoch": 1088.6184210526317,
+      "grad_norm": 1.0916163921356201,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 165470
+    },
+    {
+      "epoch": 1088.6842105263158,
+      "grad_norm": 0.970431387424469,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 165480
+    },
+    {
+      "epoch": 1088.75,
+      "grad_norm": 1.1007429361343384,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 165490
+    },
+    {
+      "epoch": 1088.8157894736842,
+      "grad_norm": 1.2328646183013916,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 165500
+    },
+    {
+      "epoch": 1088.8815789473683,
+      "grad_norm": 1.3475733995437622,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 165510
+    },
+    {
+      "epoch": 1088.9473684210527,
+      "grad_norm": 1.0184584856033325,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 165520
+    },
+    {
+      "epoch": 1089.0131578947369,
+      "grad_norm": 0.9720950722694397,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 165530
+    },
+    {
+      "epoch": 1089.078947368421,
+      "grad_norm": 1.327864646911621,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 165540
+    },
+    {
+      "epoch": 1089.1447368421052,
+      "grad_norm": 0.9376356601715088,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 165550
+    },
+    {
+      "epoch": 1089.2105263157894,
+      "grad_norm": 1.0245535373687744,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 165560
+    },
+    {
+      "epoch": 1089.2763157894738,
+      "grad_norm": 1.088114857673645,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 165570
+    },
+    {
+      "epoch": 1089.342105263158,
+      "grad_norm": 0.981323778629303,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 165580
+    },
+    {
+      "epoch": 1089.407894736842,
+      "grad_norm": 1.1533432006835938,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 165590
+    },
+    {
+      "epoch": 1089.4736842105262,
+      "grad_norm": 0.8093254566192627,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 165600
+    },
+    {
+      "epoch": 1089.5394736842106,
+      "grad_norm": 0.9100672006607056,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 165610
+    },
+    {
+      "epoch": 1089.6052631578948,
+      "grad_norm": 0.8746789693832397,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 165620
+    },
+    {
+      "epoch": 1089.671052631579,
+      "grad_norm": 1.4913291931152344,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 165630
+    },
+    {
+      "epoch": 1089.7368421052631,
+      "grad_norm": 1.1756936311721802,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 165640
+    },
+    {
+      "epoch": 1089.8026315789473,
+      "grad_norm": 1.144705891609192,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 165650
+    },
+    {
+      "epoch": 1089.8684210526317,
+      "grad_norm": 1.0096395015716553,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 165660
+    },
+    {
+      "epoch": 1089.9342105263158,
+      "grad_norm": 1.4962252378463745,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 165670
+    },
+    {
+      "epoch": 1090.0,
+      "grad_norm": 1.0682179927825928,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 165680
+    },
+    {
+      "epoch": 1090.0657894736842,
+      "grad_norm": 1.1246052980422974,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 165690
+    },
+    {
+      "epoch": 1090.1315789473683,
+      "grad_norm": 1.1388136148452759,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 165700
+    },
+    {
+      "epoch": 1090.1973684210527,
+      "grad_norm": 1.017364740371704,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 165710
+    },
+    {
+      "epoch": 1090.2631578947369,
+      "grad_norm": 1.1747958660125732,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 165720
+    },
+    {
+      "epoch": 1090.328947368421,
+      "grad_norm": 1.1064643859863281,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 165730
+    },
+    {
+      "epoch": 1090.3947368421052,
+      "grad_norm": 0.9353105425834656,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 165740
+    },
+    {
+      "epoch": 1090.4605263157894,
+      "grad_norm": 1.1124236583709717,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 165750
+    },
+    {
+      "epoch": 1090.5263157894738,
+      "grad_norm": 1.1237322092056274,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 165760
+    },
+    {
+      "epoch": 1090.592105263158,
+      "grad_norm": 0.8977600932121277,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 165770
+    },
+    {
+      "epoch": 1090.657894736842,
+      "grad_norm": 1.2883377075195312,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 165780
+    },
+    {
+      "epoch": 1090.7236842105262,
+      "grad_norm": 1.2734546661376953,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 165790
+    },
+    {
+      "epoch": 1090.7894736842106,
+      "grad_norm": 1.1414625644683838,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 165800
+    },
+    {
+      "epoch": 1090.8552631578948,
+      "grad_norm": 1.0614356994628906,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 165810
+    },
+    {
+      "epoch": 1090.921052631579,
+      "grad_norm": 1.0727033615112305,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 165820
+    },
+    {
+      "epoch": 1090.9868421052631,
+      "grad_norm": 1.1432024240493774,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 165830
+    },
+    {
+      "epoch": 1091.0526315789473,
+      "grad_norm": 1.158418893814087,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 165840
+    },
+    {
+      "epoch": 1091.1184210526317,
+      "grad_norm": 1.033921718597412,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 165850
+    },
+    {
+      "epoch": 1091.1842105263158,
+      "grad_norm": 1.0645828247070312,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 165860
+    },
+    {
+      "epoch": 1091.25,
+      "grad_norm": 0.5794889330863953,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 165870
+    },
+    {
+      "epoch": 1091.3157894736842,
+      "grad_norm": 1.1075804233551025,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 165880
+    },
+    {
+      "epoch": 1091.3815789473683,
+      "grad_norm": 0.9454206228256226,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 165890
+    },
+    {
+      "epoch": 1091.4473684210527,
+      "grad_norm": 1.1715730428695679,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 165900
+    },
+    {
+      "epoch": 1091.5131578947369,
+      "grad_norm": 1.0313235521316528,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 165910
+    },
+    {
+      "epoch": 1091.578947368421,
+      "grad_norm": 1.3936482667922974,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 165920
+    },
+    {
+      "epoch": 1091.6447368421052,
+      "grad_norm": 0.8240237832069397,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 165930
+    },
+    {
+      "epoch": 1091.7105263157894,
+      "grad_norm": 1.0866410732269287,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 165940
+    },
+    {
+      "epoch": 1091.7763157894738,
+      "grad_norm": 1.3597463369369507,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 165950
+    },
+    {
+      "epoch": 1091.842105263158,
+      "grad_norm": 1.3477791547775269,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 165960
+    },
+    {
+      "epoch": 1091.907894736842,
+      "grad_norm": 1.2000268697738647,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 165970
+    },
+    {
+      "epoch": 1091.9736842105262,
+      "grad_norm": 0.7252658009529114,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 165980
+    },
+    {
+      "epoch": 1092.0394736842106,
+      "grad_norm": 1.189745545387268,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 165990
+    },
+    {
+      "epoch": 1092.1052631578948,
+      "grad_norm": 0.7382572889328003,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 166000
+    },
+    {
+      "epoch": 1092.171052631579,
+      "grad_norm": 1.0772287845611572,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 166010
+    },
+    {
+      "epoch": 1092.2368421052631,
+      "grad_norm": 0.7473660111427307,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 166020
+    },
+    {
+      "epoch": 1092.3026315789473,
+      "grad_norm": 0.8912314176559448,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 166030
+    },
+    {
+      "epoch": 1092.3684210526317,
+      "grad_norm": 1.1493737697601318,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 166040
+    },
+    {
+      "epoch": 1092.4342105263158,
+      "grad_norm": 1.3039054870605469,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 166050
+    },
+    {
+      "epoch": 1092.5,
+      "grad_norm": 0.9073991775512695,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 166060
+    },
+    {
+      "epoch": 1092.5657894736842,
+      "grad_norm": 0.8854196667671204,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 166070
+    },
+    {
+      "epoch": 1092.6315789473683,
+      "grad_norm": 1.2598876953125,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 166080
+    },
+    {
+      "epoch": 1092.6973684210527,
+      "grad_norm": 1.2058098316192627,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 166090
+    },
+    {
+      "epoch": 1092.7631578947369,
+      "grad_norm": 1.188262701034546,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 166100
+    },
+    {
+      "epoch": 1092.828947368421,
+      "grad_norm": 1.5130947828292847,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 166110
+    },
+    {
+      "epoch": 1092.8947368421052,
+      "grad_norm": 1.2979497909545898,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 166120
+    },
+    {
+      "epoch": 1092.9605263157894,
+      "grad_norm": 1.0811450481414795,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 166130
+    },
+    {
+      "epoch": 1093.0263157894738,
+      "grad_norm": 1.0868312120437622,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 166140
+    },
+    {
+      "epoch": 1093.092105263158,
+      "grad_norm": 0.8259110450744629,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 166150
+    },
+    {
+      "epoch": 1093.157894736842,
+      "grad_norm": 1.2230782508850098,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 166160
+    },
+    {
+      "epoch": 1093.2236842105262,
+      "grad_norm": 1.296341896057129,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 166170
+    },
+    {
+      "epoch": 1093.2894736842106,
+      "grad_norm": 1.0726011991500854,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 166180
+    },
+    {
+      "epoch": 1093.3552631578948,
+      "grad_norm": 0.7867605686187744,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 166190
+    },
+    {
+      "epoch": 1093.421052631579,
+      "grad_norm": 0.8794106245040894,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 166200
+    },
+    {
+      "epoch": 1093.4868421052631,
+      "grad_norm": 1.1672301292419434,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 166210
+    },
+    {
+      "epoch": 1093.5526315789473,
+      "grad_norm": 0.9970499277114868,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 166220
+    },
+    {
+      "epoch": 1093.6184210526317,
+      "grad_norm": 1.1926120519638062,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 166230
+    },
+    {
+      "epoch": 1093.6842105263158,
+      "grad_norm": 0.935606837272644,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 166240
+    },
+    {
+      "epoch": 1093.75,
+      "grad_norm": 1.3509941101074219,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 166250
+    },
+    {
+      "epoch": 1093.8157894736842,
+      "grad_norm": 1.2598907947540283,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 166260
+    },
+    {
+      "epoch": 1093.8815789473683,
+      "grad_norm": 1.3617310523986816,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 166270
+    },
+    {
+      "epoch": 1093.9473684210527,
+      "grad_norm": 1.6481096744537354,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 166280
+    },
+    {
+      "epoch": 1094.0131578947369,
+      "grad_norm": 1.0241049528121948,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 166290
+    },
+    {
+      "epoch": 1094.078947368421,
+      "grad_norm": 1.2445513010025024,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 166300
+    },
+    {
+      "epoch": 1094.1447368421052,
+      "grad_norm": 1.102005124092102,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 166310
+    },
+    {
+      "epoch": 1094.2105263157894,
+      "grad_norm": 1.0038352012634277,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 166320
+    },
+    {
+      "epoch": 1094.2763157894738,
+      "grad_norm": 1.2613288164138794,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 166330
+    },
+    {
+      "epoch": 1094.342105263158,
+      "grad_norm": 0.9229909181594849,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 166340
+    },
+    {
+      "epoch": 1094.407894736842,
+      "grad_norm": 0.9548853635787964,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 166350
+    },
+    {
+      "epoch": 1094.4736842105262,
+      "grad_norm": 0.8751519322395325,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 166360
+    },
+    {
+      "epoch": 1094.5394736842106,
+      "grad_norm": 0.5805840492248535,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 166370
+    },
+    {
+      "epoch": 1094.6052631578948,
+      "grad_norm": 1.1700973510742188,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 166380
+    },
+    {
+      "epoch": 1094.671052631579,
+      "grad_norm": 0.8679378032684326,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 166390
+    },
+    {
+      "epoch": 1094.7368421052631,
+      "grad_norm": 0.843585193157196,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 166400
+    },
+    {
+      "epoch": 1094.8026315789473,
+      "grad_norm": 1.0492405891418457,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 166410
+    },
+    {
+      "epoch": 1094.8684210526317,
+      "grad_norm": 1.0488659143447876,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 166420
+    },
+    {
+      "epoch": 1094.9342105263158,
+      "grad_norm": 1.3835662603378296,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 166430
+    },
+    {
+      "epoch": 1095.0,
+      "grad_norm": 0.7628530859947205,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 166440
+    },
+    {
+      "epoch": 1095.0657894736842,
+      "grad_norm": 1.0557432174682617,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 166450
+    },
+    {
+      "epoch": 1095.1315789473683,
+      "grad_norm": 1.1266205310821533,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 166460
+    },
+    {
+      "epoch": 1095.1973684210527,
+      "grad_norm": 1.0164369344711304,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 166470
+    },
+    {
+      "epoch": 1095.2631578947369,
+      "grad_norm": 1.1279717683792114,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 166480
+    },
+    {
+      "epoch": 1095.328947368421,
+      "grad_norm": 1.44341242313385,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 166490
+    },
+    {
+      "epoch": 1095.3947368421052,
+      "grad_norm": 1.4984021186828613,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 166500
+    },
+    {
+      "epoch": 1095.4605263157894,
+      "grad_norm": 0.9742573499679565,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 166510
+    },
+    {
+      "epoch": 1095.5263157894738,
+      "grad_norm": 1.1598995923995972,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 166520
+    },
+    {
+      "epoch": 1095.592105263158,
+      "grad_norm": 1.0693007707595825,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 166530
+    },
+    {
+      "epoch": 1095.657894736842,
+      "grad_norm": 1.192436695098877,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 166540
+    },
+    {
+      "epoch": 1095.7236842105262,
+      "grad_norm": 1.156217098236084,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 166550
+    },
+    {
+      "epoch": 1095.7894736842106,
+      "grad_norm": 1.08530592918396,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 166560
+    },
+    {
+      "epoch": 1095.8552631578948,
+      "grad_norm": 1.1809860467910767,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 166570
+    },
+    {
+      "epoch": 1095.921052631579,
+      "grad_norm": 0.9600285887718201,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 166580
+    },
+    {
+      "epoch": 1095.9868421052631,
+      "grad_norm": 1.0501923561096191,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 166590
+    },
+    {
+      "epoch": 1096.0526315789473,
+      "grad_norm": 1.4869877099990845,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 166600
+    },
+    {
+      "epoch": 1096.1184210526317,
+      "grad_norm": 1.0109316110610962,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 166610
+    },
+    {
+      "epoch": 1096.1842105263158,
+      "grad_norm": 1.2201900482177734,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 166620
+    },
+    {
+      "epoch": 1096.25,
+      "grad_norm": 1.0076817274093628,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 166630
+    },
+    {
+      "epoch": 1096.3157894736842,
+      "grad_norm": 0.9542809128761292,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 166640
+    },
+    {
+      "epoch": 1096.3815789473683,
+      "grad_norm": 0.8175939321517944,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 166650
+    },
+    {
+      "epoch": 1096.4473684210527,
+      "grad_norm": 1.0511505603790283,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 166660
+    },
+    {
+      "epoch": 1096.5131578947369,
+      "grad_norm": 0.8790225982666016,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 166670
+    },
+    {
+      "epoch": 1096.578947368421,
+      "grad_norm": 0.8520594835281372,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 166680
+    },
+    {
+      "epoch": 1096.6447368421052,
+      "grad_norm": 0.8566333055496216,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 166690
+    },
+    {
+      "epoch": 1096.7105263157894,
+      "grad_norm": 1.1941945552825928,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 166700
+    },
+    {
+      "epoch": 1096.7763157894738,
+      "grad_norm": 0.8733620047569275,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 166710
+    },
+    {
+      "epoch": 1096.842105263158,
+      "grad_norm": 1.1847484111785889,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 166720
+    },
+    {
+      "epoch": 1096.907894736842,
+      "grad_norm": 0.9977523684501648,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 166730
+    },
+    {
+      "epoch": 1096.9736842105262,
+      "grad_norm": 1.088883638381958,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 166740
+    },
+    {
+      "epoch": 1097.0394736842106,
+      "grad_norm": 1.1181623935699463,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 166750
+    },
+    {
+      "epoch": 1097.1052631578948,
+      "grad_norm": 3.2963168621063232,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 166760
+    },
+    {
+      "epoch": 1097.171052631579,
+      "grad_norm": 1.2029335498809814,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 166770
+    },
+    {
+      "epoch": 1097.2368421052631,
+      "grad_norm": 1.2045027017593384,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 166780
+    },
+    {
+      "epoch": 1097.3026315789473,
+      "grad_norm": 1.1241041421890259,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 166790
+    },
+    {
+      "epoch": 1097.3684210526317,
+      "grad_norm": 1.700153112411499,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 166800
+    },
+    {
+      "epoch": 1097.4342105263158,
+      "grad_norm": 1.5076239109039307,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 166810
+    },
+    {
+      "epoch": 1097.5,
+      "grad_norm": 1.4871948957443237,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 166820
+    },
+    {
+      "epoch": 1097.5657894736842,
+      "grad_norm": 1.3231090307235718,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 166830
+    },
+    {
+      "epoch": 1097.6315789473683,
+      "grad_norm": 1.1733216047286987,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 166840
+    },
+    {
+      "epoch": 1097.6973684210527,
+      "grad_norm": 1.3139346837997437,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 166850
+    },
+    {
+      "epoch": 1097.7631578947369,
+      "grad_norm": 1.0993858575820923,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 166860
+    },
+    {
+      "epoch": 1097.828947368421,
+      "grad_norm": 1.0795350074768066,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 166870
+    },
+    {
+      "epoch": 1097.8947368421052,
+      "grad_norm": 1.0457212924957275,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 166880
+    },
+    {
+      "epoch": 1097.9605263157894,
+      "grad_norm": 0.9968882203102112,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 166890
+    },
+    {
+      "epoch": 1098.0263157894738,
+      "grad_norm": 1.2826480865478516,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 166900
+    },
+    {
+      "epoch": 1098.092105263158,
+      "grad_norm": 1.4935877323150635,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 166910
+    },
+    {
+      "epoch": 1098.157894736842,
+      "grad_norm": 1.1641865968704224,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 166920
+    },
+    {
+      "epoch": 1098.2236842105262,
+      "grad_norm": 1.0109764337539673,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 166930
+    },
+    {
+      "epoch": 1098.2894736842106,
+      "grad_norm": 1.3171255588531494,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 166940
+    },
+    {
+      "epoch": 1098.3552631578948,
+      "grad_norm": 1.3384495973587036,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 166950
+    },
+    {
+      "epoch": 1098.421052631579,
+      "grad_norm": 1.485098123550415,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 166960
+    },
+    {
+      "epoch": 1098.4868421052631,
+      "grad_norm": 1.4696592092514038,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 166970
+    },
+    {
+      "epoch": 1098.5526315789473,
+      "grad_norm": 0.9190577864646912,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 166980
+    },
+    {
+      "epoch": 1098.6184210526317,
+      "grad_norm": 1.0120787620544434,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 166990
+    },
+    {
+      "epoch": 1098.6842105263158,
+      "grad_norm": 1.3007423877716064,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 167000
+    },
+    {
+      "epoch": 1098.75,
+      "grad_norm": 1.21133291721344,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 167010
+    },
+    {
+      "epoch": 1098.8157894736842,
+      "grad_norm": 0.8098885416984558,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 167020
+    },
+    {
+      "epoch": 1098.8815789473683,
+      "grad_norm": 1.0924862623214722,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 167030
+    },
+    {
+      "epoch": 1098.9473684210527,
+      "grad_norm": 1.058971643447876,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 167040
+    },
+    {
+      "epoch": 1099.0131578947369,
+      "grad_norm": 1.1815778017044067,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 167050
+    },
+    {
+      "epoch": 1099.078947368421,
+      "grad_norm": 1.1275389194488525,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 167060
+    },
+    {
+      "epoch": 1099.1447368421052,
+      "grad_norm": 1.2118959426879883,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 167070
+    },
+    {
+      "epoch": 1099.2105263157894,
+      "grad_norm": 1.3578826189041138,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 167080
+    },
+    {
+      "epoch": 1099.2763157894738,
+      "grad_norm": 1.1233108043670654,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 167090
+    },
+    {
+      "epoch": 1099.342105263158,
+      "grad_norm": 0.8814030885696411,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 167100
+    },
+    {
+      "epoch": 1099.407894736842,
+      "grad_norm": 1.0305200815200806,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 167110
+    },
+    {
+      "epoch": 1099.4736842105262,
+      "grad_norm": 1.1122325658798218,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 167120
+    },
+    {
+      "epoch": 1099.5394736842106,
+      "grad_norm": 0.9248008728027344,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 167130
+    },
+    {
+      "epoch": 1099.6052631578948,
+      "grad_norm": 0.889100968837738,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 167140
+    },
+    {
+      "epoch": 1099.671052631579,
+      "grad_norm": 1.2280583381652832,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 167150
+    },
+    {
+      "epoch": 1099.7368421052631,
+      "grad_norm": 1.2300212383270264,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 167160
+    },
+    {
+      "epoch": 1099.8026315789473,
+      "grad_norm": 1.211428165435791,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 167170
+    },
+    {
+      "epoch": 1099.8684210526317,
+      "grad_norm": 1.1295406818389893,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 167180
+    },
+    {
+      "epoch": 1099.9342105263158,
+      "grad_norm": 1.1120741367340088,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 167190
+    },
+    {
+      "epoch": 1100.0,
+      "grad_norm": 0.8799479603767395,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 167200
+    },
+    {
+      "epoch": 1100.0657894736842,
+      "grad_norm": 1.0118703842163086,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 167210
+    },
+    {
+      "epoch": 1100.1315789473683,
+      "grad_norm": 1.097957968711853,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 167220
+    },
+    {
+      "epoch": 1100.1973684210527,
+      "grad_norm": 0.8075319528579712,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 167230
+    },
+    {
+      "epoch": 1100.2631578947369,
+      "grad_norm": 1.2604179382324219,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 167240
+    },
+    {
+      "epoch": 1100.328947368421,
+      "grad_norm": 1.090760350227356,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 167250
+    },
+    {
+      "epoch": 1100.3947368421052,
+      "grad_norm": 0.9950432181358337,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 167260
+    },
+    {
+      "epoch": 1100.4605263157894,
+      "grad_norm": 0.7708602547645569,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 167270
+    },
+    {
+      "epoch": 1100.5263157894738,
+      "grad_norm": 0.8636929392814636,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 167280
+    },
+    {
+      "epoch": 1100.592105263158,
+      "grad_norm": 0.7915971279144287,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 167290
+    },
+    {
+      "epoch": 1100.657894736842,
+      "grad_norm": 0.8626055121421814,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 167300
+    },
+    {
+      "epoch": 1100.7236842105262,
+      "grad_norm": 0.6943460702896118,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 167310
+    },
+    {
+      "epoch": 1100.7894736842106,
+      "grad_norm": 1.0812398195266724,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 167320
+    },
+    {
+      "epoch": 1100.8552631578948,
+      "grad_norm": 0.906474232673645,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 167330
+    },
+    {
+      "epoch": 1100.921052631579,
+      "grad_norm": 0.9703337550163269,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 167340
+    },
+    {
+      "epoch": 1100.9868421052631,
+      "grad_norm": 0.8789888024330139,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 167350
+    },
+    {
+      "epoch": 1101.0526315789473,
+      "grad_norm": 0.9876759648323059,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 167360
+    },
+    {
+      "epoch": 1101.1184210526317,
+      "grad_norm": 1.0288928747177124,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 167370
+    },
+    {
+      "epoch": 1101.1842105263158,
+      "grad_norm": 0.9884411692619324,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 167380
+    },
+    {
+      "epoch": 1101.25,
+      "grad_norm": 1.231924057006836,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 167390
+    },
+    {
+      "epoch": 1101.3157894736842,
+      "grad_norm": 1.2032873630523682,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 167400
+    },
+    {
+      "epoch": 1101.3815789473683,
+      "grad_norm": 1.1129034757614136,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 167410
+    },
+    {
+      "epoch": 1101.4473684210527,
+      "grad_norm": 1.0674737691879272,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 167420
+    },
+    {
+      "epoch": 1101.5131578947369,
+      "grad_norm": 0.9787297248840332,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 167430
+    },
+    {
+      "epoch": 1101.578947368421,
+      "grad_norm": 0.8333871364593506,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 167440
+    },
+    {
+      "epoch": 1101.6447368421052,
+      "grad_norm": 1.1701922416687012,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 167450
+    },
+    {
+      "epoch": 1101.7105263157894,
+      "grad_norm": 1.1655021905899048,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 167460
+    },
+    {
+      "epoch": 1101.7763157894738,
+      "grad_norm": 1.2170398235321045,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 167470
+    },
+    {
+      "epoch": 1101.842105263158,
+      "grad_norm": 1.1292792558670044,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 167480
+    },
+    {
+      "epoch": 1101.907894736842,
+      "grad_norm": 1.0343544483184814,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 167490
+    },
+    {
+      "epoch": 1101.9736842105262,
+      "grad_norm": 1.292896032333374,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 167500
+    },
+    {
+      "epoch": 1102.0394736842106,
+      "grad_norm": 1.3598896265029907,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 167510
+    },
+    {
+      "epoch": 1102.1052631578948,
+      "grad_norm": 1.1134599447250366,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 167520
+    },
+    {
+      "epoch": 1102.171052631579,
+      "grad_norm": 1.4782142639160156,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 167530
+    },
+    {
+      "epoch": 1102.2368421052631,
+      "grad_norm": 1.0224602222442627,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 167540
+    },
+    {
+      "epoch": 1102.3026315789473,
+      "grad_norm": 1.1568275690078735,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 167550
+    },
+    {
+      "epoch": 1102.3684210526317,
+      "grad_norm": 0.9794890284538269,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 167560
+    },
+    {
+      "epoch": 1102.4342105263158,
+      "grad_norm": 0.9508534073829651,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 167570
+    },
+    {
+      "epoch": 1102.5,
+      "grad_norm": 1.202999234199524,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 167580
+    },
+    {
+      "epoch": 1102.5657894736842,
+      "grad_norm": 1.2646609544754028,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 167590
+    },
+    {
+      "epoch": 1102.6315789473683,
+      "grad_norm": 1.1292835474014282,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 167600
+    },
+    {
+      "epoch": 1102.6973684210527,
+      "grad_norm": 0.8157707452774048,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 167610
+    },
+    {
+      "epoch": 1102.7631578947369,
+      "grad_norm": 0.9448849558830261,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 167620
+    },
+    {
+      "epoch": 1102.828947368421,
+      "grad_norm": 0.878303587436676,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 167630
+    },
+    {
+      "epoch": 1102.8947368421052,
+      "grad_norm": 1.097684383392334,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 167640
+    },
+    {
+      "epoch": 1102.9605263157894,
+      "grad_norm": 1.0110867023468018,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 167650
+    },
+    {
+      "epoch": 1103.0263157894738,
+      "grad_norm": 1.1155486106872559,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 167660
+    },
+    {
+      "epoch": 1103.092105263158,
+      "grad_norm": 1.0845309495925903,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 167670
+    },
+    {
+      "epoch": 1103.157894736842,
+      "grad_norm": 0.8324429392814636,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 167680
+    },
+    {
+      "epoch": 1103.2236842105262,
+      "grad_norm": 0.8664148449897766,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 167690
+    },
+    {
+      "epoch": 1103.2894736842106,
+      "grad_norm": 0.7451772689819336,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 167700
+    },
+    {
+      "epoch": 1103.3552631578948,
+      "grad_norm": 0.9398016929626465,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 167710
+    },
+    {
+      "epoch": 1103.421052631579,
+      "grad_norm": 1.0360288619995117,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 167720
+    },
+    {
+      "epoch": 1103.4868421052631,
+      "grad_norm": 1.274431824684143,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 167730
+    },
+    {
+      "epoch": 1103.5526315789473,
+      "grad_norm": 0.5587978363037109,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 167740
+    },
+    {
+      "epoch": 1103.6184210526317,
+      "grad_norm": 0.7810954451560974,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 167750
+    },
+    {
+      "epoch": 1103.6842105263158,
+      "grad_norm": 0.8118604421615601,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 167760
+    },
+    {
+      "epoch": 1103.75,
+      "grad_norm": 0.6753564476966858,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 167770
+    },
+    {
+      "epoch": 1103.8157894736842,
+      "grad_norm": 0.609535276889801,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 167780
+    },
+    {
+      "epoch": 1103.8815789473683,
+      "grad_norm": 0.8553572297096252,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 167790
+    },
+    {
+      "epoch": 1103.9473684210527,
+      "grad_norm": 1.0933605432510376,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 167800
+    },
+    {
+      "epoch": 1104.0131578947369,
+      "grad_norm": 1.4074715375900269,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 167810
+    },
+    {
+      "epoch": 1104.078947368421,
+      "grad_norm": 0.9313386678695679,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 167820
+    },
+    {
+      "epoch": 1104.1447368421052,
+      "grad_norm": 1.2255233526229858,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 167830
+    },
+    {
+      "epoch": 1104.2105263157894,
+      "grad_norm": 0.958168089389801,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 167840
+    },
+    {
+      "epoch": 1104.2763157894738,
+      "grad_norm": 0.7276482582092285,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 167850
+    },
+    {
+      "epoch": 1104.342105263158,
+      "grad_norm": 1.2092912197113037,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 167860
+    },
+    {
+      "epoch": 1104.407894736842,
+      "grad_norm": 0.9933399558067322,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 167870
+    },
+    {
+      "epoch": 1104.4736842105262,
+      "grad_norm": 0.8450823426246643,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 167880
+    },
+    {
+      "epoch": 1104.5394736842106,
+      "grad_norm": 1.043594479560852,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 167890
+    },
+    {
+      "epoch": 1104.6052631578948,
+      "grad_norm": 1.107570767402649,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 167900
+    },
+    {
+      "epoch": 1104.671052631579,
+      "grad_norm": 1.1680465936660767,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 167910
+    },
+    {
+      "epoch": 1104.7368421052631,
+      "grad_norm": 1.4395219087600708,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 167920
+    },
+    {
+      "epoch": 1104.8026315789473,
+      "grad_norm": 1.0182297229766846,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 167930
+    },
+    {
+      "epoch": 1104.8684210526317,
+      "grad_norm": 0.7282902598381042,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 167940
+    },
+    {
+      "epoch": 1104.9342105263158,
+      "grad_norm": 1.1767057180404663,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 167950
+    },
+    {
+      "epoch": 1105.0,
+      "grad_norm": 0.9892759323120117,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 167960
+    },
+    {
+      "epoch": 1105.0657894736842,
+      "grad_norm": 0.8956315517425537,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 167970
+    },
+    {
+      "epoch": 1105.1315789473683,
+      "grad_norm": 0.707263708114624,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 167980
+    },
+    {
+      "epoch": 1105.1973684210527,
+      "grad_norm": 0.895297110080719,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 167990
+    },
+    {
+      "epoch": 1105.2631578947369,
+      "grad_norm": 0.7965859174728394,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 168000
+    },
+    {
+      "epoch": 1105.328947368421,
+      "grad_norm": 1.3010401725769043,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 168010
+    },
+    {
+      "epoch": 1105.3947368421052,
+      "grad_norm": 1.0735414028167725,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 168020
+    },
+    {
+      "epoch": 1105.4605263157894,
+      "grad_norm": 1.0381656885147095,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 168030
+    },
+    {
+      "epoch": 1105.5263157894738,
+      "grad_norm": 0.8729901909828186,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 168040
+    },
+    {
+      "epoch": 1105.592105263158,
+      "grad_norm": 0.9868913888931274,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 168050
+    },
+    {
+      "epoch": 1105.657894736842,
+      "grad_norm": 1.1911859512329102,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 168060
+    },
+    {
+      "epoch": 1105.7236842105262,
+      "grad_norm": 1.3197442293167114,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 168070
+    },
+    {
+      "epoch": 1105.7894736842106,
+      "grad_norm": 1.240401029586792,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 168080
+    },
+    {
+      "epoch": 1105.8552631578948,
+      "grad_norm": 0.9005335569381714,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 168090
+    },
+    {
+      "epoch": 1105.921052631579,
+      "grad_norm": 0.9704708456993103,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 168100
+    },
+    {
+      "epoch": 1105.9868421052631,
+      "grad_norm": 0.759002149105072,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 168110
+    },
+    {
+      "epoch": 1106.0526315789473,
+      "grad_norm": 0.8771544694900513,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 168120
+    },
+    {
+      "epoch": 1106.1184210526317,
+      "grad_norm": 1.3510820865631104,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 168130
+    },
+    {
+      "epoch": 1106.1842105263158,
+      "grad_norm": 1.1021738052368164,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 168140
+    },
+    {
+      "epoch": 1106.25,
+      "grad_norm": 1.1089617013931274,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 168150
+    },
+    {
+      "epoch": 1106.3157894736842,
+      "grad_norm": 1.5470319986343384,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 168160
+    },
+    {
+      "epoch": 1106.3815789473683,
+      "grad_norm": 0.8695709109306335,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 168170
+    },
+    {
+      "epoch": 1106.4473684210527,
+      "grad_norm": 1.176269769668579,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 168180
+    },
+    {
+      "epoch": 1106.5131578947369,
+      "grad_norm": 1.0304797887802124,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 168190
+    },
+    {
+      "epoch": 1106.578947368421,
+      "grad_norm": 1.2818888425827026,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 168200
+    },
+    {
+      "epoch": 1106.6447368421052,
+      "grad_norm": 1.1938977241516113,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 168210
+    },
+    {
+      "epoch": 1106.7105263157894,
+      "grad_norm": 0.8075238466262817,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 168220
+    },
+    {
+      "epoch": 1106.7763157894738,
+      "grad_norm": 1.112987756729126,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 168230
+    },
+    {
+      "epoch": 1106.842105263158,
+      "grad_norm": 1.1107630729675293,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 168240
+    },
+    {
+      "epoch": 1106.907894736842,
+      "grad_norm": 0.9863311052322388,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 168250
+    },
+    {
+      "epoch": 1106.9736842105262,
+      "grad_norm": 1.1156419515609741,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 168260
+    },
+    {
+      "epoch": 1107.0394736842106,
+      "grad_norm": 1.1970049142837524,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 168270
+    },
+    {
+      "epoch": 1107.1052631578948,
+      "grad_norm": 1.1724064350128174,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 168280
+    },
+    {
+      "epoch": 1107.171052631579,
+      "grad_norm": 1.0076320171356201,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 168290
+    },
+    {
+      "epoch": 1107.2368421052631,
+      "grad_norm": 1.0176210403442383,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 168300
+    },
+    {
+      "epoch": 1107.3026315789473,
+      "grad_norm": 1.0968610048294067,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 168310
+    },
+    {
+      "epoch": 1107.3684210526317,
+      "grad_norm": 1.47584867477417,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 168320
+    },
+    {
+      "epoch": 1107.4342105263158,
+      "grad_norm": 1.49242103099823,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 168330
+    },
+    {
+      "epoch": 1107.5,
+      "grad_norm": 1.3437186479568481,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 168340
+    },
+    {
+      "epoch": 1107.5657894736842,
+      "grad_norm": 1.1982983350753784,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 168350
+    },
+    {
+      "epoch": 1107.6315789473683,
+      "grad_norm": 1.1239218711853027,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 168360
+    },
+    {
+      "epoch": 1107.6973684210527,
+      "grad_norm": 0.9983989000320435,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 168370
+    },
+    {
+      "epoch": 1107.7631578947369,
+      "grad_norm": 1.2701680660247803,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 168380
+    },
+    {
+      "epoch": 1107.828947368421,
+      "grad_norm": 1.07079017162323,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 168390
+    },
+    {
+      "epoch": 1107.8947368421052,
+      "grad_norm": 1.092276930809021,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 168400
+    },
+    {
+      "epoch": 1107.9605263157894,
+      "grad_norm": 0.7285662293434143,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 168410
+    },
+    {
+      "epoch": 1108.0263157894738,
+      "grad_norm": 1.0730990171432495,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 168420
+    },
+    {
+      "epoch": 1108.092105263158,
+      "grad_norm": 1.0767719745635986,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 168430
+    },
+    {
+      "epoch": 1108.157894736842,
+      "grad_norm": 1.1680963039398193,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 168440
+    },
+    {
+      "epoch": 1108.2236842105262,
+      "grad_norm": 1.048923373222351,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 168450
+    },
+    {
+      "epoch": 1108.2894736842106,
+      "grad_norm": 0.9691452980041504,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 168460
+    },
+    {
+      "epoch": 1108.3552631578948,
+      "grad_norm": 0.694308340549469,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 168470
+    },
+    {
+      "epoch": 1108.421052631579,
+      "grad_norm": 0.8766660094261169,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 168480
+    },
+    {
+      "epoch": 1108.4868421052631,
+      "grad_norm": 0.8163343071937561,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 168490
+    },
+    {
+      "epoch": 1108.5526315789473,
+      "grad_norm": 0.7243715524673462,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 168500
+    },
+    {
+      "epoch": 1108.6184210526317,
+      "grad_norm": 1.2673771381378174,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 168510
+    },
+    {
+      "epoch": 1108.6842105263158,
+      "grad_norm": 0.9088281989097595,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 168520
+    },
+    {
+      "epoch": 1108.75,
+      "grad_norm": 1.3097550868988037,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 168530
+    },
+    {
+      "epoch": 1108.8157894736842,
+      "grad_norm": 1.2778091430664062,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 168540
+    },
+    {
+      "epoch": 1108.8815789473683,
+      "grad_norm": 1.2676957845687866,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 168550
+    },
+    {
+      "epoch": 1108.9473684210527,
+      "grad_norm": 1.185399055480957,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 168560
+    },
+    {
+      "epoch": 1109.0131578947369,
+      "grad_norm": 0.9912101626396179,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 168570
+    },
+    {
+      "epoch": 1109.078947368421,
+      "grad_norm": 1.0790884494781494,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 168580
+    },
+    {
+      "epoch": 1109.1447368421052,
+      "grad_norm": 0.9652689695358276,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 168590
+    },
+    {
+      "epoch": 1109.2105263157894,
+      "grad_norm": 1.2606117725372314,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 168600
+    },
+    {
+      "epoch": 1109.2763157894738,
+      "grad_norm": 1.2370003461837769,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 168610
+    },
+    {
+      "epoch": 1109.342105263158,
+      "grad_norm": 0.9363791346549988,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 168620
+    },
+    {
+      "epoch": 1109.407894736842,
+      "grad_norm": 1.1172534227371216,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 168630
+    },
+    {
+      "epoch": 1109.4736842105262,
+      "grad_norm": 1.1383373737335205,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 168640
+    },
+    {
+      "epoch": 1109.5394736842106,
+      "grad_norm": 1.16228449344635,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 168650
+    },
+    {
+      "epoch": 1109.6052631578948,
+      "grad_norm": 1.1701244115829468,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 168660
+    },
+    {
+      "epoch": 1109.671052631579,
+      "grad_norm": 1.5518008470535278,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 168670
+    },
+    {
+      "epoch": 1109.7368421052631,
+      "grad_norm": 1.4797471761703491,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 168680
+    },
+    {
+      "epoch": 1109.8026315789473,
+      "grad_norm": 1.0961973667144775,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 168690
+    },
+    {
+      "epoch": 1109.8684210526317,
+      "grad_norm": 1.1600146293640137,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 168700
+    },
+    {
+      "epoch": 1109.9342105263158,
+      "grad_norm": 1.3564647436141968,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 168710
+    },
+    {
+      "epoch": 1110.0,
+      "grad_norm": 1.7049609422683716,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 168720
+    },
+    {
+      "epoch": 1110.0657894736842,
+      "grad_norm": 1.5665743350982666,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 168730
+    },
+    {
+      "epoch": 1110.1315789473683,
+      "grad_norm": 1.5646759271621704,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 168740
+    },
+    {
+      "epoch": 1110.1973684210527,
+      "grad_norm": 1.1867520809173584,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 168750
+    },
+    {
+      "epoch": 1110.2631578947369,
+      "grad_norm": 1.6456133127212524,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 168760
+    },
+    {
+      "epoch": 1110.328947368421,
+      "grad_norm": 1.2314730882644653,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 168770
+    },
+    {
+      "epoch": 1110.3947368421052,
+      "grad_norm": 1.0836608409881592,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 168780
+    },
+    {
+      "epoch": 1110.4605263157894,
+      "grad_norm": 0.9682186245918274,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 168790
+    },
+    {
+      "epoch": 1110.5263157894738,
+      "grad_norm": 0.8563161492347717,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 168800
+    },
+    {
+      "epoch": 1110.592105263158,
+      "grad_norm": 0.9814671277999878,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 168810
+    },
+    {
+      "epoch": 1110.657894736842,
+      "grad_norm": 0.7647870182991028,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 168820
+    },
+    {
+      "epoch": 1110.7236842105262,
+      "grad_norm": 1.3109147548675537,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 168830
+    },
+    {
+      "epoch": 1110.7894736842106,
+      "grad_norm": 1.0601048469543457,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 168840
+    },
+    {
+      "epoch": 1110.8552631578948,
+      "grad_norm": 1.160309910774231,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 168850
+    },
+    {
+      "epoch": 1110.921052631579,
+      "grad_norm": 1.0141721963882446,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 168860
+    },
+    {
+      "epoch": 1110.9868421052631,
+      "grad_norm": 0.9558742046356201,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 168870
+    },
+    {
+      "epoch": 1111.0526315789473,
+      "grad_norm": 1.0279165506362915,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 168880
+    },
+    {
+      "epoch": 1111.1184210526317,
+      "grad_norm": 1.179625153541565,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 168890
+    },
+    {
+      "epoch": 1111.1842105263158,
+      "grad_norm": 1.070167064666748,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 168900
+    },
+    {
+      "epoch": 1111.25,
+      "grad_norm": 1.2725509405136108,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 168910
+    },
+    {
+      "epoch": 1111.3157894736842,
+      "grad_norm": 0.9104771018028259,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 168920
+    },
+    {
+      "epoch": 1111.3815789473683,
+      "grad_norm": 0.937410831451416,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 168930
+    },
+    {
+      "epoch": 1111.4473684210527,
+      "grad_norm": 1.037760615348816,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 168940
+    },
+    {
+      "epoch": 1111.5131578947369,
+      "grad_norm": 1.1063820123672485,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 168950
+    },
+    {
+      "epoch": 1111.578947368421,
+      "grad_norm": 1.0770851373672485,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 168960
+    },
+    {
+      "epoch": 1111.6447368421052,
+      "grad_norm": 1.1745645999908447,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 168970
+    },
+    {
+      "epoch": 1111.7105263157894,
+      "grad_norm": 0.936163604259491,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 168980
+    },
+    {
+      "epoch": 1111.7763157894738,
+      "grad_norm": 1.031320333480835,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 168990
+    },
+    {
+      "epoch": 1111.842105263158,
+      "grad_norm": 0.6643326878547668,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 169000
+    },
+    {
+      "epoch": 1111.907894736842,
+      "grad_norm": 0.8796061873435974,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 169010
+    },
+    {
+      "epoch": 1111.9736842105262,
+      "grad_norm": 0.9710099697113037,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 169020
+    },
+    {
+      "epoch": 1112.0394736842106,
+      "grad_norm": 1.134057879447937,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 169030
+    },
+    {
+      "epoch": 1112.1052631578948,
+      "grad_norm": 0.7296460270881653,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 169040
+    },
+    {
+      "epoch": 1112.171052631579,
+      "grad_norm": 0.7189831733703613,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 169050
+    },
+    {
+      "epoch": 1112.2368421052631,
+      "grad_norm": 0.8894656896591187,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 169060
+    },
+    {
+      "epoch": 1112.3026315789473,
+      "grad_norm": 1.1131048202514648,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 169070
+    },
+    {
+      "epoch": 1112.3684210526317,
+      "grad_norm": 1.5262320041656494,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 169080
+    },
+    {
+      "epoch": 1112.4342105263158,
+      "grad_norm": 1.2792640924453735,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 169090
+    },
+    {
+      "epoch": 1112.5,
+      "grad_norm": 1.091755747795105,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 169100
+    },
+    {
+      "epoch": 1112.5657894736842,
+      "grad_norm": 1.1923725605010986,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 169110
+    },
+    {
+      "epoch": 1112.6315789473683,
+      "grad_norm": 1.2157421112060547,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 169120
+    },
+    {
+      "epoch": 1112.6973684210527,
+      "grad_norm": 1.0996853113174438,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 169130
+    },
+    {
+      "epoch": 1112.7631578947369,
+      "grad_norm": 1.3527673482894897,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 169140
+    },
+    {
+      "epoch": 1112.828947368421,
+      "grad_norm": 1.1760993003845215,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 169150
+    },
+    {
+      "epoch": 1112.8947368421052,
+      "grad_norm": 1.1394774913787842,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 169160
+    },
+    {
+      "epoch": 1112.9605263157894,
+      "grad_norm": 1.1786078214645386,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 169170
+    },
+    {
+      "epoch": 1113.0263157894738,
+      "grad_norm": 2.485809326171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 169180
+    },
+    {
+      "epoch": 1113.092105263158,
+      "grad_norm": 1.372261881828308,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 169190
+    },
+    {
+      "epoch": 1113.157894736842,
+      "grad_norm": 1.125187873840332,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 169200
+    },
+    {
+      "epoch": 1113.2236842105262,
+      "grad_norm": 1.4014942646026611,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 169210
+    },
+    {
+      "epoch": 1113.2894736842106,
+      "grad_norm": 1.0641945600509644,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 169220
+    },
+    {
+      "epoch": 1113.3552631578948,
+      "grad_norm": 1.023398756980896,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 169230
+    },
+    {
+      "epoch": 1113.421052631579,
+      "grad_norm": 0.8850655555725098,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 169240
+    },
+    {
+      "epoch": 1113.4868421052631,
+      "grad_norm": 1.0332579612731934,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 169250
+    },
+    {
+      "epoch": 1113.5526315789473,
+      "grad_norm": 1.3429882526397705,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 169260
+    },
+    {
+      "epoch": 1113.6184210526317,
+      "grad_norm": 1.1904481649398804,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 169270
+    },
+    {
+      "epoch": 1113.6842105263158,
+      "grad_norm": 0.8941307067871094,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 169280
+    },
+    {
+      "epoch": 1113.75,
+      "grad_norm": 1.1564146280288696,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 169290
+    },
+    {
+      "epoch": 1113.8157894736842,
+      "grad_norm": 0.770451545715332,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 169300
+    },
+    {
+      "epoch": 1113.8815789473683,
+      "grad_norm": 0.9108574986457825,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 169310
+    },
+    {
+      "epoch": 1113.9473684210527,
+      "grad_norm": 0.8220398426055908,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 169320
+    },
+    {
+      "epoch": 1114.0131578947369,
+      "grad_norm": 1.039478063583374,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 169330
+    },
+    {
+      "epoch": 1114.078947368421,
+      "grad_norm": 1.078933835029602,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 169340
+    },
+    {
+      "epoch": 1114.1447368421052,
+      "grad_norm": 1.0920974016189575,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 169350
+    },
+    {
+      "epoch": 1114.2105263157894,
+      "grad_norm": 0.9281700253486633,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 169360
+    },
+    {
+      "epoch": 1114.2763157894738,
+      "grad_norm": 1.1259106397628784,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 169370
+    },
+    {
+      "epoch": 1114.342105263158,
+      "grad_norm": 0.9117035269737244,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 169380
+    },
+    {
+      "epoch": 1114.407894736842,
+      "grad_norm": 1.178030252456665,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 169390
+    },
+    {
+      "epoch": 1114.4736842105262,
+      "grad_norm": 1.0540664196014404,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 169400
+    },
+    {
+      "epoch": 1114.5394736842106,
+      "grad_norm": 1.3689993619918823,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 169410
+    },
+    {
+      "epoch": 1114.6052631578948,
+      "grad_norm": 1.2431306838989258,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 169420
+    },
+    {
+      "epoch": 1114.671052631579,
+      "grad_norm": 1.0379416942596436,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 169430
+    },
+    {
+      "epoch": 1114.7368421052631,
+      "grad_norm": 1.2257938385009766,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 169440
+    },
+    {
+      "epoch": 1114.8026315789473,
+      "grad_norm": 0.9797086119651794,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 169450
+    },
+    {
+      "epoch": 1114.8684210526317,
+      "grad_norm": 1.3930553197860718,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 169460
+    },
+    {
+      "epoch": 1114.9342105263158,
+      "grad_norm": 1.1438504457473755,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 169470
+    },
+    {
+      "epoch": 1115.0,
+      "grad_norm": 1.0164341926574707,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 169480
+    },
+    {
+      "epoch": 1115.0657894736842,
+      "grad_norm": 0.706180989742279,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 169490
+    },
+    {
+      "epoch": 1115.1315789473683,
+      "grad_norm": 1.0323935747146606,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 169500
+    },
+    {
+      "epoch": 1115.1973684210527,
+      "grad_norm": 0.8739573359489441,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 169510
+    },
+    {
+      "epoch": 1115.2631578947369,
+      "grad_norm": 1.4711079597473145,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 169520
+    },
+    {
+      "epoch": 1115.328947368421,
+      "grad_norm": 1.3129510879516602,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 169530
+    },
+    {
+      "epoch": 1115.3947368421052,
+      "grad_norm": 1.3260138034820557,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 169540
+    },
+    {
+      "epoch": 1115.4605263157894,
+      "grad_norm": 1.2280200719833374,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 169550
+    },
+    {
+      "epoch": 1115.5263157894738,
+      "grad_norm": 1.0579249858856201,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 169560
+    },
+    {
+      "epoch": 1115.592105263158,
+      "grad_norm": 1.5779436826705933,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 169570
+    },
+    {
+      "epoch": 1115.657894736842,
+      "grad_norm": 1.2979265451431274,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 169580
+    },
+    {
+      "epoch": 1115.7236842105262,
+      "grad_norm": 1.2990245819091797,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 169590
+    },
+    {
+      "epoch": 1115.7894736842106,
+      "grad_norm": 1.1901843547821045,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 169600
+    },
+    {
+      "epoch": 1115.8552631578948,
+      "grad_norm": 1.2466164827346802,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 169610
+    },
+    {
+      "epoch": 1115.921052631579,
+      "grad_norm": 0.8972548842430115,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 169620
+    },
+    {
+      "epoch": 1115.9868421052631,
+      "grad_norm": 0.7799301743507385,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 169630
+    },
+    {
+      "epoch": 1116.0526315789473,
+      "grad_norm": 0.7389102578163147,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 169640
+    },
+    {
+      "epoch": 1116.1184210526317,
+      "grad_norm": 1.1879955530166626,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 169650
+    },
+    {
+      "epoch": 1116.1842105263158,
+      "grad_norm": 1.0571030378341675,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 169660
+    },
+    {
+      "epoch": 1116.25,
+      "grad_norm": 1.1208561658859253,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 169670
+    },
+    {
+      "epoch": 1116.3157894736842,
+      "grad_norm": 1.211946964263916,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 169680
+    },
+    {
+      "epoch": 1116.3815789473683,
+      "grad_norm": 1.0168685913085938,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 169690
+    },
+    {
+      "epoch": 1116.4473684210527,
+      "grad_norm": 1.4225287437438965,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 169700
+    },
+    {
+      "epoch": 1116.5131578947369,
+      "grad_norm": 0.7239480018615723,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 169710
+    },
+    {
+      "epoch": 1116.578947368421,
+      "grad_norm": 1.2720476388931274,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 169720
+    },
+    {
+      "epoch": 1116.6447368421052,
+      "grad_norm": 1.250244140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 169730
+    },
+    {
+      "epoch": 1116.7105263157894,
+      "grad_norm": 1.0670952796936035,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 169740
+    },
+    {
+      "epoch": 1116.7763157894738,
+      "grad_norm": 1.2624378204345703,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 169750
+    },
+    {
+      "epoch": 1116.842105263158,
+      "grad_norm": 1.0680029392242432,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 169760
+    },
+    {
+      "epoch": 1116.907894736842,
+      "grad_norm": 0.9182707071304321,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 169770
+    },
+    {
+      "epoch": 1116.9736842105262,
+      "grad_norm": 1.0356745719909668,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 169780
+    },
+    {
+      "epoch": 1117.0394736842106,
+      "grad_norm": 1.2450051307678223,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 169790
+    },
+    {
+      "epoch": 1117.1052631578948,
+      "grad_norm": 1.1483206748962402,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 169800
+    },
+    {
+      "epoch": 1117.171052631579,
+      "grad_norm": 1.0393849611282349,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 169810
+    },
+    {
+      "epoch": 1117.2368421052631,
+      "grad_norm": 1.1354923248291016,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 169820
+    },
+    {
+      "epoch": 1117.3026315789473,
+      "grad_norm": 1.2320741415023804,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 169830
+    },
+    {
+      "epoch": 1117.3684210526317,
+      "grad_norm": 1.2553433179855347,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 169840
+    },
+    {
+      "epoch": 1117.4342105263158,
+      "grad_norm": 1.142922043800354,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 169850
+    },
+    {
+      "epoch": 1117.5,
+      "grad_norm": 1.2326382398605347,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 169860
+    },
+    {
+      "epoch": 1117.5657894736842,
+      "grad_norm": 1.5461610555648804,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 169870
+    },
+    {
+      "epoch": 1117.6315789473683,
+      "grad_norm": 0.9742788076400757,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 169880
+    },
+    {
+      "epoch": 1117.6973684210527,
+      "grad_norm": 0.898607075214386,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 169890
+    },
+    {
+      "epoch": 1117.7631578947369,
+      "grad_norm": 0.6093039512634277,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 169900
+    },
+    {
+      "epoch": 1117.828947368421,
+      "grad_norm": 1.1345704793930054,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 169910
+    },
+    {
+      "epoch": 1117.8947368421052,
+      "grad_norm": 0.851069450378418,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 169920
+    },
+    {
+      "epoch": 1117.9605263157894,
+      "grad_norm": 0.9209760427474976,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 169930
+    },
+    {
+      "epoch": 1118.0263157894738,
+      "grad_norm": 0.86602783203125,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 169940
+    },
+    {
+      "epoch": 1118.092105263158,
+      "grad_norm": 0.698943555355072,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 169950
+    },
+    {
+      "epoch": 1118.157894736842,
+      "grad_norm": 0.8509097695350647,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 169960
+    },
+    {
+      "epoch": 1118.2236842105262,
+      "grad_norm": 0.893696665763855,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 169970
+    },
+    {
+      "epoch": 1118.2894736842106,
+      "grad_norm": 0.7406723499298096,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 169980
+    },
+    {
+      "epoch": 1118.3552631578948,
+      "grad_norm": 1.147310495376587,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 169990
+    },
+    {
+      "epoch": 1118.421052631579,
+      "grad_norm": 0.8037264347076416,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 170000
+    },
+    {
+      "epoch": 1118.4868421052631,
+      "grad_norm": 0.5971798300743103,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 170010
+    },
+    {
+      "epoch": 1118.5526315789473,
+      "grad_norm": 0.7508800625801086,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 170020
+    },
+    {
+      "epoch": 1118.6184210526317,
+      "grad_norm": 1.1436516046524048,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 170030
+    },
+    {
+      "epoch": 1118.6842105263158,
+      "grad_norm": 1.146233320236206,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 170040
+    },
+    {
+      "epoch": 1118.75,
+      "grad_norm": 1.1808003187179565,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 170050
+    },
+    {
+      "epoch": 1118.8157894736842,
+      "grad_norm": 0.8754172325134277,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 170060
+    },
+    {
+      "epoch": 1118.8815789473683,
+      "grad_norm": 1.1422688961029053,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 170070
+    },
+    {
+      "epoch": 1118.9473684210527,
+      "grad_norm": 0.8434144854545593,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 170080
+    },
+    {
+      "epoch": 1119.0131578947369,
+      "grad_norm": 1.1790432929992676,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 170090
+    },
+    {
+      "epoch": 1119.078947368421,
+      "grad_norm": 0.9969106316566467,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 170100
+    },
+    {
+      "epoch": 1119.1447368421052,
+      "grad_norm": 0.6672012209892273,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 170110
+    },
+    {
+      "epoch": 1119.2105263157894,
+      "grad_norm": 1.309625267982483,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 170120
+    },
+    {
+      "epoch": 1119.2763157894738,
+      "grad_norm": 1.0300084352493286,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 170130
+    },
+    {
+      "epoch": 1119.342105263158,
+      "grad_norm": 0.9487712979316711,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 170140
+    },
+    {
+      "epoch": 1119.407894736842,
+      "grad_norm": 1.0787352323532104,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 170150
+    },
+    {
+      "epoch": 1119.4736842105262,
+      "grad_norm": 1.1711540222167969,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 170160
+    },
+    {
+      "epoch": 1119.5394736842106,
+      "grad_norm": 0.9721617102622986,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 170170
+    },
+    {
+      "epoch": 1119.6052631578948,
+      "grad_norm": 1.088510513305664,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 170180
+    },
+    {
+      "epoch": 1119.671052631579,
+      "grad_norm": 1.2319920063018799,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 170190
+    },
+    {
+      "epoch": 1119.7368421052631,
+      "grad_norm": 0.7855985760688782,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 170200
+    },
+    {
+      "epoch": 1119.8026315789473,
+      "grad_norm": 0.8179022669792175,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 170210
+    },
+    {
+      "epoch": 1119.8684210526317,
+      "grad_norm": 1.2478657960891724,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 170220
+    },
+    {
+      "epoch": 1119.9342105263158,
+      "grad_norm": 0.9301165342330933,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 170230
+    },
+    {
+      "epoch": 1120.0,
+      "grad_norm": 1.096358299255371,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 170240
+    },
+    {
+      "epoch": 1120.0657894736842,
+      "grad_norm": 1.134454369544983,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 170250
+    },
+    {
+      "epoch": 1120.1315789473683,
+      "grad_norm": 1.2220538854599,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 170260
+    },
+    {
+      "epoch": 1120.1973684210527,
+      "grad_norm": 1.1378227472305298,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 170270
+    },
+    {
+      "epoch": 1120.2631578947369,
+      "grad_norm": 0.8785101771354675,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 170280
+    },
+    {
+      "epoch": 1120.328947368421,
+      "grad_norm": 0.8152796626091003,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 170290
+    },
+    {
+      "epoch": 1120.3947368421052,
+      "grad_norm": 0.6250335574150085,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 170300
+    },
+    {
+      "epoch": 1120.4605263157894,
+      "grad_norm": 1.0436937808990479,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 170310
+    },
+    {
+      "epoch": 1120.5263157894738,
+      "grad_norm": 0.9515871405601501,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 170320
+    },
+    {
+      "epoch": 1120.592105263158,
+      "grad_norm": 1.3222743272781372,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 170330
+    },
+    {
+      "epoch": 1120.657894736842,
+      "grad_norm": 0.754511833190918,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 170340
+    },
+    {
+      "epoch": 1120.7236842105262,
+      "grad_norm": 1.0005298852920532,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 170350
+    },
+    {
+      "epoch": 1120.7894736842106,
+      "grad_norm": 1.1745997667312622,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 170360
+    },
+    {
+      "epoch": 1120.8552631578948,
+      "grad_norm": 1.2459282875061035,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 170370
+    },
+    {
+      "epoch": 1120.921052631579,
+      "grad_norm": 1.0530474185943604,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 170380
+    },
+    {
+      "epoch": 1120.9868421052631,
+      "grad_norm": 1.2831578254699707,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 170390
+    },
+    {
+      "epoch": 1121.0526315789473,
+      "grad_norm": 1.1489259004592896,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 170400
+    },
+    {
+      "epoch": 1121.1184210526317,
+      "grad_norm": 0.929141104221344,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 170410
+    },
+    {
+      "epoch": 1121.1842105263158,
+      "grad_norm": 1.1473424434661865,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 170420
+    },
+    {
+      "epoch": 1121.25,
+      "grad_norm": 1.2042150497436523,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 170430
+    },
+    {
+      "epoch": 1121.3157894736842,
+      "grad_norm": 0.9099306464195251,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 170440
+    },
+    {
+      "epoch": 1121.3815789473683,
+      "grad_norm": 1.141692876815796,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 170450
+    },
+    {
+      "epoch": 1121.4473684210527,
+      "grad_norm": 1.392948865890503,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 170460
+    },
+    {
+      "epoch": 1121.5131578947369,
+      "grad_norm": 0.8575266599655151,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 170470
+    },
+    {
+      "epoch": 1121.578947368421,
+      "grad_norm": 0.9454750418663025,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 170480
+    },
+    {
+      "epoch": 1121.6447368421052,
+      "grad_norm": 0.644149899482727,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 170490
+    },
+    {
+      "epoch": 1121.7105263157894,
+      "grad_norm": 1.050371527671814,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 170500
+    },
+    {
+      "epoch": 1121.7763157894738,
+      "grad_norm": 1.0545912981033325,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 170510
+    },
+    {
+      "epoch": 1121.842105263158,
+      "grad_norm": 0.8763865828514099,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 170520
+    },
+    {
+      "epoch": 1121.907894736842,
+      "grad_norm": 1.6924924850463867,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 170530
+    },
+    {
+      "epoch": 1121.9736842105262,
+      "grad_norm": 1.3894912004470825,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 170540
+    },
+    {
+      "epoch": 1122.0394736842106,
+      "grad_norm": 1.1801551580429077,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 170550
+    },
+    {
+      "epoch": 1122.1052631578948,
+      "grad_norm": 1.1211940050125122,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 170560
+    },
+    {
+      "epoch": 1122.171052631579,
+      "grad_norm": 1.0931165218353271,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 170570
+    },
+    {
+      "epoch": 1122.2368421052631,
+      "grad_norm": 1.0842170715332031,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 170580
+    },
+    {
+      "epoch": 1122.3026315789473,
+      "grad_norm": 1.0453462600708008,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 170590
+    },
+    {
+      "epoch": 1122.3684210526317,
+      "grad_norm": 0.7612990736961365,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 170600
+    },
+    {
+      "epoch": 1122.4342105263158,
+      "grad_norm": 1.1230723857879639,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 170610
+    },
+    {
+      "epoch": 1122.5,
+      "grad_norm": 0.7165056467056274,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 170620
+    },
+    {
+      "epoch": 1122.5657894736842,
+      "grad_norm": 1.0598911046981812,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 170630
+    },
+    {
+      "epoch": 1122.6315789473683,
+      "grad_norm": 1.3187886476516724,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 170640
+    },
+    {
+      "epoch": 1122.6973684210527,
+      "grad_norm": 1.140576720237732,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 170650
+    },
+    {
+      "epoch": 1122.7631578947369,
+      "grad_norm": 0.73148512840271,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 170660
+    },
+    {
+      "epoch": 1122.828947368421,
+      "grad_norm": 1.5248452425003052,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 170670
+    },
+    {
+      "epoch": 1122.8947368421052,
+      "grad_norm": 0.9757792949676514,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 170680
+    },
+    {
+      "epoch": 1122.9605263157894,
+      "grad_norm": 0.7404382228851318,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 170690
+    },
+    {
+      "epoch": 1123.0263157894738,
+      "grad_norm": 1.0916646718978882,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 170700
+    },
+    {
+      "epoch": 1123.092105263158,
+      "grad_norm": 1.196270227432251,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 170710
+    },
+    {
+      "epoch": 1123.157894736842,
+      "grad_norm": 1.4013103246688843,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 170720
+    },
+    {
+      "epoch": 1123.2236842105262,
+      "grad_norm": 1.210837960243225,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 170730
+    },
+    {
+      "epoch": 1123.2894736842106,
+      "grad_norm": 1.1967648267745972,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 170740
+    },
+    {
+      "epoch": 1123.3552631578948,
+      "grad_norm": 1.1867121458053589,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 170750
+    },
+    {
+      "epoch": 1123.421052631579,
+      "grad_norm": 0.9490767121315002,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 170760
+    },
+    {
+      "epoch": 1123.4868421052631,
+      "grad_norm": 0.8652997016906738,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 170770
+    },
+    {
+      "epoch": 1123.5526315789473,
+      "grad_norm": 1.082627534866333,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 170780
+    },
+    {
+      "epoch": 1123.6184210526317,
+      "grad_norm": 1.2542318105697632,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 170790
+    },
+    {
+      "epoch": 1123.6842105263158,
+      "grad_norm": 1.3510973453521729,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 170800
+    },
+    {
+      "epoch": 1123.75,
+      "grad_norm": 1.249051570892334,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 170810
+    },
+    {
+      "epoch": 1123.8157894736842,
+      "grad_norm": 1.2348275184631348,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 170820
+    },
+    {
+      "epoch": 1123.8815789473683,
+      "grad_norm": 1.2740799188613892,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 170830
+    },
+    {
+      "epoch": 1123.9473684210527,
+      "grad_norm": 1.133602499961853,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 170840
+    },
+    {
+      "epoch": 1124.0131578947369,
+      "grad_norm": 0.8562770485877991,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 170850
+    },
+    {
+      "epoch": 1124.078947368421,
+      "grad_norm": 1.2140693664550781,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 170860
+    },
+    {
+      "epoch": 1124.1447368421052,
+      "grad_norm": 1.1629786491394043,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 170870
+    },
+    {
+      "epoch": 1124.2105263157894,
+      "grad_norm": 0.9350384473800659,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 170880
+    },
+    {
+      "epoch": 1124.2763157894738,
+      "grad_norm": 0.7206969261169434,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 170890
+    },
+    {
+      "epoch": 1124.342105263158,
+      "grad_norm": 1.0412704944610596,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 170900
+    },
+    {
+      "epoch": 1124.407894736842,
+      "grad_norm": 1.0339332818984985,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 170910
+    },
+    {
+      "epoch": 1124.4736842105262,
+      "grad_norm": 0.9266976714134216,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 170920
+    },
+    {
+      "epoch": 1124.5394736842106,
+      "grad_norm": 1.0310457944869995,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 170930
+    },
+    {
+      "epoch": 1124.6052631578948,
+      "grad_norm": 1.2762702703475952,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 170940
+    },
+    {
+      "epoch": 1124.671052631579,
+      "grad_norm": 1.1744694709777832,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 170950
+    },
+    {
+      "epoch": 1124.7368421052631,
+      "grad_norm": 1.1720898151397705,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 170960
+    },
+    {
+      "epoch": 1124.8026315789473,
+      "grad_norm": 0.9738285541534424,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 170970
+    },
+    {
+      "epoch": 1124.8684210526317,
+      "grad_norm": 0.9768696427345276,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 170980
+    },
+    {
+      "epoch": 1124.9342105263158,
+      "grad_norm": 1.0169459581375122,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 170990
+    },
+    {
+      "epoch": 1125.0,
+      "grad_norm": 0.9950452446937561,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 171000
+    },
+    {
+      "epoch": 1125.0657894736842,
+      "grad_norm": 0.8835775256156921,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 171010
+    },
+    {
+      "epoch": 1125.1315789473683,
+      "grad_norm": 0.8104636073112488,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 171020
+    },
+    {
+      "epoch": 1125.1973684210527,
+      "grad_norm": 0.9511426091194153,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 171030
+    },
+    {
+      "epoch": 1125.2631578947369,
+      "grad_norm": 0.873656153678894,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 171040
+    },
+    {
+      "epoch": 1125.328947368421,
+      "grad_norm": 1.1368869543075562,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 171050
+    },
+    {
+      "epoch": 1125.3947368421052,
+      "grad_norm": 0.568688154220581,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 171060
+    },
+    {
+      "epoch": 1125.4605263157894,
+      "grad_norm": 0.5979208946228027,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 171070
+    },
+    {
+      "epoch": 1125.5263157894738,
+      "grad_norm": 0.9214772582054138,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 171080
+    },
+    {
+      "epoch": 1125.592105263158,
+      "grad_norm": 0.7531283497810364,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 171090
+    },
+    {
+      "epoch": 1125.657894736842,
+      "grad_norm": 1.2580556869506836,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 171100
+    },
+    {
+      "epoch": 1125.7236842105262,
+      "grad_norm": 0.930986225605011,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 171110
+    },
+    {
+      "epoch": 1125.7894736842106,
+      "grad_norm": 1.245627999305725,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 171120
+    },
+    {
+      "epoch": 1125.8552631578948,
+      "grad_norm": 0.7773181796073914,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 171130
+    },
+    {
+      "epoch": 1125.921052631579,
+      "grad_norm": 1.173557996749878,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 171140
+    },
+    {
+      "epoch": 1125.9868421052631,
+      "grad_norm": 0.8408874869346619,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 171150
+    },
+    {
+      "epoch": 1126.0526315789473,
+      "grad_norm": 0.6218984127044678,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 171160
+    },
+    {
+      "epoch": 1126.1184210526317,
+      "grad_norm": 0.9372327923774719,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 171170
+    },
+    {
+      "epoch": 1126.1842105263158,
+      "grad_norm": 0.9183964729309082,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 171180
+    },
+    {
+      "epoch": 1126.25,
+      "grad_norm": 1.0843031406402588,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 171190
+    },
+    {
+      "epoch": 1126.3157894736842,
+      "grad_norm": 1.221227765083313,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 171200
+    },
+    {
+      "epoch": 1126.3815789473683,
+      "grad_norm": 1.0891140699386597,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 171210
+    },
+    {
+      "epoch": 1126.4473684210527,
+      "grad_norm": 1.0184756517410278,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 171220
+    },
+    {
+      "epoch": 1126.5131578947369,
+      "grad_norm": 0.9050408601760864,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 171230
+    },
+    {
+      "epoch": 1126.578947368421,
+      "grad_norm": 0.754024088382721,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 171240
+    },
+    {
+      "epoch": 1126.6447368421052,
+      "grad_norm": 1.1629414558410645,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 171250
+    },
+    {
+      "epoch": 1126.7105263157894,
+      "grad_norm": 1.1155112981796265,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 171260
+    },
+    {
+      "epoch": 1126.7763157894738,
+      "grad_norm": 1.1457107067108154,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 171270
+    },
+    {
+      "epoch": 1126.842105263158,
+      "grad_norm": 1.5924310684204102,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 171280
+    },
+    {
+      "epoch": 1126.907894736842,
+      "grad_norm": 1.267576813697815,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 171290
+    },
+    {
+      "epoch": 1126.9736842105262,
+      "grad_norm": 1.2480442523956299,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 171300
+    },
+    {
+      "epoch": 1127.0394736842106,
+      "grad_norm": 0.923325777053833,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 171310
+    },
+    {
+      "epoch": 1127.1052631578948,
+      "grad_norm": 1.2825509309768677,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 171320
+    },
+    {
+      "epoch": 1127.171052631579,
+      "grad_norm": 1.1516602039337158,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 171330
+    },
+    {
+      "epoch": 1127.2368421052631,
+      "grad_norm": 0.9105165004730225,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 171340
+    },
+    {
+      "epoch": 1127.3026315789473,
+      "grad_norm": 1.116930603981018,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 171350
+    },
+    {
+      "epoch": 1127.3684210526317,
+      "grad_norm": 1.105677604675293,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 171360
+    },
+    {
+      "epoch": 1127.4342105263158,
+      "grad_norm": 1.1379443407058716,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 171370
+    },
+    {
+      "epoch": 1127.5,
+      "grad_norm": 0.8313212394714355,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 171380
+    },
+    {
+      "epoch": 1127.5657894736842,
+      "grad_norm": 0.9692210555076599,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 171390
+    },
+    {
+      "epoch": 1127.6315789473683,
+      "grad_norm": 0.9176498651504517,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 171400
+    },
+    {
+      "epoch": 1127.6973684210527,
+      "grad_norm": 1.2443631887435913,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 171410
+    },
+    {
+      "epoch": 1127.7631578947369,
+      "grad_norm": 1.4867162704467773,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 171420
+    },
+    {
+      "epoch": 1127.828947368421,
+      "grad_norm": 1.2587157487869263,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 171430
+    },
+    {
+      "epoch": 1127.8947368421052,
+      "grad_norm": 1.1008973121643066,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 171440
+    },
+    {
+      "epoch": 1127.9605263157894,
+      "grad_norm": 1.035931944847107,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 171450
+    },
+    {
+      "epoch": 1128.0263157894738,
+      "grad_norm": 0.9197618961334229,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 171460
+    },
+    {
+      "epoch": 1128.092105263158,
+      "grad_norm": 0.9170363545417786,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 171470
+    },
+    {
+      "epoch": 1128.157894736842,
+      "grad_norm": 1.0054898262023926,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 171480
+    },
+    {
+      "epoch": 1128.2236842105262,
+      "grad_norm": 0.6774921417236328,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 171490
+    },
+    {
+      "epoch": 1128.2894736842106,
+      "grad_norm": 1.0647435188293457,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 171500
+    },
+    {
+      "epoch": 1128.3552631578948,
+      "grad_norm": 0.9801605343818665,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 171510
+    },
+    {
+      "epoch": 1128.421052631579,
+      "grad_norm": 0.9069273471832275,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 171520
+    },
+    {
+      "epoch": 1128.4868421052631,
+      "grad_norm": 0.8292950987815857,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 171530
+    },
+    {
+      "epoch": 1128.5526315789473,
+      "grad_norm": 1.0863277912139893,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 171540
+    },
+    {
+      "epoch": 1128.6184210526317,
+      "grad_norm": 1.3724346160888672,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 171550
+    },
+    {
+      "epoch": 1128.6842105263158,
+      "grad_norm": 0.905957043170929,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 171560
+    },
+    {
+      "epoch": 1128.75,
+      "grad_norm": 0.5370518565177917,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 171570
+    },
+    {
+      "epoch": 1128.8157894736842,
+      "grad_norm": 1.248266339302063,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 171580
+    },
+    {
+      "epoch": 1128.8815789473683,
+      "grad_norm": 1.00150728225708,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 171590
+    },
+    {
+      "epoch": 1128.9473684210527,
+      "grad_norm": 1.0399243831634521,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 171600
+    },
+    {
+      "epoch": 1129.0131578947369,
+      "grad_norm": 1.1588037014007568,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 171610
+    },
+    {
+      "epoch": 1129.078947368421,
+      "grad_norm": 1.0992803573608398,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 171620
+    },
+    {
+      "epoch": 1129.1447368421052,
+      "grad_norm": 1.2114496231079102,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 171630
+    },
+    {
+      "epoch": 1129.2105263157894,
+      "grad_norm": 1.3456045389175415,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 171640
+    },
+    {
+      "epoch": 1129.2763157894738,
+      "grad_norm": 1.063812017440796,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 171650
+    },
+    {
+      "epoch": 1129.342105263158,
+      "grad_norm": 0.7482918500900269,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 171660
+    },
+    {
+      "epoch": 1129.407894736842,
+      "grad_norm": 0.9320112466812134,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 171670
+    },
+    {
+      "epoch": 1129.4736842105262,
+      "grad_norm": 1.2601550817489624,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 171680
+    },
+    {
+      "epoch": 1129.5394736842106,
+      "grad_norm": 0.9702651500701904,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 171690
+    },
+    {
+      "epoch": 1129.6052631578948,
+      "grad_norm": 0.6806167960166931,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 171700
+    },
+    {
+      "epoch": 1129.671052631579,
+      "grad_norm": 0.8016168475151062,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 171710
+    },
+    {
+      "epoch": 1129.7368421052631,
+      "grad_norm": 1.0033786296844482,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 171720
+    },
+    {
+      "epoch": 1129.8026315789473,
+      "grad_norm": 1.152343988418579,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 171730
+    },
+    {
+      "epoch": 1129.8684210526317,
+      "grad_norm": 1.0898712873458862,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 171740
+    },
+    {
+      "epoch": 1129.9342105263158,
+      "grad_norm": 1.0133626461029053,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 171750
+    },
+    {
+      "epoch": 1130.0,
+      "grad_norm": 1.0823217630386353,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 171760
+    },
+    {
+      "epoch": 1130.0657894736842,
+      "grad_norm": 0.853507936000824,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 171770
+    },
+    {
+      "epoch": 1130.1315789473683,
+      "grad_norm": 1.2531830072402954,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 171780
+    },
+    {
+      "epoch": 1130.1973684210527,
+      "grad_norm": 1.178091287612915,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 171790
+    },
+    {
+      "epoch": 1130.2631578947369,
+      "grad_norm": 1.0549789667129517,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 171800
+    },
+    {
+      "epoch": 1130.328947368421,
+      "grad_norm": 1.3365063667297363,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 171810
+    },
+    {
+      "epoch": 1130.3947368421052,
+      "grad_norm": 1.022502064704895,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 171820
+    },
+    {
+      "epoch": 1130.4605263157894,
+      "grad_norm": 0.9836164116859436,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 171830
+    },
+    {
+      "epoch": 1130.5263157894738,
+      "grad_norm": 1.1745048761367798,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 171840
+    },
+    {
+      "epoch": 1130.592105263158,
+      "grad_norm": 1.0340749025344849,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 171850
+    },
+    {
+      "epoch": 1130.657894736842,
+      "grad_norm": 0.8751736283302307,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 171860
+    },
+    {
+      "epoch": 1130.7236842105262,
+      "grad_norm": 1.1623742580413818,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 171870
+    },
+    {
+      "epoch": 1130.7894736842106,
+      "grad_norm": 0.9135785102844238,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 171880
+    },
+    {
+      "epoch": 1130.8552631578948,
+      "grad_norm": 0.9940324425697327,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 171890
+    },
+    {
+      "epoch": 1130.921052631579,
+      "grad_norm": 0.944202721118927,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 171900
+    },
+    {
+      "epoch": 1130.9868421052631,
+      "grad_norm": 1.2017103433609009,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 171910
+    },
+    {
+      "epoch": 1131.0526315789473,
+      "grad_norm": 0.7654322385787964,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 171920
+    },
+    {
+      "epoch": 1131.1184210526317,
+      "grad_norm": 0.7834600210189819,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 171930
+    },
+    {
+      "epoch": 1131.1842105263158,
+      "grad_norm": 0.751923143863678,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 171940
+    },
+    {
+      "epoch": 1131.25,
+      "grad_norm": 0.9888975024223328,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 171950
+    },
+    {
+      "epoch": 1131.3157894736842,
+      "grad_norm": 1.2195971012115479,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 171960
+    },
+    {
+      "epoch": 1131.3815789473683,
+      "grad_norm": 1.110573410987854,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 171970
+    },
+    {
+      "epoch": 1131.4473684210527,
+      "grad_norm": 1.1125825643539429,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 171980
+    },
+    {
+      "epoch": 1131.5131578947369,
+      "grad_norm": 0.6448672413825989,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 171990
+    },
+    {
+      "epoch": 1131.578947368421,
+      "grad_norm": 1.2890968322753906,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 172000
+    },
+    {
+      "epoch": 1131.6447368421052,
+      "grad_norm": 1.0350221395492554,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 172010
+    },
+    {
+      "epoch": 1131.7105263157894,
+      "grad_norm": 0.9488757252693176,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 172020
+    },
+    {
+      "epoch": 1131.7763157894738,
+      "grad_norm": 0.7833364009857178,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 172030
+    },
+    {
+      "epoch": 1131.842105263158,
+      "grad_norm": 1.1103259325027466,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 172040
+    },
+    {
+      "epoch": 1131.907894736842,
+      "grad_norm": 1.4510931968688965,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 172050
+    },
+    {
+      "epoch": 1131.9736842105262,
+      "grad_norm": 1.293509840965271,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 172060
+    },
+    {
+      "epoch": 1132.0394736842106,
+      "grad_norm": 1.0810400247573853,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 172070
+    },
+    {
+      "epoch": 1132.1052631578948,
+      "grad_norm": 1.3170888423919678,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 172080
+    },
+    {
+      "epoch": 1132.171052631579,
+      "grad_norm": 0.8184434771537781,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 172090
+    },
+    {
+      "epoch": 1132.2368421052631,
+      "grad_norm": 1.073915958404541,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 172100
+    },
+    {
+      "epoch": 1132.3026315789473,
+      "grad_norm": 1.019290566444397,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 172110
+    },
+    {
+      "epoch": 1132.3684210526317,
+      "grad_norm": 1.3928847312927246,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 172120
+    },
+    {
+      "epoch": 1132.4342105263158,
+      "grad_norm": 1.158503770828247,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 172130
+    },
+    {
+      "epoch": 1132.5,
+      "grad_norm": 1.2527748346328735,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 172140
+    },
+    {
+      "epoch": 1132.5657894736842,
+      "grad_norm": 1.0773019790649414,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 172150
+    },
+    {
+      "epoch": 1132.6315789473683,
+      "grad_norm": 1.3126847743988037,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 172160
+    },
+    {
+      "epoch": 1132.6973684210527,
+      "grad_norm": 0.9994803071022034,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 172170
+    },
+    {
+      "epoch": 1132.7631578947369,
+      "grad_norm": 1.4703785181045532,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 172180
+    },
+    {
+      "epoch": 1132.828947368421,
+      "grad_norm": 0.6720287799835205,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 172190
+    },
+    {
+      "epoch": 1132.8947368421052,
+      "grad_norm": 0.6046428084373474,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 172200
+    },
+    {
+      "epoch": 1132.9605263157894,
+      "grad_norm": 1.031443476676941,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 172210
+    },
+    {
+      "epoch": 1133.0263157894738,
+      "grad_norm": 1.2193036079406738,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 172220
+    },
+    {
+      "epoch": 1133.092105263158,
+      "grad_norm": 1.1105955839157104,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 172230
+    },
+    {
+      "epoch": 1133.157894736842,
+      "grad_norm": 1.0140831470489502,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 172240
+    },
+    {
+      "epoch": 1133.2236842105262,
+      "grad_norm": 0.8403400778770447,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 172250
+    },
+    {
+      "epoch": 1133.2894736842106,
+      "grad_norm": 1.143967628479004,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 172260
+    },
+    {
+      "epoch": 1133.3552631578948,
+      "grad_norm": 0.6676881909370422,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 172270
+    },
+    {
+      "epoch": 1133.421052631579,
+      "grad_norm": 0.8999795913696289,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 172280
+    },
+    {
+      "epoch": 1133.4868421052631,
+      "grad_norm": 0.9858129024505615,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 172290
+    },
+    {
+      "epoch": 1133.5526315789473,
+      "grad_norm": 0.9326895475387573,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 172300
+    },
+    {
+      "epoch": 1133.6184210526317,
+      "grad_norm": 1.069669246673584,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 172310
+    },
+    {
+      "epoch": 1133.6842105263158,
+      "grad_norm": 0.8025325536727905,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 172320
+    },
+    {
+      "epoch": 1133.75,
+      "grad_norm": 0.5994061231613159,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 172330
+    },
+    {
+      "epoch": 1133.8157894736842,
+      "grad_norm": 0.6835926175117493,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 172340
+    },
+    {
+      "epoch": 1133.8815789473683,
+      "grad_norm": 0.6774199604988098,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 172350
+    },
+    {
+      "epoch": 1133.9473684210527,
+      "grad_norm": 1.1998493671417236,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 172360
+    },
+    {
+      "epoch": 1134.0131578947369,
+      "grad_norm": 1.2091012001037598,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 172370
+    },
+    {
+      "epoch": 1134.078947368421,
+      "grad_norm": 0.9642314910888672,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 172380
+    },
+    {
+      "epoch": 1134.1447368421052,
+      "grad_norm": 0.9943333864212036,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 172390
+    },
+    {
+      "epoch": 1134.2105263157894,
+      "grad_norm": 1.449242115020752,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 172400
+    },
+    {
+      "epoch": 1134.2763157894738,
+      "grad_norm": 1.4561964273452759,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 172410
+    },
+    {
+      "epoch": 1134.342105263158,
+      "grad_norm": 1.191004991531372,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 172420
+    },
+    {
+      "epoch": 1134.407894736842,
+      "grad_norm": 1.171547293663025,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 172430
+    },
+    {
+      "epoch": 1134.4736842105262,
+      "grad_norm": 1.08799409866333,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 172440
+    },
+    {
+      "epoch": 1134.5394736842106,
+      "grad_norm": 1.141973614692688,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 172450
+    },
+    {
+      "epoch": 1134.6052631578948,
+      "grad_norm": 1.2193361520767212,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 172460
+    },
+    {
+      "epoch": 1134.671052631579,
+      "grad_norm": 1.011452078819275,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 172470
+    },
+    {
+      "epoch": 1134.7368421052631,
+      "grad_norm": 1.046140432357788,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 172480
+    },
+    {
+      "epoch": 1134.8026315789473,
+      "grad_norm": 0.9603179693222046,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 172490
+    },
+    {
+      "epoch": 1134.8684210526317,
+      "grad_norm": 1.3100394010543823,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 172500
+    },
+    {
+      "epoch": 1134.9342105263158,
+      "grad_norm": 0.8967265486717224,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 172510
+    },
+    {
+      "epoch": 1135.0,
+      "grad_norm": 0.750058650970459,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 172520
+    },
+    {
+      "epoch": 1135.0657894736842,
+      "grad_norm": 0.9799665808677673,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 172530
+    },
+    {
+      "epoch": 1135.1315789473683,
+      "grad_norm": 0.871820330619812,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 172540
+    },
+    {
+      "epoch": 1135.1973684210527,
+      "grad_norm": 1.1122573614120483,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 172550
+    },
+    {
+      "epoch": 1135.2631578947369,
+      "grad_norm": 0.9300641417503357,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 172560
+    },
+    {
+      "epoch": 1135.328947368421,
+      "grad_norm": 1.2777584791183472,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 172570
+    },
+    {
+      "epoch": 1135.3947368421052,
+      "grad_norm": 0.886059582233429,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 172580
+    },
+    {
+      "epoch": 1135.4605263157894,
+      "grad_norm": 1.118418574333191,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 172590
+    },
+    {
+      "epoch": 1135.5263157894738,
+      "grad_norm": 1.219286561012268,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 172600
+    },
+    {
+      "epoch": 1135.592105263158,
+      "grad_norm": 1.4709956645965576,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 172610
+    },
+    {
+      "epoch": 1135.657894736842,
+      "grad_norm": 1.124354600906372,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 172620
+    },
+    {
+      "epoch": 1135.7236842105262,
+      "grad_norm": 0.929551362991333,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 172630
+    },
+    {
+      "epoch": 1135.7894736842106,
+      "grad_norm": 1.3097996711730957,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 172640
+    },
+    {
+      "epoch": 1135.8552631578948,
+      "grad_norm": 1.0899239778518677,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 172650
+    },
+    {
+      "epoch": 1135.921052631579,
+      "grad_norm": 1.0806041955947876,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "step": 172660
+    },
+    {
+      "epoch": 1135.9868421052631,
+      "grad_norm": 1.1731756925582886,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 172670
+    },
+    {
+      "epoch": 1136.0526315789473,
+      "grad_norm": 0.7187557816505432,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 172680
+    },
+    {
+      "epoch": 1136.1184210526317,
+      "grad_norm": 1.014794945716858,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 172690
+    },
+    {
+      "epoch": 1136.1842105263158,
+      "grad_norm": 0.9141129851341248,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 172700
+    },
+    {
+      "epoch": 1136.25,
+      "grad_norm": 0.8806213736534119,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 172710
+    },
+    {
+      "epoch": 1136.3157894736842,
+      "grad_norm": 1.0453318357467651,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 172720
+    },
+    {
+      "epoch": 1136.3815789473683,
+      "grad_norm": 1.0713512897491455,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 172730
+    },
+    {
+      "epoch": 1136.4473684210527,
+      "grad_norm": 0.8867670893669128,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 172740
+    },
+    {
+      "epoch": 1136.5131578947369,
+      "grad_norm": 1.3366000652313232,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 172750
+    },
+    {
+      "epoch": 1136.578947368421,
+      "grad_norm": 1.2236058712005615,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 172760
+    },
+    {
+      "epoch": 1136.6447368421052,
+      "grad_norm": 1.2140995264053345,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 172770
+    },
+    {
+      "epoch": 1136.7105263157894,
+      "grad_norm": 1.1668179035186768,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 172780
+    },
+    {
+      "epoch": 1136.7763157894738,
+      "grad_norm": 0.9051017761230469,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 172790
+    },
+    {
+      "epoch": 1136.842105263158,
+      "grad_norm": 1.346951961517334,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 172800
+    },
+    {
+      "epoch": 1136.907894736842,
+      "grad_norm": 0.9546437859535217,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 172810
+    },
+    {
+      "epoch": 1136.9736842105262,
+      "grad_norm": 1.1562156677246094,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 172820
+    },
+    {
+      "epoch": 1137.0394736842106,
+      "grad_norm": 1.1238209009170532,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 172830
+    },
+    {
+      "epoch": 1137.1052631578948,
+      "grad_norm": 1.0962235927581787,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 172840
+    },
+    {
+      "epoch": 1137.171052631579,
+      "grad_norm": 1.1757200956344604,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 172850
+    },
+    {
+      "epoch": 1137.2368421052631,
+      "grad_norm": 0.6541777849197388,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 172860
+    },
+    {
+      "epoch": 1137.3026315789473,
+      "grad_norm": 0.7562914490699768,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 172870
+    },
+    {
+      "epoch": 1137.3684210526317,
+      "grad_norm": 0.8109030723571777,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 172880
+    },
+    {
+      "epoch": 1137.4342105263158,
+      "grad_norm": 0.924420952796936,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 172890
+    },
+    {
+      "epoch": 1137.5,
+      "grad_norm": 0.7896656394004822,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 172900
+    },
+    {
+      "epoch": 1137.5657894736842,
+      "grad_norm": 1.00142502784729,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 172910
+    },
+    {
+      "epoch": 1137.6315789473683,
+      "grad_norm": 0.8568731546401978,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 172920
+    },
+    {
+      "epoch": 1137.6973684210527,
+      "grad_norm": 1.0580912828445435,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 172930
+    },
+    {
+      "epoch": 1137.7631578947369,
+      "grad_norm": 0.8699727058410645,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 172940
+    },
+    {
+      "epoch": 1137.828947368421,
+      "grad_norm": 0.8225347399711609,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 172950
+    },
+    {
+      "epoch": 1137.8947368421052,
+      "grad_norm": 1.0345213413238525,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 172960
+    },
+    {
+      "epoch": 1137.9605263157894,
+      "grad_norm": 1.1980303525924683,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 172970
+    },
+    {
+      "epoch": 1138.0263157894738,
+      "grad_norm": 1.3538132905960083,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 172980
+    },
+    {
+      "epoch": 1138.092105263158,
+      "grad_norm": 1.0862656831741333,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 172990
+    },
+    {
+      "epoch": 1138.157894736842,
+      "grad_norm": 1.2043201923370361,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 173000
+    },
+    {
+      "epoch": 1138.2236842105262,
+      "grad_norm": 1.0616967678070068,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 173010
+    },
+    {
+      "epoch": 1138.2894736842106,
+      "grad_norm": 1.2403620481491089,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 173020
+    },
+    {
+      "epoch": 1138.3552631578948,
+      "grad_norm": 1.2264503240585327,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 173030
+    },
+    {
+      "epoch": 1138.421052631579,
+      "grad_norm": 0.9958550333976746,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 173040
+    },
+    {
+      "epoch": 1138.4868421052631,
+      "grad_norm": 0.9058167934417725,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 173050
+    },
+    {
+      "epoch": 1138.5526315789473,
+      "grad_norm": 0.7711786031723022,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 173060
+    },
+    {
+      "epoch": 1138.6184210526317,
+      "grad_norm": 1.0303704738616943,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 173070
+    },
+    {
+      "epoch": 1138.6842105263158,
+      "grad_norm": 1.2294689416885376,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 173080
+    },
+    {
+      "epoch": 1138.75,
+      "grad_norm": 0.888333797454834,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 173090
+    },
+    {
+      "epoch": 1138.8157894736842,
+      "grad_norm": 0.8880026936531067,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 173100
+    },
+    {
+      "epoch": 1138.8815789473683,
+      "grad_norm": 0.8499730229377747,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 173110
+    },
+    {
+      "epoch": 1138.9473684210527,
+      "grad_norm": 1.0166504383087158,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 173120
+    },
+    {
+      "epoch": 1139.0131578947369,
+      "grad_norm": 1.199263334274292,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 173130
+    },
+    {
+      "epoch": 1139.078947368421,
+      "grad_norm": 0.9333541393280029,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 173140
+    },
+    {
+      "epoch": 1139.1447368421052,
+      "grad_norm": 1.0832608938217163,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 173150
+    },
+    {
+      "epoch": 1139.2105263157894,
+      "grad_norm": 1.0656371116638184,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 173160
+    },
+    {
+      "epoch": 1139.2763157894738,
+      "grad_norm": 0.7567489147186279,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 173170
+    },
+    {
+      "epoch": 1139.342105263158,
+      "grad_norm": 1.146492600440979,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 173180
+    },
+    {
+      "epoch": 1139.407894736842,
+      "grad_norm": 1.0995131731033325,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 173190
+    },
+    {
+      "epoch": 1139.4736842105262,
+      "grad_norm": 1.1685936450958252,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 173200
+    },
+    {
+      "epoch": 1139.5394736842106,
+      "grad_norm": 0.8442726731300354,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 173210
+    },
+    {
+      "epoch": 1139.6052631578948,
+      "grad_norm": 0.9669975638389587,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 173220
+    },
+    {
+      "epoch": 1139.671052631579,
+      "grad_norm": 1.19166100025177,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 173230
+    },
+    {
+      "epoch": 1139.7368421052631,
+      "grad_norm": 0.6347991228103638,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 173240
+    },
+    {
+      "epoch": 1139.8026315789473,
+      "grad_norm": 0.877065122127533,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 173250
+    },
+    {
+      "epoch": 1139.8684210526317,
+      "grad_norm": 0.9608317017555237,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 173260
+    },
+    {
+      "epoch": 1139.9342105263158,
+      "grad_norm": 1.243171215057373,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 173270
+    },
+    {
+      "epoch": 1140.0,
+      "grad_norm": 0.7330761551856995,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 173280
+    },
+    {
+      "epoch": 1140.0657894736842,
+      "grad_norm": 1.0452905893325806,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 173290
+    },
+    {
+      "epoch": 1140.1315789473683,
+      "grad_norm": 1.095406413078308,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 173300
+    },
+    {
+      "epoch": 1140.1973684210527,
+      "grad_norm": 0.8945803642272949,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 173310
+    },
+    {
+      "epoch": 1140.2631578947369,
+      "grad_norm": 1.0646826028823853,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 173320
+    },
+    {
+      "epoch": 1140.328947368421,
+      "grad_norm": 0.9003171324729919,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 173330
+    },
+    {
+      "epoch": 1140.3947368421052,
+      "grad_norm": 1.0605850219726562,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 173340
+    },
+    {
+      "epoch": 1140.4605263157894,
+      "grad_norm": 1.0738884210586548,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 173350
+    },
+    {
+      "epoch": 1140.5263157894738,
+      "grad_norm": 1.067591667175293,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 173360
+    },
+    {
+      "epoch": 1140.592105263158,
+      "grad_norm": 1.4844350814819336,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 173370
+    },
+    {
+      "epoch": 1140.657894736842,
+      "grad_norm": 1.0386178493499756,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 173380
+    },
+    {
+      "epoch": 1140.7236842105262,
+      "grad_norm": 1.306445837020874,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 173390
+    },
+    {
+      "epoch": 1140.7894736842106,
+      "grad_norm": 1.4519286155700684,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 173400
+    },
+    {
+      "epoch": 1140.8552631578948,
+      "grad_norm": 1.0992698669433594,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 173410
+    },
+    {
+      "epoch": 1140.921052631579,
+      "grad_norm": 1.0953896045684814,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 173420
+    },
+    {
+      "epoch": 1140.9868421052631,
+      "grad_norm": 1.086705207824707,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 173430
+    },
+    {
+      "epoch": 1141.0526315789473,
+      "grad_norm": 0.9892106056213379,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 173440
+    },
+    {
+      "epoch": 1141.1184210526317,
+      "grad_norm": 0.8099537491798401,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 173450
+    },
+    {
+      "epoch": 1141.1842105263158,
+      "grad_norm": 1.2578682899475098,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 173460
+    },
+    {
+      "epoch": 1141.25,
+      "grad_norm": 1.0416673421859741,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 173470
+    },
+    {
+      "epoch": 1141.3157894736842,
+      "grad_norm": 0.9099404215812683,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 173480
+    },
+    {
+      "epoch": 1141.3815789473683,
+      "grad_norm": 1.2768899202346802,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 173490
+    },
+    {
+      "epoch": 1141.4473684210527,
+      "grad_norm": 1.1993544101715088,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 173500
+    },
+    {
+      "epoch": 1141.5131578947369,
+      "grad_norm": 1.094346046447754,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 173510
+    },
+    {
+      "epoch": 1141.578947368421,
+      "grad_norm": 0.8274684548377991,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 173520
+    },
+    {
+      "epoch": 1141.6447368421052,
+      "grad_norm": 1.1275930404663086,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 173530
+    },
+    {
+      "epoch": 1141.7105263157894,
+      "grad_norm": 1.065172553062439,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 173540
+    },
+    {
+      "epoch": 1141.7763157894738,
+      "grad_norm": 0.9865522980690002,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 173550
+    },
+    {
+      "epoch": 1141.842105263158,
+      "grad_norm": 0.7311487793922424,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 173560
+    },
+    {
+      "epoch": 1141.907894736842,
+      "grad_norm": 0.9053587317466736,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 173570
+    },
+    {
+      "epoch": 1141.9736842105262,
+      "grad_norm": 1.1042989492416382,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 173580
+    },
+    {
+      "epoch": 1142.0394736842106,
+      "grad_norm": 1.2687427997589111,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 173590
+    },
+    {
+      "epoch": 1142.1052631578948,
+      "grad_norm": 0.8820356726646423,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 173600
+    },
+    {
+      "epoch": 1142.171052631579,
+      "grad_norm": 1.3813494443893433,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 173610
+    },
+    {
+      "epoch": 1142.2368421052631,
+      "grad_norm": 0.895897626876831,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 173620
+    },
+    {
+      "epoch": 1142.3026315789473,
+      "grad_norm": 1.020932674407959,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 173630
+    },
+    {
+      "epoch": 1142.3684210526317,
+      "grad_norm": 1.1267672777175903,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 173640
+    },
+    {
+      "epoch": 1142.4342105263158,
+      "grad_norm": 0.8028258085250854,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 173650
+    },
+    {
+      "epoch": 1142.5,
+      "grad_norm": 1.3700445890426636,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 173660
+    },
+    {
+      "epoch": 1142.5657894736842,
+      "grad_norm": 0.9799003005027771,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 173670
+    },
+    {
+      "epoch": 1142.6315789473683,
+      "grad_norm": 0.9328917264938354,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 173680
+    },
+    {
+      "epoch": 1142.6973684210527,
+      "grad_norm": 0.8575257658958435,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 173690
+    },
+    {
+      "epoch": 1142.7631578947369,
+      "grad_norm": 0.9841787219047546,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 173700
+    },
+    {
+      "epoch": 1142.828947368421,
+      "grad_norm": 1.0814399719238281,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 173710
+    },
+    {
+      "epoch": 1142.8947368421052,
+      "grad_norm": 1.0300018787384033,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 173720
+    },
+    {
+      "epoch": 1142.9605263157894,
+      "grad_norm": 0.9641240239143372,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 173730
+    },
+    {
+      "epoch": 1143.0263157894738,
+      "grad_norm": 1.5655041933059692,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 173740
+    },
+    {
+      "epoch": 1143.092105263158,
+      "grad_norm": 1.1116403341293335,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 173750
+    },
+    {
+      "epoch": 1143.157894736842,
+      "grad_norm": 0.8238686919212341,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 173760
+    },
+    {
+      "epoch": 1143.2236842105262,
+      "grad_norm": 1.036594033241272,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 173770
+    },
+    {
+      "epoch": 1143.2894736842106,
+      "grad_norm": 1.2605559825897217,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 173780
+    },
+    {
+      "epoch": 1143.3552631578948,
+      "grad_norm": 0.9970645904541016,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 173790
+    },
+    {
+      "epoch": 1143.421052631579,
+      "grad_norm": 0.645720362663269,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 173800
+    },
+    {
+      "epoch": 1143.4868421052631,
+      "grad_norm": 1.0435887575149536,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 173810
+    },
+    {
+      "epoch": 1143.5526315789473,
+      "grad_norm": 1.0935068130493164,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 173820
+    },
+    {
+      "epoch": 1143.6184210526317,
+      "grad_norm": 0.9884122610092163,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 173830
+    },
+    {
+      "epoch": 1143.6842105263158,
+      "grad_norm": 1.100342035293579,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 173840
+    },
+    {
+      "epoch": 1143.75,
+      "grad_norm": 0.95659339427948,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 173850
+    },
+    {
+      "epoch": 1143.8157894736842,
+      "grad_norm": 0.9715131521224976,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 173860
+    },
+    {
+      "epoch": 1143.8815789473683,
+      "grad_norm": 0.7474709153175354,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 173870
+    },
+    {
+      "epoch": 1143.9473684210527,
+      "grad_norm": 0.9470142126083374,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 173880
+    },
+    {
+      "epoch": 1144.0131578947369,
+      "grad_norm": 0.9878813624382019,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 173890
+    },
+    {
+      "epoch": 1144.078947368421,
+      "grad_norm": 0.6697216629981995,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 173900
+    },
+    {
+      "epoch": 1144.1447368421052,
+      "grad_norm": 1.1067986488342285,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 173910
+    },
+    {
+      "epoch": 1144.2105263157894,
+      "grad_norm": 1.1652238368988037,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 173920
+    },
+    {
+      "epoch": 1144.2763157894738,
+      "grad_norm": 1.5187675952911377,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 173930
+    },
+    {
+      "epoch": 1144.342105263158,
+      "grad_norm": 0.8939927220344543,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 173940
+    },
+    {
+      "epoch": 1144.407894736842,
+      "grad_norm": 1.0155541896820068,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 173950
+    },
+    {
+      "epoch": 1144.4736842105262,
+      "grad_norm": 1.0913894176483154,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 173960
+    },
+    {
+      "epoch": 1144.5394736842106,
+      "grad_norm": 0.9633883237838745,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 173970
+    },
+    {
+      "epoch": 1144.6052631578948,
+      "grad_norm": 0.845365047454834,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 173980
+    },
+    {
+      "epoch": 1144.671052631579,
+      "grad_norm": 0.9477745890617371,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 173990
+    },
+    {
+      "epoch": 1144.7368421052631,
+      "grad_norm": 1.1216838359832764,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 174000
+    },
+    {
+      "epoch": 1144.8026315789473,
+      "grad_norm": 1.5633032321929932,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 174010
+    },
+    {
+      "epoch": 1144.8684210526317,
+      "grad_norm": 1.2574188709259033,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 174020
+    },
+    {
+      "epoch": 1144.9342105263158,
+      "grad_norm": 1.3190844058990479,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 174030
+    },
+    {
+      "epoch": 1145.0,
+      "grad_norm": 1.2496864795684814,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 174040
+    },
+    {
+      "epoch": 1145.0657894736842,
+      "grad_norm": 1.188367486000061,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 174050
+    },
+    {
+      "epoch": 1145.1315789473683,
+      "grad_norm": 1.2431753873825073,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 174060
+    },
+    {
+      "epoch": 1145.1973684210527,
+      "grad_norm": 1.5690101385116577,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 174070
+    },
+    {
+      "epoch": 1145.2631578947369,
+      "grad_norm": 1.4982401132583618,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 174080
+    },
+    {
+      "epoch": 1145.328947368421,
+      "grad_norm": 1.3239631652832031,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 174090
+    },
+    {
+      "epoch": 1145.3947368421052,
+      "grad_norm": 1.079105257987976,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 174100
+    },
+    {
+      "epoch": 1145.4605263157894,
+      "grad_norm": 1.0493974685668945,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 174110
+    },
+    {
+      "epoch": 1145.5263157894738,
+      "grad_norm": 1.0567365884780884,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 174120
+    },
+    {
+      "epoch": 1145.592105263158,
+      "grad_norm": 1.0888392925262451,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 174130
+    },
+    {
+      "epoch": 1145.657894736842,
+      "grad_norm": 1.156753420829773,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 174140
+    },
+    {
+      "epoch": 1145.7236842105262,
+      "grad_norm": 0.7009889483451843,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 174150
+    },
+    {
+      "epoch": 1145.7894736842106,
+      "grad_norm": 1.242331862449646,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 174160
+    },
+    {
+      "epoch": 1145.8552631578948,
+      "grad_norm": 1.1424720287322998,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 174170
+    },
+    {
+      "epoch": 1145.921052631579,
+      "grad_norm": 1.1428437232971191,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 174180
+    },
+    {
+      "epoch": 1145.9868421052631,
+      "grad_norm": 1.0217548608779907,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 174190
+    },
+    {
+      "epoch": 1146.0526315789473,
+      "grad_norm": 0.9081407189369202,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 174200
+    },
+    {
+      "epoch": 1146.1184210526317,
+      "grad_norm": 1.0314104557037354,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 174210
+    },
+    {
+      "epoch": 1146.1842105263158,
+      "grad_norm": 0.9840436577796936,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 174220
+    },
+    {
+      "epoch": 1146.25,
+      "grad_norm": 1.1434142589569092,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 174230
+    },
+    {
+      "epoch": 1146.3157894736842,
+      "grad_norm": 0.9102608561515808,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 174240
+    },
+    {
+      "epoch": 1146.3815789473683,
+      "grad_norm": 1.0038232803344727,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 174250
+    },
+    {
+      "epoch": 1146.4473684210527,
+      "grad_norm": 0.7302023768424988,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 174260
+    },
+    {
+      "epoch": 1146.5131578947369,
+      "grad_norm": 1.0826613903045654,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 174270
+    },
+    {
+      "epoch": 1146.578947368421,
+      "grad_norm": 1.095320224761963,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 174280
+    },
+    {
+      "epoch": 1146.6447368421052,
+      "grad_norm": 1.0644302368164062,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 174290
+    },
+    {
+      "epoch": 1146.7105263157894,
+      "grad_norm": 1.1615333557128906,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 174300
+    },
+    {
+      "epoch": 1146.7763157894738,
+      "grad_norm": 1.3274446725845337,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 174310
+    },
+    {
+      "epoch": 1146.842105263158,
+      "grad_norm": 1.1300023794174194,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 174320
+    },
+    {
+      "epoch": 1146.907894736842,
+      "grad_norm": 1.0761096477508545,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 174330
+    },
+    {
+      "epoch": 1146.9736842105262,
+      "grad_norm": 0.9165819883346558,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 174340
+    },
+    {
+      "epoch": 1147.0394736842106,
+      "grad_norm": 0.8142982721328735,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 174350
+    },
+    {
+      "epoch": 1147.1052631578948,
+      "grad_norm": 0.7436504364013672,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 174360
+    },
+    {
+      "epoch": 1147.171052631579,
+      "grad_norm": 0.9768326878547668,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 174370
+    },
+    {
+      "epoch": 1147.2368421052631,
+      "grad_norm": 1.2096675634384155,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 174380
+    },
+    {
+      "epoch": 1147.3026315789473,
+      "grad_norm": 1.1367160081863403,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 174390
+    },
+    {
+      "epoch": 1147.3684210526317,
+      "grad_norm": 0.680374801158905,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 174400
+    },
+    {
+      "epoch": 1147.4342105263158,
+      "grad_norm": 0.9637120962142944,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 174410
+    },
+    {
+      "epoch": 1147.5,
+      "grad_norm": 0.9413122534751892,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 174420
+    },
+    {
+      "epoch": 1147.5657894736842,
+      "grad_norm": 1.165696620941162,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 174430
+    },
+    {
+      "epoch": 1147.6315789473683,
+      "grad_norm": 1.0719096660614014,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 174440
+    },
+    {
+      "epoch": 1147.6973684210527,
+      "grad_norm": 0.9276707768440247,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 174450
+    },
+    {
+      "epoch": 1147.7631578947369,
+      "grad_norm": 0.7522660493850708,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 174460
+    },
+    {
+      "epoch": 1147.828947368421,
+      "grad_norm": 0.9146892428398132,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 174470
+    },
+    {
+      "epoch": 1147.8947368421052,
+      "grad_norm": 1.5214803218841553,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 174480
+    },
+    {
+      "epoch": 1147.9605263157894,
+      "grad_norm": 1.152605652809143,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 174490
+    },
+    {
+      "epoch": 1148.0263157894738,
+      "grad_norm": 1.000752568244934,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 174500
+    },
+    {
+      "epoch": 1148.092105263158,
+      "grad_norm": 1.0302700996398926,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 174510
+    },
+    {
+      "epoch": 1148.157894736842,
+      "grad_norm": 1.0237771272659302,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 174520
+    },
+    {
+      "epoch": 1148.2236842105262,
+      "grad_norm": 1.1548134088516235,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 174530
+    },
+    {
+      "epoch": 1148.2894736842106,
+      "grad_norm": 1.4400047063827515,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 174540
+    },
+    {
+      "epoch": 1148.3552631578948,
+      "grad_norm": 0.9746676087379456,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 174550
+    },
+    {
+      "epoch": 1148.421052631579,
+      "grad_norm": 0.8200980424880981,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 174560
+    },
+    {
+      "epoch": 1148.4868421052631,
+      "grad_norm": 1.2456996440887451,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 174570
+    },
+    {
+      "epoch": 1148.5526315789473,
+      "grad_norm": 1.0017765760421753,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 174580
+    },
+    {
+      "epoch": 1148.6184210526317,
+      "grad_norm": 0.7824232578277588,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 174590
+    },
+    {
+      "epoch": 1148.6842105263158,
+      "grad_norm": 1.0675746202468872,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 174600
+    },
+    {
+      "epoch": 1148.75,
+      "grad_norm": 0.8915340304374695,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 174610
+    },
+    {
+      "epoch": 1148.8157894736842,
+      "grad_norm": 1.1041761636734009,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 174620
+    },
+    {
+      "epoch": 1148.8815789473683,
+      "grad_norm": 0.9389930367469788,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 174630
+    },
+    {
+      "epoch": 1148.9473684210527,
+      "grad_norm": 0.7159057259559631,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 174640
+    },
+    {
+      "epoch": 1149.0131578947369,
+      "grad_norm": 0.8780013918876648,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 174650
+    },
+    {
+      "epoch": 1149.078947368421,
+      "grad_norm": 0.7802185416221619,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 174660
+    },
+    {
+      "epoch": 1149.1447368421052,
+      "grad_norm": 1.1969313621520996,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 174670
+    },
+    {
+      "epoch": 1149.2105263157894,
+      "grad_norm": 1.0634304285049438,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 174680
+    },
+    {
+      "epoch": 1149.2763157894738,
+      "grad_norm": 0.9525381326675415,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 174690
+    },
+    {
+      "epoch": 1149.342105263158,
+      "grad_norm": 0.895878791809082,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 174700
+    },
+    {
+      "epoch": 1149.407894736842,
+      "grad_norm": 0.641148030757904,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 174710
+    },
+    {
+      "epoch": 1149.4736842105262,
+      "grad_norm": 1.0341429710388184,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 174720
+    },
+    {
+      "epoch": 1149.5394736842106,
+      "grad_norm": 0.9922910928726196,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 174730
+    },
+    {
+      "epoch": 1149.6052631578948,
+      "grad_norm": 1.1714662313461304,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 174740
+    },
+    {
+      "epoch": 1149.671052631579,
+      "grad_norm": 2.198077440261841,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 174750
+    },
+    {
+      "epoch": 1149.7368421052631,
+      "grad_norm": 1.9137755632400513,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 174760
+    },
+    {
+      "epoch": 1149.8026315789473,
+      "grad_norm": 1.6982780694961548,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 174770
+    },
+    {
+      "epoch": 1149.8684210526317,
+      "grad_norm": 1.3892414569854736,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 174780
+    },
+    {
+      "epoch": 1149.9342105263158,
+      "grad_norm": 1.2968597412109375,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 174790
+    },
+    {
+      "epoch": 1150.0,
+      "grad_norm": 1.1849431991577148,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 174800
+    },
+    {
+      "epoch": 1150.0657894736842,
+      "grad_norm": 1.133933663368225,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 174810
+    },
+    {
+      "epoch": 1150.1315789473683,
+      "grad_norm": 1.385817050933838,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 174820
+    },
+    {
+      "epoch": 1150.1973684210527,
+      "grad_norm": 1.2556873559951782,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 174830
+    },
+    {
+      "epoch": 1150.2631578947369,
+      "grad_norm": 1.142346978187561,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 174840
+    },
+    {
+      "epoch": 1150.328947368421,
+      "grad_norm": 1.2064703702926636,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 174850
+    },
+    {
+      "epoch": 1150.3947368421052,
+      "grad_norm": 1.0505454540252686,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 174860
+    },
+    {
+      "epoch": 1150.4605263157894,
+      "grad_norm": 1.084917664527893,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 174870
+    },
+    {
+      "epoch": 1150.5263157894738,
+      "grad_norm": 1.289945125579834,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 174880
+    },
+    {
+      "epoch": 1150.592105263158,
+      "grad_norm": 0.8230373859405518,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 174890
+    },
+    {
+      "epoch": 1150.657894736842,
+      "grad_norm": 1.1008986234664917,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 174900
+    },
+    {
+      "epoch": 1150.7236842105262,
+      "grad_norm": 1.0355833768844604,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 174910
+    },
+    {
+      "epoch": 1150.7894736842106,
+      "grad_norm": 1.206417441368103,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 174920
+    },
+    {
+      "epoch": 1150.8552631578948,
+      "grad_norm": 1.393118977546692,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 174930
+    },
+    {
+      "epoch": 1150.921052631579,
+      "grad_norm": 1.0129419565200806,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 174940
+    },
+    {
+      "epoch": 1150.9868421052631,
+      "grad_norm": 0.9136284589767456,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 174950
+    },
+    {
+      "epoch": 1151.0526315789473,
+      "grad_norm": 0.9406630992889404,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 174960
+    },
+    {
+      "epoch": 1151.1184210526317,
+      "grad_norm": 1.0768885612487793,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 174970
+    },
+    {
+      "epoch": 1151.1842105263158,
+      "grad_norm": 0.9179930686950684,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 174980
+    },
+    {
+      "epoch": 1151.25,
+      "grad_norm": 0.6786910891532898,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 174990
+    },
+    {
+      "epoch": 1151.3157894736842,
+      "grad_norm": 1.1306889057159424,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 175000
+    },
+    {
+      "epoch": 1151.3815789473683,
+      "grad_norm": 1.0117748975753784,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 175010
+    },
+    {
+      "epoch": 1151.4473684210527,
+      "grad_norm": 0.661830484867096,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 175020
+    },
+    {
+      "epoch": 1151.5131578947369,
+      "grad_norm": 1.0096147060394287,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 175030
+    },
+    {
+      "epoch": 1151.578947368421,
+      "grad_norm": 1.1573773622512817,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 175040
+    },
+    {
+      "epoch": 1151.6447368421052,
+      "grad_norm": 1.0948268175125122,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 175050
+    },
+    {
+      "epoch": 1151.7105263157894,
+      "grad_norm": 0.9008634090423584,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 175060
+    },
+    {
+      "epoch": 1151.7763157894738,
+      "grad_norm": 0.8874291181564331,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 175070
+    },
+    {
+      "epoch": 1151.842105263158,
+      "grad_norm": 1.33906090259552,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 175080
+    },
+    {
+      "epoch": 1151.907894736842,
+      "grad_norm": 1.3272651433944702,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 175090
+    },
+    {
+      "epoch": 1151.9736842105262,
+      "grad_norm": 1.1627230644226074,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 175100
+    },
+    {
+      "epoch": 1152.0394736842106,
+      "grad_norm": 0.9578849077224731,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 175110
+    },
+    {
+      "epoch": 1152.1052631578948,
+      "grad_norm": 1.016489863395691,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 175120
+    },
+    {
+      "epoch": 1152.171052631579,
+      "grad_norm": 0.9531344771385193,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 175130
+    },
+    {
+      "epoch": 1152.2368421052631,
+      "grad_norm": 0.9982447028160095,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 175140
+    },
+    {
+      "epoch": 1152.3026315789473,
+      "grad_norm": 1.3460252285003662,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 175150
+    },
+    {
+      "epoch": 1152.3684210526317,
+      "grad_norm": 1.0662846565246582,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 175160
+    },
+    {
+      "epoch": 1152.4342105263158,
+      "grad_norm": 0.7010901570320129,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 175170
+    },
+    {
+      "epoch": 1152.5,
+      "grad_norm": 0.8454627990722656,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 175180
+    },
+    {
+      "epoch": 1152.5657894736842,
+      "grad_norm": 0.858896017074585,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 175190
+    },
+    {
+      "epoch": 1152.6315789473683,
+      "grad_norm": 1.0066238641738892,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 175200
+    },
+    {
+      "epoch": 1152.6973684210527,
+      "grad_norm": 1.1515685319900513,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 175210
+    },
+    {
+      "epoch": 1152.7631578947369,
+      "grad_norm": 0.8878205418586731,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 175220
+    },
+    {
+      "epoch": 1152.828947368421,
+      "grad_norm": 1.0133531093597412,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 175230
+    },
+    {
+      "epoch": 1152.8947368421052,
+      "grad_norm": 1.2004742622375488,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 175240
+    },
+    {
+      "epoch": 1152.9605263157894,
+      "grad_norm": 0.8571202754974365,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 175250
+    },
+    {
+      "epoch": 1153.0263157894738,
+      "grad_norm": 1.2629226446151733,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 175260
+    },
+    {
+      "epoch": 1153.092105263158,
+      "grad_norm": 1.3697537183761597,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 175270
+    },
+    {
+      "epoch": 1153.157894736842,
+      "grad_norm": 1.2100975513458252,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 175280
+    },
+    {
+      "epoch": 1153.2236842105262,
+      "grad_norm": 1.1854854822158813,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 175290
+    },
+    {
+      "epoch": 1153.2894736842106,
+      "grad_norm": 0.7720757126808167,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 175300
+    },
+    {
+      "epoch": 1153.3552631578948,
+      "grad_norm": 1.1668815612792969,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 175310
+    },
+    {
+      "epoch": 1153.421052631579,
+      "grad_norm": 1.0737518072128296,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 175320
+    },
+    {
+      "epoch": 1153.4868421052631,
+      "grad_norm": 1.1186960935592651,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 175330
+    },
+    {
+      "epoch": 1153.5526315789473,
+      "grad_norm": 0.9653819799423218,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 175340
+    },
+    {
+      "epoch": 1153.6184210526317,
+      "grad_norm": 0.9481167793273926,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 175350
+    },
+    {
+      "epoch": 1153.6842105263158,
+      "grad_norm": 1.0212024450302124,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 175360
+    },
+    {
+      "epoch": 1153.75,
+      "grad_norm": 1.3355201482772827,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 175370
+    },
+    {
+      "epoch": 1153.8157894736842,
+      "grad_norm": 0.727803111076355,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 175380
+    },
+    {
+      "epoch": 1153.8815789473683,
+      "grad_norm": 0.894546627998352,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 175390
+    },
+    {
+      "epoch": 1153.9473684210527,
+      "grad_norm": 1.0446175336837769,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 175400
+    },
+    {
+      "epoch": 1154.0131578947369,
+      "grad_norm": 1.3837840557098389,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 175410
+    },
+    {
+      "epoch": 1154.078947368421,
+      "grad_norm": 1.131395936012268,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 175420
+    },
+    {
+      "epoch": 1154.1447368421052,
+      "grad_norm": 1.1270420551300049,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 175430
+    },
+    {
+      "epoch": 1154.2105263157894,
+      "grad_norm": 0.8780666589736938,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 175440
+    },
+    {
+      "epoch": 1154.2763157894738,
+      "grad_norm": 1.054305911064148,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 175450
+    },
+    {
+      "epoch": 1154.342105263158,
+      "grad_norm": 0.7585221529006958,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 175460
+    },
+    {
+      "epoch": 1154.407894736842,
+      "grad_norm": 0.8007470965385437,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 175470
+    },
+    {
+      "epoch": 1154.4736842105262,
+      "grad_norm": 1.0967533588409424,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 175480
+    },
+    {
+      "epoch": 1154.5394736842106,
+      "grad_norm": 1.0877718925476074,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 175490
+    },
+    {
+      "epoch": 1154.6052631578948,
+      "grad_norm": 0.965691864490509,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 175500
+    },
+    {
+      "epoch": 1154.671052631579,
+      "grad_norm": 0.9504064917564392,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 175510
+    },
+    {
+      "epoch": 1154.7368421052631,
+      "grad_norm": 1.0463320016860962,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 175520
+    },
+    {
+      "epoch": 1154.8026315789473,
+      "grad_norm": 1.1356477737426758,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 175530
+    },
+    {
+      "epoch": 1154.8684210526317,
+      "grad_norm": 0.937968909740448,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 175540
+    },
+    {
+      "epoch": 1154.9342105263158,
+      "grad_norm": 1.2012879848480225,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 175550
+    },
+    {
+      "epoch": 1155.0,
+      "grad_norm": 1.171942949295044,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 175560
+    },
+    {
+      "epoch": 1155.0657894736842,
+      "grad_norm": 0.92462557554245,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 175570
+    },
+    {
+      "epoch": 1155.1315789473683,
+      "grad_norm": 1.2336483001708984,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 175580
+    },
+    {
+      "epoch": 1155.1973684210527,
+      "grad_norm": 1.019647240638733,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 175590
+    },
+    {
+      "epoch": 1155.2631578947369,
+      "grad_norm": 1.0296791791915894,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 175600
+    },
+    {
+      "epoch": 1155.328947368421,
+      "grad_norm": 1.081726312637329,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 175610
+    },
+    {
+      "epoch": 1155.3947368421052,
+      "grad_norm": 0.6053563356399536,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 175620
+    },
+    {
+      "epoch": 1155.4605263157894,
+      "grad_norm": 0.908021867275238,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 175630
+    },
+    {
+      "epoch": 1155.5263157894738,
+      "grad_norm": 0.8451210856437683,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 175640
+    },
+    {
+      "epoch": 1155.592105263158,
+      "grad_norm": 1.068265676498413,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 175650
+    },
+    {
+      "epoch": 1155.657894736842,
+      "grad_norm": 1.213500738143921,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 175660
+    },
+    {
+      "epoch": 1155.7236842105262,
+      "grad_norm": 0.8982007503509521,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 175670
+    },
+    {
+      "epoch": 1155.7894736842106,
+      "grad_norm": 1.0422526597976685,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 175680
+    },
+    {
+      "epoch": 1155.8552631578948,
+      "grad_norm": 0.9385272264480591,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 175690
+    },
+    {
+      "epoch": 1155.921052631579,
+      "grad_norm": 0.9605172872543335,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 175700
+    },
+    {
+      "epoch": 1155.9868421052631,
+      "grad_norm": 1.1489660739898682,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 175710
+    },
+    {
+      "epoch": 1156.0526315789473,
+      "grad_norm": 1.1852326393127441,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 175720
+    },
+    {
+      "epoch": 1156.1184210526317,
+      "grad_norm": 1.5356533527374268,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 175730
+    },
+    {
+      "epoch": 1156.1842105263158,
+      "grad_norm": 0.974061131477356,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 175740
+    },
+    {
+      "epoch": 1156.25,
+      "grad_norm": 1.0487468242645264,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 175750
+    },
+    {
+      "epoch": 1156.3157894736842,
+      "grad_norm": 1.5691155195236206,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 175760
+    },
+    {
+      "epoch": 1156.3815789473683,
+      "grad_norm": 1.4372376203536987,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 175770
+    },
+    {
+      "epoch": 1156.4473684210527,
+      "grad_norm": 1.508100152015686,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 175780
+    },
+    {
+      "epoch": 1156.5131578947369,
+      "grad_norm": 1.0872918367385864,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 175790
+    },
+    {
+      "epoch": 1156.578947368421,
+      "grad_norm": 1.0140292644500732,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 175800
+    },
+    {
+      "epoch": 1156.6447368421052,
+      "grad_norm": 0.9516561031341553,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 175810
+    },
+    {
+      "epoch": 1156.7105263157894,
+      "grad_norm": 1.1463650465011597,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 175820
+    },
+    {
+      "epoch": 1156.7763157894738,
+      "grad_norm": 1.2068666219711304,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 175830
+    },
+    {
+      "epoch": 1156.842105263158,
+      "grad_norm": 0.8994778394699097,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 175840
+    },
+    {
+      "epoch": 1156.907894736842,
+      "grad_norm": 1.115564227104187,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 175850
+    },
+    {
+      "epoch": 1156.9736842105262,
+      "grad_norm": 1.5093863010406494,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 175860
+    },
+    {
+      "epoch": 1157.0394736842106,
+      "grad_norm": 1.3060439825057983,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 175870
+    },
+    {
+      "epoch": 1157.1052631578948,
+      "grad_norm": 1.0380960702896118,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 175880
+    },
+    {
+      "epoch": 1157.171052631579,
+      "grad_norm": 1.381862998008728,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 175890
+    },
+    {
+      "epoch": 1157.2368421052631,
+      "grad_norm": 0.9802920818328857,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 175900
+    },
+    {
+      "epoch": 1157.3026315789473,
+      "grad_norm": 1.1407344341278076,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 175910
+    },
+    {
+      "epoch": 1157.3684210526317,
+      "grad_norm": 1.3060024976730347,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 175920
+    },
+    {
+      "epoch": 1157.4342105263158,
+      "grad_norm": 1.3562568426132202,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 175930
+    },
+    {
+      "epoch": 1157.5,
+      "grad_norm": 0.958991527557373,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 175940
+    },
+    {
+      "epoch": 1157.5657894736842,
+      "grad_norm": 1.1298139095306396,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 175950
+    },
+    {
+      "epoch": 1157.6315789473683,
+      "grad_norm": 1.4933613538742065,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 175960
+    },
+    {
+      "epoch": 1157.6973684210527,
+      "grad_norm": 1.4154495000839233,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 175970
+    },
+    {
+      "epoch": 1157.7631578947369,
+      "grad_norm": 1.487290382385254,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 175980
+    },
+    {
+      "epoch": 1157.828947368421,
+      "grad_norm": 1.0696330070495605,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 175990
+    },
+    {
+      "epoch": 1157.8947368421052,
+      "grad_norm": 1.3961727619171143,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 176000
+    },
+    {
+      "epoch": 1157.9605263157894,
+      "grad_norm": 0.8423775434494019,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 176010
+    },
+    {
+      "epoch": 1158.0263157894738,
+      "grad_norm": 1.1189703941345215,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 176020
+    },
+    {
+      "epoch": 1158.092105263158,
+      "grad_norm": 1.04086434841156,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 176030
+    },
+    {
+      "epoch": 1158.157894736842,
+      "grad_norm": 0.8030532002449036,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 176040
+    },
+    {
+      "epoch": 1158.2236842105262,
+      "grad_norm": 1.1623908281326294,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 176050
+    },
+    {
+      "epoch": 1158.2894736842106,
+      "grad_norm": 0.7452375292778015,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 176060
+    },
+    {
+      "epoch": 1158.3552631578948,
+      "grad_norm": 0.625102162361145,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 176070
+    },
+    {
+      "epoch": 1158.421052631579,
+      "grad_norm": 0.6674894690513611,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 176080
+    },
+    {
+      "epoch": 1158.4868421052631,
+      "grad_norm": 0.7138943076133728,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 176090
+    },
+    {
+      "epoch": 1158.5526315789473,
+      "grad_norm": 0.9019205570220947,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 176100
+    },
+    {
+      "epoch": 1158.6184210526317,
+      "grad_norm": 1.1200653314590454,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 176110
+    },
+    {
+      "epoch": 1158.6842105263158,
+      "grad_norm": 1.2586793899536133,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 176120
+    },
+    {
+      "epoch": 1158.75,
+      "grad_norm": 0.7260456085205078,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 176130
+    },
+    {
+      "epoch": 1158.8157894736842,
+      "grad_norm": 1.0455162525177002,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 176140
+    },
+    {
+      "epoch": 1158.8815789473683,
+      "grad_norm": 1.5511831045150757,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 176150
+    },
+    {
+      "epoch": 1158.9473684210527,
+      "grad_norm": 0.8422067165374756,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 176160
+    },
+    {
+      "epoch": 1159.0131578947369,
+      "grad_norm": 1.0608569383621216,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 176170
+    },
+    {
+      "epoch": 1159.078947368421,
+      "grad_norm": 0.9599637985229492,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 176180
+    },
+    {
+      "epoch": 1159.1447368421052,
+      "grad_norm": 1.0301839113235474,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 176190
+    },
+    {
+      "epoch": 1159.2105263157894,
+      "grad_norm": 1.3413852453231812,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 176200
+    },
+    {
+      "epoch": 1159.2763157894738,
+      "grad_norm": 1.2173924446105957,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 176210
+    },
+    {
+      "epoch": 1159.342105263158,
+      "grad_norm": 1.487339735031128,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 176220
+    },
+    {
+      "epoch": 1159.407894736842,
+      "grad_norm": 1.342084527015686,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 176230
+    },
+    {
+      "epoch": 1159.4736842105262,
+      "grad_norm": 1.627140998840332,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 176240
+    },
+    {
+      "epoch": 1159.5394736842106,
+      "grad_norm": 1.2196797132492065,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 176250
+    },
+    {
+      "epoch": 1159.6052631578948,
+      "grad_norm": 1.0979526042938232,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 176260
+    },
+    {
+      "epoch": 1159.671052631579,
+      "grad_norm": 1.0794157981872559,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 176270
+    },
+    {
+      "epoch": 1159.7368421052631,
+      "grad_norm": 1.344915747642517,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 176280
+    },
+    {
+      "epoch": 1159.8026315789473,
+      "grad_norm": 0.9110754728317261,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 176290
+    },
+    {
+      "epoch": 1159.8684210526317,
+      "grad_norm": 0.9595354795455933,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 176300
+    },
+    {
+      "epoch": 1159.9342105263158,
+      "grad_norm": 1.3430472612380981,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 176310
+    },
+    {
+      "epoch": 1160.0,
+      "grad_norm": 0.8704079985618591,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 176320
+    },
+    {
+      "epoch": 1160.0657894736842,
+      "grad_norm": 1.069422721862793,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 176330
+    },
+    {
+      "epoch": 1160.1315789473683,
+      "grad_norm": 1.1828012466430664,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 176340
+    },
+    {
+      "epoch": 1160.1973684210527,
+      "grad_norm": 1.0887290239334106,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 176350
+    },
+    {
+      "epoch": 1160.2631578947369,
+      "grad_norm": 0.9942824244499207,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 176360
+    },
+    {
+      "epoch": 1160.328947368421,
+      "grad_norm": 0.9858255982398987,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 176370
+    },
+    {
+      "epoch": 1160.3947368421052,
+      "grad_norm": 0.7866831421852112,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 176380
+    },
+    {
+      "epoch": 1160.4605263157894,
+      "grad_norm": 0.8747726082801819,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 176390
+    },
+    {
+      "epoch": 1160.5263157894738,
+      "grad_norm": 1.004457712173462,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 176400
+    },
+    {
+      "epoch": 1160.592105263158,
+      "grad_norm": 0.9716295599937439,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 176410
+    },
+    {
+      "epoch": 1160.657894736842,
+      "grad_norm": 1.2135223150253296,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 176420
+    },
+    {
+      "epoch": 1160.7236842105262,
+      "grad_norm": 1.085508942604065,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 176430
+    },
+    {
+      "epoch": 1160.7894736842106,
+      "grad_norm": 1.4997029304504395,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 176440
+    },
+    {
+      "epoch": 1160.8552631578948,
+      "grad_norm": 1.238146185874939,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 176450
+    },
+    {
+      "epoch": 1160.921052631579,
+      "grad_norm": 0.9297826290130615,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 176460
+    },
+    {
+      "epoch": 1160.9868421052631,
+      "grad_norm": 0.8368382453918457,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 176470
+    },
+    {
+      "epoch": 1161.0526315789473,
+      "grad_norm": 0.933539092540741,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 176480
+    },
+    {
+      "epoch": 1161.1184210526317,
+      "grad_norm": 1.1759393215179443,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 176490
+    },
+    {
+      "epoch": 1161.1842105263158,
+      "grad_norm": 1.2678519487380981,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 176500
+    },
+    {
+      "epoch": 1161.25,
+      "grad_norm": 1.227339506149292,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 176510
+    },
+    {
+      "epoch": 1161.3157894736842,
+      "grad_norm": 1.1603562831878662,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 176520
+    },
+    {
+      "epoch": 1161.3815789473683,
+      "grad_norm": 1.2368062734603882,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 176530
+    },
+    {
+      "epoch": 1161.4473684210527,
+      "grad_norm": 1.3081183433532715,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 176540
+    },
+    {
+      "epoch": 1161.5131578947369,
+      "grad_norm": 1.4025623798370361,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 176550
+    },
+    {
+      "epoch": 1161.578947368421,
+      "grad_norm": 1.1546919345855713,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "step": 176560
+    },
+    {
+      "epoch": 1161.6447368421052,
+      "grad_norm": 0.9056599736213684,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 176570
+    },
+    {
+      "epoch": 1161.7105263157894,
+      "grad_norm": 1.22064208984375,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 176580
+    },
+    {
+      "epoch": 1161.7763157894738,
+      "grad_norm": 0.9950551390647888,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 176590
+    },
+    {
+      "epoch": 1161.842105263158,
+      "grad_norm": 1.0174490213394165,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 176600
+    },
+    {
+      "epoch": 1161.907894736842,
+      "grad_norm": 1.21022629737854,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 176610
+    },
+    {
+      "epoch": 1161.9736842105262,
+      "grad_norm": 0.6943854093551636,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 176620
+    },
+    {
+      "epoch": 1162.0394736842106,
+      "grad_norm": 1.3050177097320557,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 176630
+    },
+    {
+      "epoch": 1162.1052631578948,
+      "grad_norm": 1.040520429611206,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 176640
+    },
+    {
+      "epoch": 1162.171052631579,
+      "grad_norm": 1.3651999235153198,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 176650
+    },
+    {
+      "epoch": 1162.2368421052631,
+      "grad_norm": 1.1482195854187012,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 176660
+    },
+    {
+      "epoch": 1162.3026315789473,
+      "grad_norm": 1.137037754058838,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 176670
+    },
+    {
+      "epoch": 1162.3684210526317,
+      "grad_norm": 1.381179690361023,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "step": 176680
+    },
+    {
+      "epoch": 1162.4342105263158,
+      "grad_norm": 1.0104210376739502,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 176690
+    },
+    {
+      "epoch": 1162.5,
+      "grad_norm": 1.2539334297180176,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 176700
+    },
+    {
+      "epoch": 1162.5657894736842,
+      "grad_norm": 1.0029774904251099,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 176710
+    },
+    {
+      "epoch": 1162.6315789473683,
+      "grad_norm": 0.8852845430374146,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 176720
+    },
+    {
+      "epoch": 1162.6973684210527,
+      "grad_norm": 1.0239177942276,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 176730
+    },
+    {
+      "epoch": 1162.7631578947369,
+      "grad_norm": 0.8226960897445679,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 176740
+    },
+    {
+      "epoch": 1162.828947368421,
+      "grad_norm": 0.9960774779319763,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 176750
+    },
+    {
+      "epoch": 1162.8947368421052,
+      "grad_norm": 1.1968601942062378,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 176760
+    },
+    {
+      "epoch": 1162.9605263157894,
+      "grad_norm": 1.1123733520507812,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 176770
+    },
+    {
+      "epoch": 1163.0263157894738,
+      "grad_norm": 1.0978655815124512,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 176780
+    },
+    {
+      "epoch": 1163.092105263158,
+      "grad_norm": 0.8756230473518372,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 176790
+    },
+    {
+      "epoch": 1163.157894736842,
+      "grad_norm": 1.0591200590133667,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 176800
+    },
+    {
+      "epoch": 1163.2236842105262,
+      "grad_norm": 0.7175271511077881,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 176810
+    },
+    {
+      "epoch": 1163.2894736842106,
+      "grad_norm": 1.0549403429031372,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 176820
+    },
+    {
+      "epoch": 1163.3552631578948,
+      "grad_norm": 1.0469660758972168,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 176830
+    },
+    {
+      "epoch": 1163.421052631579,
+      "grad_norm": 0.8197566270828247,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 176840
+    },
+    {
+      "epoch": 1163.4868421052631,
+      "grad_norm": 1.2558757066726685,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 176850
+    },
+    {
+      "epoch": 1163.5526315789473,
+      "grad_norm": 1.1719180345535278,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 176860
+    },
+    {
+      "epoch": 1163.6184210526317,
+      "grad_norm": 0.817057728767395,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 176870
+    },
+    {
+      "epoch": 1163.6842105263158,
+      "grad_norm": 1.178382158279419,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 176880
+    },
+    {
+      "epoch": 1163.75,
+      "grad_norm": 1.0112900733947754,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 176890
+    },
+    {
+      "epoch": 1163.8157894736842,
+      "grad_norm": 0.7621281743049622,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 176900
+    },
+    {
+      "epoch": 1163.8815789473683,
+      "grad_norm": 0.6787197589874268,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 176910
+    },
+    {
+      "epoch": 1163.9473684210527,
+      "grad_norm": 1.0531107187271118,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 176920
+    },
+    {
+      "epoch": 1164.0131578947369,
+      "grad_norm": 0.5585743188858032,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 176930
+    },
+    {
+      "epoch": 1164.078947368421,
+      "grad_norm": 0.8682098388671875,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 176940
+    },
+    {
+      "epoch": 1164.1447368421052,
+      "grad_norm": 0.7695916295051575,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 176950
+    },
+    {
+      "epoch": 1164.2105263157894,
+      "grad_norm": 1.2778445482254028,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 176960
+    },
+    {
+      "epoch": 1164.2763157894738,
+      "grad_norm": 1.0614190101623535,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 176970
+    },
+    {
+      "epoch": 1164.342105263158,
+      "grad_norm": 0.7645097374916077,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 176980
+    },
+    {
+      "epoch": 1164.407894736842,
+      "grad_norm": 1.0178488492965698,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 176990
+    },
+    {
+      "epoch": 1164.4736842105262,
+      "grad_norm": 1.3799580335617065,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 177000
+    },
+    {
+      "epoch": 1164.5394736842106,
+      "grad_norm": 1.1391440629959106,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 177010
+    },
+    {
+      "epoch": 1164.6052631578948,
+      "grad_norm": 1.0461714267730713,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 177020
+    },
+    {
+      "epoch": 1164.671052631579,
+      "grad_norm": 1.119397521018982,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 177030
+    },
+    {
+      "epoch": 1164.7368421052631,
+      "grad_norm": 1.3247554302215576,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 177040
+    },
+    {
+      "epoch": 1164.8026315789473,
+      "grad_norm": 0.9581509828567505,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 177050
+    },
+    {
+      "epoch": 1164.8684210526317,
+      "grad_norm": 1.0519541501998901,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 177060
+    },
+    {
+      "epoch": 1164.9342105263158,
+      "grad_norm": 0.7811246514320374,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 177070
+    },
+    {
+      "epoch": 1165.0,
+      "grad_norm": 0.8886945843696594,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 177080
+    },
+    {
+      "epoch": 1165.0657894736842,
+      "grad_norm": 1.0160313844680786,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 177090
+    },
+    {
+      "epoch": 1165.1315789473683,
+      "grad_norm": 1.3136074542999268,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 177100
+    },
+    {
+      "epoch": 1165.1973684210527,
+      "grad_norm": 1.1090233325958252,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 177110
+    },
+    {
+      "epoch": 1165.2631578947369,
+      "grad_norm": 1.0425786972045898,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 177120
+    },
+    {
+      "epoch": 1165.328947368421,
+      "grad_norm": 1.0848468542099,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 177130
+    },
+    {
+      "epoch": 1165.3947368421052,
+      "grad_norm": 0.8333765268325806,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 177140
+    },
+    {
+      "epoch": 1165.4605263157894,
+      "grad_norm": 0.8007072806358337,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 177150
+    },
+    {
+      "epoch": 1165.5263157894738,
+      "grad_norm": 1.0657711029052734,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 177160
+    },
+    {
+      "epoch": 1165.592105263158,
+      "grad_norm": 0.8942683339118958,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 177170
+    },
+    {
+      "epoch": 1165.657894736842,
+      "grad_norm": 0.7064841985702515,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 177180
+    },
+    {
+      "epoch": 1165.7236842105262,
+      "grad_norm": 0.673353910446167,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 177190
+    },
+    {
+      "epoch": 1165.7894736842106,
+      "grad_norm": 1.201241135597229,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 177200
+    },
+    {
+      "epoch": 1165.8552631578948,
+      "grad_norm": 1.1128406524658203,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 177210
+    },
+    {
+      "epoch": 1165.921052631579,
+      "grad_norm": 1.0577025413513184,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 177220
+    },
+    {
+      "epoch": 1165.9868421052631,
+      "grad_norm": 1.063428282737732,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 177230
+    },
+    {
+      "epoch": 1166.0526315789473,
+      "grad_norm": 0.6475265622138977,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 177240
+    },
+    {
+      "epoch": 1166.1184210526317,
+      "grad_norm": 1.0673624277114868,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 177250
+    },
+    {
+      "epoch": 1166.1842105263158,
+      "grad_norm": 1.2042251825332642,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 177260
+    },
+    {
+      "epoch": 1166.25,
+      "grad_norm": 1.023403286933899,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 177270
+    },
+    {
+      "epoch": 1166.3157894736842,
+      "grad_norm": 1.0554587841033936,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 177280
+    },
+    {
+      "epoch": 1166.3815789473683,
+      "grad_norm": 1.2385607957839966,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 177290
+    },
+    {
+      "epoch": 1166.4473684210527,
+      "grad_norm": 1.058170199394226,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 177300
+    },
+    {
+      "epoch": 1166.5131578947369,
+      "grad_norm": 0.7787966728210449,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 177310
+    },
+    {
+      "epoch": 1166.578947368421,
+      "grad_norm": 0.8278951048851013,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 177320
+    },
+    {
+      "epoch": 1166.6447368421052,
+      "grad_norm": 0.945210337638855,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 177330
+    },
+    {
+      "epoch": 1166.7105263157894,
+      "grad_norm": 0.8070635795593262,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 177340
+    },
+    {
+      "epoch": 1166.7763157894738,
+      "grad_norm": 1.1100777387619019,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 177350
+    },
+    {
+      "epoch": 1166.842105263158,
+      "grad_norm": 1.3282232284545898,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 177360
+    },
+    {
+      "epoch": 1166.907894736842,
+      "grad_norm": 1.6687939167022705,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 177370
+    },
+    {
+      "epoch": 1166.9736842105262,
+      "grad_norm": 1.4565303325653076,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 177380
+    },
+    {
+      "epoch": 1167.0394736842106,
+      "grad_norm": 0.7736444473266602,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 177390
+    },
+    {
+      "epoch": 1167.1052631578948,
+      "grad_norm": 1.261600375175476,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 177400
+    },
+    {
+      "epoch": 1167.171052631579,
+      "grad_norm": 1.2047302722930908,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 177410
+    },
+    {
+      "epoch": 1167.2368421052631,
+      "grad_norm": 1.3666282892227173,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 177420
+    },
+    {
+      "epoch": 1167.3026315789473,
+      "grad_norm": 1.41166353225708,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 177430
+    },
+    {
+      "epoch": 1167.3684210526317,
+      "grad_norm": 0.9169084429740906,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 177440
+    },
+    {
+      "epoch": 1167.4342105263158,
+      "grad_norm": 1.0971057415008545,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 177450
+    },
+    {
+      "epoch": 1167.5,
+      "grad_norm": 1.0250972509384155,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 177460
+    },
+    {
+      "epoch": 1167.5657894736842,
+      "grad_norm": 1.008253812789917,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 177470
+    },
+    {
+      "epoch": 1167.6315789473683,
+      "grad_norm": 1.0217920541763306,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 177480
+    },
+    {
+      "epoch": 1167.6973684210527,
+      "grad_norm": 1.0006431341171265,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 177490
+    },
+    {
+      "epoch": 1167.7631578947369,
+      "grad_norm": 0.8362782001495361,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 177500
+    },
+    {
+      "epoch": 1167.828947368421,
+      "grad_norm": 1.3864368200302124,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 177510
+    },
+    {
+      "epoch": 1167.8947368421052,
+      "grad_norm": 1.3755213022232056,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 177520
+    },
+    {
+      "epoch": 1167.9605263157894,
+      "grad_norm": 1.060314416885376,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 177530
+    },
+    {
+      "epoch": 1168.0263157894738,
+      "grad_norm": 1.2582261562347412,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 177540
+    },
+    {
+      "epoch": 1168.092105263158,
+      "grad_norm": 1.0791796445846558,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 177550
+    },
+    {
+      "epoch": 1168.157894736842,
+      "grad_norm": 1.2090390920639038,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 177560
+    },
+    {
+      "epoch": 1168.2236842105262,
+      "grad_norm": 0.7330916523933411,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 177570
+    },
+    {
+      "epoch": 1168.2894736842106,
+      "grad_norm": 0.9363530874252319,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 177580
+    },
+    {
+      "epoch": 1168.3552631578948,
+      "grad_norm": 0.9685211777687073,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 177590
+    },
+    {
+      "epoch": 1168.421052631579,
+      "grad_norm": 1.0913938283920288,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 177600
+    },
+    {
+      "epoch": 1168.4868421052631,
+      "grad_norm": 1.035131812095642,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 177610
+    },
+    {
+      "epoch": 1168.5526315789473,
+      "grad_norm": 1.0176951885223389,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 177620
+    },
+    {
+      "epoch": 1168.6184210526317,
+      "grad_norm": 0.8525509834289551,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 177630
+    },
+    {
+      "epoch": 1168.6842105263158,
+      "grad_norm": 1.0058506727218628,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 177640
+    },
+    {
+      "epoch": 1168.75,
+      "grad_norm": 1.1106820106506348,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 177650
+    },
+    {
+      "epoch": 1168.8157894736842,
+      "grad_norm": 0.9219995737075806,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 177660
+    },
+    {
+      "epoch": 1168.8815789473683,
+      "grad_norm": 0.9162684082984924,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 177670
+    },
+    {
+      "epoch": 1168.9473684210527,
+      "grad_norm": 1.331979513168335,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 177680
+    },
+    {
+      "epoch": 1169.0131578947369,
+      "grad_norm": 0.8545548915863037,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 177690
+    },
+    {
+      "epoch": 1169.078947368421,
+      "grad_norm": 0.989903450012207,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 177700
+    },
+    {
+      "epoch": 1169.1447368421052,
+      "grad_norm": 0.9872443079948425,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 177710
+    },
+    {
+      "epoch": 1169.2105263157894,
+      "grad_norm": 1.1131963729858398,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 177720
+    },
+    {
+      "epoch": 1169.2763157894738,
+      "grad_norm": 0.7441876530647278,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 177730
+    },
+    {
+      "epoch": 1169.342105263158,
+      "grad_norm": 1.1648389101028442,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 177740
+    },
+    {
+      "epoch": 1169.407894736842,
+      "grad_norm": 1.0851612091064453,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 177750
+    },
+    {
+      "epoch": 1169.4736842105262,
+      "grad_norm": 0.9316413402557373,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 177760
+    },
+    {
+      "epoch": 1169.5394736842106,
+      "grad_norm": 0.8847388029098511,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 177770
+    },
+    {
+      "epoch": 1169.6052631578948,
+      "grad_norm": 1.1352200508117676,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 177780
+    },
+    {
+      "epoch": 1169.671052631579,
+      "grad_norm": 0.9478182792663574,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 177790
+    },
+    {
+      "epoch": 1169.7368421052631,
+      "grad_norm": 0.7659247517585754,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 177800
+    },
+    {
+      "epoch": 1169.8026315789473,
+      "grad_norm": 0.7848868370056152,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 177810
+    },
+    {
+      "epoch": 1169.8684210526317,
+      "grad_norm": 1.0635731220245361,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 177820
+    },
+    {
+      "epoch": 1169.9342105263158,
+      "grad_norm": 1.3124653100967407,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 177830
+    },
+    {
+      "epoch": 1170.0,
+      "grad_norm": 0.9985669255256653,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 177840
+    },
+    {
+      "epoch": 1170.0657894736842,
+      "grad_norm": 1.075380563735962,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 177850
+    },
+    {
+      "epoch": 1170.1315789473683,
+      "grad_norm": 0.7401938438415527,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 177860
+    },
+    {
+      "epoch": 1170.1973684210527,
+      "grad_norm": 1.1559667587280273,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 177870
+    },
+    {
+      "epoch": 1170.2631578947369,
+      "grad_norm": 1.152563452720642,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 177880
+    },
+    {
+      "epoch": 1170.328947368421,
+      "grad_norm": 0.8326547145843506,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 177890
+    },
+    {
+      "epoch": 1170.3947368421052,
+      "grad_norm": 1.1279677152633667,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 177900
+    },
+    {
+      "epoch": 1170.4605263157894,
+      "grad_norm": 1.192924976348877,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 177910
+    },
+    {
+      "epoch": 1170.5263157894738,
+      "grad_norm": 1.3840651512145996,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 177920
+    },
+    {
+      "epoch": 1170.592105263158,
+      "grad_norm": 1.016034483909607,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 177930
+    },
+    {
+      "epoch": 1170.657894736842,
+      "grad_norm": 1.2145895957946777,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 177940
+    },
+    {
+      "epoch": 1170.7236842105262,
+      "grad_norm": 0.9925963878631592,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 177950
+    },
+    {
+      "epoch": 1170.7894736842106,
+      "grad_norm": 0.9628350734710693,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 177960
+    },
+    {
+      "epoch": 1170.8552631578948,
+      "grad_norm": 0.8890961408615112,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 177970
+    },
+    {
+      "epoch": 1170.921052631579,
+      "grad_norm": 0.8459914922714233,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 177980
+    },
+    {
+      "epoch": 1170.9868421052631,
+      "grad_norm": 0.828228235244751,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 177990
+    },
+    {
+      "epoch": 1171.0526315789473,
+      "grad_norm": 0.8761423230171204,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 178000
+    },
+    {
+      "epoch": 1171.1184210526317,
+      "grad_norm": 0.9206410050392151,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 178010
+    },
+    {
+      "epoch": 1171.1842105263158,
+      "grad_norm": 0.7937294840812683,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 178020
+    },
+    {
+      "epoch": 1171.25,
+      "grad_norm": 0.9361086487770081,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178030
+    },
+    {
+      "epoch": 1171.3157894736842,
+      "grad_norm": 1.000895380973816,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 178040
+    },
+    {
+      "epoch": 1171.3815789473683,
+      "grad_norm": 1.2588021755218506,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 178050
+    },
+    {
+      "epoch": 1171.4473684210527,
+      "grad_norm": 1.0151118040084839,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 178060
+    },
+    {
+      "epoch": 1171.5131578947369,
+      "grad_norm": 1.0695226192474365,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 178070
+    },
+    {
+      "epoch": 1171.578947368421,
+      "grad_norm": 0.9043155908584595,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 178080
+    },
+    {
+      "epoch": 1171.6447368421052,
+      "grad_norm": 0.9957794547080994,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 178090
+    },
+    {
+      "epoch": 1171.7105263157894,
+      "grad_norm": 0.9040594100952148,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 178100
+    },
+    {
+      "epoch": 1171.7763157894738,
+      "grad_norm": 0.9921096563339233,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 178110
+    },
+    {
+      "epoch": 1171.842105263158,
+      "grad_norm": 1.4011787176132202,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 178120
+    },
+    {
+      "epoch": 1171.907894736842,
+      "grad_norm": 1.1821715831756592,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 178130
+    },
+    {
+      "epoch": 1171.9736842105262,
+      "grad_norm": 1.4412405490875244,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 178140
+    },
+    {
+      "epoch": 1172.0394736842106,
+      "grad_norm": 1.3233447074890137,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 178150
+    },
+    {
+      "epoch": 1172.1052631578948,
+      "grad_norm": 1.3075172901153564,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 178160
+    },
+    {
+      "epoch": 1172.171052631579,
+      "grad_norm": 1.162680983543396,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178170
+    },
+    {
+      "epoch": 1172.2368421052631,
+      "grad_norm": 0.6982220411300659,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178180
+    },
+    {
+      "epoch": 1172.3026315789473,
+      "grad_norm": 1.1954939365386963,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 178190
+    },
+    {
+      "epoch": 1172.3684210526317,
+      "grad_norm": 1.6869287490844727,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 178200
+    },
+    {
+      "epoch": 1172.4342105263158,
+      "grad_norm": 1.649885654449463,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 178210
+    },
+    {
+      "epoch": 1172.5,
+      "grad_norm": 1.5803041458129883,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 178220
+    },
+    {
+      "epoch": 1172.5657894736842,
+      "grad_norm": 1.7074077129364014,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 178230
+    },
+    {
+      "epoch": 1172.6315789473683,
+      "grad_norm": 1.3916934728622437,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 178240
+    },
+    {
+      "epoch": 1172.6973684210527,
+      "grad_norm": 1.668851613998413,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 178250
+    },
+    {
+      "epoch": 1172.7631578947369,
+      "grad_norm": 1.1936376094818115,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 178260
+    },
+    {
+      "epoch": 1172.828947368421,
+      "grad_norm": 1.4797992706298828,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 178270
+    },
+    {
+      "epoch": 1172.8947368421052,
+      "grad_norm": 1.1677192449569702,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 178280
+    },
+    {
+      "epoch": 1172.9605263157894,
+      "grad_norm": 1.2152477502822876,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 178290
+    },
+    {
+      "epoch": 1173.0263157894738,
+      "grad_norm": 1.248241662979126,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 178300
+    },
+    {
+      "epoch": 1173.092105263158,
+      "grad_norm": 1.2586883306503296,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178310
+    },
+    {
+      "epoch": 1173.157894736842,
+      "grad_norm": 1.2856673002243042,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 178320
+    },
+    {
+      "epoch": 1173.2236842105262,
+      "grad_norm": 1.19162917137146,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 178330
+    },
+    {
+      "epoch": 1173.2894736842106,
+      "grad_norm": 0.8176268935203552,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178340
+    },
+    {
+      "epoch": 1173.3552631578948,
+      "grad_norm": 0.7139531970024109,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 178350
+    },
+    {
+      "epoch": 1173.421052631579,
+      "grad_norm": 0.7016176581382751,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 178360
+    },
+    {
+      "epoch": 1173.4868421052631,
+      "grad_norm": 1.0311745405197144,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 178370
+    },
+    {
+      "epoch": 1173.5526315789473,
+      "grad_norm": 0.8565056920051575,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 178380
+    },
+    {
+      "epoch": 1173.6184210526317,
+      "grad_norm": 1.1574746370315552,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 178390
+    },
+    {
+      "epoch": 1173.6842105263158,
+      "grad_norm": 1.0975223779678345,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 178400
+    },
+    {
+      "epoch": 1173.75,
+      "grad_norm": 0.849980890750885,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 178410
+    },
+    {
+      "epoch": 1173.8157894736842,
+      "grad_norm": 0.8010478019714355,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 178420
+    },
+    {
+      "epoch": 1173.8815789473683,
+      "grad_norm": 0.7263656258583069,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 178430
+    },
+    {
+      "epoch": 1173.9473684210527,
+      "grad_norm": 0.9674796462059021,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 178440
+    },
+    {
+      "epoch": 1174.0131578947369,
+      "grad_norm": 0.8462510108947754,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 178450
+    },
+    {
+      "epoch": 1174.078947368421,
+      "grad_norm": 0.921765923500061,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 178460
+    },
+    {
+      "epoch": 1174.1447368421052,
+      "grad_norm": 1.0591344833374023,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 178470
+    },
+    {
+      "epoch": 1174.2105263157894,
+      "grad_norm": 1.122679352760315,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 178480
+    },
+    {
+      "epoch": 1174.2763157894738,
+      "grad_norm": 0.8474733233451843,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 178490
+    },
+    {
+      "epoch": 1174.342105263158,
+      "grad_norm": 0.9375333189964294,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 178500
+    },
+    {
+      "epoch": 1174.407894736842,
+      "grad_norm": 0.9587281942367554,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 178510
+    },
+    {
+      "epoch": 1174.4736842105262,
+      "grad_norm": 0.998461902141571,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 178520
+    },
+    {
+      "epoch": 1174.5394736842106,
+      "grad_norm": 1.0688536167144775,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 178530
+    },
+    {
+      "epoch": 1174.6052631578948,
+      "grad_norm": 0.768993079662323,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 178540
+    },
+    {
+      "epoch": 1174.671052631579,
+      "grad_norm": 0.9926314353942871,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 178550
+    },
+    {
+      "epoch": 1174.7368421052631,
+      "grad_norm": 1.068148136138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 178560
+    },
+    {
+      "epoch": 1174.8026315789473,
+      "grad_norm": 0.9381814002990723,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 178570
+    },
+    {
+      "epoch": 1174.8684210526317,
+      "grad_norm": 1.102551817893982,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178580
+    },
+    {
+      "epoch": 1174.9342105263158,
+      "grad_norm": 0.8753682374954224,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 178590
+    },
+    {
+      "epoch": 1175.0,
+      "grad_norm": 1.0447521209716797,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 178600
+    },
+    {
+      "epoch": 1175.0657894736842,
+      "grad_norm": 1.3192112445831299,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 178610
+    },
+    {
+      "epoch": 1175.1315789473683,
+      "grad_norm": 1.2548503875732422,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 178620
+    },
+    {
+      "epoch": 1175.1973684210527,
+      "grad_norm": 1.1852110624313354,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 178630
+    },
+    {
+      "epoch": 1175.2631578947369,
+      "grad_norm": 1.1667160987854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 178640
+    },
+    {
+      "epoch": 1175.328947368421,
+      "grad_norm": 0.9693193435668945,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 178650
+    },
+    {
+      "epoch": 1175.3947368421052,
+      "grad_norm": 1.279008150100708,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 178660
+    },
+    {
+      "epoch": 1175.4605263157894,
+      "grad_norm": 0.8803616762161255,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 178670
+    },
+    {
+      "epoch": 1175.5263157894738,
+      "grad_norm": 0.7935562133789062,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 178680
+    },
+    {
+      "epoch": 1175.592105263158,
+      "grad_norm": 0.8752424120903015,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 178690
+    },
+    {
+      "epoch": 1175.657894736842,
+      "grad_norm": 1.2736647129058838,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 178700
+    },
+    {
+      "epoch": 1175.7236842105262,
+      "grad_norm": 0.9050102829933167,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 178710
+    },
+    {
+      "epoch": 1175.7894736842106,
+      "grad_norm": 0.9366145730018616,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 178720
+    },
+    {
+      "epoch": 1175.8552631578948,
+      "grad_norm": 1.0141912698745728,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 178730
+    },
+    {
+      "epoch": 1175.921052631579,
+      "grad_norm": 0.8957210183143616,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178740
+    },
+    {
+      "epoch": 1175.9868421052631,
+      "grad_norm": 0.743053674697876,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 178750
+    },
+    {
+      "epoch": 1176.0526315789473,
+      "grad_norm": 0.8863754868507385,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 178760
+    },
+    {
+      "epoch": 1176.1184210526317,
+      "grad_norm": 0.7674370408058167,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 178770
+    },
+    {
+      "epoch": 1176.1842105263158,
+      "grad_norm": 0.6406316757202148,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 178780
+    },
+    {
+      "epoch": 1176.25,
+      "grad_norm": 0.5896703600883484,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178790
+    },
+    {
+      "epoch": 1176.3157894736842,
+      "grad_norm": 0.7623488903045654,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 178800
+    },
+    {
+      "epoch": 1176.3815789473683,
+      "grad_norm": 0.7528663277626038,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 178810
+    },
+    {
+      "epoch": 1176.4473684210527,
+      "grad_norm": 0.9548066854476929,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 178820
+    },
+    {
+      "epoch": 1176.5131578947369,
+      "grad_norm": 0.880696177482605,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 178830
+    },
+    {
+      "epoch": 1176.578947368421,
+      "grad_norm": 1.145169734954834,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 178840
+    },
+    {
+      "epoch": 1176.6447368421052,
+      "grad_norm": 1.2576274871826172,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 178850
+    },
+    {
+      "epoch": 1176.7105263157894,
+      "grad_norm": 1.057856798171997,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 178860
+    },
+    {
+      "epoch": 1176.7763157894738,
+      "grad_norm": 0.9875898957252502,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 178870
+    },
+    {
+      "epoch": 1176.842105263158,
+      "grad_norm": 1.0104905366897583,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 178880
+    },
+    {
+      "epoch": 1176.907894736842,
+      "grad_norm": 0.7599539756774902,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178890
+    },
+    {
+      "epoch": 1176.9736842105262,
+      "grad_norm": 0.6184892058372498,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 178900
+    },
+    {
+      "epoch": 1177.0394736842106,
+      "grad_norm": 1.0564872026443481,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 178910
+    },
+    {
+      "epoch": 1177.1052631578948,
+      "grad_norm": 0.9271602034568787,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 178920
+    },
+    {
+      "epoch": 1177.171052631579,
+      "grad_norm": 0.9372125267982483,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 178930
+    },
+    {
+      "epoch": 1177.2368421052631,
+      "grad_norm": 0.8659113645553589,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 178940
+    },
+    {
+      "epoch": 1177.3026315789473,
+      "grad_norm": 0.9990260601043701,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 178950
+    },
+    {
+      "epoch": 1177.3684210526317,
+      "grad_norm": 0.9762459993362427,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 178960
+    },
+    {
+      "epoch": 1177.4342105263158,
+      "grad_norm": 0.8299205899238586,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 178970
+    },
+    {
+      "epoch": 1177.5,
+      "grad_norm": 1.0260875225067139,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 178980
+    },
+    {
+      "epoch": 1177.5657894736842,
+      "grad_norm": 0.8214814066886902,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 178990
+    },
+    {
+      "epoch": 1177.6315789473683,
+      "grad_norm": 0.8492797017097473,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 179000
+    },
+    {
+      "epoch": 1177.6973684210527,
+      "grad_norm": 1.093478798866272,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 179010
+    },
+    {
+      "epoch": 1177.7631578947369,
+      "grad_norm": 1.2081449031829834,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 179020
+    },
+    {
+      "epoch": 1177.828947368421,
+      "grad_norm": 1.1940277814865112,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 179030
+    },
+    {
+      "epoch": 1177.8947368421052,
+      "grad_norm": 0.8136548399925232,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 179040
+    },
+    {
+      "epoch": 1177.9605263157894,
+      "grad_norm": 0.9597692489624023,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 179050
+    },
+    {
+      "epoch": 1178.0263157894738,
+      "grad_norm": 1.0133211612701416,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 179060
+    },
+    {
+      "epoch": 1178.092105263158,
+      "grad_norm": 1.08301842212677,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 179070
+    },
+    {
+      "epoch": 1178.157894736842,
+      "grad_norm": 0.9614161252975464,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 179080
+    },
+    {
+      "epoch": 1178.2236842105262,
+      "grad_norm": 1.0885823965072632,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 179090
+    },
+    {
+      "epoch": 1178.2894736842106,
+      "grad_norm": 0.9387257695198059,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 179100
+    },
+    {
+      "epoch": 1178.3552631578948,
+      "grad_norm": 0.7102392911911011,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 179110
+    },
+    {
+      "epoch": 1178.421052631579,
+      "grad_norm": 0.9195134043693542,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 179120
+    },
+    {
+      "epoch": 1178.4868421052631,
+      "grad_norm": 1.0749733448028564,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 179130
+    },
+    {
+      "epoch": 1178.5526315789473,
+      "grad_norm": 1.2915416955947876,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 179140
+    },
+    {
+      "epoch": 1178.6184210526317,
+      "grad_norm": 0.8285413980484009,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 179150
+    },
+    {
+      "epoch": 1178.6842105263158,
+      "grad_norm": 0.7751388549804688,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 179160
+    },
+    {
+      "epoch": 1178.75,
+      "grad_norm": 0.8202561736106873,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 179170
+    },
+    {
+      "epoch": 1178.8157894736842,
+      "grad_norm": 0.9537473320960999,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 179180
+    },
+    {
+      "epoch": 1178.8815789473683,
+      "grad_norm": 0.9641271829605103,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 179190
+    },
+    {
+      "epoch": 1178.9473684210527,
+      "grad_norm": 0.962111234664917,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 179200
+    },
+    {
+      "epoch": 1179.0131578947369,
+      "grad_norm": 0.8706178665161133,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 179210
+    },
+    {
+      "epoch": 1179.078947368421,
+      "grad_norm": 1.0734243392944336,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 179220
+    },
+    {
+      "epoch": 1179.1447368421052,
+      "grad_norm": 1.075181484222412,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 179230
+    },
+    {
+      "epoch": 1179.2105263157894,
+      "grad_norm": 1.1446199417114258,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 179240
+    },
+    {
+      "epoch": 1179.2763157894738,
+      "grad_norm": 0.9987981915473938,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 179250
+    },
+    {
+      "epoch": 1179.342105263158,
+      "grad_norm": 0.8230014443397522,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 179260
+    },
+    {
+      "epoch": 1179.407894736842,
+      "grad_norm": 0.7945327162742615,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 179270
+    },
+    {
+      "epoch": 1179.4736842105262,
+      "grad_norm": 0.7237135171890259,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 179280
+    },
+    {
+      "epoch": 1179.5394736842106,
+      "grad_norm": 0.8276739716529846,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 179290
+    },
+    {
+      "epoch": 1179.6052631578948,
+      "grad_norm": 1.0555814504623413,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 179300
+    },
+    {
+      "epoch": 1179.671052631579,
+      "grad_norm": 1.0211066007614136,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 179310
+    },
+    {
+      "epoch": 1179.7368421052631,
+      "grad_norm": 1.0281317234039307,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 179320
+    },
+    {
+      "epoch": 1179.8026315789473,
+      "grad_norm": 1.0497641563415527,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 179330
+    },
+    {
+      "epoch": 1179.8684210526317,
+      "grad_norm": 1.2660293579101562,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 179340
+    },
+    {
+      "epoch": 1179.9342105263158,
+      "grad_norm": 0.6678763628005981,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 179350
+    },
+    {
+      "epoch": 1180.0,
+      "grad_norm": 1.2870025634765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 179360
+    },
+    {
+      "epoch": 1180.0657894736842,
+      "grad_norm": 1.0261781215667725,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 179370
+    },
+    {
+      "epoch": 1180.1315789473683,
+      "grad_norm": 0.9007861614227295,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 179380
+    },
+    {
+      "epoch": 1180.1973684210527,
+      "grad_norm": 0.9087187647819519,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 179390
+    },
+    {
+      "epoch": 1180.2631578947369,
+      "grad_norm": 1.1201528310775757,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 179400
+    },
+    {
+      "epoch": 1180.328947368421,
+      "grad_norm": 1.268721342086792,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 179410
+    },
+    {
+      "epoch": 1180.3947368421052,
+      "grad_norm": 1.1803576946258545,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 179420
+    },
+    {
+      "epoch": 1180.4605263157894,
+      "grad_norm": 0.8738627433776855,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 179430
+    },
+    {
+      "epoch": 1180.5263157894738,
+      "grad_norm": 0.9588783383369446,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 179440
+    },
+    {
+      "epoch": 1180.592105263158,
+      "grad_norm": 0.8148517608642578,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 179450
+    },
+    {
+      "epoch": 1180.657894736842,
+      "grad_norm": 0.8594145774841309,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 179460
+    },
+    {
+      "epoch": 1180.7236842105262,
+      "grad_norm": 0.658765435218811,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 179470
+    },
+    {
+      "epoch": 1180.7894736842106,
+      "grad_norm": 0.9419595003128052,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 179480
+    },
+    {
+      "epoch": 1180.8552631578948,
+      "grad_norm": 0.9967935681343079,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 179490
+    },
+    {
+      "epoch": 1180.921052631579,
+      "grad_norm": 1.151293158531189,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 179500
+    },
+    {
+      "epoch": 1180.9868421052631,
+      "grad_norm": 1.1010416746139526,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 179510
+    },
+    {
+      "epoch": 1181.0526315789473,
+      "grad_norm": 1.184219241142273,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 179520
+    },
+    {
+      "epoch": 1181.1184210526317,
+      "grad_norm": 1.5013480186462402,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 179530
+    },
+    {
+      "epoch": 1181.1842105263158,
+      "grad_norm": 1.2943904399871826,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 179540
+    },
+    {
+      "epoch": 1181.25,
+      "grad_norm": 1.3031779527664185,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 179550
+    },
+    {
+      "epoch": 1181.3157894736842,
+      "grad_norm": 0.875571072101593,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 179560
+    },
+    {
+      "epoch": 1181.3815789473683,
+      "grad_norm": 1.010248064994812,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 179570
+    },
+    {
+      "epoch": 1181.4473684210527,
+      "grad_norm": 1.0535480976104736,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 179580
+    },
+    {
+      "epoch": 1181.5131578947369,
+      "grad_norm": 0.9515687823295593,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 179590
+    },
+    {
+      "epoch": 1181.578947368421,
+      "grad_norm": 0.5328004360198975,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 179600
+    },
+    {
+      "epoch": 1181.6447368421052,
+      "grad_norm": 1.2475734949111938,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 179610
+    },
+    {
+      "epoch": 1181.7105263157894,
+      "grad_norm": 0.8466393351554871,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 179620
+    },
+    {
+      "epoch": 1181.7763157894738,
+      "grad_norm": 0.8406497240066528,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 179630
+    },
+    {
+      "epoch": 1181.842105263158,
+      "grad_norm": 1.0086272954940796,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 179640
+    },
+    {
+      "epoch": 1181.907894736842,
+      "grad_norm": 0.9933298826217651,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 179650
+    },
+    {
+      "epoch": 1181.9736842105262,
+      "grad_norm": 1.049748420715332,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 179660
+    },
+    {
+      "epoch": 1182.0394736842106,
+      "grad_norm": 1.0045021772384644,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 179670
+    },
+    {
+      "epoch": 1182.1052631578948,
+      "grad_norm": 0.8957365155220032,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 179680
+    },
+    {
+      "epoch": 1182.171052631579,
+      "grad_norm": 0.8856828212738037,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 179690
+    },
+    {
+      "epoch": 1182.2368421052631,
+      "grad_norm": 1.2047641277313232,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 179700
+    },
+    {
+      "epoch": 1182.3026315789473,
+      "grad_norm": 1.1860121488571167,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 179710
+    },
+    {
+      "epoch": 1182.3684210526317,
+      "grad_norm": 1.2543531656265259,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 179720
+    },
+    {
+      "epoch": 1182.4342105263158,
+      "grad_norm": 1.3099639415740967,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 179730
+    },
+    {
+      "epoch": 1182.5,
+      "grad_norm": 1.1823962926864624,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 179740
+    },
+    {
+      "epoch": 1182.5657894736842,
+      "grad_norm": 0.9950479865074158,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 179750
+    },
+    {
+      "epoch": 1182.6315789473683,
+      "grad_norm": 1.14492928981781,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 179760
+    },
+    {
+      "epoch": 1182.6973684210527,
+      "grad_norm": 0.985609233379364,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 179770
+    },
+    {
+      "epoch": 1182.7631578947369,
+      "grad_norm": 1.0433059930801392,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 179780
+    },
+    {
+      "epoch": 1182.828947368421,
+      "grad_norm": 0.9120721220970154,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 179790
+    },
+    {
+      "epoch": 1182.8947368421052,
+      "grad_norm": 0.8850807547569275,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 179800
+    },
+    {
+      "epoch": 1182.9605263157894,
+      "grad_norm": 0.6139315962791443,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 179810
+    },
+    {
+      "epoch": 1183.0263157894738,
+      "grad_norm": 1.1187626123428345,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 179820
+    },
+    {
+      "epoch": 1183.092105263158,
+      "grad_norm": 1.0548551082611084,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 179830
+    },
+    {
+      "epoch": 1183.157894736842,
+      "grad_norm": 0.9442790746688843,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 179840
+    },
+    {
+      "epoch": 1183.2236842105262,
+      "grad_norm": 0.7005117535591125,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 179850
+    },
+    {
+      "epoch": 1183.2894736842106,
+      "grad_norm": 0.5735732913017273,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 179860
+    },
+    {
+      "epoch": 1183.3552631578948,
+      "grad_norm": 1.027298092842102,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 179870
+    },
+    {
+      "epoch": 1183.421052631579,
+      "grad_norm": 1.3360284566879272,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 179880
+    },
+    {
+      "epoch": 1183.4868421052631,
+      "grad_norm": 0.8555198907852173,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 179890
+    },
+    {
+      "epoch": 1183.5526315789473,
+      "grad_norm": 1.2100673913955688,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 179900
+    },
+    {
+      "epoch": 1183.6184210526317,
+      "grad_norm": 0.8696701526641846,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 179910
+    },
+    {
+      "epoch": 1183.6842105263158,
+      "grad_norm": 1.044405460357666,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 179920
+    },
+    {
+      "epoch": 1183.75,
+      "grad_norm": 0.6577524542808533,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 179930
+    },
+    {
+      "epoch": 1183.8157894736842,
+      "grad_norm": 0.8520231246948242,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 179940
+    },
+    {
+      "epoch": 1183.8815789473683,
+      "grad_norm": 0.5790631175041199,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "step": 179950
+    },
+    {
+      "epoch": 1183.9473684210527,
+      "grad_norm": 0.6821280717849731,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 179960
+    },
+    {
+      "epoch": 1184.0131578947369,
+      "grad_norm": 1.1946996450424194,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 179970
+    },
+    {
+      "epoch": 1184.078947368421,
+      "grad_norm": 0.8206828832626343,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 179980
+    },
+    {
+      "epoch": 1184.1447368421052,
+      "grad_norm": 1.200190782546997,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 179990
+    },
+    {
+      "epoch": 1184.2105263157894,
+      "grad_norm": 1.0176090002059937,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 180000
+    },
+    {
+      "epoch": 1184.2763157894738,
+      "grad_norm": 1.791040301322937,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 180010
+    },
+    {
+      "epoch": 1184.342105263158,
+      "grad_norm": 1.1457858085632324,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 180020
+    },
+    {
+      "epoch": 1184.407894736842,
+      "grad_norm": 1.3513076305389404,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 180030
+    },
+    {
+      "epoch": 1184.4736842105262,
+      "grad_norm": 1.1888035535812378,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 180040
+    },
+    {
+      "epoch": 1184.5394736842106,
+      "grad_norm": 1.3418779373168945,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 180050
+    },
+    {
+      "epoch": 1184.6052631578948,
+      "grad_norm": 1.2430517673492432,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 180060
+    },
+    {
+      "epoch": 1184.671052631579,
+      "grad_norm": 1.299909234046936,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 180070
+    },
+    {
+      "epoch": 1184.7368421052631,
+      "grad_norm": 1.187268614768982,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 180080
+    },
+    {
+      "epoch": 1184.8026315789473,
+      "grad_norm": 1.0313128232955933,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 180090
+    },
+    {
+      "epoch": 1184.8684210526317,
+      "grad_norm": 0.9773435592651367,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 180100
+    },
+    {
+      "epoch": 1184.9342105263158,
+      "grad_norm": 0.9462606310844421,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 180110
+    },
+    {
+      "epoch": 1185.0,
+      "grad_norm": 1.008164882659912,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 180120
+    },
+    {
+      "epoch": 1185.0657894736842,
+      "grad_norm": 1.0677521228790283,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 180130
+    },
+    {
+      "epoch": 1185.1315789473683,
+      "grad_norm": 1.056257963180542,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 180140
+    },
+    {
+      "epoch": 1185.1973684210527,
+      "grad_norm": 1.3330624103546143,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 180150
+    },
+    {
+      "epoch": 1185.2631578947369,
+      "grad_norm": 1.2008347511291504,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 180160
+    },
+    {
+      "epoch": 1185.328947368421,
+      "grad_norm": 1.0260757207870483,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 180170
+    },
+    {
+      "epoch": 1185.3947368421052,
+      "grad_norm": 0.9635398983955383,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 180180
+    },
+    {
+      "epoch": 1185.4605263157894,
+      "grad_norm": 0.9844223260879517,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 180190
+    },
+    {
+      "epoch": 1185.5263157894738,
+      "grad_norm": 0.9323477745056152,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 180200
+    },
+    {
+      "epoch": 1185.592105263158,
+      "grad_norm": 0.7835142612457275,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 180210
+    },
+    {
+      "epoch": 1185.657894736842,
+      "grad_norm": 0.8820164799690247,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 180220
+    },
+    {
+      "epoch": 1185.7236842105262,
+      "grad_norm": 1.1750354766845703,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 180230
+    },
+    {
+      "epoch": 1185.7894736842106,
+      "grad_norm": 1.058611512184143,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 180240
+    },
+    {
+      "epoch": 1185.8552631578948,
+      "grad_norm": 1.0352505445480347,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 180250
+    },
+    {
+      "epoch": 1185.921052631579,
+      "grad_norm": 0.8906688690185547,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 180260
+    },
+    {
+      "epoch": 1185.9868421052631,
+      "grad_norm": 1.0255763530731201,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 180270
+    },
+    {
+      "epoch": 1186.0526315789473,
+      "grad_norm": 0.9030811190605164,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 180280
+    },
+    {
+      "epoch": 1186.1184210526317,
+      "grad_norm": 0.7949104905128479,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 180290
+    },
+    {
+      "epoch": 1186.1842105263158,
+      "grad_norm": 1.2452713251113892,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 180300
+    },
+    {
+      "epoch": 1186.25,
+      "grad_norm": 0.8592615127563477,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 180310
+    },
+    {
+      "epoch": 1186.3157894736842,
+      "grad_norm": 1.0579206943511963,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 180320
+    },
+    {
+      "epoch": 1186.3815789473683,
+      "grad_norm": 1.0151481628417969,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 180330
+    },
+    {
+      "epoch": 1186.4473684210527,
+      "grad_norm": 1.2412348985671997,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 180340
+    },
+    {
+      "epoch": 1186.5131578947369,
+      "grad_norm": 0.9836140275001526,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 180350
+    },
+    {
+      "epoch": 1186.578947368421,
+      "grad_norm": 0.8268551826477051,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 180360
+    },
+    {
+      "epoch": 1186.6447368421052,
+      "grad_norm": 1.1718807220458984,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 180370
+    },
+    {
+      "epoch": 1186.7105263157894,
+      "grad_norm": 0.9186016321182251,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 180380
+    },
+    {
+      "epoch": 1186.7763157894738,
+      "grad_norm": 1.0187981128692627,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 180390
+    },
+    {
+      "epoch": 1186.842105263158,
+      "grad_norm": 0.7766441106796265,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 180400
+    },
+    {
+      "epoch": 1186.907894736842,
+      "grad_norm": 1.1324797868728638,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 180410
+    },
+    {
+      "epoch": 1186.9736842105262,
+      "grad_norm": 1.3124502897262573,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 180420
+    },
+    {
+      "epoch": 1187.0394736842106,
+      "grad_norm": 1.0421258211135864,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 180430
+    },
+    {
+      "epoch": 1187.1052631578948,
+      "grad_norm": 0.9878388047218323,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 180440
+    },
+    {
+      "epoch": 1187.171052631579,
+      "grad_norm": 1.404099941253662,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 180450
+    },
+    {
+      "epoch": 1187.2368421052631,
+      "grad_norm": 0.8869650959968567,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 180460
+    },
+    {
+      "epoch": 1187.3026315789473,
+      "grad_norm": 0.7417500615119934,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 180470
+    },
+    {
+      "epoch": 1187.3684210526317,
+      "grad_norm": 0.8958377838134766,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 180480
+    },
+    {
+      "epoch": 1187.4342105263158,
+      "grad_norm": 1.3181413412094116,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 180490
+    },
+    {
+      "epoch": 1187.5,
+      "grad_norm": 1.1831432580947876,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 180500
+    },
+    {
+      "epoch": 1187.5657894736842,
+      "grad_norm": 0.9017171263694763,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 180510
+    },
+    {
+      "epoch": 1187.6315789473683,
+      "grad_norm": 1.0367988348007202,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 180520
+    },
+    {
+      "epoch": 1187.6973684210527,
+      "grad_norm": 0.9045135974884033,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 180530
+    },
+    {
+      "epoch": 1187.7631578947369,
+      "grad_norm": 0.8042669296264648,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 180540
+    },
+    {
+      "epoch": 1187.828947368421,
+      "grad_norm": 1.19108247756958,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 180550
+    },
+    {
+      "epoch": 1187.8947368421052,
+      "grad_norm": 1.238985300064087,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 180560
+    },
+    {
+      "epoch": 1187.9605263157894,
+      "grad_norm": 1.5841645002365112,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 180570
+    },
+    {
+      "epoch": 1188.0263157894738,
+      "grad_norm": 1.1452237367630005,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 180580
+    },
+    {
+      "epoch": 1188.092105263158,
+      "grad_norm": 1.4237656593322754,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 180590
+    },
+    {
+      "epoch": 1188.157894736842,
+      "grad_norm": 1.343073844909668,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 180600
+    },
+    {
+      "epoch": 1188.2236842105262,
+      "grad_norm": 0.9501973986625671,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 180610
+    },
+    {
+      "epoch": 1188.2894736842106,
+      "grad_norm": 0.8718429803848267,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 180620
+    },
+    {
+      "epoch": 1188.3552631578948,
+      "grad_norm": 1.1177846193313599,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 180630
+    },
+    {
+      "epoch": 1188.421052631579,
+      "grad_norm": 0.7141475677490234,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 180640
+    },
+    {
+      "epoch": 1188.4868421052631,
+      "grad_norm": 0.9596945643424988,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 180650
+    },
+    {
+      "epoch": 1188.5526315789473,
+      "grad_norm": 1.1708953380584717,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 180660
+    },
+    {
+      "epoch": 1188.6184210526317,
+      "grad_norm": 1.1487609148025513,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 180670
+    },
+    {
+      "epoch": 1188.6842105263158,
+      "grad_norm": 0.9982112050056458,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 180680
+    },
+    {
+      "epoch": 1188.75,
+      "grad_norm": 1.1738011837005615,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 180690
+    },
+    {
+      "epoch": 1188.8157894736842,
+      "grad_norm": 1.2586804628372192,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 180700
+    },
+    {
+      "epoch": 1188.8815789473683,
+      "grad_norm": 1.6938642263412476,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 180710
+    },
+    {
+      "epoch": 1188.9473684210527,
+      "grad_norm": 1.3859210014343262,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 180720
+    },
+    {
+      "epoch": 1189.0131578947369,
+      "grad_norm": 1.3656494617462158,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 180730
+    },
+    {
+      "epoch": 1189.078947368421,
+      "grad_norm": 1.3069894313812256,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 180740
+    },
+    {
+      "epoch": 1189.1447368421052,
+      "grad_norm": 1.5368648767471313,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 180750
+    },
+    {
+      "epoch": 1189.2105263157894,
+      "grad_norm": 0.9646233916282654,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 180760
+    },
+    {
+      "epoch": 1189.2763157894738,
+      "grad_norm": 1.1789246797561646,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 180770
+    },
+    {
+      "epoch": 1189.342105263158,
+      "grad_norm": 0.9047682881355286,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 180780
+    },
+    {
+      "epoch": 1189.407894736842,
+      "grad_norm": 1.1768347024917603,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 180790
+    },
+    {
+      "epoch": 1189.4736842105262,
+      "grad_norm": 0.8927736878395081,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 180800
+    },
+    {
+      "epoch": 1189.5394736842106,
+      "grad_norm": 1.3227018117904663,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 180810
+    },
+    {
+      "epoch": 1189.6052631578948,
+      "grad_norm": 1.1930198669433594,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 180820
+    },
+    {
+      "epoch": 1189.671052631579,
+      "grad_norm": 0.9631125926971436,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 180830
+    },
+    {
+      "epoch": 1189.7368421052631,
+      "grad_norm": 0.9095155000686646,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 180840
+    },
+    {
+      "epoch": 1189.8026315789473,
+      "grad_norm": 0.9870935082435608,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 180850
+    },
+    {
+      "epoch": 1189.8684210526317,
+      "grad_norm": 0.9359210133552551,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 180860
+    },
+    {
+      "epoch": 1189.9342105263158,
+      "grad_norm": 1.2630891799926758,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 180870
+    },
+    {
+      "epoch": 1190.0,
+      "grad_norm": 0.9118783473968506,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 180880
+    },
+    {
+      "epoch": 1190.0657894736842,
+      "grad_norm": 0.9447109699249268,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 180890
+    },
+    {
+      "epoch": 1190.1315789473683,
+      "grad_norm": 0.8619827032089233,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 180900
+    },
+    {
+      "epoch": 1190.1973684210527,
+      "grad_norm": 1.0921404361724854,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 180910
+    },
+    {
+      "epoch": 1190.2631578947369,
+      "grad_norm": 0.7819488048553467,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 180920
+    },
+    {
+      "epoch": 1190.328947368421,
+      "grad_norm": 0.9675580263137817,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 180930
+    },
+    {
+      "epoch": 1190.3947368421052,
+      "grad_norm": 0.9420514106750488,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 180940
+    },
+    {
+      "epoch": 1190.4605263157894,
+      "grad_norm": 1.3097189664840698,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 180950
+    },
+    {
+      "epoch": 1190.5263157894738,
+      "grad_norm": 1.0128648281097412,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 180960
+    },
+    {
+      "epoch": 1190.592105263158,
+      "grad_norm": 0.833601713180542,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 180970
+    },
+    {
+      "epoch": 1190.657894736842,
+      "grad_norm": 0.7364785671234131,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 180980
+    },
+    {
+      "epoch": 1190.7236842105262,
+      "grad_norm": 1.0391943454742432,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 180990
+    },
+    {
+      "epoch": 1190.7894736842106,
+      "grad_norm": 0.9360489845275879,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 181000
+    },
+    {
+      "epoch": 1190.8552631578948,
+      "grad_norm": 0.8230446577072144,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 181010
+    },
+    {
+      "epoch": 1190.921052631579,
+      "grad_norm": 0.6308891773223877,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 181020
+    },
+    {
+      "epoch": 1190.9868421052631,
+      "grad_norm": 0.9240980744361877,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 181030
+    },
+    {
+      "epoch": 1191.0526315789473,
+      "grad_norm": 0.9708207845687866,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 181040
+    },
+    {
+      "epoch": 1191.1184210526317,
+      "grad_norm": 1.2800960540771484,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 181050
+    },
+    {
+      "epoch": 1191.1842105263158,
+      "grad_norm": 0.9510250687599182,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 181060
+    },
+    {
+      "epoch": 1191.25,
+      "grad_norm": 0.8674155473709106,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 181070
+    },
+    {
+      "epoch": 1191.3157894736842,
+      "grad_norm": 1.276930332183838,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 181080
+    },
+    {
+      "epoch": 1191.3815789473683,
+      "grad_norm": 1.3920434713363647,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 181090
+    },
+    {
+      "epoch": 1191.4473684210527,
+      "grad_norm": 0.7923642992973328,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 181100
+    },
+    {
+      "epoch": 1191.5131578947369,
+      "grad_norm": 1.4579869508743286,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 181110
+    },
+    {
+      "epoch": 1191.578947368421,
+      "grad_norm": 0.9362051486968994,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 181120
+    },
+    {
+      "epoch": 1191.6447368421052,
+      "grad_norm": 1.371772050857544,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 181130
+    },
+    {
+      "epoch": 1191.7105263157894,
+      "grad_norm": 1.1880755424499512,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 181140
+    },
+    {
+      "epoch": 1191.7763157894738,
+      "grad_norm": 0.8417680859565735,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 181150
+    },
+    {
+      "epoch": 1191.842105263158,
+      "grad_norm": 0.9738719463348389,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 181160
+    },
+    {
+      "epoch": 1191.907894736842,
+      "grad_norm": 0.8332647085189819,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 181170
+    },
+    {
+      "epoch": 1191.9736842105262,
+      "grad_norm": 0.8110590577125549,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 181180
+    },
+    {
+      "epoch": 1192.0394736842106,
+      "grad_norm": 0.7889165878295898,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 181190
+    },
+    {
+      "epoch": 1192.1052631578948,
+      "grad_norm": 0.9631662368774414,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 181200
+    },
+    {
+      "epoch": 1192.171052631579,
+      "grad_norm": 0.9708088040351868,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 181210
+    },
+    {
+      "epoch": 1192.2368421052631,
+      "grad_norm": 0.9359713196754456,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 181220
+    },
+    {
+      "epoch": 1192.3026315789473,
+      "grad_norm": 0.9870832562446594,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 181230
+    },
+    {
+      "epoch": 1192.3684210526317,
+      "grad_norm": 1.1724073886871338,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 181240
+    },
+    {
+      "epoch": 1192.4342105263158,
+      "grad_norm": 0.7455830574035645,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 181250
+    },
+    {
+      "epoch": 1192.5,
+      "grad_norm": 0.9996033906936646,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 181260
+    },
+    {
+      "epoch": 1192.5657894736842,
+      "grad_norm": 1.123407006263733,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 181270
+    },
+    {
+      "epoch": 1192.6315789473683,
+      "grad_norm": 0.9888333678245544,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 181280
+    },
+    {
+      "epoch": 1192.6973684210527,
+      "grad_norm": 1.7157537937164307,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 181290
+    },
+    {
+      "epoch": 1192.7631578947369,
+      "grad_norm": 1.2598707675933838,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 181300
+    },
+    {
+      "epoch": 1192.828947368421,
+      "grad_norm": 1.1619699001312256,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 181310
+    },
+    {
+      "epoch": 1192.8947368421052,
+      "grad_norm": 0.9161400198936462,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 181320
+    },
+    {
+      "epoch": 1192.9605263157894,
+      "grad_norm": 0.7341195344924927,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 181330
+    },
+    {
+      "epoch": 1193.0263157894738,
+      "grad_norm": 1.1331208944320679,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 181340
+    },
+    {
+      "epoch": 1193.092105263158,
+      "grad_norm": 1.149287223815918,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 181350
+    },
+    {
+      "epoch": 1193.157894736842,
+      "grad_norm": 0.9564964175224304,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 181360
+    },
+    {
+      "epoch": 1193.2236842105262,
+      "grad_norm": 1.252368450164795,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 181370
+    },
+    {
+      "epoch": 1193.2894736842106,
+      "grad_norm": 1.285278558731079,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 181380
+    },
+    {
+      "epoch": 1193.3552631578948,
+      "grad_norm": 1.0179182291030884,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 181390
+    },
+    {
+      "epoch": 1193.421052631579,
+      "grad_norm": 0.8512329459190369,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 181400
+    },
+    {
+      "epoch": 1193.4868421052631,
+      "grad_norm": 1.3857789039611816,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 181410
+    },
+    {
+      "epoch": 1193.5526315789473,
+      "grad_norm": 1.1037719249725342,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 181420
+    },
+    {
+      "epoch": 1193.6184210526317,
+      "grad_norm": 1.2100577354431152,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 181430
+    },
+    {
+      "epoch": 1193.6842105263158,
+      "grad_norm": 1.3522088527679443,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 181440
+    },
+    {
+      "epoch": 1193.75,
+      "grad_norm": 1.1661986112594604,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 181450
+    },
+    {
+      "epoch": 1193.8157894736842,
+      "grad_norm": 1.2195343971252441,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 181460
+    },
+    {
+      "epoch": 1193.8815789473683,
+      "grad_norm": 1.2342994213104248,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 181470
+    },
+    {
+      "epoch": 1193.9473684210527,
+      "grad_norm": 1.23150634765625,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 181480
+    },
+    {
+      "epoch": 1194.0131578947369,
+      "grad_norm": 1.1036239862442017,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 181490
+    },
+    {
+      "epoch": 1194.078947368421,
+      "grad_norm": 0.7650617361068726,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 181500
+    },
+    {
+      "epoch": 1194.1447368421052,
+      "grad_norm": 0.7626112699508667,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 181510
+    },
+    {
+      "epoch": 1194.2105263157894,
+      "grad_norm": 1.1375484466552734,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 181520
+    },
+    {
+      "epoch": 1194.2763157894738,
+      "grad_norm": 0.9652150869369507,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 181530
+    },
+    {
+      "epoch": 1194.342105263158,
+      "grad_norm": 0.9495952725410461,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 181540
+    },
+    {
+      "epoch": 1194.407894736842,
+      "grad_norm": 1.1295645236968994,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 181550
+    },
+    {
+      "epoch": 1194.4736842105262,
+      "grad_norm": 0.7198823094367981,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 181560
+    },
+    {
+      "epoch": 1194.5394736842106,
+      "grad_norm": 1.061182975769043,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 181570
+    },
+    {
+      "epoch": 1194.6052631578948,
+      "grad_norm": 0.6715997457504272,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 181580
+    },
+    {
+      "epoch": 1194.671052631579,
+      "grad_norm": 0.8791295289993286,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 181590
+    },
+    {
+      "epoch": 1194.7368421052631,
+      "grad_norm": 1.9613913297653198,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 181600
+    },
+    {
+      "epoch": 1194.8026315789473,
+      "grad_norm": 2.2601711750030518,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 181610
+    },
+    {
+      "epoch": 1194.8684210526317,
+      "grad_norm": 2.3166682720184326,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 181620
+    },
+    {
+      "epoch": 1194.9342105263158,
+      "grad_norm": 1.4985716342926025,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 181630
+    },
+    {
+      "epoch": 1195.0,
+      "grad_norm": 1.3064441680908203,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 181640
+    },
+    {
+      "epoch": 1195.0657894736842,
+      "grad_norm": 0.9696851372718811,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 181650
+    },
+    {
+      "epoch": 1195.1315789473683,
+      "grad_norm": 1.3249530792236328,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 181660
+    },
+    {
+      "epoch": 1195.1973684210527,
+      "grad_norm": 1.042757511138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 181670
+    },
+    {
+      "epoch": 1195.2631578947369,
+      "grad_norm": 0.9609398245811462,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 181680
+    },
+    {
+      "epoch": 1195.328947368421,
+      "grad_norm": 1.0072219371795654,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 181690
+    },
+    {
+      "epoch": 1195.3947368421052,
+      "grad_norm": 1.1560006141662598,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 181700
+    },
+    {
+      "epoch": 1195.4605263157894,
+      "grad_norm": 0.9536867737770081,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 181710
+    },
+    {
+      "epoch": 1195.5263157894738,
+      "grad_norm": 0.8431065678596497,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 181720
+    },
+    {
+      "epoch": 1195.592105263158,
+      "grad_norm": 1.448400855064392,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 181730
+    },
+    {
+      "epoch": 1195.657894736842,
+      "grad_norm": 1.354210615158081,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 181740
+    },
+    {
+      "epoch": 1195.7236842105262,
+      "grad_norm": 1.0818076133728027,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 181750
+    },
+    {
+      "epoch": 1195.7894736842106,
+      "grad_norm": 1.0603787899017334,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 181760
+    },
+    {
+      "epoch": 1195.8552631578948,
+      "grad_norm": 1.0252753496170044,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 181770
+    },
+    {
+      "epoch": 1195.921052631579,
+      "grad_norm": 0.9185944199562073,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 181780
+    },
+    {
+      "epoch": 1195.9868421052631,
+      "grad_norm": 1.012385368347168,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 181790
+    },
+    {
+      "epoch": 1196.0526315789473,
+      "grad_norm": 1.1767807006835938,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 181800
+    },
+    {
+      "epoch": 1196.1184210526317,
+      "grad_norm": 0.9487605094909668,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 181810
+    },
+    {
+      "epoch": 1196.1842105263158,
+      "grad_norm": 1.0731463432312012,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 181820
+    },
+    {
+      "epoch": 1196.25,
+      "grad_norm": 1.1926857233047485,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 181830
+    },
+    {
+      "epoch": 1196.3157894736842,
+      "grad_norm": 1.1458866596221924,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 181840
+    },
+    {
+      "epoch": 1196.3815789473683,
+      "grad_norm": 0.7594784498214722,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 181850
+    },
+    {
+      "epoch": 1196.4473684210527,
+      "grad_norm": 0.7159861326217651,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 181860
+    },
+    {
+      "epoch": 1196.5131578947369,
+      "grad_norm": 0.9384440183639526,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 181870
+    },
+    {
+      "epoch": 1196.578947368421,
+      "grad_norm": 0.7661808729171753,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 181880
+    },
+    {
+      "epoch": 1196.6447368421052,
+      "grad_norm": 0.8647274971008301,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 181890
+    },
+    {
+      "epoch": 1196.7105263157894,
+      "grad_norm": 0.8873572945594788,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 181900
+    },
+    {
+      "epoch": 1196.7763157894738,
+      "grad_norm": 1.2721664905548096,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 181910
+    },
+    {
+      "epoch": 1196.842105263158,
+      "grad_norm": 0.8040602207183838,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 181920
+    },
+    {
+      "epoch": 1196.907894736842,
+      "grad_norm": 1.413497805595398,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 181930
+    },
+    {
+      "epoch": 1196.9736842105262,
+      "grad_norm": 1.1012861728668213,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 181940
+    },
+    {
+      "epoch": 1197.0394736842106,
+      "grad_norm": 1.0302972793579102,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 181950
+    },
+    {
+      "epoch": 1197.1052631578948,
+      "grad_norm": 0.9582118391990662,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 181960
+    },
+    {
+      "epoch": 1197.171052631579,
+      "grad_norm": 1.0968737602233887,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 181970
+    },
+    {
+      "epoch": 1197.2368421052631,
+      "grad_norm": 1.0602166652679443,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 181980
+    },
+    {
+      "epoch": 1197.3026315789473,
+      "grad_norm": 0.890673816204071,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 181990
+    },
+    {
+      "epoch": 1197.3684210526317,
+      "grad_norm": 0.9959932565689087,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 182000
+    },
+    {
+      "epoch": 1197.4342105263158,
+      "grad_norm": 1.0189555883407593,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 182010
+    },
+    {
+      "epoch": 1197.5,
+      "grad_norm": 1.3010244369506836,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 182020
+    },
+    {
+      "epoch": 1197.5657894736842,
+      "grad_norm": 0.6490765810012817,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 182030
+    },
+    {
+      "epoch": 1197.6315789473683,
+      "grad_norm": 0.9288578629493713,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 182040
+    },
+    {
+      "epoch": 1197.6973684210527,
+      "grad_norm": 0.9825459122657776,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 182050
+    },
+    {
+      "epoch": 1197.7631578947369,
+      "grad_norm": 1.1905490159988403,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 182060
+    },
+    {
+      "epoch": 1197.828947368421,
+      "grad_norm": 0.7103608846664429,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 182070
+    },
+    {
+      "epoch": 1197.8947368421052,
+      "grad_norm": 0.9443063735961914,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 182080
+    },
+    {
+      "epoch": 1197.9605263157894,
+      "grad_norm": 1.0908981561660767,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 182090
+    },
+    {
+      "epoch": 1198.0263157894738,
+      "grad_norm": 1.0232374668121338,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 182100
+    },
+    {
+      "epoch": 1198.092105263158,
+      "grad_norm": 0.8261113166809082,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 182110
+    },
+    {
+      "epoch": 1198.157894736842,
+      "grad_norm": 0.9809024930000305,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 182120
+    },
+    {
+      "epoch": 1198.2236842105262,
+      "grad_norm": 1.3282575607299805,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 182130
+    },
+    {
+      "epoch": 1198.2894736842106,
+      "grad_norm": 0.756084144115448,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 182140
+    },
+    {
+      "epoch": 1198.3552631578948,
+      "grad_norm": 1.1468161344528198,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 182150
+    },
+    {
+      "epoch": 1198.421052631579,
+      "grad_norm": 1.3206136226654053,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 182160
+    },
+    {
+      "epoch": 1198.4868421052631,
+      "grad_norm": 1.339464545249939,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 182170
+    },
+    {
+      "epoch": 1198.5526315789473,
+      "grad_norm": 1.1766377687454224,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 182180
+    },
+    {
+      "epoch": 1198.6184210526317,
+      "grad_norm": 1.342929482460022,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 182190
+    },
+    {
+      "epoch": 1198.6842105263158,
+      "grad_norm": 1.5300862789154053,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 182200
+    },
+    {
+      "epoch": 1198.75,
+      "grad_norm": 1.2704681158065796,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 182210
+    },
+    {
+      "epoch": 1198.8157894736842,
+      "grad_norm": 0.9477614164352417,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 182220
+    },
+    {
+      "epoch": 1198.8815789473683,
+      "grad_norm": 1.055249810218811,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 182230
+    },
+    {
+      "epoch": 1198.9473684210527,
+      "grad_norm": 1.057106375694275,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 182240
+    },
+    {
+      "epoch": 1199.0131578947369,
+      "grad_norm": 1.096887469291687,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 182250
+    },
+    {
+      "epoch": 1199.078947368421,
+      "grad_norm": 1.1574534177780151,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 182260
+    },
+    {
+      "epoch": 1199.1447368421052,
+      "grad_norm": 1.2946901321411133,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 182270
+    },
+    {
+      "epoch": 1199.2105263157894,
+      "grad_norm": 1.0303276777267456,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 182280
+    },
+    {
+      "epoch": 1199.2763157894738,
+      "grad_norm": 0.9668704867362976,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 182290
+    },
+    {
+      "epoch": 1199.342105263158,
+      "grad_norm": 1.6884338855743408,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 182300
+    },
+    {
+      "epoch": 1199.407894736842,
+      "grad_norm": 1.1282835006713867,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 182310
+    },
+    {
+      "epoch": 1199.4736842105262,
+      "grad_norm": 1.2376283407211304,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 182320
+    },
+    {
+      "epoch": 1199.5394736842106,
+      "grad_norm": 1.081881046295166,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 182330
+    },
+    {
+      "epoch": 1199.6052631578948,
+      "grad_norm": 1.080176830291748,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 182340
+    },
+    {
+      "epoch": 1199.671052631579,
+      "grad_norm": 1.160550832748413,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 182350
+    },
+    {
+      "epoch": 1199.7368421052631,
+      "grad_norm": 1.1847903728485107,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 182360
+    },
+    {
+      "epoch": 1199.8026315789473,
+      "grad_norm": 1.0338932275772095,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 182370
+    },
+    {
+      "epoch": 1199.8684210526317,
+      "grad_norm": 0.7202876806259155,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 182380
+    },
+    {
+      "epoch": 1199.9342105263158,
+      "grad_norm": 0.9999865293502808,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 182390
+    },
+    {
+      "epoch": 1200.0,
+      "grad_norm": 0.7838411331176758,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 182400
+    },
+    {
+      "epoch": 1200.0657894736842,
+      "grad_norm": 0.8725451827049255,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 182410
+    },
+    {
+      "epoch": 1200.1315789473683,
+      "grad_norm": 0.8963289856910706,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 182420
+    },
+    {
+      "epoch": 1200.1973684210527,
+      "grad_norm": 0.8601619601249695,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 182430
+    },
+    {
+      "epoch": 1200.2631578947369,
+      "grad_norm": 0.9362215399742126,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 182440
+    },
+    {
+      "epoch": 1200.328947368421,
+      "grad_norm": 0.7226735949516296,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 182450
+    },
+    {
+      "epoch": 1200.3947368421052,
+      "grad_norm": 1.1684337854385376,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 182460
+    },
+    {
+      "epoch": 1200.4605263157894,
+      "grad_norm": 1.1520575284957886,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 182470
+    },
+    {
+      "epoch": 1200.5263157894738,
+      "grad_norm": 1.0707088708877563,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 182480
+    },
+    {
+      "epoch": 1200.592105263158,
+      "grad_norm": 0.8464767336845398,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 182490
+    },
+    {
+      "epoch": 1200.657894736842,
+      "grad_norm": 1.186996579170227,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 182500
+    },
+    {
+      "epoch": 1200.7236842105262,
+      "grad_norm": 1.0348249673843384,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 182510
+    },
+    {
+      "epoch": 1200.7894736842106,
+      "grad_norm": 1.1279314756393433,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 182520
+    },
+    {
+      "epoch": 1200.8552631578948,
+      "grad_norm": 0.6997672319412231,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 182530
+    },
+    {
+      "epoch": 1200.921052631579,
+      "grad_norm": 0.8470837473869324,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 182540
+    },
+    {
+      "epoch": 1200.9868421052631,
+      "grad_norm": 1.178248405456543,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 182550
+    },
+    {
+      "epoch": 1201.0526315789473,
+      "grad_norm": 0.958922266960144,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 182560
+    },
+    {
+      "epoch": 1201.1184210526317,
+      "grad_norm": 0.8927515745162964,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 182570
+    },
+    {
+      "epoch": 1201.1842105263158,
+      "grad_norm": 1.0204429626464844,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 182580
+    },
+    {
+      "epoch": 1201.25,
+      "grad_norm": 0.7796499729156494,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 182590
+    },
+    {
+      "epoch": 1201.3157894736842,
+      "grad_norm": 0.7244349718093872,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 182600
+    },
+    {
+      "epoch": 1201.3815789473683,
+      "grad_norm": 0.921933114528656,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 182610
+    },
+    {
+      "epoch": 1201.4473684210527,
+      "grad_norm": 1.1822832822799683,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 182620
+    },
+    {
+      "epoch": 1201.5131578947369,
+      "grad_norm": 0.8840344548225403,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 182630
+    },
+    {
+      "epoch": 1201.578947368421,
+      "grad_norm": 1.2693527936935425,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 182640
+    },
+    {
+      "epoch": 1201.6447368421052,
+      "grad_norm": 1.1028697490692139,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 182650
+    },
+    {
+      "epoch": 1201.7105263157894,
+      "grad_norm": 0.9558727145195007,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 182660
+    },
+    {
+      "epoch": 1201.7763157894738,
+      "grad_norm": 1.0722416639328003,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 182670
+    },
+    {
+      "epoch": 1201.842105263158,
+      "grad_norm": 1.0818010568618774,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 182680
+    },
+    {
+      "epoch": 1201.907894736842,
+      "grad_norm": 0.9133005142211914,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 182690
+    },
+    {
+      "epoch": 1201.9736842105262,
+      "grad_norm": 0.8337648510932922,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 182700
+    },
+    {
+      "epoch": 1202.0394736842106,
+      "grad_norm": 1.0010408163070679,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 182710
+    },
+    {
+      "epoch": 1202.1052631578948,
+      "grad_norm": 1.2349421977996826,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 182720
+    },
+    {
+      "epoch": 1202.171052631579,
+      "grad_norm": 0.9251318573951721,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 182730
+    },
+    {
+      "epoch": 1202.2368421052631,
+      "grad_norm": 1.2574422359466553,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 182740
+    },
+    {
+      "epoch": 1202.3026315789473,
+      "grad_norm": 1.0148321390151978,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 182750
+    },
+    {
+      "epoch": 1202.3684210526317,
+      "grad_norm": 0.4935104548931122,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 182760
+    },
+    {
+      "epoch": 1202.4342105263158,
+      "grad_norm": 0.8940833806991577,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 182770
+    },
+    {
+      "epoch": 1202.5,
+      "grad_norm": 0.8071444034576416,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 182780
+    },
+    {
+      "epoch": 1202.5657894736842,
+      "grad_norm": 0.8158524036407471,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 182790
+    },
+    {
+      "epoch": 1202.6315789473683,
+      "grad_norm": 1.1093024015426636,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 182800
+    },
+    {
+      "epoch": 1202.6973684210527,
+      "grad_norm": 1.2366114854812622,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 182810
+    },
+    {
+      "epoch": 1202.7631578947369,
+      "grad_norm": 1.1967785358428955,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 182820
+    },
+    {
+      "epoch": 1202.828947368421,
+      "grad_norm": 1.1936818361282349,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 182830
+    },
+    {
+      "epoch": 1202.8947368421052,
+      "grad_norm": 1.7148412466049194,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 182840
+    },
+    {
+      "epoch": 1202.9605263157894,
+      "grad_norm": 1.4570777416229248,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 182850
+    },
+    {
+      "epoch": 1203.0263157894738,
+      "grad_norm": 1.157327651977539,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 182860
+    },
+    {
+      "epoch": 1203.092105263158,
+      "grad_norm": 1.0844436883926392,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 182870
+    },
+    {
+      "epoch": 1203.157894736842,
+      "grad_norm": 1.446609616279602,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 182880
+    },
+    {
+      "epoch": 1203.2236842105262,
+      "grad_norm": 0.9722105264663696,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 182890
+    },
+    {
+      "epoch": 1203.2894736842106,
+      "grad_norm": 0.864902675151825,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 182900
+    },
+    {
+      "epoch": 1203.3552631578948,
+      "grad_norm": 0.9735894203186035,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 182910
+    },
+    {
+      "epoch": 1203.421052631579,
+      "grad_norm": 0.9694196581840515,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 182920
+    },
+    {
+      "epoch": 1203.4868421052631,
+      "grad_norm": 1.1758695840835571,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 182930
+    },
+    {
+      "epoch": 1203.5526315789473,
+      "grad_norm": 0.9834887981414795,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 182940
+    },
+    {
+      "epoch": 1203.6184210526317,
+      "grad_norm": 1.0233968496322632,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 182950
+    },
+    {
+      "epoch": 1203.6842105263158,
+      "grad_norm": 0.826490044593811,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 182960
+    },
+    {
+      "epoch": 1203.75,
+      "grad_norm": 0.7306733727455139,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 182970
+    },
+    {
+      "epoch": 1203.8157894736842,
+      "grad_norm": 1.019582986831665,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 182980
+    },
+    {
+      "epoch": 1203.8815789473683,
+      "grad_norm": 1.1735548973083496,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 182990
+    },
+    {
+      "epoch": 1203.9473684210527,
+      "grad_norm": 1.2922629117965698,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 183000
+    },
+    {
+      "epoch": 1204.0131578947369,
+      "grad_norm": 1.170229434967041,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 183010
+    },
+    {
+      "epoch": 1204.078947368421,
+      "grad_norm": 1.1170730590820312,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 183020
+    },
+    {
+      "epoch": 1204.1447368421052,
+      "grad_norm": 1.4809566736221313,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 183030
+    },
+    {
+      "epoch": 1204.2105263157894,
+      "grad_norm": 1.3102599382400513,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 183040
+    },
+    {
+      "epoch": 1204.2763157894738,
+      "grad_norm": 1.1147892475128174,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 183050
+    },
+    {
+      "epoch": 1204.342105263158,
+      "grad_norm": 1.0742824077606201,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 183060
+    },
+    {
+      "epoch": 1204.407894736842,
+      "grad_norm": 1.0442990064620972,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 183070
+    },
+    {
+      "epoch": 1204.4736842105262,
+      "grad_norm": 0.6480495929718018,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 183080
+    },
+    {
+      "epoch": 1204.5394736842106,
+      "grad_norm": 1.12603759765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 183090
+    },
+    {
+      "epoch": 1204.6052631578948,
+      "grad_norm": 0.9349260926246643,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 183100
+    },
+    {
+      "epoch": 1204.671052631579,
+      "grad_norm": 1.4842894077301025,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 183110
+    },
+    {
+      "epoch": 1204.7368421052631,
+      "grad_norm": 1.003530502319336,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 183120
+    },
+    {
+      "epoch": 1204.8026315789473,
+      "grad_norm": 1.3181918859481812,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 183130
+    },
+    {
+      "epoch": 1204.8684210526317,
+      "grad_norm": 0.9079726934432983,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 183140
+    },
+    {
+      "epoch": 1204.9342105263158,
+      "grad_norm": 1.0107637643814087,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 183150
+    },
+    {
+      "epoch": 1205.0,
+      "grad_norm": 1.043506145477295,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 183160
+    },
+    {
+      "epoch": 1205.0657894736842,
+      "grad_norm": 0.9948956966400146,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 183170
+    },
+    {
+      "epoch": 1205.1315789473683,
+      "grad_norm": 0.8911522626876831,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 183180
+    },
+    {
+      "epoch": 1205.1973684210527,
+      "grad_norm": 0.9188016057014465,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 183190
+    },
+    {
+      "epoch": 1205.2631578947369,
+      "grad_norm": 0.9119938611984253,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 183200
+    },
+    {
+      "epoch": 1205.328947368421,
+      "grad_norm": 1.064100742340088,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 183210
+    },
+    {
+      "epoch": 1205.3947368421052,
+      "grad_norm": 1.1177867650985718,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 183220
+    },
+    {
+      "epoch": 1205.4605263157894,
+      "grad_norm": 0.7622190117835999,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 183230
+    },
+    {
+      "epoch": 1205.5263157894738,
+      "grad_norm": 0.9194398522377014,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 183240
+    },
+    {
+      "epoch": 1205.592105263158,
+      "grad_norm": 1.1614681482315063,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 183250
+    },
+    {
+      "epoch": 1205.657894736842,
+      "grad_norm": 1.296439290046692,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 183260
+    },
+    {
+      "epoch": 1205.7236842105262,
+      "grad_norm": 1.368787169456482,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 183270
+    },
+    {
+      "epoch": 1205.7894736842106,
+      "grad_norm": 1.1311159133911133,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 183280
+    },
+    {
+      "epoch": 1205.8552631578948,
+      "grad_norm": 1.0988332033157349,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 183290
+    },
+    {
+      "epoch": 1205.921052631579,
+      "grad_norm": 0.6983150839805603,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 183300
+    },
+    {
+      "epoch": 1205.9868421052631,
+      "grad_norm": 0.8750330805778503,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 183310
+    },
+    {
+      "epoch": 1206.0526315789473,
+      "grad_norm": 1.087065577507019,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 183320
+    },
+    {
+      "epoch": 1206.1184210526317,
+      "grad_norm": 0.9251524806022644,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 183330
+    },
+    {
+      "epoch": 1206.1842105263158,
+      "grad_norm": 0.998687744140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 183340
+    },
+    {
+      "epoch": 1206.25,
+      "grad_norm": 0.6075114011764526,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 183350
+    },
+    {
+      "epoch": 1206.3157894736842,
+      "grad_norm": 0.7300073504447937,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 183360
+    },
+    {
+      "epoch": 1206.3815789473683,
+      "grad_norm": 0.7422103881835938,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 183370
+    },
+    {
+      "epoch": 1206.4473684210527,
+      "grad_norm": 0.9821145534515381,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 183380
+    },
+    {
+      "epoch": 1206.5131578947369,
+      "grad_norm": 0.9073783159255981,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 183390
+    },
+    {
+      "epoch": 1206.578947368421,
+      "grad_norm": 0.854741096496582,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 183400
+    },
+    {
+      "epoch": 1206.6447368421052,
+      "grad_norm": 1.1214978694915771,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 183410
+    },
+    {
+      "epoch": 1206.7105263157894,
+      "grad_norm": 0.9807927012443542,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 183420
+    },
+    {
+      "epoch": 1206.7763157894738,
+      "grad_norm": 0.8988727927207947,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 183430
+    },
+    {
+      "epoch": 1206.842105263158,
+      "grad_norm": 0.840644896030426,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 183440
+    },
+    {
+      "epoch": 1206.907894736842,
+      "grad_norm": 1.0273395776748657,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 183450
+    },
+    {
+      "epoch": 1206.9736842105262,
+      "grad_norm": 1.0159443616867065,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 183460
+    },
+    {
+      "epoch": 1207.0394736842106,
+      "grad_norm": 1.0389869213104248,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 183470
+    },
+    {
+      "epoch": 1207.1052631578948,
+      "grad_norm": 0.9145666360855103,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 183480
+    },
+    {
+      "epoch": 1207.171052631579,
+      "grad_norm": 0.9767587184906006,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 183490
+    },
+    {
+      "epoch": 1207.2368421052631,
+      "grad_norm": 1.1191926002502441,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 183500
+    },
+    {
+      "epoch": 1207.3026315789473,
+      "grad_norm": 0.898288905620575,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 183510
+    },
+    {
+      "epoch": 1207.3684210526317,
+      "grad_norm": 0.7566203474998474,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 183520
+    },
+    {
+      "epoch": 1207.4342105263158,
+      "grad_norm": 1.134048342704773,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 183530
+    },
+    {
+      "epoch": 1207.5,
+      "grad_norm": 1.08811354637146,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 183540
+    },
+    {
+      "epoch": 1207.5657894736842,
+      "grad_norm": 1.2516783475875854,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 183550
+    },
+    {
+      "epoch": 1207.6315789473683,
+      "grad_norm": 1.2968010902404785,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 183560
+    },
+    {
+      "epoch": 1207.6973684210527,
+      "grad_norm": 0.9975584149360657,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 183570
+    },
+    {
+      "epoch": 1207.7631578947369,
+      "grad_norm": 1.1204050779342651,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 183580
+    },
+    {
+      "epoch": 1207.828947368421,
+      "grad_norm": 1.050008773803711,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 183590
+    },
+    {
+      "epoch": 1207.8947368421052,
+      "grad_norm": 0.832999050617218,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 183600
+    },
+    {
+      "epoch": 1207.9605263157894,
+      "grad_norm": 0.9619593024253845,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 183610
+    },
+    {
+      "epoch": 1208.0263157894738,
+      "grad_norm": 0.7968380451202393,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 183620
+    },
+    {
+      "epoch": 1208.092105263158,
+      "grad_norm": 0.9513537883758545,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 183630
+    },
+    {
+      "epoch": 1208.157894736842,
+      "grad_norm": 1.0646263360977173,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 183640
+    },
+    {
+      "epoch": 1208.2236842105262,
+      "grad_norm": 0.8924401998519897,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 183650
+    },
+    {
+      "epoch": 1208.2894736842106,
+      "grad_norm": 0.9576424360275269,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 183660
+    },
+    {
+      "epoch": 1208.3552631578948,
+      "grad_norm": 1.0064516067504883,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 183670
+    },
+    {
+      "epoch": 1208.421052631579,
+      "grad_norm": 1.2385472059249878,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 183680
+    },
+    {
+      "epoch": 1208.4868421052631,
+      "grad_norm": 1.1335607767105103,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 183690
+    },
+    {
+      "epoch": 1208.5526315789473,
+      "grad_norm": 1.0447804927825928,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 183700
+    },
+    {
+      "epoch": 1208.6184210526317,
+      "grad_norm": 0.9965493679046631,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 183710
+    },
+    {
+      "epoch": 1208.6842105263158,
+      "grad_norm": 0.8367345929145813,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 183720
+    },
+    {
+      "epoch": 1208.75,
+      "grad_norm": 1.0351852178573608,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 183730
+    },
+    {
+      "epoch": 1208.8157894736842,
+      "grad_norm": 0.9388834834098816,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 183740
+    },
+    {
+      "epoch": 1208.8815789473683,
+      "grad_norm": 0.9737825393676758,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 183750
+    },
+    {
+      "epoch": 1208.9473684210527,
+      "grad_norm": 1.0484927892684937,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 183760
+    },
+    {
+      "epoch": 1209.0131578947369,
+      "grad_norm": 0.9659373760223389,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 183770
+    },
+    {
+      "epoch": 1209.078947368421,
+      "grad_norm": 1.433643102645874,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 183780
+    },
+    {
+      "epoch": 1209.1447368421052,
+      "grad_norm": 0.7290152311325073,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 183790
+    },
+    {
+      "epoch": 1209.2105263157894,
+      "grad_norm": 1.0148568153381348,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 183800
+    },
+    {
+      "epoch": 1209.2763157894738,
+      "grad_norm": 1.4373056888580322,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 183810
+    },
+    {
+      "epoch": 1209.342105263158,
+      "grad_norm": 0.8225641846656799,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 183820
+    },
+    {
+      "epoch": 1209.407894736842,
+      "grad_norm": 1.236804485321045,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 183830
+    },
+    {
+      "epoch": 1209.4736842105262,
+      "grad_norm": 1.466613531112671,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 183840
+    },
+    {
+      "epoch": 1209.5394736842106,
+      "grad_norm": 1.2647264003753662,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 183850
+    },
+    {
+      "epoch": 1209.6052631578948,
+      "grad_norm": 0.9265692234039307,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 183860
+    },
+    {
+      "epoch": 1209.671052631579,
+      "grad_norm": 1.0921601057052612,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 183870
+    },
+    {
+      "epoch": 1209.7368421052631,
+      "grad_norm": 1.2621147632598877,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 183880
+    },
+    {
+      "epoch": 1209.8026315789473,
+      "grad_norm": 1.0996668338775635,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 183890
+    },
+    {
+      "epoch": 1209.8684210526317,
+      "grad_norm": 1.0238312482833862,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 183900
+    },
+    {
+      "epoch": 1209.9342105263158,
+      "grad_norm": 1.0571686029434204,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 183910
+    },
+    {
+      "epoch": 1210.0,
+      "grad_norm": 1.0116900205612183,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 183920
+    },
+    {
+      "epoch": 1210.0657894736842,
+      "grad_norm": 1.5669469833374023,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 183930
+    },
+    {
+      "epoch": 1210.1315789473683,
+      "grad_norm": 1.1383708715438843,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 183940
+    },
+    {
+      "epoch": 1210.1973684210527,
+      "grad_norm": 1.053849697113037,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 183950
+    },
+    {
+      "epoch": 1210.2631578947369,
+      "grad_norm": 0.9576135873794556,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 183960
+    },
+    {
+      "epoch": 1210.328947368421,
+      "grad_norm": 1.0139400959014893,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "step": 183970
+    },
+    {
+      "epoch": 1210.3947368421052,
+      "grad_norm": 0.9644301533699036,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 183980
+    },
+    {
+      "epoch": 1210.4605263157894,
+      "grad_norm": 0.9945560693740845,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 183990
+    },
+    {
+      "epoch": 1210.5263157894738,
+      "grad_norm": 1.0940089225769043,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 184000
+    },
+    {
+      "epoch": 1210.592105263158,
+      "grad_norm": 0.9017602801322937,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 184010
+    },
+    {
+      "epoch": 1210.657894736842,
+      "grad_norm": 0.8119890093803406,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 184020
+    },
+    {
+      "epoch": 1210.7236842105262,
+      "grad_norm": 0.7484346628189087,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 184030
+    },
+    {
+      "epoch": 1210.7894736842106,
+      "grad_norm": 0.6174297332763672,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 184040
+    },
+    {
+      "epoch": 1210.8552631578948,
+      "grad_norm": 0.9243563413619995,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 184050
+    },
+    {
+      "epoch": 1210.921052631579,
+      "grad_norm": 1.0928142070770264,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 184060
+    },
+    {
+      "epoch": 1210.9868421052631,
+      "grad_norm": 1.0260541439056396,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 184070
+    },
+    {
+      "epoch": 1211.0526315789473,
+      "grad_norm": 0.8601813316345215,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 184080
+    },
+    {
+      "epoch": 1211.1184210526317,
+      "grad_norm": 1.062245488166809,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 184090
+    },
+    {
+      "epoch": 1211.1842105263158,
+      "grad_norm": 0.8199011087417603,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 184100
+    },
+    {
+      "epoch": 1211.25,
+      "grad_norm": 0.9992640614509583,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 184110
+    },
+    {
+      "epoch": 1211.3157894736842,
+      "grad_norm": 1.1344804763793945,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 184120
+    },
+    {
+      "epoch": 1211.3815789473683,
+      "grad_norm": 0.9198042750358582,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 184130
+    },
+    {
+      "epoch": 1211.4473684210527,
+      "grad_norm": 1.2158046960830688,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 184140
+    },
+    {
+      "epoch": 1211.5131578947369,
+      "grad_norm": 1.368562936782837,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 184150
+    },
+    {
+      "epoch": 1211.578947368421,
+      "grad_norm": 0.9911054372787476,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 184160
+    },
+    {
+      "epoch": 1211.6447368421052,
+      "grad_norm": 1.076811671257019,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 184170
+    },
+    {
+      "epoch": 1211.7105263157894,
+      "grad_norm": 1.632210373878479,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 184180
+    },
+    {
+      "epoch": 1211.7763157894738,
+      "grad_norm": 1.2261195182800293,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 184190
+    },
+    {
+      "epoch": 1211.842105263158,
+      "grad_norm": 1.1790894269943237,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 184200
+    },
+    {
+      "epoch": 1211.907894736842,
+      "grad_norm": 0.8446111083030701,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 184210
+    },
+    {
+      "epoch": 1211.9736842105262,
+      "grad_norm": 0.72014319896698,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 184220
+    },
+    {
+      "epoch": 1212.0394736842106,
+      "grad_norm": 0.8168789744377136,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 184230
+    },
+    {
+      "epoch": 1212.1052631578948,
+      "grad_norm": 0.8248627781867981,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 184240
+    },
+    {
+      "epoch": 1212.171052631579,
+      "grad_norm": 0.6290561556816101,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 184250
+    },
+    {
+      "epoch": 1212.2368421052631,
+      "grad_norm": 0.9064732789993286,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 184260
+    },
+    {
+      "epoch": 1212.3026315789473,
+      "grad_norm": 1.5276134014129639,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 184270
+    },
+    {
+      "epoch": 1212.3684210526317,
+      "grad_norm": 1.2133339643478394,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 184280
+    },
+    {
+      "epoch": 1212.4342105263158,
+      "grad_norm": 1.096923589706421,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 184290
+    },
+    {
+      "epoch": 1212.5,
+      "grad_norm": 1.0609931945800781,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 184300
+    },
+    {
+      "epoch": 1212.5657894736842,
+      "grad_norm": 1.0884696245193481,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 184310
+    },
+    {
+      "epoch": 1212.6315789473683,
+      "grad_norm": 1.2373757362365723,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 184320
+    },
+    {
+      "epoch": 1212.6973684210527,
+      "grad_norm": 0.8663111925125122,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 184330
+    },
+    {
+      "epoch": 1212.7631578947369,
+      "grad_norm": 0.8895679116249084,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 184340
+    },
+    {
+      "epoch": 1212.828947368421,
+      "grad_norm": 0.9693534970283508,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 184350
+    },
+    {
+      "epoch": 1212.8947368421052,
+      "grad_norm": 1.173888921737671,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 184360
+    },
+    {
+      "epoch": 1212.9605263157894,
+      "grad_norm": 0.8452108502388,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 184370
+    },
+    {
+      "epoch": 1213.0263157894738,
+      "grad_norm": 1.301694393157959,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 184380
+    },
+    {
+      "epoch": 1213.092105263158,
+      "grad_norm": 1.157357931137085,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 184390
+    },
+    {
+      "epoch": 1213.157894736842,
+      "grad_norm": 0.9105615019798279,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 184400
+    },
+    {
+      "epoch": 1213.2236842105262,
+      "grad_norm": 1.3666666746139526,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 184410
+    },
+    {
+      "epoch": 1213.2894736842106,
+      "grad_norm": 1.4840264320373535,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 184420
+    },
+    {
+      "epoch": 1213.3552631578948,
+      "grad_norm": 1.0419845581054688,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 184430
+    },
+    {
+      "epoch": 1213.421052631579,
+      "grad_norm": 1.0189485549926758,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 184440
+    },
+    {
+      "epoch": 1213.4868421052631,
+      "grad_norm": 1.2533512115478516,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 184450
+    },
+    {
+      "epoch": 1213.5526315789473,
+      "grad_norm": 0.8926238417625427,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 184460
+    },
+    {
+      "epoch": 1213.6184210526317,
+      "grad_norm": 1.0614726543426514,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 184470
+    },
+    {
+      "epoch": 1213.6842105263158,
+      "grad_norm": 1.0628024339675903,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 184480
+    },
+    {
+      "epoch": 1213.75,
+      "grad_norm": 1.3046667575836182,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 184490
+    },
+    {
+      "epoch": 1213.8157894736842,
+      "grad_norm": 1.021803379058838,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 184500
+    },
+    {
+      "epoch": 1213.8815789473683,
+      "grad_norm": 1.1435785293579102,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 184510
+    },
+    {
+      "epoch": 1213.9473684210527,
+      "grad_norm": 1.1126259565353394,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 184520
+    },
+    {
+      "epoch": 1214.0131578947369,
+      "grad_norm": 1.026728630065918,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 184530
+    },
+    {
+      "epoch": 1214.078947368421,
+      "grad_norm": 0.6541023254394531,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 184540
+    },
+    {
+      "epoch": 1214.1447368421052,
+      "grad_norm": 0.8954967260360718,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 184550
+    },
+    {
+      "epoch": 1214.2105263157894,
+      "grad_norm": 1.3695642948150635,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 184560
+    },
+    {
+      "epoch": 1214.2763157894738,
+      "grad_norm": 1.1998350620269775,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 184570
+    },
+    {
+      "epoch": 1214.342105263158,
+      "grad_norm": 1.03053617477417,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 184580
+    },
+    {
+      "epoch": 1214.407894736842,
+      "grad_norm": 1.0831443071365356,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 184590
+    },
+    {
+      "epoch": 1214.4736842105262,
+      "grad_norm": 0.8419860005378723,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 184600
+    },
+    {
+      "epoch": 1214.5394736842106,
+      "grad_norm": 1.1064237356185913,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 184610
+    },
+    {
+      "epoch": 1214.6052631578948,
+      "grad_norm": 1.0451138019561768,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 184620
+    },
+    {
+      "epoch": 1214.671052631579,
+      "grad_norm": 0.9039570093154907,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 184630
+    },
+    {
+      "epoch": 1214.7368421052631,
+      "grad_norm": 1.0026644468307495,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 184640
+    },
+    {
+      "epoch": 1214.8026315789473,
+      "grad_norm": 0.8823190927505493,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 184650
+    },
+    {
+      "epoch": 1214.8684210526317,
+      "grad_norm": 1.510465383529663,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 184660
+    },
+    {
+      "epoch": 1214.9342105263158,
+      "grad_norm": 0.9225495457649231,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 184670
+    },
+    {
+      "epoch": 1215.0,
+      "grad_norm": 0.866723895072937,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 184680
+    },
+    {
+      "epoch": 1215.0657894736842,
+      "grad_norm": 1.0360679626464844,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 184690
+    },
+    {
+      "epoch": 1215.1315789473683,
+      "grad_norm": 1.0180871486663818,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 184700
+    },
+    {
+      "epoch": 1215.1973684210527,
+      "grad_norm": 0.8230453729629517,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 184710
+    },
+    {
+      "epoch": 1215.2631578947369,
+      "grad_norm": 1.398345947265625,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 184720
+    },
+    {
+      "epoch": 1215.328947368421,
+      "grad_norm": 1.5286283493041992,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 184730
+    },
+    {
+      "epoch": 1215.3947368421052,
+      "grad_norm": 1.097556710243225,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 184740
+    },
+    {
+      "epoch": 1215.4605263157894,
+      "grad_norm": 1.1276756525039673,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 184750
+    },
+    {
+      "epoch": 1215.5263157894738,
+      "grad_norm": 1.013379693031311,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 184760
+    },
+    {
+      "epoch": 1215.592105263158,
+      "grad_norm": 1.1437040567398071,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 184770
+    },
+    {
+      "epoch": 1215.657894736842,
+      "grad_norm": 0.8518174290657043,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 184780
+    },
+    {
+      "epoch": 1215.7236842105262,
+      "grad_norm": 1.0094599723815918,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 184790
+    },
+    {
+      "epoch": 1215.7894736842106,
+      "grad_norm": 1.0971324443817139,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 184800
+    },
+    {
+      "epoch": 1215.8552631578948,
+      "grad_norm": 1.0762872695922852,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 184810
+    },
+    {
+      "epoch": 1215.921052631579,
+      "grad_norm": 0.6933231353759766,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 184820
+    },
+    {
+      "epoch": 1215.9868421052631,
+      "grad_norm": 1.1836282014846802,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 184830
+    },
+    {
+      "epoch": 1216.0526315789473,
+      "grad_norm": 1.4843765497207642,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 184840
+    },
+    {
+      "epoch": 1216.1184210526317,
+      "grad_norm": 1.573395013809204,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 184850
+    },
+    {
+      "epoch": 1216.1842105263158,
+      "grad_norm": 1.349432110786438,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 184860
+    },
+    {
+      "epoch": 1216.25,
+      "grad_norm": 0.9781970381736755,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 184870
+    },
+    {
+      "epoch": 1216.3157894736842,
+      "grad_norm": 0.9339350461959839,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 184880
+    },
+    {
+      "epoch": 1216.3815789473683,
+      "grad_norm": 1.192179799079895,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 184890
+    },
+    {
+      "epoch": 1216.4473684210527,
+      "grad_norm": 0.9777132272720337,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 184900
+    },
+    {
+      "epoch": 1216.5131578947369,
+      "grad_norm": 1.1878960132598877,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 184910
+    },
+    {
+      "epoch": 1216.578947368421,
+      "grad_norm": 1.1332019567489624,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 184920
+    },
+    {
+      "epoch": 1216.6447368421052,
+      "grad_norm": 1.2628083229064941,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 184930
+    },
+    {
+      "epoch": 1216.7105263157894,
+      "grad_norm": 1.167966365814209,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 184940
+    },
+    {
+      "epoch": 1216.7763157894738,
+      "grad_norm": 1.3506724834442139,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 184950
+    },
+    {
+      "epoch": 1216.842105263158,
+      "grad_norm": 1.2869248390197754,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 184960
+    },
+    {
+      "epoch": 1216.907894736842,
+      "grad_norm": 1.2013112306594849,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 184970
+    },
+    {
+      "epoch": 1216.9736842105262,
+      "grad_norm": 0.8679819107055664,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 184980
+    },
+    {
+      "epoch": 1217.0394736842106,
+      "grad_norm": 1.0381109714508057,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 184990
+    },
+    {
+      "epoch": 1217.1052631578948,
+      "grad_norm": 1.11387038230896,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 185000
+    },
+    {
+      "epoch": 1217.171052631579,
+      "grad_norm": 1.2010740041732788,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 185010
+    },
+    {
+      "epoch": 1217.2368421052631,
+      "grad_norm": 1.2932566404342651,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 185020
+    },
+    {
+      "epoch": 1217.3026315789473,
+      "grad_norm": 0.7184022665023804,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 185030
+    },
+    {
+      "epoch": 1217.3684210526317,
+      "grad_norm": 0.8820980787277222,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 185040
+    },
+    {
+      "epoch": 1217.4342105263158,
+      "grad_norm": 1.4294477701187134,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 185050
+    },
+    {
+      "epoch": 1217.5,
+      "grad_norm": 1.082810878753662,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 185060
+    },
+    {
+      "epoch": 1217.5657894736842,
+      "grad_norm": 1.033191442489624,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 185070
+    },
+    {
+      "epoch": 1217.6315789473683,
+      "grad_norm": 1.1163650751113892,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 185080
+    },
+    {
+      "epoch": 1217.6973684210527,
+      "grad_norm": 1.2836352586746216,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 185090
+    },
+    {
+      "epoch": 1217.7631578947369,
+      "grad_norm": 1.1807303428649902,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 185100
+    },
+    {
+      "epoch": 1217.828947368421,
+      "grad_norm": 1.0362707376480103,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 185110
+    },
+    {
+      "epoch": 1217.8947368421052,
+      "grad_norm": 1.2636171579360962,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 185120
+    },
+    {
+      "epoch": 1217.9605263157894,
+      "grad_norm": 1.2539921998977661,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 185130
+    },
+    {
+      "epoch": 1218.0263157894738,
+      "grad_norm": 1.5820645093917847,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 185140
+    },
+    {
+      "epoch": 1218.092105263158,
+      "grad_norm": 0.8721255660057068,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 185150
+    },
+    {
+      "epoch": 1218.157894736842,
+      "grad_norm": 1.0245574712753296,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 185160
+    },
+    {
+      "epoch": 1218.2236842105262,
+      "grad_norm": 1.1424921751022339,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 185170
+    },
+    {
+      "epoch": 1218.2894736842106,
+      "grad_norm": 1.0817832946777344,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 185180
+    },
+    {
+      "epoch": 1218.3552631578948,
+      "grad_norm": 0.9851661920547485,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 185190
+    },
+    {
+      "epoch": 1218.421052631579,
+      "grad_norm": 1.0707749128341675,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 185200
+    },
+    {
+      "epoch": 1218.4868421052631,
+      "grad_norm": 0.903893768787384,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 185210
+    },
+    {
+      "epoch": 1218.5526315789473,
+      "grad_norm": 0.8631360530853271,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 185220
+    },
+    {
+      "epoch": 1218.6184210526317,
+      "grad_norm": 0.8446455001831055,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 185230
+    },
+    {
+      "epoch": 1218.6842105263158,
+      "grad_norm": 1.238858938217163,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 185240
+    },
+    {
+      "epoch": 1218.75,
+      "grad_norm": 1.4525588750839233,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 185250
+    },
+    {
+      "epoch": 1218.8157894736842,
+      "grad_norm": 0.7506535649299622,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 185260
+    },
+    {
+      "epoch": 1218.8815789473683,
+      "grad_norm": 1.0195599794387817,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 185270
+    },
+    {
+      "epoch": 1218.9473684210527,
+      "grad_norm": 1.0614231824874878,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 185280
+    },
+    {
+      "epoch": 1219.0131578947369,
+      "grad_norm": 0.9390943050384521,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 185290
+    },
+    {
+      "epoch": 1219.078947368421,
+      "grad_norm": 1.175963282585144,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 185300
+    },
+    {
+      "epoch": 1219.1447368421052,
+      "grad_norm": 1.5767515897750854,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 185310
+    },
+    {
+      "epoch": 1219.2105263157894,
+      "grad_norm": 1.0393364429473877,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 185320
+    },
+    {
+      "epoch": 1219.2763157894738,
+      "grad_norm": 1.084050178527832,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 185330
+    },
+    {
+      "epoch": 1219.342105263158,
+      "grad_norm": 1.1142622232437134,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 185340
+    },
+    {
+      "epoch": 1219.407894736842,
+      "grad_norm": 1.1076176166534424,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 185350
+    },
+    {
+      "epoch": 1219.4736842105262,
+      "grad_norm": 0.9142469167709351,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 185360
+    },
+    {
+      "epoch": 1219.5394736842106,
+      "grad_norm": 0.9830288887023926,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 185370
+    },
+    {
+      "epoch": 1219.6052631578948,
+      "grad_norm": 1.4786913394927979,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 185380
+    },
+    {
+      "epoch": 1219.671052631579,
+      "grad_norm": 1.2091387510299683,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 185390
+    },
+    {
+      "epoch": 1219.7368421052631,
+      "grad_norm": 1.051276683807373,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 185400
+    },
+    {
+      "epoch": 1219.8026315789473,
+      "grad_norm": 1.0199636220932007,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 185410
+    },
+    {
+      "epoch": 1219.8684210526317,
+      "grad_norm": 0.9174479246139526,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 185420
+    },
+    {
+      "epoch": 1219.9342105263158,
+      "grad_norm": 0.9192227721214294,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 185430
+    },
+    {
+      "epoch": 1220.0,
+      "grad_norm": 0.7987057566642761,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 185440
+    },
+    {
+      "epoch": 1220.0657894736842,
+      "grad_norm": 0.6848961710929871,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 185450
+    },
+    {
+      "epoch": 1220.1315789473683,
+      "grad_norm": 0.7822224497795105,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 185460
+    },
+    {
+      "epoch": 1220.1973684210527,
+      "grad_norm": 0.8950042724609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 185470
+    },
+    {
+      "epoch": 1220.2631578947369,
+      "grad_norm": 0.982462465763092,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 185480
+    },
+    {
+      "epoch": 1220.328947368421,
+      "grad_norm": 1.3260400295257568,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 185490
+    },
+    {
+      "epoch": 1220.3947368421052,
+      "grad_norm": 1.1258111000061035,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 185500
+    },
+    {
+      "epoch": 1220.4605263157894,
+      "grad_norm": 0.9215371608734131,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 185510
+    },
+    {
+      "epoch": 1220.5263157894738,
+      "grad_norm": 1.1126407384872437,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 185520
+    },
+    {
+      "epoch": 1220.592105263158,
+      "grad_norm": 0.8571876287460327,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 185530
+    },
+    {
+      "epoch": 1220.657894736842,
+      "grad_norm": 1.0084202289581299,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 185540
+    },
+    {
+      "epoch": 1220.7236842105262,
+      "grad_norm": 1.0304898023605347,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 185550
+    },
+    {
+      "epoch": 1220.7894736842106,
+      "grad_norm": 1.4203600883483887,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 185560
+    },
+    {
+      "epoch": 1220.8552631578948,
+      "grad_norm": 0.695814311504364,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 185570
+    },
+    {
+      "epoch": 1220.921052631579,
+      "grad_norm": 1.066893219947815,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 185580
+    },
+    {
+      "epoch": 1220.9868421052631,
+      "grad_norm": 1.0553237199783325,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 185590
+    },
+    {
+      "epoch": 1221.0526315789473,
+      "grad_norm": 1.2047573328018188,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 185600
+    },
+    {
+      "epoch": 1221.1184210526317,
+      "grad_norm": 1.031988263130188,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 185610
+    },
+    {
+      "epoch": 1221.1842105263158,
+      "grad_norm": 1.2671812772750854,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 185620
+    },
+    {
+      "epoch": 1221.25,
+      "grad_norm": 1.1938546895980835,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 185630
+    },
+    {
+      "epoch": 1221.3157894736842,
+      "grad_norm": 0.8404895067214966,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 185640
+    },
+    {
+      "epoch": 1221.3815789473683,
+      "grad_norm": 1.2843692302703857,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 185650
+    },
+    {
+      "epoch": 1221.4473684210527,
+      "grad_norm": 1.1143099069595337,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 185660
+    },
+    {
+      "epoch": 1221.5131578947369,
+      "grad_norm": 1.0575257539749146,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 185670
+    },
+    {
+      "epoch": 1221.578947368421,
+      "grad_norm": 0.7534637451171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 185680
+    },
+    {
+      "epoch": 1221.6447368421052,
+      "grad_norm": 0.7828577756881714,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 185690
+    },
+    {
+      "epoch": 1221.7105263157894,
+      "grad_norm": 0.9473252296447754,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 185700
+    },
+    {
+      "epoch": 1221.7763157894738,
+      "grad_norm": 1.1639002561569214,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 185710
+    },
+    {
+      "epoch": 1221.842105263158,
+      "grad_norm": 0.9955407977104187,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 185720
+    },
+    {
+      "epoch": 1221.907894736842,
+      "grad_norm": 1.1388436555862427,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 185730
+    },
+    {
+      "epoch": 1221.9736842105262,
+      "grad_norm": 0.793958842754364,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 185740
+    },
+    {
+      "epoch": 1222.0394736842106,
+      "grad_norm": 0.7527614235877991,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 185750
+    },
+    {
+      "epoch": 1222.1052631578948,
+      "grad_norm": 1.2795637845993042,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 185760
+    },
+    {
+      "epoch": 1222.171052631579,
+      "grad_norm": 0.7451528906822205,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 185770
+    },
+    {
+      "epoch": 1222.2368421052631,
+      "grad_norm": 0.9207596778869629,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 185780
+    },
+    {
+      "epoch": 1222.3026315789473,
+      "grad_norm": 0.8541901111602783,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 185790
+    },
+    {
+      "epoch": 1222.3684210526317,
+      "grad_norm": 0.9875356554985046,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 185800
+    },
+    {
+      "epoch": 1222.4342105263158,
+      "grad_norm": 0.8189533948898315,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 185810
+    },
+    {
+      "epoch": 1222.5,
+      "grad_norm": 1.3536303043365479,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 185820
+    },
+    {
+      "epoch": 1222.5657894736842,
+      "grad_norm": 1.253621220588684,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 185830
+    },
+    {
+      "epoch": 1222.6315789473683,
+      "grad_norm": 0.9833441376686096,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 185840
+    },
+    {
+      "epoch": 1222.6973684210527,
+      "grad_norm": 1.02719247341156,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 185850
+    },
+    {
+      "epoch": 1222.7631578947369,
+      "grad_norm": 0.8571970462799072,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 185860
+    },
+    {
+      "epoch": 1222.828947368421,
+      "grad_norm": 1.270446538925171,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 185870
+    },
+    {
+      "epoch": 1222.8947368421052,
+      "grad_norm": 1.1519957780838013,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 185880
+    },
+    {
+      "epoch": 1222.9605263157894,
+      "grad_norm": 1.149924397468567,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 185890
+    },
+    {
+      "epoch": 1223.0263157894738,
+      "grad_norm": 1.0503630638122559,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 185900
+    },
+    {
+      "epoch": 1223.092105263158,
+      "grad_norm": 1.033896565437317,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 185910
+    },
+    {
+      "epoch": 1223.157894736842,
+      "grad_norm": 1.530225396156311,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 185920
+    },
+    {
+      "epoch": 1223.2236842105262,
+      "grad_norm": 1.317832112312317,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 185930
+    },
+    {
+      "epoch": 1223.2894736842106,
+      "grad_norm": 1.1852115392684937,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 185940
+    },
+    {
+      "epoch": 1223.3552631578948,
+      "grad_norm": 1.0423250198364258,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 185950
+    },
+    {
+      "epoch": 1223.421052631579,
+      "grad_norm": 1.0078924894332886,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 185960
+    },
+    {
+      "epoch": 1223.4868421052631,
+      "grad_norm": 1.1069872379302979,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 185970
+    },
+    {
+      "epoch": 1223.5526315789473,
+      "grad_norm": 1.058595061302185,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 185980
+    },
+    {
+      "epoch": 1223.6184210526317,
+      "grad_norm": 1.1351420879364014,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 185990
+    },
+    {
+      "epoch": 1223.6842105263158,
+      "grad_norm": 1.0550578832626343,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 186000
+    },
+    {
+      "epoch": 1223.75,
+      "grad_norm": 0.7912912964820862,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 186010
+    },
+    {
+      "epoch": 1223.8157894736842,
+      "grad_norm": 1.267266035079956,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 186020
+    },
+    {
+      "epoch": 1223.8815789473683,
+      "grad_norm": 0.7863309979438782,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 186030
+    },
+    {
+      "epoch": 1223.9473684210527,
+      "grad_norm": 0.8405379056930542,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "step": 186040
+    },
+    {
+      "epoch": 1224.0131578947369,
+      "grad_norm": 0.7136426568031311,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 186050
+    },
+    {
+      "epoch": 1224.078947368421,
+      "grad_norm": 1.2677913904190063,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 186060
+    },
+    {
+      "epoch": 1224.1447368421052,
+      "grad_norm": 1.0475759506225586,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 186070
+    },
+    {
+      "epoch": 1224.2105263157894,
+      "grad_norm": 0.9506308436393738,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 186080
+    },
+    {
+      "epoch": 1224.2763157894738,
+      "grad_norm": 0.9816416501998901,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 186090
+    },
+    {
+      "epoch": 1224.342105263158,
+      "grad_norm": 0.6930526494979858,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 186100
+    },
+    {
+      "epoch": 1224.407894736842,
+      "grad_norm": 0.9830896258354187,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 186110
+    },
+    {
+      "epoch": 1224.4736842105262,
+      "grad_norm": 0.9465078115463257,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 186120
+    },
+    {
+      "epoch": 1224.5394736842106,
+      "grad_norm": 0.8693153858184814,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 186130
+    },
+    {
+      "epoch": 1224.6052631578948,
+      "grad_norm": 1.0287957191467285,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 186140
+    },
+    {
+      "epoch": 1224.671052631579,
+      "grad_norm": 0.7358331084251404,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 186150
+    },
+    {
+      "epoch": 1224.7368421052631,
+      "grad_norm": 0.6375886797904968,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 186160
+    },
+    {
+      "epoch": 1224.8026315789473,
+      "grad_norm": 1.0094578266143799,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 186170
+    },
+    {
+      "epoch": 1224.8684210526317,
+      "grad_norm": 1.2570643424987793,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 186180
+    },
+    {
+      "epoch": 1224.9342105263158,
+      "grad_norm": 0.9317885637283325,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 186190
+    },
+    {
+      "epoch": 1225.0,
+      "grad_norm": 0.8335903286933899,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 186200
+    },
+    {
+      "epoch": 1225.0657894736842,
+      "grad_norm": 1.1470184326171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 186210
+    },
+    {
+      "epoch": 1225.1315789473683,
+      "grad_norm": 1.1468976736068726,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 186220
+    },
+    {
+      "epoch": 1225.1973684210527,
+      "grad_norm": 1.1850436925888062,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 186230
+    },
+    {
+      "epoch": 1225.2631578947369,
+      "grad_norm": 1.3890918493270874,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 186240
+    },
+    {
+      "epoch": 1225.328947368421,
+      "grad_norm": 0.9903066754341125,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 186250
+    },
+    {
+      "epoch": 1225.3947368421052,
+      "grad_norm": 0.9706083536148071,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 186260
+    },
+    {
+      "epoch": 1225.4605263157894,
+      "grad_norm": 1.4619114398956299,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 186270
+    },
+    {
+      "epoch": 1225.5263157894738,
+      "grad_norm": 0.8819053173065186,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 186280
+    },
+    {
+      "epoch": 1225.592105263158,
+      "grad_norm": 0.8979442715644836,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 186290
+    },
+    {
+      "epoch": 1225.657894736842,
+      "grad_norm": 0.9516298770904541,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 186300
+    },
+    {
+      "epoch": 1225.7236842105262,
+      "grad_norm": 1.1198757886886597,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 186310
+    },
+    {
+      "epoch": 1225.7894736842106,
+      "grad_norm": 1.2486748695373535,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 186320
+    },
+    {
+      "epoch": 1225.8552631578948,
+      "grad_norm": 1.025698184967041,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 186330
+    },
+    {
+      "epoch": 1225.921052631579,
+      "grad_norm": 0.6521353125572205,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 186340
+    },
+    {
+      "epoch": 1225.9868421052631,
+      "grad_norm": 0.7902568578720093,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 186350
+    },
+    {
+      "epoch": 1226.0526315789473,
+      "grad_norm": 1.071388840675354,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 186360
+    },
+    {
+      "epoch": 1226.1184210526317,
+      "grad_norm": 0.9144065976142883,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 186370
+    },
+    {
+      "epoch": 1226.1842105263158,
+      "grad_norm": 1.1299687623977661,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 186380
+    },
+    {
+      "epoch": 1226.25,
+      "grad_norm": 1.0512938499450684,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 186390
+    },
+    {
+      "epoch": 1226.3157894736842,
+      "grad_norm": 0.8974717259407043,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 186400
+    },
+    {
+      "epoch": 1226.3815789473683,
+      "grad_norm": 0.9535547494888306,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 186410
+    },
+    {
+      "epoch": 1226.4473684210527,
+      "grad_norm": 1.0984748601913452,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 186420
+    },
+    {
+      "epoch": 1226.5131578947369,
+      "grad_norm": 1.2447376251220703,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 186430
+    },
+    {
+      "epoch": 1226.578947368421,
+      "grad_norm": 1.0115010738372803,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 186440
+    },
+    {
+      "epoch": 1226.6447368421052,
+      "grad_norm": 1.0430307388305664,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 186450
+    },
+    {
+      "epoch": 1226.7105263157894,
+      "grad_norm": 0.9099544882774353,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 186460
+    },
+    {
+      "epoch": 1226.7763157894738,
+      "grad_norm": 0.878151535987854,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 186470
+    },
+    {
+      "epoch": 1226.842105263158,
+      "grad_norm": 0.6946768164634705,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 186480
+    },
+    {
+      "epoch": 1226.907894736842,
+      "grad_norm": 1.0372909307479858,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 186490
+    },
+    {
+      "epoch": 1226.9736842105262,
+      "grad_norm": 0.8411084413528442,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 186500
+    },
+    {
+      "epoch": 1227.0394736842106,
+      "grad_norm": 0.7090082764625549,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 186510
+    },
+    {
+      "epoch": 1227.1052631578948,
+      "grad_norm": 0.802839457988739,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 186520
+    },
+    {
+      "epoch": 1227.171052631579,
+      "grad_norm": 0.8215923309326172,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 186530
+    },
+    {
+      "epoch": 1227.2368421052631,
+      "grad_norm": 0.8801704049110413,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 186540
+    },
+    {
+      "epoch": 1227.3026315789473,
+      "grad_norm": 1.15470552444458,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 186550
+    },
+    {
+      "epoch": 1227.3684210526317,
+      "grad_norm": 1.114592432975769,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 186560
+    },
+    {
+      "epoch": 1227.4342105263158,
+      "grad_norm": 1.192267656326294,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 186570
+    },
+    {
+      "epoch": 1227.5,
+      "grad_norm": 0.8330889940261841,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 186580
+    },
+    {
+      "epoch": 1227.5657894736842,
+      "grad_norm": 1.0712906122207642,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 186590
+    },
+    {
+      "epoch": 1227.6315789473683,
+      "grad_norm": 1.0091384649276733,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 186600
+    },
+    {
+      "epoch": 1227.6973684210527,
+      "grad_norm": 0.8430705070495605,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 186610
+    },
+    {
+      "epoch": 1227.7631578947369,
+      "grad_norm": 1.0369099378585815,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 186620
+    },
+    {
+      "epoch": 1227.828947368421,
+      "grad_norm": 1.0996822118759155,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 186630
+    },
+    {
+      "epoch": 1227.8947368421052,
+      "grad_norm": 1.3173720836639404,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 186640
+    },
+    {
+      "epoch": 1227.9605263157894,
+      "grad_norm": 1.1714130640029907,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 186650
+    },
+    {
+      "epoch": 1228.0263157894738,
+      "grad_norm": 1.1778959035873413,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 186660
+    },
+    {
+      "epoch": 1228.092105263158,
+      "grad_norm": 1.0300238132476807,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 186670
+    },
+    {
+      "epoch": 1228.157894736842,
+      "grad_norm": 0.9473615288734436,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 186680
+    },
+    {
+      "epoch": 1228.2236842105262,
+      "grad_norm": 1.171370267868042,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 186690
+    },
+    {
+      "epoch": 1228.2894736842106,
+      "grad_norm": 0.8373340964317322,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 186700
+    },
+    {
+      "epoch": 1228.3552631578948,
+      "grad_norm": 0.9212744832038879,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 186710
+    },
+    {
+      "epoch": 1228.421052631579,
+      "grad_norm": 1.379812240600586,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 186720
+    },
+    {
+      "epoch": 1228.4868421052631,
+      "grad_norm": 1.5933481454849243,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 186730
+    },
+    {
+      "epoch": 1228.5526315789473,
+      "grad_norm": 1.3223012685775757,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 186740
+    },
+    {
+      "epoch": 1228.6184210526317,
+      "grad_norm": 1.4536499977111816,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 186750
+    },
+    {
+      "epoch": 1228.6842105263158,
+      "grad_norm": 1.071915626525879,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 186760
+    },
+    {
+      "epoch": 1228.75,
+      "grad_norm": 1.0385130643844604,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 186770
+    },
+    {
+      "epoch": 1228.8157894736842,
+      "grad_norm": 1.3961035013198853,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 186780
+    },
+    {
+      "epoch": 1228.8815789473683,
+      "grad_norm": 1.3020026683807373,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 186790
+    },
+    {
+      "epoch": 1228.9473684210527,
+      "grad_norm": 1.341375708580017,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 186800
+    },
+    {
+      "epoch": 1229.0131578947369,
+      "grad_norm": 1.4074076414108276,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 186810
+    },
+    {
+      "epoch": 1229.078947368421,
+      "grad_norm": 0.9496380090713501,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 186820
+    },
+    {
+      "epoch": 1229.1447368421052,
+      "grad_norm": 1.0725762844085693,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 186830
+    },
+    {
+      "epoch": 1229.2105263157894,
+      "grad_norm": 0.9692988991737366,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 186840
+    },
+    {
+      "epoch": 1229.2763157894738,
+      "grad_norm": 0.9738875031471252,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 186850
+    },
+    {
+      "epoch": 1229.342105263158,
+      "grad_norm": 1.2129768133163452,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 186860
+    },
+    {
+      "epoch": 1229.407894736842,
+      "grad_norm": 0.7136619091033936,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 186870
+    },
+    {
+      "epoch": 1229.4736842105262,
+      "grad_norm": 1.1828364133834839,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 186880
+    },
+    {
+      "epoch": 1229.5394736842106,
+      "grad_norm": 1.1648519039154053,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 186890
+    },
+    {
+      "epoch": 1229.6052631578948,
+      "grad_norm": 0.9487463235855103,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "step": 186900
+    },
+    {
+      "epoch": 1229.671052631579,
+      "grad_norm": 1.2276160717010498,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 186910
+    },
+    {
+      "epoch": 1229.7368421052631,
+      "grad_norm": 0.9763864278793335,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 186920
+    },
+    {
+      "epoch": 1229.8026315789473,
+      "grad_norm": 0.8818470239639282,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 186930
+    },
+    {
+      "epoch": 1229.8684210526317,
+      "grad_norm": 0.7005442380905151,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 186940
+    },
+    {
+      "epoch": 1229.9342105263158,
+      "grad_norm": 0.5166844129562378,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 186950
+    },
+    {
+      "epoch": 1230.0,
+      "grad_norm": 0.9323204755783081,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 186960
+    },
+    {
+      "epoch": 1230.0657894736842,
+      "grad_norm": 1.0536624193191528,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 186970
+    },
+    {
+      "epoch": 1230.1315789473683,
+      "grad_norm": 0.9364405274391174,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 186980
+    },
+    {
+      "epoch": 1230.1973684210527,
+      "grad_norm": 0.9843827486038208,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 186990
+    },
+    {
+      "epoch": 1230.2631578947369,
+      "grad_norm": 1.1612672805786133,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 187000
+    },
+    {
+      "epoch": 1230.328947368421,
+      "grad_norm": 1.0113646984100342,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 187010
+    },
+    {
+      "epoch": 1230.3947368421052,
+      "grad_norm": 0.8987430334091187,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 187020
+    },
+    {
+      "epoch": 1230.4605263157894,
+      "grad_norm": 1.0051621198654175,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 187030
+    },
+    {
+      "epoch": 1230.5263157894738,
+      "grad_norm": 1.093727469444275,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 187040
+    },
+    {
+      "epoch": 1230.592105263158,
+      "grad_norm": 0.9883322715759277,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 187050
+    },
+    {
+      "epoch": 1230.657894736842,
+      "grad_norm": 0.8294075131416321,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 187060
+    },
+    {
+      "epoch": 1230.7236842105262,
+      "grad_norm": 0.8644272685050964,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 187070
+    },
+    {
+      "epoch": 1230.7894736842106,
+      "grad_norm": 0.7558962106704712,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 187080
+    },
+    {
+      "epoch": 1230.8552631578948,
+      "grad_norm": 0.6551307439804077,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 187090
+    },
+    {
+      "epoch": 1230.921052631579,
+      "grad_norm": 0.8390691876411438,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 187100
+    },
+    {
+      "epoch": 1230.9868421052631,
+      "grad_norm": 1.2046642303466797,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "step": 187110
+    },
+    {
+      "epoch": 1231.0526315789473,
+      "grad_norm": 0.6695582866668701,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 187120
+    },
+    {
+      "epoch": 1231.1184210526317,
+      "grad_norm": 0.7703352570533752,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 187130
+    },
+    {
+      "epoch": 1231.1842105263158,
+      "grad_norm": 0.9928059577941895,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 187140
+    },
+    {
+      "epoch": 1231.25,
+      "grad_norm": 0.8067262768745422,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 187150
+    },
+    {
+      "epoch": 1231.3157894736842,
+      "grad_norm": 0.8623690009117126,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 187160
+    },
+    {
+      "epoch": 1231.3815789473683,
+      "grad_norm": 0.8938329219818115,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 187170
+    },
+    {
+      "epoch": 1231.4473684210527,
+      "grad_norm": 1.010252594947815,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 187180
+    },
+    {
+      "epoch": 1231.5131578947369,
+      "grad_norm": 1.1144150495529175,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 187190
+    },
+    {
+      "epoch": 1231.578947368421,
+      "grad_norm": 0.7431076765060425,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 187200
+    },
+    {
+      "epoch": 1231.6447368421052,
+      "grad_norm": 0.8599720597267151,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 187210
+    },
+    {
+      "epoch": 1231.7105263157894,
+      "grad_norm": 1.280928134918213,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 187220
+    },
+    {
+      "epoch": 1231.7763157894738,
+      "grad_norm": 0.8571041822433472,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 187230
+    },
+    {
+      "epoch": 1231.842105263158,
+      "grad_norm": 0.6886258125305176,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 187240
+    },
+    {
+      "epoch": 1231.907894736842,
+      "grad_norm": 0.9688119888305664,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 187250
+    },
+    {
+      "epoch": 1231.9736842105262,
+      "grad_norm": 1.306778907775879,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 187260
+    },
+    {
+      "epoch": 1232.0394736842106,
+      "grad_norm": 1.23270845413208,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 187270
+    },
+    {
+      "epoch": 1232.1052631578948,
+      "grad_norm": 1.2252572774887085,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 187280
+    },
+    {
+      "epoch": 1232.171052631579,
+      "grad_norm": 0.9559009075164795,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 187290
+    },
+    {
+      "epoch": 1232.2368421052631,
+      "grad_norm": 0.895037829875946,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 187300
+    },
+    {
+      "epoch": 1232.3026315789473,
+      "grad_norm": 1.1871989965438843,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 187310
+    },
+    {
+      "epoch": 1232.3684210526317,
+      "grad_norm": 1.0235185623168945,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 187320
+    },
+    {
+      "epoch": 1232.4342105263158,
+      "grad_norm": 0.5289400815963745,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 187330
+    },
+    {
+      "epoch": 1232.5,
+      "grad_norm": 1.077367901802063,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 187340
+    },
+    {
+      "epoch": 1232.5657894736842,
+      "grad_norm": 1.0103833675384521,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 187350
+    },
+    {
+      "epoch": 1232.6315789473683,
+      "grad_norm": 1.0976219177246094,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 187360
+    },
+    {
+      "epoch": 1232.6973684210527,
+      "grad_norm": 0.8010794520378113,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 187370
+    },
+    {
+      "epoch": 1232.7631578947369,
+      "grad_norm": 0.8650757074356079,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 187380
+    },
+    {
+      "epoch": 1232.828947368421,
+      "grad_norm": 0.9312655329704285,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 187390
+    },
+    {
+      "epoch": 1232.8947368421052,
+      "grad_norm": 0.9070029854774475,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 187400
+    },
+    {
+      "epoch": 1232.9605263157894,
+      "grad_norm": 0.7533953189849854,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 187410
+    },
+    {
+      "epoch": 1233.0263157894738,
+      "grad_norm": 0.9384796023368835,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 187420
+    },
+    {
+      "epoch": 1233.092105263158,
+      "grad_norm": 1.097226858139038,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 187430
+    },
+    {
+      "epoch": 1233.157894736842,
+      "grad_norm": 1.156470537185669,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 187440
+    },
+    {
+      "epoch": 1233.2236842105262,
+      "grad_norm": 0.7621616125106812,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 187450
+    },
+    {
+      "epoch": 1233.2894736842106,
+      "grad_norm": 1.3856050968170166,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 187460
+    },
+    {
+      "epoch": 1233.3552631578948,
+      "grad_norm": 0.9687317609786987,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 187470
+    },
+    {
+      "epoch": 1233.421052631579,
+      "grad_norm": 0.9934854507446289,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 187480
+    },
+    {
+      "epoch": 1233.4868421052631,
+      "grad_norm": 1.1821582317352295,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 187490
+    },
+    {
+      "epoch": 1233.5526315789473,
+      "grad_norm": 0.8907933235168457,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 187500
+    },
+    {
+      "epoch": 1233.6184210526317,
+      "grad_norm": 0.7803938388824463,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 187510
+    },
+    {
+      "epoch": 1233.6842105263158,
+      "grad_norm": 0.8478217124938965,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 187520
+    },
+    {
+      "epoch": 1233.75,
+      "grad_norm": 0.9070845246315002,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 187530
+    },
+    {
+      "epoch": 1233.8157894736842,
+      "grad_norm": 0.7729270458221436,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 187540
+    },
+    {
+      "epoch": 1233.8815789473683,
+      "grad_norm": 0.689509391784668,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 187550
+    },
+    {
+      "epoch": 1233.9473684210527,
+      "grad_norm": 1.0114260911941528,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 187560
+    },
+    {
+      "epoch": 1234.0131578947369,
+      "grad_norm": 0.8790208697319031,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 187570
+    },
+    {
+      "epoch": 1234.078947368421,
+      "grad_norm": 0.7868586182594299,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 187580
+    },
+    {
+      "epoch": 1234.1447368421052,
+      "grad_norm": 1.1828399896621704,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 187590
+    },
+    {
+      "epoch": 1234.2105263157894,
+      "grad_norm": 2.568375825881958,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 187600
+    },
+    {
+      "epoch": 1234.2763157894738,
+      "grad_norm": 1.1999150514602661,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 187610
+    },
+    {
+      "epoch": 1234.342105263158,
+      "grad_norm": 1.172085165977478,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 187620
+    },
+    {
+      "epoch": 1234.407894736842,
+      "grad_norm": 1.260359287261963,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 187630
+    },
+    {
+      "epoch": 1234.4736842105262,
+      "grad_norm": 0.9659546613693237,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 187640
+    },
+    {
+      "epoch": 1234.5394736842106,
+      "grad_norm": 1.1871107816696167,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 187650
+    },
+    {
+      "epoch": 1234.6052631578948,
+      "grad_norm": 0.8396361470222473,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 187660
+    },
+    {
+      "epoch": 1234.671052631579,
+      "grad_norm": 0.862738311290741,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 187670
+    },
+    {
+      "epoch": 1234.7368421052631,
+      "grad_norm": 0.8556384444236755,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 187680
+    },
+    {
+      "epoch": 1234.8026315789473,
+      "grad_norm": 1.10312819480896,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 187690
+    },
+    {
+      "epoch": 1234.8684210526317,
+      "grad_norm": 0.9752662777900696,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 187700
+    },
+    {
+      "epoch": 1234.9342105263158,
+      "grad_norm": 0.7204729914665222,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 187710
+    },
+    {
+      "epoch": 1235.0,
+      "grad_norm": 0.7936938405036926,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 187720
+    },
+    {
+      "epoch": 1235.0657894736842,
+      "grad_norm": 0.8540555238723755,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 187730
+    },
+    {
+      "epoch": 1235.1315789473683,
+      "grad_norm": 1.0517683029174805,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 187740
+    },
+    {
+      "epoch": 1235.1973684210527,
+      "grad_norm": 0.9040259718894958,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 187750
+    },
+    {
+      "epoch": 1235.2631578947369,
+      "grad_norm": 1.0970168113708496,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 187760
+    },
+    {
+      "epoch": 1235.328947368421,
+      "grad_norm": 0.7497133016586304,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 187770
+    },
+    {
+      "epoch": 1235.3947368421052,
+      "grad_norm": 1.210512638092041,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 187780
+    },
+    {
+      "epoch": 1235.4605263157894,
+      "grad_norm": 1.1530228853225708,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 187790
+    },
+    {
+      "epoch": 1235.5263157894738,
+      "grad_norm": 1.0944901704788208,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 187800
+    },
+    {
+      "epoch": 1235.592105263158,
+      "grad_norm": 0.8939099311828613,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 187810
+    },
+    {
+      "epoch": 1235.657894736842,
+      "grad_norm": 0.756377100944519,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 187820
+    },
+    {
+      "epoch": 1235.7236842105262,
+      "grad_norm": 0.9245010614395142,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 187830
+    },
+    {
+      "epoch": 1235.7894736842106,
+      "grad_norm": 1.1319273710250854,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 187840
+    },
+    {
+      "epoch": 1235.8552631578948,
+      "grad_norm": 1.1174309253692627,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 187850
+    },
+    {
+      "epoch": 1235.921052631579,
+      "grad_norm": 0.8425665497779846,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 187860
+    },
+    {
+      "epoch": 1235.9868421052631,
+      "grad_norm": 0.9204140901565552,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 187870
+    },
+    {
+      "epoch": 1236.0526315789473,
+      "grad_norm": 0.7384737730026245,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 187880
+    },
+    {
+      "epoch": 1236.1184210526317,
+      "grad_norm": 1.1339808702468872,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 187890
+    },
+    {
+      "epoch": 1236.1842105263158,
+      "grad_norm": 1.197279453277588,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 187900
+    },
+    {
+      "epoch": 1236.25,
+      "grad_norm": 0.8099234700202942,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 187910
+    },
+    {
+      "epoch": 1236.3157894736842,
+      "grad_norm": 0.9687013626098633,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 187920
+    },
+    {
+      "epoch": 1236.3815789473683,
+      "grad_norm": 0.9288181066513062,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 187930
+    },
+    {
+      "epoch": 1236.4473684210527,
+      "grad_norm": 1.2456539869308472,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 187940
+    },
+    {
+      "epoch": 1236.5131578947369,
+      "grad_norm": 1.1154669523239136,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 187950
+    },
+    {
+      "epoch": 1236.578947368421,
+      "grad_norm": 1.3738938570022583,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 187960
+    },
+    {
+      "epoch": 1236.6447368421052,
+      "grad_norm": 1.0610955953598022,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 187970
+    },
+    {
+      "epoch": 1236.7105263157894,
+      "grad_norm": 1.108542799949646,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 187980
+    },
+    {
+      "epoch": 1236.7763157894738,
+      "grad_norm": 0.896397054195404,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 187990
+    },
+    {
+      "epoch": 1236.842105263158,
+      "grad_norm": 0.7802370190620422,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 188000
+    },
+    {
+      "epoch": 1236.907894736842,
+      "grad_norm": 0.5918530225753784,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 188010
+    },
+    {
+      "epoch": 1236.9736842105262,
+      "grad_norm": 1.0109541416168213,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 188020
+    },
+    {
+      "epoch": 1237.0394736842106,
+      "grad_norm": 0.9148463010787964,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 188030
+    },
+    {
+      "epoch": 1237.1052631578948,
+      "grad_norm": 0.9371543526649475,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 188040
+    },
+    {
+      "epoch": 1237.171052631579,
+      "grad_norm": 0.9670152068138123,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 188050
+    },
+    {
+      "epoch": 1237.2368421052631,
+      "grad_norm": 0.973000705242157,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 188060
+    },
+    {
+      "epoch": 1237.3026315789473,
+      "grad_norm": 0.9884973764419556,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 188070
+    },
+    {
+      "epoch": 1237.3684210526317,
+      "grad_norm": 1.0578480958938599,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 188080
+    },
+    {
+      "epoch": 1237.4342105263158,
+      "grad_norm": 0.9570189118385315,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 188090
+    },
+    {
+      "epoch": 1237.5,
+      "grad_norm": 1.118773102760315,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 188100
+    },
+    {
+      "epoch": 1237.5657894736842,
+      "grad_norm": 1.1802111864089966,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 188110
+    },
+    {
+      "epoch": 1237.6315789473683,
+      "grad_norm": 0.8796427845954895,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 188120
+    },
+    {
+      "epoch": 1237.6973684210527,
+      "grad_norm": 1.2683212757110596,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 188130
+    },
+    {
+      "epoch": 1237.7631578947369,
+      "grad_norm": 1.0439941883087158,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 188140
+    },
+    {
+      "epoch": 1237.828947368421,
+      "grad_norm": 0.969838559627533,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 188150
+    },
+    {
+      "epoch": 1237.8947368421052,
+      "grad_norm": 1.0440396070480347,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 188160
+    },
+    {
+      "epoch": 1237.9605263157894,
+      "grad_norm": 1.1314268112182617,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 188170
+    },
+    {
+      "epoch": 1238.0263157894738,
+      "grad_norm": 1.2484370470046997,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 188180
+    },
+    {
+      "epoch": 1238.092105263158,
+      "grad_norm": 0.7492143511772156,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 188190
+    },
+    {
+      "epoch": 1238.157894736842,
+      "grad_norm": 1.2526135444641113,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 188200
+    },
+    {
+      "epoch": 1238.2236842105262,
+      "grad_norm": 1.2862526178359985,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 188210
+    },
+    {
+      "epoch": 1238.2894736842106,
+      "grad_norm": 1.1841567754745483,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 188220
+    },
+    {
+      "epoch": 1238.3552631578948,
+      "grad_norm": 1.1451932191848755,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 188230
+    },
+    {
+      "epoch": 1238.421052631579,
+      "grad_norm": 0.9696784019470215,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 188240
+    },
+    {
+      "epoch": 1238.4868421052631,
+      "grad_norm": 1.0971126556396484,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 188250
+    },
+    {
+      "epoch": 1238.5526315789473,
+      "grad_norm": 0.7748174071311951,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 188260
+    },
+    {
+      "epoch": 1238.6184210526317,
+      "grad_norm": 0.851367712020874,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 188270
+    },
+    {
+      "epoch": 1238.6842105263158,
+      "grad_norm": 0.7852613925933838,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 188280
+    },
+    {
+      "epoch": 1238.75,
+      "grad_norm": 0.9901844263076782,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 188290
+    },
+    {
+      "epoch": 1238.8157894736842,
+      "grad_norm": 1.0415254831314087,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 188300
+    },
+    {
+      "epoch": 1238.8815789473683,
+      "grad_norm": 1.2643136978149414,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 188310
+    },
+    {
+      "epoch": 1238.9473684210527,
+      "grad_norm": 1.2420594692230225,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 188320
+    },
+    {
+      "epoch": 1239.0131578947369,
+      "grad_norm": 0.9030695557594299,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 188330
+    },
+    {
+      "epoch": 1239.078947368421,
+      "grad_norm": 0.8474811911582947,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 188340
+    },
+    {
+      "epoch": 1239.1447368421052,
+      "grad_norm": 0.9705641865730286,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 188350
+    },
+    {
+      "epoch": 1239.2105263157894,
+      "grad_norm": 0.9619945883750916,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 188360
+    },
+    {
+      "epoch": 1239.2763157894738,
+      "grad_norm": 1.1420098543167114,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 188370
+    },
+    {
+      "epoch": 1239.342105263158,
+      "grad_norm": 1.0340029001235962,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 188380
+    },
+    {
+      "epoch": 1239.407894736842,
+      "grad_norm": 0.9889524579048157,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 188390
+    },
+    {
+      "epoch": 1239.4736842105262,
+      "grad_norm": 1.768046259880066,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 188400
+    },
+    {
+      "epoch": 1239.5394736842106,
+      "grad_norm": 1.4465337991714478,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 188410
+    },
+    {
+      "epoch": 1239.6052631578948,
+      "grad_norm": 1.0993311405181885,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 188420
+    },
+    {
+      "epoch": 1239.671052631579,
+      "grad_norm": 1.098059892654419,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 188430
+    },
+    {
+      "epoch": 1239.7368421052631,
+      "grad_norm": 1.0775386095046997,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 188440
+    },
+    {
+      "epoch": 1239.8026315789473,
+      "grad_norm": 0.6998803019523621,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 188450
+    },
+    {
+      "epoch": 1239.8684210526317,
+      "grad_norm": 0.9273382425308228,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 188460
+    },
+    {
+      "epoch": 1239.9342105263158,
+      "grad_norm": 1.2026686668395996,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 188470
+    },
+    {
+      "epoch": 1240.0,
+      "grad_norm": 0.9668415784835815,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 188480
+    },
+    {
+      "epoch": 1240.0657894736842,
+      "grad_norm": 1.1113210916519165,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 188490
+    },
+    {
+      "epoch": 1240.1315789473683,
+      "grad_norm": 1.2677863836288452,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 188500
+    },
+    {
+      "epoch": 1240.1973684210527,
+      "grad_norm": 1.4234356880187988,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 188510
+    },
+    {
+      "epoch": 1240.2631578947369,
+      "grad_norm": 1.1672581434249878,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 188520
+    },
+    {
+      "epoch": 1240.328947368421,
+      "grad_norm": 1.075741171836853,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 188530
+    },
+    {
+      "epoch": 1240.3947368421052,
+      "grad_norm": 0.8153349161148071,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 188540
+    },
+    {
+      "epoch": 1240.4605263157894,
+      "grad_norm": 1.0249247550964355,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 188550
+    },
+    {
+      "epoch": 1240.5263157894738,
+      "grad_norm": 1.0031867027282715,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 188560
+    },
+    {
+      "epoch": 1240.592105263158,
+      "grad_norm": 1.103637933731079,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 188570
+    },
+    {
+      "epoch": 1240.657894736842,
+      "grad_norm": 1.0994689464569092,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 188580
+    },
+    {
+      "epoch": 1240.7236842105262,
+      "grad_norm": 1.0447285175323486,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 188590
+    },
+    {
+      "epoch": 1240.7894736842106,
+      "grad_norm": 0.9495458006858826,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 188600
+    },
+    {
+      "epoch": 1240.8552631578948,
+      "grad_norm": 0.9653858542442322,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 188610
+    },
+    {
+      "epoch": 1240.921052631579,
+      "grad_norm": 1.0502097606658936,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 188620
+    },
+    {
+      "epoch": 1240.9868421052631,
+      "grad_norm": 0.9432629942893982,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 188630
+    },
+    {
+      "epoch": 1241.0526315789473,
+      "grad_norm": 0.9657227396965027,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 188640
+    },
+    {
+      "epoch": 1241.1184210526317,
+      "grad_norm": 0.8866987228393555,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 188650
+    },
+    {
+      "epoch": 1241.1842105263158,
+      "grad_norm": 0.9814696907997131,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 188660
+    },
+    {
+      "epoch": 1241.25,
+      "grad_norm": 1.1600935459136963,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 188670
+    },
+    {
+      "epoch": 1241.3157894736842,
+      "grad_norm": 1.2622950077056885,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 188680
+    },
+    {
+      "epoch": 1241.3815789473683,
+      "grad_norm": 0.7318976521492004,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 188690
+    },
+    {
+      "epoch": 1241.4473684210527,
+      "grad_norm": 1.3622748851776123,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 188700
+    },
+    {
+      "epoch": 1241.5131578947369,
+      "grad_norm": 1.0387088060379028,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 188710
+    },
+    {
+      "epoch": 1241.578947368421,
+      "grad_norm": 0.8503026366233826,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 188720
+    },
+    {
+      "epoch": 1241.6447368421052,
+      "grad_norm": 0.909067690372467,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 188730
+    },
+    {
+      "epoch": 1241.7105263157894,
+      "grad_norm": 0.9165451526641846,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 188740
+    },
+    {
+      "epoch": 1241.7763157894738,
+      "grad_norm": 0.7640308737754822,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 188750
+    },
+    {
+      "epoch": 1241.842105263158,
+      "grad_norm": 0.9033883810043335,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 188760
+    },
+    {
+      "epoch": 1241.907894736842,
+      "grad_norm": 1.0701746940612793,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 188770
+    },
+    {
+      "epoch": 1241.9736842105262,
+      "grad_norm": 0.9798920154571533,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 188780
+    },
+    {
+      "epoch": 1242.0394736842106,
+      "grad_norm": 1.3275381326675415,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 188790
+    },
+    {
+      "epoch": 1242.1052631578948,
+      "grad_norm": 1.1964999437332153,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 188800
+    },
+    {
+      "epoch": 1242.171052631579,
+      "grad_norm": 1.0140482187271118,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 188810
+    },
+    {
+      "epoch": 1242.2368421052631,
+      "grad_norm": 0.8349454998970032,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 188820
+    },
+    {
+      "epoch": 1242.3026315789473,
+      "grad_norm": 1.2491860389709473,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 188830
+    },
+    {
+      "epoch": 1242.3684210526317,
+      "grad_norm": 1.1465342044830322,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 188840
+    },
+    {
+      "epoch": 1242.4342105263158,
+      "grad_norm": 0.7169458866119385,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 188850
+    },
+    {
+      "epoch": 1242.5,
+      "grad_norm": 0.887615978717804,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 188860
+    },
+    {
+      "epoch": 1242.5657894736842,
+      "grad_norm": 1.243327260017395,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 188870
+    },
+    {
+      "epoch": 1242.6315789473683,
+      "grad_norm": 1.3320894241333008,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 188880
+    },
+    {
+      "epoch": 1242.6973684210527,
+      "grad_norm": 1.0858453512191772,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 188890
+    },
+    {
+      "epoch": 1242.7631578947369,
+      "grad_norm": 1.123133659362793,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 188900
+    },
+    {
+      "epoch": 1242.828947368421,
+      "grad_norm": 0.7207273840904236,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 188910
+    },
+    {
+      "epoch": 1242.8947368421052,
+      "grad_norm": 0.6698071956634521,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 188920
+    },
+    {
+      "epoch": 1242.9605263157894,
+      "grad_norm": 1.2865219116210938,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 188930
+    },
+    {
+      "epoch": 1243.0263157894738,
+      "grad_norm": 0.8424243927001953,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 188940
+    },
+    {
+      "epoch": 1243.092105263158,
+      "grad_norm": 1.2220584154129028,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 188950
+    },
+    {
+      "epoch": 1243.157894736842,
+      "grad_norm": 1.234723687171936,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 188960
+    },
+    {
+      "epoch": 1243.2236842105262,
+      "grad_norm": 1.3230215311050415,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 188970
+    },
+    {
+      "epoch": 1243.2894736842106,
+      "grad_norm": 1.1419968605041504,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 188980
+    },
+    {
+      "epoch": 1243.3552631578948,
+      "grad_norm": 1.2557247877120972,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 188990
+    },
+    {
+      "epoch": 1243.421052631579,
+      "grad_norm": 1.0815895795822144,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 189000
+    },
+    {
+      "epoch": 1243.4868421052631,
+      "grad_norm": 0.997104823589325,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 189010
+    },
+    {
+      "epoch": 1243.5526315789473,
+      "grad_norm": 0.776548445224762,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 189020
+    },
+    {
+      "epoch": 1243.6184210526317,
+      "grad_norm": 0.8415836691856384,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 189030
+    },
+    {
+      "epoch": 1243.6842105263158,
+      "grad_norm": 0.7960478663444519,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 189040
+    },
+    {
+      "epoch": 1243.75,
+      "grad_norm": 0.7596374750137329,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 189050
+    },
+    {
+      "epoch": 1243.8157894736842,
+      "grad_norm": 1.2631161212921143,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 189060
+    },
+    {
+      "epoch": 1243.8815789473683,
+      "grad_norm": 0.959852397441864,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 189070
+    },
+    {
+      "epoch": 1243.9473684210527,
+      "grad_norm": 0.6801808476448059,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 189080
+    },
+    {
+      "epoch": 1244.0131578947369,
+      "grad_norm": 1.1035255193710327,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 189090
+    },
+    {
+      "epoch": 1244.078947368421,
+      "grad_norm": 0.7732435464859009,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 189100
+    },
+    {
+      "epoch": 1244.1447368421052,
+      "grad_norm": 1.119423747062683,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 189110
+    },
+    {
+      "epoch": 1244.2105263157894,
+      "grad_norm": 1.1290099620819092,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 189120
+    },
+    {
+      "epoch": 1244.2763157894738,
+      "grad_norm": 0.9945160150527954,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 189130
+    },
+    {
+      "epoch": 1244.342105263158,
+      "grad_norm": 1.1095064878463745,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 189140
+    },
+    {
+      "epoch": 1244.407894736842,
+      "grad_norm": 0.930774450302124,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 189150
+    },
+    {
+      "epoch": 1244.4736842105262,
+      "grad_norm": 0.7640677094459534,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 189160
+    },
+    {
+      "epoch": 1244.5394736842106,
+      "grad_norm": 0.8537974953651428,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 189170
+    },
+    {
+      "epoch": 1244.6052631578948,
+      "grad_norm": 0.8385094404220581,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 189180
+    },
+    {
+      "epoch": 1244.671052631579,
+      "grad_norm": 0.8964349627494812,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 189190
+    },
+    {
+      "epoch": 1244.7368421052631,
+      "grad_norm": 0.9953314065933228,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 189200
+    },
+    {
+      "epoch": 1244.8026315789473,
+      "grad_norm": 1.099317193031311,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 189210
+    },
+    {
+      "epoch": 1244.8684210526317,
+      "grad_norm": 1.2956123352050781,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 189220
+    },
+    {
+      "epoch": 1244.9342105263158,
+      "grad_norm": 1.0341782569885254,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 189230
+    },
+    {
+      "epoch": 1245.0,
+      "grad_norm": 1.3250845670700073,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 189240
+    },
+    {
+      "epoch": 1245.0657894736842,
+      "grad_norm": 0.8931382298469543,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 189250
+    },
+    {
+      "epoch": 1245.1315789473683,
+      "grad_norm": 0.8672465682029724,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 189260
+    },
+    {
+      "epoch": 1245.1973684210527,
+      "grad_norm": 1.1268553733825684,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 189270
+    },
+    {
+      "epoch": 1245.2631578947369,
+      "grad_norm": 0.9678467512130737,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 189280
+    },
+    {
+      "epoch": 1245.328947368421,
+      "grad_norm": 1.09657883644104,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 189290
+    },
+    {
+      "epoch": 1245.3947368421052,
+      "grad_norm": 1.1582295894622803,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 189300
+    },
+    {
+      "epoch": 1245.4605263157894,
+      "grad_norm": 0.817107617855072,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 189310
+    },
+    {
+      "epoch": 1245.5263157894738,
+      "grad_norm": 1.2293273210525513,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 189320
+    },
+    {
+      "epoch": 1245.592105263158,
+      "grad_norm": 0.8553192019462585,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 189330
+    },
+    {
+      "epoch": 1245.657894736842,
+      "grad_norm": 1.2988234758377075,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 189340
+    },
+    {
+      "epoch": 1245.7236842105262,
+      "grad_norm": 1.1800487041473389,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 189350
+    },
+    {
+      "epoch": 1245.7894736842106,
+      "grad_norm": 1.1548720598220825,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 189360
+    },
+    {
+      "epoch": 1245.8552631578948,
+      "grad_norm": 1.1566232442855835,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 189370
+    },
+    {
+      "epoch": 1245.921052631579,
+      "grad_norm": 1.2034006118774414,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 189380
+    },
+    {
+      "epoch": 1245.9868421052631,
+      "grad_norm": 1.0221554040908813,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 189390
+    },
+    {
+      "epoch": 1246.0526315789473,
+      "grad_norm": 0.8397990465164185,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 189400
+    },
+    {
+      "epoch": 1246.1184210526317,
+      "grad_norm": 1.0938118696212769,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 189410
+    },
+    {
+      "epoch": 1246.1842105263158,
+      "grad_norm": 0.7125457525253296,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 189420
+    },
+    {
+      "epoch": 1246.25,
+      "grad_norm": 1.0024956464767456,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 189430
+    },
+    {
+      "epoch": 1246.3157894736842,
+      "grad_norm": 0.7446456551551819,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 189440
+    },
+    {
+      "epoch": 1246.3815789473683,
+      "grad_norm": 0.9926702380180359,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 189450
+    },
+    {
+      "epoch": 1246.4473684210527,
+      "grad_norm": 1.1048181056976318,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 189460
+    },
+    {
+      "epoch": 1246.5131578947369,
+      "grad_norm": 1.3652530908584595,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 189470
+    },
+    {
+      "epoch": 1246.578947368421,
+      "grad_norm": 1.2224739789962769,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 189480
+    },
+    {
+      "epoch": 1246.6447368421052,
+      "grad_norm": 1.0956324338912964,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 189490
+    },
+    {
+      "epoch": 1246.7105263157894,
+      "grad_norm": 0.8215267658233643,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 189500
+    },
+    {
+      "epoch": 1246.7763157894738,
+      "grad_norm": 1.3271948099136353,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 189510
+    },
+    {
+      "epoch": 1246.842105263158,
+      "grad_norm": 1.0132023096084595,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 189520
+    },
+    {
+      "epoch": 1246.907894736842,
+      "grad_norm": 1.202150821685791,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 189530
+    },
+    {
+      "epoch": 1246.9736842105262,
+      "grad_norm": 1.2758268117904663,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 189540
+    },
+    {
+      "epoch": 1247.0394736842106,
+      "grad_norm": 1.0313605070114136,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 189550
+    },
+    {
+      "epoch": 1247.1052631578948,
+      "grad_norm": 1.0390371084213257,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 189560
+    },
+    {
+      "epoch": 1247.171052631579,
+      "grad_norm": 0.8408495187759399,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 189570
+    },
+    {
+      "epoch": 1247.2368421052631,
+      "grad_norm": 1.059502124786377,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 189580
+    },
+    {
+      "epoch": 1247.3026315789473,
+      "grad_norm": 1.236569881439209,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 189590
+    },
+    {
+      "epoch": 1247.3684210526317,
+      "grad_norm": 1.1000412702560425,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 189600
+    },
+    {
+      "epoch": 1247.4342105263158,
+      "grad_norm": 1.3047072887420654,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 189610
+    },
+    {
+      "epoch": 1247.5,
+      "grad_norm": 1.1245132684707642,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 189620
+    },
+    {
+      "epoch": 1247.5657894736842,
+      "grad_norm": 1.1942509412765503,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 189630
+    },
+    {
+      "epoch": 1247.6315789473683,
+      "grad_norm": 1.323743462562561,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 189640
+    },
+    {
+      "epoch": 1247.6973684210527,
+      "grad_norm": 1.3108394145965576,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 189650
+    },
+    {
+      "epoch": 1247.7631578947369,
+      "grad_norm": 1.1331356763839722,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 189660
+    },
+    {
+      "epoch": 1247.828947368421,
+      "grad_norm": 1.2502070665359497,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 189670
+    },
+    {
+      "epoch": 1247.8947368421052,
+      "grad_norm": 0.7357776165008545,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 189680
+    },
+    {
+      "epoch": 1247.9605263157894,
+      "grad_norm": 1.009076714515686,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 189690
+    },
+    {
+      "epoch": 1248.0263157894738,
+      "grad_norm": 0.9644945859909058,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 189700
+    },
+    {
+      "epoch": 1248.092105263158,
+      "grad_norm": 0.8089728355407715,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 189710
+    },
+    {
+      "epoch": 1248.157894736842,
+      "grad_norm": 1.2572730779647827,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 189720
+    },
+    {
+      "epoch": 1248.2236842105262,
+      "grad_norm": 1.3105838298797607,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 189730
+    },
+    {
+      "epoch": 1248.2894736842106,
+      "grad_norm": 0.7587280869483948,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 189740
+    },
+    {
+      "epoch": 1248.3552631578948,
+      "grad_norm": 0.8644336462020874,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 189750
+    },
+    {
+      "epoch": 1248.421052631579,
+      "grad_norm": 0.8798325657844543,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 189760
+    },
+    {
+      "epoch": 1248.4868421052631,
+      "grad_norm": 1.5717847347259521,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 189770
+    },
+    {
+      "epoch": 1248.5526315789473,
+      "grad_norm": 0.9737314581871033,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 189780
+    },
+    {
+      "epoch": 1248.6184210526317,
+      "grad_norm": 0.884158194065094,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 189790
+    },
+    {
+      "epoch": 1248.6842105263158,
+      "grad_norm": 0.8840878009796143,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 189800
+    },
+    {
+      "epoch": 1248.75,
+      "grad_norm": 0.7942262291908264,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 189810
+    },
+    {
+      "epoch": 1248.8157894736842,
+      "grad_norm": 1.0044453144073486,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 189820
+    },
+    {
+      "epoch": 1248.8815789473683,
+      "grad_norm": 1.1081262826919556,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 189830
+    },
+    {
+      "epoch": 1248.9473684210527,
+      "grad_norm": 0.7932543754577637,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 189840
+    },
+    {
+      "epoch": 1249.0131578947369,
+      "grad_norm": 1.0459403991699219,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 189850
+    },
+    {
+      "epoch": 1249.078947368421,
+      "grad_norm": 1.1057350635528564,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 189860
+    },
+    {
+      "epoch": 1249.1447368421052,
+      "grad_norm": 0.8458811640739441,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 189870
+    },
+    {
+      "epoch": 1249.2105263157894,
+      "grad_norm": 0.7159251570701599,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 189880
+    },
+    {
+      "epoch": 1249.2763157894738,
+      "grad_norm": 1.037407636642456,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 189890
+    },
+    {
+      "epoch": 1249.342105263158,
+      "grad_norm": 0.9437480568885803,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 189900
+    },
+    {
+      "epoch": 1249.407894736842,
+      "grad_norm": 0.7939618229866028,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 189910
+    },
+    {
+      "epoch": 1249.4736842105262,
+      "grad_norm": 1.138870358467102,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 189920
+    },
+    {
+      "epoch": 1249.5394736842106,
+      "grad_norm": 0.7695747017860413,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 189930
+    },
+    {
+      "epoch": 1249.6052631578948,
+      "grad_norm": 1.0361729860305786,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 189940
+    },
+    {
+      "epoch": 1249.671052631579,
+      "grad_norm": 0.9108820557594299,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 189950
+    },
+    {
+      "epoch": 1249.7368421052631,
+      "grad_norm": 1.085376501083374,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 189960
+    },
+    {
+      "epoch": 1249.8026315789473,
+      "grad_norm": 1.2229604721069336,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 189970
+    },
+    {
+      "epoch": 1249.8684210526317,
+      "grad_norm": 0.9447299242019653,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 189980
+    },
+    {
+      "epoch": 1249.9342105263158,
+      "grad_norm": 0.8941596746444702,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 189990
+    },
+    {
+      "epoch": 1250.0,
+      "grad_norm": 0.8298214673995972,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 190000
+    },
+    {
+      "epoch": 1250.0657894736842,
+      "grad_norm": 0.8514341711997986,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 190010
+    },
+    {
+      "epoch": 1250.1315789473683,
+      "grad_norm": 1.1375254392623901,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 190020
+    },
+    {
+      "epoch": 1250.1973684210527,
+      "grad_norm": 0.9819296598434448,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 190030
+    },
+    {
+      "epoch": 1250.2631578947369,
+      "grad_norm": 0.9162175059318542,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 190040
+    },
+    {
+      "epoch": 1250.328947368421,
+      "grad_norm": 1.0678719282150269,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 190050
+    },
+    {
+      "epoch": 1250.3947368421052,
+      "grad_norm": 1.3395860195159912,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 190060
+    },
+    {
+      "epoch": 1250.4605263157894,
+      "grad_norm": 0.8305793404579163,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 190070
+    },
+    {
+      "epoch": 1250.5263157894738,
+      "grad_norm": 1.182700514793396,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 190080
+    },
+    {
+      "epoch": 1250.592105263158,
+      "grad_norm": 0.9990321397781372,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 190090
+    },
+    {
+      "epoch": 1250.657894736842,
+      "grad_norm": 0.9627804756164551,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 190100
+    },
+    {
+      "epoch": 1250.7236842105262,
+      "grad_norm": 0.7819858193397522,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 190110
+    },
+    {
+      "epoch": 1250.7894736842106,
+      "grad_norm": 0.8119869232177734,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 190120
+    },
+    {
+      "epoch": 1250.8552631578948,
+      "grad_norm": 1.0260744094848633,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 190130
+    },
+    {
+      "epoch": 1250.921052631579,
+      "grad_norm": 1.2213586568832397,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 190140
+    },
+    {
+      "epoch": 1250.9868421052631,
+      "grad_norm": 1.0840460062026978,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 190150
+    },
+    {
+      "epoch": 1251.0526315789473,
+      "grad_norm": 1.0639532804489136,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 190160
+    },
+    {
+      "epoch": 1251.1184210526317,
+      "grad_norm": 1.0878303050994873,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 190170
+    },
+    {
+      "epoch": 1251.1842105263158,
+      "grad_norm": 0.787267804145813,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 190180
+    },
+    {
+      "epoch": 1251.25,
+      "grad_norm": 1.283235788345337,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 190190
+    },
+    {
+      "epoch": 1251.3157894736842,
+      "grad_norm": 1.402185320854187,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 190200
+    },
+    {
+      "epoch": 1251.3815789473683,
+      "grad_norm": 0.9384086728096008,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 190210
+    },
+    {
+      "epoch": 1251.4473684210527,
+      "grad_norm": 1.2631064653396606,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 190220
+    },
+    {
+      "epoch": 1251.5131578947369,
+      "grad_norm": 1.1802366971969604,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 190230
+    },
+    {
+      "epoch": 1251.578947368421,
+      "grad_norm": 1.0927965641021729,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 190240
+    },
+    {
+      "epoch": 1251.6447368421052,
+      "grad_norm": 1.3226182460784912,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 190250
+    },
+    {
+      "epoch": 1251.7105263157894,
+      "grad_norm": 1.1238583326339722,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 190260
+    },
+    {
+      "epoch": 1251.7763157894738,
+      "grad_norm": 1.0334173440933228,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 190270
+    },
+    {
+      "epoch": 1251.842105263158,
+      "grad_norm": 0.7510884404182434,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 190280
+    },
+    {
+      "epoch": 1251.907894736842,
+      "grad_norm": 1.0210719108581543,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 190290
+    },
+    {
+      "epoch": 1251.9736842105262,
+      "grad_norm": 1.1662030220031738,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 190300
+    },
+    {
+      "epoch": 1252.0394736842106,
+      "grad_norm": 0.8975061774253845,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 190310
+    },
+    {
+      "epoch": 1252.1052631578948,
+      "grad_norm": 1.1046485900878906,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 190320
+    },
+    {
+      "epoch": 1252.171052631579,
+      "grad_norm": 0.740519106388092,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 190330
+    },
+    {
+      "epoch": 1252.2368421052631,
+      "grad_norm": 0.7718915939331055,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 190340
+    },
+    {
+      "epoch": 1252.3026315789473,
+      "grad_norm": 0.7174364328384399,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 190350
+    },
+    {
+      "epoch": 1252.3684210526317,
+      "grad_norm": 0.7045212388038635,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 190360
+    },
+    {
+      "epoch": 1252.4342105263158,
+      "grad_norm": 1.1751577854156494,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 190370
+    },
+    {
+      "epoch": 1252.5,
+      "grad_norm": 1.107194423675537,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 190380
+    },
+    {
+      "epoch": 1252.5657894736842,
+      "grad_norm": 1.6648300886154175,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 190390
+    },
+    {
+      "epoch": 1252.6315789473683,
+      "grad_norm": 1.0686296224594116,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 190400
+    },
+    {
+      "epoch": 1252.6973684210527,
+      "grad_norm": 1.0623538494110107,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 190410
+    },
+    {
+      "epoch": 1252.7631578947369,
+      "grad_norm": 1.1247642040252686,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 190420
+    },
+    {
+      "epoch": 1252.828947368421,
+      "grad_norm": 1.0035089254379272,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 190430
+    },
+    {
+      "epoch": 1252.8947368421052,
+      "grad_norm": 1.277760624885559,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 190440
+    },
+    {
+      "epoch": 1252.9605263157894,
+      "grad_norm": 1.3326517343521118,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 190450
+    },
+    {
+      "epoch": 1253.0263157894738,
+      "grad_norm": 1.0908524990081787,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 190460
+    },
+    {
+      "epoch": 1253.092105263158,
+      "grad_norm": 1.1250838041305542,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 190470
+    },
+    {
+      "epoch": 1253.157894736842,
+      "grad_norm": 1.1381192207336426,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 190480
+    },
+    {
+      "epoch": 1253.2236842105262,
+      "grad_norm": 1.1070564985275269,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 190490
+    },
+    {
+      "epoch": 1253.2894736842106,
+      "grad_norm": 0.9667354226112366,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 190500
+    },
+    {
+      "epoch": 1253.3552631578948,
+      "grad_norm": 0.9640436172485352,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 190510
+    },
+    {
+      "epoch": 1253.421052631579,
+      "grad_norm": 1.0675090551376343,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 190520
+    },
+    {
+      "epoch": 1253.4868421052631,
+      "grad_norm": 0.9862016439437866,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 190530
+    },
+    {
+      "epoch": 1253.5526315789473,
+      "grad_norm": 0.98412024974823,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 190540
+    },
+    {
+      "epoch": 1253.6184210526317,
+      "grad_norm": 0.9340914487838745,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 190550
+    },
+    {
+      "epoch": 1253.6842105263158,
+      "grad_norm": 1.179900884628296,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 190560
+    },
+    {
+      "epoch": 1253.75,
+      "grad_norm": 1.0011744499206543,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 190570
+    },
+    {
+      "epoch": 1253.8157894736842,
+      "grad_norm": 0.9150945544242859,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 190580
+    },
+    {
+      "epoch": 1253.8815789473683,
+      "grad_norm": 1.025476098060608,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 190590
+    },
+    {
+      "epoch": 1253.9473684210527,
+      "grad_norm": 0.7332446575164795,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 190600
+    },
+    {
+      "epoch": 1254.0131578947369,
+      "grad_norm": 0.9624845385551453,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 190610
+    },
+    {
+      "epoch": 1254.078947368421,
+      "grad_norm": 0.9401580691337585,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 190620
+    },
+    {
+      "epoch": 1254.1447368421052,
+      "grad_norm": 1.178200602531433,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 190630
+    },
+    {
+      "epoch": 1254.2105263157894,
+      "grad_norm": 0.8018969297409058,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 190640
+    },
+    {
+      "epoch": 1254.2763157894738,
+      "grad_norm": 1.0022549629211426,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 190650
+    },
+    {
+      "epoch": 1254.342105263158,
+      "grad_norm": 1.0292798280715942,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 190660
+    },
+    {
+      "epoch": 1254.407894736842,
+      "grad_norm": 0.9748183488845825,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 190670
+    },
+    {
+      "epoch": 1254.4736842105262,
+      "grad_norm": 1.1642075777053833,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 190680
+    },
+    {
+      "epoch": 1254.5394736842106,
+      "grad_norm": 1.0161871910095215,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 190690
+    },
+    {
+      "epoch": 1254.6052631578948,
+      "grad_norm": 1.2374141216278076,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 190700
+    },
+    {
+      "epoch": 1254.671052631579,
+      "grad_norm": 0.6536117196083069,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 190710
+    },
+    {
+      "epoch": 1254.7368421052631,
+      "grad_norm": 0.9045120477676392,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 190720
+    },
+    {
+      "epoch": 1254.8026315789473,
+      "grad_norm": 0.969821035861969,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "step": 190730
+    },
+    {
+      "epoch": 1254.8684210526317,
+      "grad_norm": 0.9699207544326782,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 190740
+    },
+    {
+      "epoch": 1254.9342105263158,
+      "grad_norm": 1.0693953037261963,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 190750
+    },
+    {
+      "epoch": 1255.0,
+      "grad_norm": 1.0095750093460083,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 190760
+    },
+    {
+      "epoch": 1255.0657894736842,
+      "grad_norm": 1.1202609539031982,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 190770
+    },
+    {
+      "epoch": 1255.1315789473683,
+      "grad_norm": 1.095145344734192,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 190780
+    },
+    {
+      "epoch": 1255.1973684210527,
+      "grad_norm": 0.8262442350387573,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 190790
+    },
+    {
+      "epoch": 1255.2631578947369,
+      "grad_norm": 1.1005247831344604,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 190800
+    },
+    {
+      "epoch": 1255.328947368421,
+      "grad_norm": 1.0275697708129883,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 190810
+    },
+    {
+      "epoch": 1255.3947368421052,
+      "grad_norm": 0.8765363097190857,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 190820
+    },
+    {
+      "epoch": 1255.4605263157894,
+      "grad_norm": 0.8573766946792603,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 190830
+    },
+    {
+      "epoch": 1255.5263157894738,
+      "grad_norm": 1.2033637762069702,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 190840
+    },
+    {
+      "epoch": 1255.592105263158,
+      "grad_norm": 0.8023409843444824,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 190850
+    },
+    {
+      "epoch": 1255.657894736842,
+      "grad_norm": 0.9117615222930908,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 190860
+    },
+    {
+      "epoch": 1255.7236842105262,
+      "grad_norm": 1.0993090867996216,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 190870
+    },
+    {
+      "epoch": 1255.7894736842106,
+      "grad_norm": 1.0232125520706177,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 190880
+    },
+    {
+      "epoch": 1255.8552631578948,
+      "grad_norm": 1.2864488363265991,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 190890
+    },
+    {
+      "epoch": 1255.921052631579,
+      "grad_norm": 0.8744462132453918,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 190900
+    },
+    {
+      "epoch": 1255.9868421052631,
+      "grad_norm": 0.8951719403266907,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 190910
+    },
+    {
+      "epoch": 1256.0526315789473,
+      "grad_norm": 1.058536410331726,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 190920
+    },
+    {
+      "epoch": 1256.1184210526317,
+      "grad_norm": 0.9087652564048767,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 190930
+    },
+    {
+      "epoch": 1256.1842105263158,
+      "grad_norm": 1.1572648286819458,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 190940
+    },
+    {
+      "epoch": 1256.25,
+      "grad_norm": 1.497307300567627,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 190950
+    },
+    {
+      "epoch": 1256.3157894736842,
+      "grad_norm": 1.1288655996322632,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 190960
+    },
+    {
+      "epoch": 1256.3815789473683,
+      "grad_norm": 1.1092500686645508,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 190970
+    },
+    {
+      "epoch": 1256.4473684210527,
+      "grad_norm": 0.9487274885177612,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 190980
+    },
+    {
+      "epoch": 1256.5131578947369,
+      "grad_norm": 1.2593556642532349,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 190990
+    },
+    {
+      "epoch": 1256.578947368421,
+      "grad_norm": 1.0957036018371582,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 191000
+    },
+    {
+      "epoch": 1256.6447368421052,
+      "grad_norm": 1.3564567565917969,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 191010
+    },
+    {
+      "epoch": 1256.7105263157894,
+      "grad_norm": 1.0658830404281616,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 191020
+    },
+    {
+      "epoch": 1256.7763157894738,
+      "grad_norm": 1.4636614322662354,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 191030
+    },
+    {
+      "epoch": 1256.842105263158,
+      "grad_norm": 0.8907326459884644,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 191040
+    },
+    {
+      "epoch": 1256.907894736842,
+      "grad_norm": 0.6591998934745789,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 191050
+    },
+    {
+      "epoch": 1256.9736842105262,
+      "grad_norm": 1.2480213642120361,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 191060
+    },
+    {
+      "epoch": 1257.0394736842106,
+      "grad_norm": 1.0565965175628662,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 191070
+    },
+    {
+      "epoch": 1257.1052631578948,
+      "grad_norm": 0.7769169807434082,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 191080
+    },
+    {
+      "epoch": 1257.171052631579,
+      "grad_norm": 1.2173153162002563,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 191090
+    },
+    {
+      "epoch": 1257.2368421052631,
+      "grad_norm": 1.1387425661087036,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 191100
+    },
+    {
+      "epoch": 1257.3026315789473,
+      "grad_norm": 1.099096417427063,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 191110
+    },
+    {
+      "epoch": 1257.3684210526317,
+      "grad_norm": 0.9480217695236206,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 191120
+    },
+    {
+      "epoch": 1257.4342105263158,
+      "grad_norm": 0.8401440382003784,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 191130
+    },
+    {
+      "epoch": 1257.5,
+      "grad_norm": 0.9184837341308594,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 191140
+    },
+    {
+      "epoch": 1257.5657894736842,
+      "grad_norm": 0.9384816288948059,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 191150
+    },
+    {
+      "epoch": 1257.6315789473683,
+      "grad_norm": 0.8797683119773865,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 191160
+    },
+    {
+      "epoch": 1257.6973684210527,
+      "grad_norm": 0.781887412071228,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 191170
+    },
+    {
+      "epoch": 1257.7631578947369,
+      "grad_norm": 0.9485580325126648,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 191180
+    },
+    {
+      "epoch": 1257.828947368421,
+      "grad_norm": 0.7003353834152222,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 191190
+    },
+    {
+      "epoch": 1257.8947368421052,
+      "grad_norm": 0.9563307166099548,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 191200
+    },
+    {
+      "epoch": 1257.9605263157894,
+      "grad_norm": 0.787761926651001,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 191210
+    },
+    {
+      "epoch": 1258.0263157894738,
+      "grad_norm": 1.2450077533721924,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 191220
+    },
+    {
+      "epoch": 1258.092105263158,
+      "grad_norm": 1.280927062034607,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 191230
+    },
+    {
+      "epoch": 1258.157894736842,
+      "grad_norm": 1.0110337734222412,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 191240
+    },
+    {
+      "epoch": 1258.2236842105262,
+      "grad_norm": 1.0578523874282837,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 191250
+    },
+    {
+      "epoch": 1258.2894736842106,
+      "grad_norm": 1.5664889812469482,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 191260
+    },
+    {
+      "epoch": 1258.3552631578948,
+      "grad_norm": 1.2868566513061523,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 191270
+    },
+    {
+      "epoch": 1258.421052631579,
+      "grad_norm": 1.0075466632843018,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 191280
+    },
+    {
+      "epoch": 1258.4868421052631,
+      "grad_norm": 1.2369372844696045,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 191290
+    },
+    {
+      "epoch": 1258.5526315789473,
+      "grad_norm": 1.0888575315475464,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 191300
+    },
+    {
+      "epoch": 1258.6184210526317,
+      "grad_norm": 1.023420810699463,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 191310
+    },
+    {
+      "epoch": 1258.6842105263158,
+      "grad_norm": 1.248289704322815,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 191320
+    },
+    {
+      "epoch": 1258.75,
+      "grad_norm": 1.0365052223205566,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 191330
+    },
+    {
+      "epoch": 1258.8157894736842,
+      "grad_norm": 1.4416338205337524,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 191340
+    },
+    {
+      "epoch": 1258.8815789473683,
+      "grad_norm": 1.3358416557312012,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 191350
+    },
+    {
+      "epoch": 1258.9473684210527,
+      "grad_norm": 1.1580281257629395,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 191360
+    },
+    {
+      "epoch": 1259.0131578947369,
+      "grad_norm": 1.4239611625671387,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 191370
+    },
+    {
+      "epoch": 1259.078947368421,
+      "grad_norm": 0.7980217337608337,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 191380
+    },
+    {
+      "epoch": 1259.1447368421052,
+      "grad_norm": 0.9824530482292175,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 191390
+    },
+    {
+      "epoch": 1259.2105263157894,
+      "grad_norm": 0.770296573638916,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 191400
+    },
+    {
+      "epoch": 1259.2763157894738,
+      "grad_norm": 0.780419647693634,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 191410
+    },
+    {
+      "epoch": 1259.342105263158,
+      "grad_norm": 0.8854151964187622,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 191420
+    },
+    {
+      "epoch": 1259.407894736842,
+      "grad_norm": 0.6854922771453857,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 191430
+    },
+    {
+      "epoch": 1259.4736842105262,
+      "grad_norm": 0.6790972948074341,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 191440
+    },
+    {
+      "epoch": 1259.5394736842106,
+      "grad_norm": 0.5925989747047424,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 191450
+    },
+    {
+      "epoch": 1259.6052631578948,
+      "grad_norm": 1.307077407836914,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 191460
+    },
+    {
+      "epoch": 1259.671052631579,
+      "grad_norm": 0.9079083800315857,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 191470
+    },
+    {
+      "epoch": 1259.7368421052631,
+      "grad_norm": 1.258397102355957,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 191480
+    },
+    {
+      "epoch": 1259.8026315789473,
+      "grad_norm": 1.126745343208313,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 191490
+    },
+    {
+      "epoch": 1259.8684210526317,
+      "grad_norm": 0.9693688154220581,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 191500
+    },
+    {
+      "epoch": 1259.9342105263158,
+      "grad_norm": 1.0901410579681396,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 191510
+    },
+    {
+      "epoch": 1260.0,
+      "grad_norm": 1.0153412818908691,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 191520
+    },
+    {
+      "epoch": 1260.0657894736842,
+      "grad_norm": 1.094736099243164,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 191530
+    },
+    {
+      "epoch": 1260.1315789473683,
+      "grad_norm": 1.3164678812026978,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 191540
+    },
+    {
+      "epoch": 1260.1973684210527,
+      "grad_norm": 1.0634385347366333,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 191550
+    },
+    {
+      "epoch": 1260.2631578947369,
+      "grad_norm": 1.0030722618103027,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 191560
+    },
+    {
+      "epoch": 1260.328947368421,
+      "grad_norm": 1.0075671672821045,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 191570
+    },
+    {
+      "epoch": 1260.3947368421052,
+      "grad_norm": 1.2293148040771484,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 191580
+    },
+    {
+      "epoch": 1260.4605263157894,
+      "grad_norm": 0.9783899784088135,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 191590
+    },
+    {
+      "epoch": 1260.5263157894738,
+      "grad_norm": 1.0404366254806519,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 191600
+    },
+    {
+      "epoch": 1260.592105263158,
+      "grad_norm": 1.1873605251312256,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 191610
+    },
+    {
+      "epoch": 1260.657894736842,
+      "grad_norm": 0.8735949397087097,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 191620
+    },
+    {
+      "epoch": 1260.7236842105262,
+      "grad_norm": 1.2900155782699585,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 191630
+    },
+    {
+      "epoch": 1260.7894736842106,
+      "grad_norm": 1.2164313793182373,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 191640
+    },
+    {
+      "epoch": 1260.8552631578948,
+      "grad_norm": 1.0002045631408691,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 191650
+    },
+    {
+      "epoch": 1260.921052631579,
+      "grad_norm": 1.3534318208694458,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 191660
+    },
+    {
+      "epoch": 1260.9868421052631,
+      "grad_norm": 1.0810010433197021,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 191670
+    },
+    {
+      "epoch": 1261.0526315789473,
+      "grad_norm": 1.0390048027038574,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 191680
+    },
+    {
+      "epoch": 1261.1184210526317,
+      "grad_norm": 0.8042935729026794,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 191690
+    },
+    {
+      "epoch": 1261.1842105263158,
+      "grad_norm": 0.8306818008422852,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 191700
+    },
+    {
+      "epoch": 1261.25,
+      "grad_norm": 0.769751787185669,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 191710
+    },
+    {
+      "epoch": 1261.3157894736842,
+      "grad_norm": 0.807361364364624,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 191720
+    },
+    {
+      "epoch": 1261.3815789473683,
+      "grad_norm": 0.6746041178703308,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 191730
+    },
+    {
+      "epoch": 1261.4473684210527,
+      "grad_norm": 0.6681817173957825,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 191740
+    },
+    {
+      "epoch": 1261.5131578947369,
+      "grad_norm": 0.5453993082046509,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 191750
+    },
+    {
+      "epoch": 1261.578947368421,
+      "grad_norm": 0.9125059247016907,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 191760
+    },
+    {
+      "epoch": 1261.6447368421052,
+      "grad_norm": 0.5915899276733398,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 191770
+    },
+    {
+      "epoch": 1261.7105263157894,
+      "grad_norm": 0.9361303448677063,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 191780
+    },
+    {
+      "epoch": 1261.7763157894738,
+      "grad_norm": 1.2069631814956665,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 191790
+    },
+    {
+      "epoch": 1261.842105263158,
+      "grad_norm": 1.1386783123016357,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 191800
+    },
+    {
+      "epoch": 1261.907894736842,
+      "grad_norm": 1.0958198308944702,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 191810
+    },
+    {
+      "epoch": 1261.9736842105262,
+      "grad_norm": 0.7236484885215759,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 191820
+    },
+    {
+      "epoch": 1262.0394736842106,
+      "grad_norm": 1.066319465637207,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 191830
+    },
+    {
+      "epoch": 1262.1052631578948,
+      "grad_norm": 1.1681187152862549,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 191840
+    },
+    {
+      "epoch": 1262.171052631579,
+      "grad_norm": 0.9392849206924438,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 191850
+    },
+    {
+      "epoch": 1262.2368421052631,
+      "grad_norm": 1.0458626747131348,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 191860
+    },
+    {
+      "epoch": 1262.3026315789473,
+      "grad_norm": 0.7124366164207458,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 191870
+    },
+    {
+      "epoch": 1262.3684210526317,
+      "grad_norm": 0.9901890158653259,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 191880
+    },
+    {
+      "epoch": 1262.4342105263158,
+      "grad_norm": 1.156256914138794,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 191890
+    },
+    {
+      "epoch": 1262.5,
+      "grad_norm": 1.051873803138733,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 191900
+    },
+    {
+      "epoch": 1262.5657894736842,
+      "grad_norm": 1.0366870164871216,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 191910
+    },
+    {
+      "epoch": 1262.6315789473683,
+      "grad_norm": 0.843497633934021,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 191920
+    },
+    {
+      "epoch": 1262.6973684210527,
+      "grad_norm": 0.8482083678245544,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 191930
+    },
+    {
+      "epoch": 1262.7631578947369,
+      "grad_norm": 0.7245591282844543,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 191940
+    },
+    {
+      "epoch": 1262.828947368421,
+      "grad_norm": 0.7226964831352234,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 191950
+    },
+    {
+      "epoch": 1262.8947368421052,
+      "grad_norm": 1.0393447875976562,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 191960
+    },
+    {
+      "epoch": 1262.9605263157894,
+      "grad_norm": 0.9947481155395508,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 191970
+    },
+    {
+      "epoch": 1263.0263157894738,
+      "grad_norm": 1.087018609046936,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 191980
+    },
+    {
+      "epoch": 1263.092105263158,
+      "grad_norm": 1.0347148180007935,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 191990
+    },
+    {
+      "epoch": 1263.157894736842,
+      "grad_norm": 0.8795108795166016,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 192000
+    },
+    {
+      "epoch": 1263.2236842105262,
+      "grad_norm": 0.9262383580207825,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 192010
+    },
+    {
+      "epoch": 1263.2894736842106,
+      "grad_norm": 0.9647265076637268,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 192020
+    },
+    {
+      "epoch": 1263.3552631578948,
+      "grad_norm": 1.0269702672958374,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 192030
+    },
+    {
+      "epoch": 1263.421052631579,
+      "grad_norm": 0.9326910972595215,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 192040
+    },
+    {
+      "epoch": 1263.4868421052631,
+      "grad_norm": 1.2422527074813843,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 192050
+    },
+    {
+      "epoch": 1263.5526315789473,
+      "grad_norm": 1.082977056503296,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 192060
+    },
+    {
+      "epoch": 1263.6184210526317,
+      "grad_norm": 1.06131112575531,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 192070
+    },
+    {
+      "epoch": 1263.6842105263158,
+      "grad_norm": 1.269061803817749,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 192080
+    },
+    {
+      "epoch": 1263.75,
+      "grad_norm": 0.993816614151001,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 192090
+    },
+    {
+      "epoch": 1263.8157894736842,
+      "grad_norm": 0.9369046688079834,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 192100
+    },
+    {
+      "epoch": 1263.8815789473683,
+      "grad_norm": 1.1662249565124512,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 192110
+    },
+    {
+      "epoch": 1263.9473684210527,
+      "grad_norm": 0.8694877028465271,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 192120
+    },
+    {
+      "epoch": 1264.0131578947369,
+      "grad_norm": 1.0363537073135376,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 192130
+    },
+    {
+      "epoch": 1264.078947368421,
+      "grad_norm": 0.8358498215675354,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 192140
+    },
+    {
+      "epoch": 1264.1447368421052,
+      "grad_norm": 0.7027297616004944,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 192150
+    },
+    {
+      "epoch": 1264.2105263157894,
+      "grad_norm": 0.8918916583061218,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 192160
+    },
+    {
+      "epoch": 1264.2763157894738,
+      "grad_norm": 0.9327743053436279,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 192170
+    },
+    {
+      "epoch": 1264.342105263158,
+      "grad_norm": 1.0644272565841675,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 192180
+    },
+    {
+      "epoch": 1264.407894736842,
+      "grad_norm": 1.1869374513626099,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 192190
+    },
+    {
+      "epoch": 1264.4736842105262,
+      "grad_norm": 0.9971968531608582,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 192200
+    },
+    {
+      "epoch": 1264.5394736842106,
+      "grad_norm": 0.745521068572998,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 192210
+    },
+    {
+      "epoch": 1264.6052631578948,
+      "grad_norm": 1.1426355838775635,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 192220
+    },
+    {
+      "epoch": 1264.671052631579,
+      "grad_norm": 1.11016845703125,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 192230
+    },
+    {
+      "epoch": 1264.7368421052631,
+      "grad_norm": 0.9929028153419495,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 192240
+    },
+    {
+      "epoch": 1264.8026315789473,
+      "grad_norm": 0.9268110394477844,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 192250
+    },
+    {
+      "epoch": 1264.8684210526317,
+      "grad_norm": 0.8309184312820435,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 192260
+    },
+    {
+      "epoch": 1264.9342105263158,
+      "grad_norm": 1.2837961912155151,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 192270
+    },
+    {
+      "epoch": 1265.0,
+      "grad_norm": 0.8434421420097351,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 192280
+    },
+    {
+      "epoch": 1265.0657894736842,
+      "grad_norm": 1.0207735300064087,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 192290
+    },
+    {
+      "epoch": 1265.1315789473683,
+      "grad_norm": 0.9556507468223572,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 192300
+    },
+    {
+      "epoch": 1265.1973684210527,
+      "grad_norm": 1.1968109607696533,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 192310
+    },
+    {
+      "epoch": 1265.2631578947369,
+      "grad_norm": 1.016579270362854,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 192320
+    },
+    {
+      "epoch": 1265.328947368421,
+      "grad_norm": 1.248950481414795,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 192330
+    },
+    {
+      "epoch": 1265.3947368421052,
+      "grad_norm": 1.086604118347168,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 192340
+    },
+    {
+      "epoch": 1265.4605263157894,
+      "grad_norm": 1.240754246711731,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 192350
+    },
+    {
+      "epoch": 1265.5263157894738,
+      "grad_norm": 1.088182806968689,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 192360
+    },
+    {
+      "epoch": 1265.592105263158,
+      "grad_norm": 1.2884389162063599,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 192370
+    },
+    {
+      "epoch": 1265.657894736842,
+      "grad_norm": 0.9595872759819031,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 192380
+    },
+    {
+      "epoch": 1265.7236842105262,
+      "grad_norm": 0.7716920375823975,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 192390
+    },
+    {
+      "epoch": 1265.7894736842106,
+      "grad_norm": 1.2177424430847168,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 192400
+    },
+    {
+      "epoch": 1265.8552631578948,
+      "grad_norm": 1.1005266904830933,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 192410
+    },
+    {
+      "epoch": 1265.921052631579,
+      "grad_norm": 0.8893570899963379,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 192420
+    },
+    {
+      "epoch": 1265.9868421052631,
+      "grad_norm": 1.0129755735397339,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 192430
+    },
+    {
+      "epoch": 1266.0526315789473,
+      "grad_norm": 1.1341480016708374,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 192440
+    },
+    {
+      "epoch": 1266.1184210526317,
+      "grad_norm": 0.907776951789856,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 192450
+    },
+    {
+      "epoch": 1266.1842105263158,
+      "grad_norm": 0.811395525932312,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 192460
+    },
+    {
+      "epoch": 1266.25,
+      "grad_norm": 1.0590198040008545,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 192470
+    },
+    {
+      "epoch": 1266.3157894736842,
+      "grad_norm": 1.0062720775604248,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 192480
+    },
+    {
+      "epoch": 1266.3815789473683,
+      "grad_norm": 0.9959651827812195,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 192490
+    },
+    {
+      "epoch": 1266.4473684210527,
+      "grad_norm": 1.1516484022140503,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 192500
+    },
+    {
+      "epoch": 1266.5131578947369,
+      "grad_norm": 0.8557493090629578,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 192510
+    },
+    {
+      "epoch": 1266.578947368421,
+      "grad_norm": 1.0222887992858887,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 192520
+    },
+    {
+      "epoch": 1266.6447368421052,
+      "grad_norm": 1.160636067390442,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 192530
+    },
+    {
+      "epoch": 1266.7105263157894,
+      "grad_norm": 0.7170030474662781,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 192540
+    },
+    {
+      "epoch": 1266.7763157894738,
+      "grad_norm": 1.1373835802078247,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 192550
+    },
+    {
+      "epoch": 1266.842105263158,
+      "grad_norm": 0.9175273776054382,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 192560
+    },
+    {
+      "epoch": 1266.907894736842,
+      "grad_norm": 1.1107465028762817,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 192570
+    },
+    {
+      "epoch": 1266.9736842105262,
+      "grad_norm": 1.1369487047195435,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 192580
+    },
+    {
+      "epoch": 1267.0394736842106,
+      "grad_norm": 1.0154566764831543,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 192590
+    },
+    {
+      "epoch": 1267.1052631578948,
+      "grad_norm": 0.7436667084693909,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 192600
+    },
+    {
+      "epoch": 1267.171052631579,
+      "grad_norm": 1.1619658470153809,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 192610
+    },
+    {
+      "epoch": 1267.2368421052631,
+      "grad_norm": 1.0842703580856323,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 192620
+    },
+    {
+      "epoch": 1267.3026315789473,
+      "grad_norm": 1.0601835250854492,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 192630
+    },
+    {
+      "epoch": 1267.3684210526317,
+      "grad_norm": 1.2090107202529907,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 192640
+    },
+    {
+      "epoch": 1267.4342105263158,
+      "grad_norm": 0.9081798195838928,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 192650
+    },
+    {
+      "epoch": 1267.5,
+      "grad_norm": 0.960776686668396,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 192660
+    },
+    {
+      "epoch": 1267.5657894736842,
+      "grad_norm": 0.7369850277900696,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 192670
+    },
+    {
+      "epoch": 1267.6315789473683,
+      "grad_norm": 0.9591938853263855,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 192680
+    },
+    {
+      "epoch": 1267.6973684210527,
+      "grad_norm": 0.9522967338562012,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 192690
+    },
+    {
+      "epoch": 1267.7631578947369,
+      "grad_norm": 1.203101396560669,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 192700
+    },
+    {
+      "epoch": 1267.828947368421,
+      "grad_norm": 0.974402666091919,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 192710
+    },
+    {
+      "epoch": 1267.8947368421052,
+      "grad_norm": 0.6702833771705627,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 192720
+    },
+    {
+      "epoch": 1267.9605263157894,
+      "grad_norm": 0.7159844636917114,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 192730
+    },
+    {
+      "epoch": 1268.0263157894738,
+      "grad_norm": 0.7871840596199036,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 192740
+    },
+    {
+      "epoch": 1268.092105263158,
+      "grad_norm": 1.0035476684570312,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 192750
+    },
+    {
+      "epoch": 1268.157894736842,
+      "grad_norm": 1.134181022644043,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 192760
+    },
+    {
+      "epoch": 1268.2236842105262,
+      "grad_norm": 1.1951532363891602,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 192770
+    },
+    {
+      "epoch": 1268.2894736842106,
+      "grad_norm": 0.7619327902793884,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 192780
+    },
+    {
+      "epoch": 1268.3552631578948,
+      "grad_norm": 1.1061769723892212,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 192790
+    },
+    {
+      "epoch": 1268.421052631579,
+      "grad_norm": 0.7267657518386841,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 192800
+    },
+    {
+      "epoch": 1268.4868421052631,
+      "grad_norm": 1.2342023849487305,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 192810
+    },
+    {
+      "epoch": 1268.5526315789473,
+      "grad_norm": 0.790826141834259,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 192820
+    },
+    {
+      "epoch": 1268.6184210526317,
+      "grad_norm": 0.762381374835968,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 192830
+    },
+    {
+      "epoch": 1268.6842105263158,
+      "grad_norm": 1.3754265308380127,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 192840
+    },
+    {
+      "epoch": 1268.75,
+      "grad_norm": 1.1752783060073853,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 192850
+    },
+    {
+      "epoch": 1268.8157894736842,
+      "grad_norm": 1.3542336225509644,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 192860
+    },
+    {
+      "epoch": 1268.8815789473683,
+      "grad_norm": 1.3783767223358154,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 192870
+    },
+    {
+      "epoch": 1268.9473684210527,
+      "grad_norm": 0.9072512984275818,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 192880
+    },
+    {
+      "epoch": 1269.0131578947369,
+      "grad_norm": 1.1490826606750488,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 192890
+    },
+    {
+      "epoch": 1269.078947368421,
+      "grad_norm": 1.1428316831588745,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 192900
+    },
+    {
+      "epoch": 1269.1447368421052,
+      "grad_norm": 1.2847250699996948,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 192910
+    },
+    {
+      "epoch": 1269.2105263157894,
+      "grad_norm": 0.9286653995513916,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 192920
+    },
+    {
+      "epoch": 1269.2763157894738,
+      "grad_norm": 0.81577467918396,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 192930
+    },
+    {
+      "epoch": 1269.342105263158,
+      "grad_norm": 0.6981829404830933,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 192940
+    },
+    {
+      "epoch": 1269.407894736842,
+      "grad_norm": 0.916454553604126,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 192950
+    },
+    {
+      "epoch": 1269.4736842105262,
+      "grad_norm": 0.8955065608024597,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 192960
+    },
+    {
+      "epoch": 1269.5394736842106,
+      "grad_norm": 1.2273223400115967,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 192970
+    },
+    {
+      "epoch": 1269.6052631578948,
+      "grad_norm": 1.2199914455413818,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 192980
+    },
+    {
+      "epoch": 1269.671052631579,
+      "grad_norm": 1.2514991760253906,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 192990
+    },
+    {
+      "epoch": 1269.7368421052631,
+      "grad_norm": 1.1191940307617188,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 193000
+    },
+    {
+      "epoch": 1269.8026315789473,
+      "grad_norm": 1.1365535259246826,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 193010
+    },
+    {
+      "epoch": 1269.8684210526317,
+      "grad_norm": 1.1664906740188599,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 193020
+    },
+    {
+      "epoch": 1269.9342105263158,
+      "grad_norm": 1.2412478923797607,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 193030
+    },
+    {
+      "epoch": 1270.0,
+      "grad_norm": 1.1647230386734009,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 193040
+    },
+    {
+      "epoch": 1270.0657894736842,
+      "grad_norm": 1.4412037134170532,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 193050
+    },
+    {
+      "epoch": 1270.1315789473683,
+      "grad_norm": 1.1689636707305908,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 193060
+    },
+    {
+      "epoch": 1270.1973684210527,
+      "grad_norm": 0.9392591118812561,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 193070
+    },
+    {
+      "epoch": 1270.2631578947369,
+      "grad_norm": 1.0944244861602783,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 193080
+    },
+    {
+      "epoch": 1270.328947368421,
+      "grad_norm": 0.9576421976089478,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 193090
+    },
+    {
+      "epoch": 1270.3947368421052,
+      "grad_norm": 0.7435740828514099,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "step": 193100
+    },
+    {
+      "epoch": 1270.4605263157894,
+      "grad_norm": 0.8688104152679443,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 193110
+    },
+    {
+      "epoch": 1270.5263157894738,
+      "grad_norm": 0.8202447295188904,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 193120
+    },
+    {
+      "epoch": 1270.592105263158,
+      "grad_norm": 0.7284652590751648,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 193130
+    },
+    {
+      "epoch": 1270.657894736842,
+      "grad_norm": 0.9387590289115906,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 193140
+    },
+    {
+      "epoch": 1270.7236842105262,
+      "grad_norm": 1.2218446731567383,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 193150
+    },
+    {
+      "epoch": 1270.7894736842106,
+      "grad_norm": 0.6693400144577026,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 193160
+    },
+    {
+      "epoch": 1270.8552631578948,
+      "grad_norm": 1.0726978778839111,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 193170
+    },
+    {
+      "epoch": 1270.921052631579,
+      "grad_norm": 1.0772373676300049,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 193180
+    },
+    {
+      "epoch": 1270.9868421052631,
+      "grad_norm": 1.2476823329925537,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 193190
+    },
+    {
+      "epoch": 1271.0526315789473,
+      "grad_norm": 1.162440299987793,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 193200
+    },
+    {
+      "epoch": 1271.1184210526317,
+      "grad_norm": 0.7666929960250854,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 193210
+    },
+    {
+      "epoch": 1271.1842105263158,
+      "grad_norm": 1.0631024837493896,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 193220
+    },
+    {
+      "epoch": 1271.25,
+      "grad_norm": 0.8989170789718628,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 193230
+    },
+    {
+      "epoch": 1271.3157894736842,
+      "grad_norm": 1.1688525676727295,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 193240
+    },
+    {
+      "epoch": 1271.3815789473683,
+      "grad_norm": 1.0052133798599243,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 193250
+    },
+    {
+      "epoch": 1271.4473684210527,
+      "grad_norm": 0.9958306550979614,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 193260
+    },
+    {
+      "epoch": 1271.5131578947369,
+      "grad_norm": 1.302128553390503,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 193270
+    },
+    {
+      "epoch": 1271.578947368421,
+      "grad_norm": 0.8450573086738586,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 193280
+    },
+    {
+      "epoch": 1271.6447368421052,
+      "grad_norm": 1.1243925094604492,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 193290
+    },
+    {
+      "epoch": 1271.7105263157894,
+      "grad_norm": 1.0591107606887817,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 193300
+    },
+    {
+      "epoch": 1271.7763157894738,
+      "grad_norm": 1.174922227859497,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 193310
+    },
+    {
+      "epoch": 1271.842105263158,
+      "grad_norm": 0.8074167966842651,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 193320
+    },
+    {
+      "epoch": 1271.907894736842,
+      "grad_norm": 0.9306144714355469,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 193330
+    },
+    {
+      "epoch": 1271.9736842105262,
+      "grad_norm": 0.8695573806762695,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 193340
+    },
+    {
+      "epoch": 1272.0394736842106,
+      "grad_norm": 0.9562103152275085,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 193350
+    },
+    {
+      "epoch": 1272.1052631578948,
+      "grad_norm": 0.7408207058906555,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 193360
+    },
+    {
+      "epoch": 1272.171052631579,
+      "grad_norm": 0.8762840628623962,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 193370
+    },
+    {
+      "epoch": 1272.2368421052631,
+      "grad_norm": 1.0107637643814087,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 193380
+    },
+    {
+      "epoch": 1272.3026315789473,
+      "grad_norm": 0.9742252230644226,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 193390
+    },
+    {
+      "epoch": 1272.3684210526317,
+      "grad_norm": 1.0357519388198853,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 193400
+    },
+    {
+      "epoch": 1272.4342105263158,
+      "grad_norm": 1.0647118091583252,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 193410
+    },
+    {
+      "epoch": 1272.5,
+      "grad_norm": 1.2739837169647217,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 193420
+    },
+    {
+      "epoch": 1272.5657894736842,
+      "grad_norm": 1.3156754970550537,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 193430
+    },
+    {
+      "epoch": 1272.6315789473683,
+      "grad_norm": 1.3240246772766113,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 193440
+    },
+    {
+      "epoch": 1272.6973684210527,
+      "grad_norm": 1.2527782917022705,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 193450
+    },
+    {
+      "epoch": 1272.7631578947369,
+      "grad_norm": 1.097855567932129,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 193460
+    },
+    {
+      "epoch": 1272.828947368421,
+      "grad_norm": 0.7602299451828003,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 193470
+    },
+    {
+      "epoch": 1272.8947368421052,
+      "grad_norm": 0.9691528677940369,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 193480
+    },
+    {
+      "epoch": 1272.9605263157894,
+      "grad_norm": 0.8665773868560791,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 193490
+    },
+    {
+      "epoch": 1273.0263157894738,
+      "grad_norm": 0.9522979855537415,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 193500
+    },
+    {
+      "epoch": 1273.092105263158,
+      "grad_norm": 1.1000076532363892,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 193510
+    },
+    {
+      "epoch": 1273.157894736842,
+      "grad_norm": 1.0523219108581543,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 193520
+    },
+    {
+      "epoch": 1273.2236842105262,
+      "grad_norm": 0.7484111785888672,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 193530
+    },
+    {
+      "epoch": 1273.2894736842106,
+      "grad_norm": 0.6804783344268799,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 193540
+    },
+    {
+      "epoch": 1273.3552631578948,
+      "grad_norm": 1.133036494255066,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 193550
+    },
+    {
+      "epoch": 1273.421052631579,
+      "grad_norm": 0.9207621812820435,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 193560
+    },
+    {
+      "epoch": 1273.4868421052631,
+      "grad_norm": 0.9967194199562073,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 193570
+    },
+    {
+      "epoch": 1273.5526315789473,
+      "grad_norm": 0.6356325745582581,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 193580
+    },
+    {
+      "epoch": 1273.6184210526317,
+      "grad_norm": 0.9811447262763977,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 193590
+    },
+    {
+      "epoch": 1273.6842105263158,
+      "grad_norm": 0.9721751809120178,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 193600
+    },
+    {
+      "epoch": 1273.75,
+      "grad_norm": 0.8101963996887207,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 193610
+    },
+    {
+      "epoch": 1273.8157894736842,
+      "grad_norm": 1.3644793033599854,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 193620
+    },
+    {
+      "epoch": 1273.8815789473683,
+      "grad_norm": 0.827865719795227,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 193630
+    },
+    {
+      "epoch": 1273.9473684210527,
+      "grad_norm": 1.0075935125350952,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 193640
+    },
+    {
+      "epoch": 1274.0131578947369,
+      "grad_norm": 1.2060139179229736,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 193650
+    },
+    {
+      "epoch": 1274.078947368421,
+      "grad_norm": 1.1146782636642456,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 193660
+    },
+    {
+      "epoch": 1274.1447368421052,
+      "grad_norm": 1.0539047718048096,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 193670
+    },
+    {
+      "epoch": 1274.2105263157894,
+      "grad_norm": 0.9264514446258545,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 193680
+    },
+    {
+      "epoch": 1274.2763157894738,
+      "grad_norm": 0.6521208882331848,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 193690
+    },
+    {
+      "epoch": 1274.342105263158,
+      "grad_norm": 0.6478739380836487,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 193700
+    },
+    {
+      "epoch": 1274.407894736842,
+      "grad_norm": 0.8767127990722656,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 193710
+    },
+    {
+      "epoch": 1274.4736842105262,
+      "grad_norm": 1.1907069683074951,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 193720
+    },
+    {
+      "epoch": 1274.5394736842106,
+      "grad_norm": 1.0393046140670776,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 193730
+    },
+    {
+      "epoch": 1274.6052631578948,
+      "grad_norm": 0.9379701614379883,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 193740
+    },
+    {
+      "epoch": 1274.671052631579,
+      "grad_norm": 0.976507306098938,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 193750
+    },
+    {
+      "epoch": 1274.7368421052631,
+      "grad_norm": 0.986788272857666,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 193760
+    },
+    {
+      "epoch": 1274.8026315789473,
+      "grad_norm": 1.1617448329925537,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 193770
+    },
+    {
+      "epoch": 1274.8684210526317,
+      "grad_norm": 1.0417400598526,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 193780
+    },
+    {
+      "epoch": 1274.9342105263158,
+      "grad_norm": 0.8060194253921509,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 193790
+    },
+    {
+      "epoch": 1275.0,
+      "grad_norm": 1.053001880645752,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 193800
+    },
+    {
+      "epoch": 1275.0657894736842,
+      "grad_norm": 1.0343172550201416,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 193810
+    },
+    {
+      "epoch": 1275.1315789473683,
+      "grad_norm": 0.8656930923461914,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 193820
+    },
+    {
+      "epoch": 1275.1973684210527,
+      "grad_norm": 1.0269204378128052,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 193830
+    },
+    {
+      "epoch": 1275.2631578947369,
+      "grad_norm": 1.2087546586990356,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 193840
+    },
+    {
+      "epoch": 1275.328947368421,
+      "grad_norm": 0.8350012898445129,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 193850
+    },
+    {
+      "epoch": 1275.3947368421052,
+      "grad_norm": 0.9556059241294861,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 193860
+    },
+    {
+      "epoch": 1275.4605263157894,
+      "grad_norm": 1.050558090209961,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 193870
+    },
+    {
+      "epoch": 1275.5263157894738,
+      "grad_norm": 1.1796860694885254,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 193880
+    },
+    {
+      "epoch": 1275.592105263158,
+      "grad_norm": 0.9585551023483276,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 193890
+    },
+    {
+      "epoch": 1275.657894736842,
+      "grad_norm": 1.2745286226272583,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 193900
+    },
+    {
+      "epoch": 1275.7236842105262,
+      "grad_norm": 0.8697680234909058,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 193910
+    },
+    {
+      "epoch": 1275.7894736842106,
+      "grad_norm": 1.0319898128509521,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 193920
+    },
+    {
+      "epoch": 1275.8552631578948,
+      "grad_norm": 0.8313315510749817,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 193930
+    },
+    {
+      "epoch": 1275.921052631579,
+      "grad_norm": 1.148795247077942,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 193940
+    },
+    {
+      "epoch": 1275.9868421052631,
+      "grad_norm": 0.8364288210868835,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 193950
+    },
+    {
+      "epoch": 1276.0526315789473,
+      "grad_norm": 0.909010648727417,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 193960
+    },
+    {
+      "epoch": 1276.1184210526317,
+      "grad_norm": 0.9541910290718079,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 193970
+    },
+    {
+      "epoch": 1276.1842105263158,
+      "grad_norm": 1.6058369874954224,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 193980
+    },
+    {
+      "epoch": 1276.25,
+      "grad_norm": 3.058375835418701,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 193990
+    },
+    {
+      "epoch": 1276.3157894736842,
+      "grad_norm": 2.173565149307251,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 194000
+    },
+    {
+      "epoch": 1276.3815789473683,
+      "grad_norm": 1.741365671157837,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 194010
+    },
+    {
+      "epoch": 1276.4473684210527,
+      "grad_norm": 1.5380631685256958,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 194020
+    },
+    {
+      "epoch": 1276.5131578947369,
+      "grad_norm": 1.3168596029281616,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 194030
+    },
+    {
+      "epoch": 1276.578947368421,
+      "grad_norm": 1.2907633781433105,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 194040
+    },
+    {
+      "epoch": 1276.6447368421052,
+      "grad_norm": 1.0855131149291992,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 194050
+    },
+    {
+      "epoch": 1276.7105263157894,
+      "grad_norm": 1.2930248975753784,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 194060
+    },
+    {
+      "epoch": 1276.7763157894738,
+      "grad_norm": 1.222780466079712,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 194070
+    },
+    {
+      "epoch": 1276.842105263158,
+      "grad_norm": 1.4316956996917725,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 194080
+    },
+    {
+      "epoch": 1276.907894736842,
+      "grad_norm": 1.0720902681350708,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 194090
+    },
+    {
+      "epoch": 1276.9736842105262,
+      "grad_norm": 1.6173840761184692,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 194100
+    },
+    {
+      "epoch": 1277.0394736842106,
+      "grad_norm": 1.3037574291229248,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 194110
+    },
+    {
+      "epoch": 1277.1052631578948,
+      "grad_norm": 1.301659345626831,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 194120
+    },
+    {
+      "epoch": 1277.171052631579,
+      "grad_norm": 1.402639627456665,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 194130
+    },
+    {
+      "epoch": 1277.2368421052631,
+      "grad_norm": 1.272783637046814,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 194140
+    },
+    {
+      "epoch": 1277.3026315789473,
+      "grad_norm": 0.7505586743354797,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 194150
+    },
+    {
+      "epoch": 1277.3684210526317,
+      "grad_norm": 0.9745710492134094,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 194160
+    },
+    {
+      "epoch": 1277.4342105263158,
+      "grad_norm": 1.12748384475708,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 194170
+    },
+    {
+      "epoch": 1277.5,
+      "grad_norm": 1.296922206878662,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 194180
+    },
+    {
+      "epoch": 1277.5657894736842,
+      "grad_norm": 1.1418737173080444,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 194190
+    },
+    {
+      "epoch": 1277.6315789473683,
+      "grad_norm": 0.7893008589744568,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 194200
+    },
+    {
+      "epoch": 1277.6973684210527,
+      "grad_norm": 0.9640665650367737,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 194210
+    },
+    {
+      "epoch": 1277.7631578947369,
+      "grad_norm": 1.2299898862838745,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 194220
+    },
+    {
+      "epoch": 1277.828947368421,
+      "grad_norm": 0.9324010610580444,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 194230
+    },
+    {
+      "epoch": 1277.8947368421052,
+      "grad_norm": 0.841152012348175,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 194240
+    },
+    {
+      "epoch": 1277.9605263157894,
+      "grad_norm": 1.026451826095581,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 194250
+    },
+    {
+      "epoch": 1278.0263157894738,
+      "grad_norm": 0.8263790011405945,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 194260
+    },
+    {
+      "epoch": 1278.092105263158,
+      "grad_norm": 0.8429759740829468,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 194270
+    },
+    {
+      "epoch": 1278.157894736842,
+      "grad_norm": 0.9431121945381165,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 194280
+    },
+    {
+      "epoch": 1278.2236842105262,
+      "grad_norm": 0.8703896999359131,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 194290
+    },
+    {
+      "epoch": 1278.2894736842106,
+      "grad_norm": 0.7083398699760437,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 194300
+    },
+    {
+      "epoch": 1278.3552631578948,
+      "grad_norm": 1.1701858043670654,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 194310
+    },
+    {
+      "epoch": 1278.421052631579,
+      "grad_norm": 1.2647616863250732,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 194320
+    },
+    {
+      "epoch": 1278.4868421052631,
+      "grad_norm": 0.9830572009086609,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 194330
+    },
+    {
+      "epoch": 1278.5526315789473,
+      "grad_norm": 0.9235027432441711,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 194340
+    },
+    {
+      "epoch": 1278.6184210526317,
+      "grad_norm": 0.9852763414382935,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 194350
+    },
+    {
+      "epoch": 1278.6842105263158,
+      "grad_norm": 0.9919136762619019,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 194360
+    },
+    {
+      "epoch": 1278.75,
+      "grad_norm": 0.9679889678955078,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 194370
+    },
+    {
+      "epoch": 1278.8157894736842,
+      "grad_norm": 1.1155283451080322,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 194380
+    },
+    {
+      "epoch": 1278.8815789473683,
+      "grad_norm": 1.1698013544082642,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 194390
+    },
+    {
+      "epoch": 1278.9473684210527,
+      "grad_norm": 1.2240992784500122,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 194400
+    },
+    {
+      "epoch": 1279.0131578947369,
+      "grad_norm": 1.3296053409576416,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 194410
+    },
+    {
+      "epoch": 1279.078947368421,
+      "grad_norm": 1.4155553579330444,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 194420
+    },
+    {
+      "epoch": 1279.1447368421052,
+      "grad_norm": 1.3145627975463867,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 194430
+    },
+    {
+      "epoch": 1279.2105263157894,
+      "grad_norm": 1.4205524921417236,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 194440
+    },
+    {
+      "epoch": 1279.2763157894738,
+      "grad_norm": 0.8720544576644897,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 194450
+    },
+    {
+      "epoch": 1279.342105263158,
+      "grad_norm": 0.9181710481643677,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 194460
+    },
+    {
+      "epoch": 1279.407894736842,
+      "grad_norm": 1.1017323732376099,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 194470
+    },
+    {
+      "epoch": 1279.4736842105262,
+      "grad_norm": 0.9298954606056213,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 194480
+    },
+    {
+      "epoch": 1279.5394736842106,
+      "grad_norm": 0.9003967046737671,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 194490
+    },
+    {
+      "epoch": 1279.6052631578948,
+      "grad_norm": 1.0150399208068848,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 194500
+    },
+    {
+      "epoch": 1279.671052631579,
+      "grad_norm": 0.9780325889587402,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 194510
+    },
+    {
+      "epoch": 1279.7368421052631,
+      "grad_norm": 0.9293487668037415,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 194520
+    },
+    {
+      "epoch": 1279.8026315789473,
+      "grad_norm": 0.82585608959198,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 194530
+    },
+    {
+      "epoch": 1279.8684210526317,
+      "grad_norm": 0.9033555388450623,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 194540
+    },
+    {
+      "epoch": 1279.9342105263158,
+      "grad_norm": 1.1674561500549316,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 194550
+    },
+    {
+      "epoch": 1280.0,
+      "grad_norm": 0.9872167110443115,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 194560
+    },
+    {
+      "epoch": 1280.0657894736842,
+      "grad_norm": 0.7526688575744629,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 194570
+    },
+    {
+      "epoch": 1280.1315789473683,
+      "grad_norm": 1.100469708442688,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 194580
+    },
+    {
+      "epoch": 1280.1973684210527,
+      "grad_norm": 1.1157853603363037,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 194590
+    },
+    {
+      "epoch": 1280.2631578947369,
+      "grad_norm": 1.3560817241668701,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 194600
+    },
+    {
+      "epoch": 1280.328947368421,
+      "grad_norm": 1.132397174835205,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 194610
+    },
+    {
+      "epoch": 1280.3947368421052,
+      "grad_norm": 1.0123302936553955,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 194620
+    },
+    {
+      "epoch": 1280.4605263157894,
+      "grad_norm": 0.9225643873214722,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 194630
+    },
+    {
+      "epoch": 1280.5263157894738,
+      "grad_norm": 0.7562076449394226,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 194640
+    },
+    {
+      "epoch": 1280.592105263158,
+      "grad_norm": 0.5493844747543335,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 194650
+    },
+    {
+      "epoch": 1280.657894736842,
+      "grad_norm": 0.9773281216621399,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 194660
+    },
+    {
+      "epoch": 1280.7236842105262,
+      "grad_norm": 0.8132543563842773,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 194670
+    },
+    {
+      "epoch": 1280.7894736842106,
+      "grad_norm": 1.0290168523788452,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 194680
+    },
+    {
+      "epoch": 1280.8552631578948,
+      "grad_norm": 1.0300383567810059,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 194690
+    },
+    {
+      "epoch": 1280.921052631579,
+      "grad_norm": 0.831333339214325,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 194700
+    },
+    {
+      "epoch": 1280.9868421052631,
+      "grad_norm": 1.1776654720306396,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 194710
+    },
+    {
+      "epoch": 1281.0526315789473,
+      "grad_norm": 1.3399312496185303,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 194720
+    },
+    {
+      "epoch": 1281.1184210526317,
+      "grad_norm": 1.3882694244384766,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 194730
+    },
+    {
+      "epoch": 1281.1842105263158,
+      "grad_norm": 1.010141372680664,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 194740
+    },
+    {
+      "epoch": 1281.25,
+      "grad_norm": 1.1796314716339111,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 194750
+    },
+    {
+      "epoch": 1281.3157894736842,
+      "grad_norm": 0.9938530325889587,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 194760
+    },
+    {
+      "epoch": 1281.3815789473683,
+      "grad_norm": 0.8807727098464966,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 194770
+    },
+    {
+      "epoch": 1281.4473684210527,
+      "grad_norm": 0.9626410603523254,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 194780
+    },
+    {
+      "epoch": 1281.5131578947369,
+      "grad_norm": 0.9949114322662354,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 194790
+    },
+    {
+      "epoch": 1281.578947368421,
+      "grad_norm": 0.6117430925369263,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 194800
+    },
+    {
+      "epoch": 1281.6447368421052,
+      "grad_norm": 0.8383743762969971,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 194810
+    },
+    {
+      "epoch": 1281.7105263157894,
+      "grad_norm": 1.091646432876587,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 194820
+    },
+    {
+      "epoch": 1281.7763157894738,
+      "grad_norm": 0.7561153769493103,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 194830
+    },
+    {
+      "epoch": 1281.842105263158,
+      "grad_norm": 0.7222174406051636,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 194840
+    },
+    {
+      "epoch": 1281.907894736842,
+      "grad_norm": 0.7637395262718201,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 194850
+    },
+    {
+      "epoch": 1281.9736842105262,
+      "grad_norm": 1.0813342332839966,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 194860
+    },
+    {
+      "epoch": 1282.0394736842106,
+      "grad_norm": 1.0132391452789307,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 194870
+    },
+    {
+      "epoch": 1282.1052631578948,
+      "grad_norm": 0.9855432510375977,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 194880
+    },
+    {
+      "epoch": 1282.171052631579,
+      "grad_norm": 1.1103159189224243,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 194890
+    },
+    {
+      "epoch": 1282.2368421052631,
+      "grad_norm": 1.3338035345077515,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 194900
+    },
+    {
+      "epoch": 1282.3026315789473,
+      "grad_norm": 1.1458650827407837,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 194910
+    },
+    {
+      "epoch": 1282.3684210526317,
+      "grad_norm": 1.2011997699737549,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 194920
+    },
+    {
+      "epoch": 1282.4342105263158,
+      "grad_norm": 0.8735988736152649,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 194930
+    },
+    {
+      "epoch": 1282.5,
+      "grad_norm": 0.8202658295631409,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 194940
+    },
+    {
+      "epoch": 1282.5657894736842,
+      "grad_norm": 1.0125209093093872,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 194950
+    },
+    {
+      "epoch": 1282.6315789473683,
+      "grad_norm": 1.1227607727050781,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 194960
+    },
+    {
+      "epoch": 1282.6973684210527,
+      "grad_norm": 1.05240797996521,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 194970
+    },
+    {
+      "epoch": 1282.7631578947369,
+      "grad_norm": 1.0156041383743286,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 194980
+    },
+    {
+      "epoch": 1282.828947368421,
+      "grad_norm": 1.0174468755722046,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 194990
+    },
+    {
+      "epoch": 1282.8947368421052,
+      "grad_norm": 1.300093412399292,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 195000
+    },
+    {
+      "epoch": 1282.9605263157894,
+      "grad_norm": 1.298466444015503,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 195010
+    },
+    {
+      "epoch": 1283.0263157894738,
+      "grad_norm": 1.0919368267059326,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 195020
+    },
+    {
+      "epoch": 1283.092105263158,
+      "grad_norm": 1.0547558069229126,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 195030
+    },
+    {
+      "epoch": 1283.157894736842,
+      "grad_norm": 1.1440446376800537,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 195040
+    },
+    {
+      "epoch": 1283.2236842105262,
+      "grad_norm": 0.7482719421386719,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 195050
+    },
+    {
+      "epoch": 1283.2894736842106,
+      "grad_norm": 1.0653194189071655,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 195060
+    },
+    {
+      "epoch": 1283.3552631578948,
+      "grad_norm": 1.172141432762146,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 195070
+    },
+    {
+      "epoch": 1283.421052631579,
+      "grad_norm": 0.7871286273002625,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 195080
+    },
+    {
+      "epoch": 1283.4868421052631,
+      "grad_norm": 0.9559378027915955,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 195090
+    },
+    {
+      "epoch": 1283.5526315789473,
+      "grad_norm": 1.331890344619751,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 195100
+    },
+    {
+      "epoch": 1283.6184210526317,
+      "grad_norm": 0.9049778580665588,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 195110
+    },
+    {
+      "epoch": 1283.6842105263158,
+      "grad_norm": 0.6896979808807373,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 195120
+    },
+    {
+      "epoch": 1283.75,
+      "grad_norm": 0.6551532745361328,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 195130
+    },
+    {
+      "epoch": 1283.8157894736842,
+      "grad_norm": 0.8988133072853088,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 195140
+    },
+    {
+      "epoch": 1283.8815789473683,
+      "grad_norm": 0.7520022392272949,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 195150
+    },
+    {
+      "epoch": 1283.9473684210527,
+      "grad_norm": 1.120882511138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 195160
+    },
+    {
+      "epoch": 1284.0131578947369,
+      "grad_norm": 0.8684665560722351,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 195170
+    },
+    {
+      "epoch": 1284.078947368421,
+      "grad_norm": 0.9502143859863281,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 195180
+    },
+    {
+      "epoch": 1284.1447368421052,
+      "grad_norm": 1.1483829021453857,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 195190
+    },
+    {
+      "epoch": 1284.2105263157894,
+      "grad_norm": 1.1794594526290894,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 195200
+    },
+    {
+      "epoch": 1284.2763157894738,
+      "grad_norm": 0.681542694568634,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 195210
+    },
+    {
+      "epoch": 1284.342105263158,
+      "grad_norm": 0.8862521648406982,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 195220
+    },
+    {
+      "epoch": 1284.407894736842,
+      "grad_norm": 0.9744770526885986,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 195230
+    },
+    {
+      "epoch": 1284.4736842105262,
+      "grad_norm": 0.7451329827308655,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 195240
+    },
+    {
+      "epoch": 1284.5394736842106,
+      "grad_norm": 0.7159759402275085,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 195250
+    },
+    {
+      "epoch": 1284.6052631578948,
+      "grad_norm": 1.3413786888122559,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 195260
+    },
+    {
+      "epoch": 1284.671052631579,
+      "grad_norm": 1.1536619663238525,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 195270
+    },
+    {
+      "epoch": 1284.7368421052631,
+      "grad_norm": 0.8164981007575989,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 195280
+    },
+    {
+      "epoch": 1284.8026315789473,
+      "grad_norm": 0.990378737449646,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 195290
+    },
+    {
+      "epoch": 1284.8684210526317,
+      "grad_norm": 0.8975867629051208,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 195300
+    },
+    {
+      "epoch": 1284.9342105263158,
+      "grad_norm": 0.9112573862075806,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 195310
+    },
+    {
+      "epoch": 1285.0,
+      "grad_norm": 0.8136677145957947,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 195320
+    },
+    {
+      "epoch": 1285.0657894736842,
+      "grad_norm": 0.9761696457862854,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 195330
+    },
+    {
+      "epoch": 1285.1315789473683,
+      "grad_norm": 0.5923354029655457,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 195340
+    },
+    {
+      "epoch": 1285.1973684210527,
+      "grad_norm": 0.8915498852729797,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 195350
+    },
+    {
+      "epoch": 1285.2631578947369,
+      "grad_norm": 1.0586600303649902,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 195360
+    },
+    {
+      "epoch": 1285.328947368421,
+      "grad_norm": 1.0132033824920654,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 195370
+    },
+    {
+      "epoch": 1285.3947368421052,
+      "grad_norm": 1.0920168161392212,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 195380
+    },
+    {
+      "epoch": 1285.4605263157894,
+      "grad_norm": 0.8626097440719604,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 195390
+    },
+    {
+      "epoch": 1285.5263157894738,
+      "grad_norm": 0.9792760014533997,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 195400
+    },
+    {
+      "epoch": 1285.592105263158,
+      "grad_norm": 1.0166680812835693,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 195410
+    },
+    {
+      "epoch": 1285.657894736842,
+      "grad_norm": 0.970467209815979,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 195420
+    },
+    {
+      "epoch": 1285.7236842105262,
+      "grad_norm": 1.0889147520065308,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 195430
+    },
+    {
+      "epoch": 1285.7894736842106,
+      "grad_norm": 0.7250308394432068,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 195440
+    },
+    {
+      "epoch": 1285.8552631578948,
+      "grad_norm": 0.8987478613853455,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 195450
+    },
+    {
+      "epoch": 1285.921052631579,
+      "grad_norm": 0.7027731537818909,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 195460
+    },
+    {
+      "epoch": 1285.9868421052631,
+      "grad_norm": 1.0836803913116455,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 195470
+    },
+    {
+      "epoch": 1286.0526315789473,
+      "grad_norm": 0.9599834084510803,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 195480
+    },
+    {
+      "epoch": 1286.1184210526317,
+      "grad_norm": 1.119584083557129,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 195490
+    },
+    {
+      "epoch": 1286.1842105263158,
+      "grad_norm": 1.0971158742904663,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 195500
+    },
+    {
+      "epoch": 1286.25,
+      "grad_norm": 0.7488318085670471,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 195510
+    },
+    {
+      "epoch": 1286.3157894736842,
+      "grad_norm": 0.7044021487236023,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 195520
+    },
+    {
+      "epoch": 1286.3815789473683,
+      "grad_norm": 1.1740487813949585,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 195530
+    },
+    {
+      "epoch": 1286.4473684210527,
+      "grad_norm": 1.206044316291809,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 195540
+    },
+    {
+      "epoch": 1286.5131578947369,
+      "grad_norm": 0.9043120741844177,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 195550
+    },
+    {
+      "epoch": 1286.578947368421,
+      "grad_norm": 1.060771107673645,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 195560
+    },
+    {
+      "epoch": 1286.6447368421052,
+      "grad_norm": 1.237007975578308,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 195570
+    },
+    {
+      "epoch": 1286.7105263157894,
+      "grad_norm": 1.2525277137756348,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 195580
+    },
+    {
+      "epoch": 1286.7763157894738,
+      "grad_norm": 0.9757577180862427,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 195590
+    },
+    {
+      "epoch": 1286.842105263158,
+      "grad_norm": 0.8785386085510254,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 195600
+    },
+    {
+      "epoch": 1286.907894736842,
+      "grad_norm": 1.1863728761672974,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 195610
+    },
+    {
+      "epoch": 1286.9736842105262,
+      "grad_norm": 0.9649335741996765,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 195620
+    },
+    {
+      "epoch": 1287.0394736842106,
+      "grad_norm": 0.8843129873275757,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 195630
+    },
+    {
+      "epoch": 1287.1052631578948,
+      "grad_norm": 1.302202820777893,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 195640
+    },
+    {
+      "epoch": 1287.171052631579,
+      "grad_norm": 1.2117173671722412,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 195650
+    },
+    {
+      "epoch": 1287.2368421052631,
+      "grad_norm": 1.168351411819458,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 195660
+    },
+    {
+      "epoch": 1287.3026315789473,
+      "grad_norm": 1.0641189813613892,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 195670
+    },
+    {
+      "epoch": 1287.3684210526317,
+      "grad_norm": 1.3389551639556885,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 195680
+    },
+    {
+      "epoch": 1287.4342105263158,
+      "grad_norm": 0.8608621954917908,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 195690
+    },
+    {
+      "epoch": 1287.5,
+      "grad_norm": 0.9058868885040283,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 195700
+    },
+    {
+      "epoch": 1287.5657894736842,
+      "grad_norm": 1.3241692781448364,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 195710
+    },
+    {
+      "epoch": 1287.6315789473683,
+      "grad_norm": 1.0901117324829102,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 195720
+    },
+    {
+      "epoch": 1287.6973684210527,
+      "grad_norm": 1.2744159698486328,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 195730
+    },
+    {
+      "epoch": 1287.7631578947369,
+      "grad_norm": 0.9215327501296997,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 195740
+    },
+    {
+      "epoch": 1287.828947368421,
+      "grad_norm": 0.8423858284950256,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 195750
+    },
+    {
+      "epoch": 1287.8947368421052,
+      "grad_norm": 1.16117262840271,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 195760
+    },
+    {
+      "epoch": 1287.9605263157894,
+      "grad_norm": 0.973017156124115,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 195770
+    },
+    {
+      "epoch": 1288.0263157894738,
+      "grad_norm": 0.9248650074005127,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 195780
+    },
+    {
+      "epoch": 1288.092105263158,
+      "grad_norm": 1.0818302631378174,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 195790
+    },
+    {
+      "epoch": 1288.157894736842,
+      "grad_norm": 0.8217904567718506,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 195800
+    },
+    {
+      "epoch": 1288.2236842105262,
+      "grad_norm": 1.021702527999878,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 195810
+    },
+    {
+      "epoch": 1288.2894736842106,
+      "grad_norm": 0.8209040760993958,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 195820
+    },
+    {
+      "epoch": 1288.3552631578948,
+      "grad_norm": 0.7926214933395386,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 195830
+    },
+    {
+      "epoch": 1288.421052631579,
+      "grad_norm": 0.9760251045227051,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 195840
+    },
+    {
+      "epoch": 1288.4868421052631,
+      "grad_norm": 0.9303892850875854,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 195850
+    },
+    {
+      "epoch": 1288.5526315789473,
+      "grad_norm": 1.2029811143875122,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 195860
+    },
+    {
+      "epoch": 1288.6184210526317,
+      "grad_norm": 1.0326858758926392,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 195870
+    },
+    {
+      "epoch": 1288.6842105263158,
+      "grad_norm": 0.8558474779129028,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 195880
+    },
+    {
+      "epoch": 1288.75,
+      "grad_norm": 0.9762908220291138,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 195890
+    },
+    {
+      "epoch": 1288.8157894736842,
+      "grad_norm": 0.9533197283744812,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 195900
+    },
+    {
+      "epoch": 1288.8815789473683,
+      "grad_norm": 1.0016499757766724,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 195910
+    },
+    {
+      "epoch": 1288.9473684210527,
+      "grad_norm": 1.404975414276123,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 195920
+    },
+    {
+      "epoch": 1289.0131578947369,
+      "grad_norm": 1.2138851881027222,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 195930
+    },
+    {
+      "epoch": 1289.078947368421,
+      "grad_norm": 1.0937894582748413,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 195940
+    },
+    {
+      "epoch": 1289.1447368421052,
+      "grad_norm": 0.9936800599098206,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 195950
+    },
+    {
+      "epoch": 1289.2105263157894,
+      "grad_norm": 1.2059030532836914,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 195960
+    },
+    {
+      "epoch": 1289.2763157894738,
+      "grad_norm": 1.204706072807312,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 195970
+    },
+    {
+      "epoch": 1289.342105263158,
+      "grad_norm": 0.8758791089057922,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 195980
+    },
+    {
+      "epoch": 1289.407894736842,
+      "grad_norm": 0.9688547849655151,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 195990
+    },
+    {
+      "epoch": 1289.4736842105262,
+      "grad_norm": 0.8228425979614258,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 196000
+    },
+    {
+      "epoch": 1289.5394736842106,
+      "grad_norm": 0.6902533173561096,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 196010
+    },
+    {
+      "epoch": 1289.6052631578948,
+      "grad_norm": 1.0823955535888672,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 196020
+    },
+    {
+      "epoch": 1289.671052631579,
+      "grad_norm": 1.114900827407837,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 196030
+    },
+    {
+      "epoch": 1289.7368421052631,
+      "grad_norm": 1.076406717300415,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 196040
+    },
+    {
+      "epoch": 1289.8026315789473,
+      "grad_norm": 0.841337263584137,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 196050
+    },
+    {
+      "epoch": 1289.8684210526317,
+      "grad_norm": 0.953133761882782,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 196060
+    },
+    {
+      "epoch": 1289.9342105263158,
+      "grad_norm": 0.9329866766929626,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 196070
+    },
+    {
+      "epoch": 1290.0,
+      "grad_norm": 1.4121922254562378,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 196080
+    },
+    {
+      "epoch": 1290.0657894736842,
+      "grad_norm": 1.1717429161071777,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 196090
+    },
+    {
+      "epoch": 1290.1315789473683,
+      "grad_norm": 1.0481109619140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 196100
+    },
+    {
+      "epoch": 1290.1973684210527,
+      "grad_norm": 1.3172156810760498,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 196110
+    },
+    {
+      "epoch": 1290.2631578947369,
+      "grad_norm": 1.1237008571624756,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 196120
+    },
+    {
+      "epoch": 1290.328947368421,
+      "grad_norm": 1.2435554265975952,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 196130
+    },
+    {
+      "epoch": 1290.3947368421052,
+      "grad_norm": 0.9842655062675476,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 196140
+    },
+    {
+      "epoch": 1290.4605263157894,
+      "grad_norm": 0.9428750276565552,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "step": 196150
+    },
+    {
+      "epoch": 1290.5263157894738,
+      "grad_norm": 0.8579443693161011,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 196160
+    },
+    {
+      "epoch": 1290.592105263158,
+      "grad_norm": 0.9661730527877808,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 196170
+    },
+    {
+      "epoch": 1290.657894736842,
+      "grad_norm": 1.2594717741012573,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 196180
+    },
+    {
+      "epoch": 1290.7236842105262,
+      "grad_norm": 1.2238619327545166,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 196190
+    },
+    {
+      "epoch": 1290.7894736842106,
+      "grad_norm": 1.045121431350708,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 196200
+    },
+    {
+      "epoch": 1290.8552631578948,
+      "grad_norm": 1.2637920379638672,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 196210
+    },
+    {
+      "epoch": 1290.921052631579,
+      "grad_norm": 1.0736345052719116,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "step": 196220
+    },
+    {
+      "epoch": 1290.9868421052631,
+      "grad_norm": 1.1902934312820435,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 196230
+    },
+    {
+      "epoch": 1291.0526315789473,
+      "grad_norm": 1.206912875175476,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 196240
+    },
+    {
+      "epoch": 1291.1184210526317,
+      "grad_norm": 0.6837294697761536,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 196250
+    },
+    {
+      "epoch": 1291.1842105263158,
+      "grad_norm": 0.8661401271820068,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 196260
+    },
+    {
+      "epoch": 1291.25,
+      "grad_norm": 0.7421029210090637,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 196270
+    },
+    {
+      "epoch": 1291.3157894736842,
+      "grad_norm": 1.000644326210022,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 196280
+    },
+    {
+      "epoch": 1291.3815789473683,
+      "grad_norm": 0.9609134793281555,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 196290
+    },
+    {
+      "epoch": 1291.4473684210527,
+      "grad_norm": 0.8988838791847229,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 196300
+    },
+    {
+      "epoch": 1291.5131578947369,
+      "grad_norm": 1.2548774480819702,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 196310
+    },
+    {
+      "epoch": 1291.578947368421,
+      "grad_norm": 1.2236343622207642,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 196320
+    },
+    {
+      "epoch": 1291.6447368421052,
+      "grad_norm": 0.982333242893219,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 196330
+    },
+    {
+      "epoch": 1291.7105263157894,
+      "grad_norm": 1.1261245012283325,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 196340
+    },
+    {
+      "epoch": 1291.7763157894738,
+      "grad_norm": 1.1063497066497803,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 196350
+    },
+    {
+      "epoch": 1291.842105263158,
+      "grad_norm": 0.8248695731163025,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 196360
+    },
+    {
+      "epoch": 1291.907894736842,
+      "grad_norm": 1.017749309539795,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 196370
+    },
+    {
+      "epoch": 1291.9736842105262,
+      "grad_norm": 0.7296943664550781,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 196380
+    },
+    {
+      "epoch": 1292.0394736842106,
+      "grad_norm": 0.9445720314979553,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 196390
+    },
+    {
+      "epoch": 1292.1052631578948,
+      "grad_norm": 2.0632238388061523,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 196400
+    },
+    {
+      "epoch": 1292.171052631579,
+      "grad_norm": 1.4882488250732422,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 196410
+    },
+    {
+      "epoch": 1292.2368421052631,
+      "grad_norm": 1.1946419477462769,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 196420
+    },
+    {
+      "epoch": 1292.3026315789473,
+      "grad_norm": 1.3664677143096924,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 196430
+    },
+    {
+      "epoch": 1292.3684210526317,
+      "grad_norm": 1.171695590019226,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 196440
+    },
+    {
+      "epoch": 1292.4342105263158,
+      "grad_norm": 1.2387498617172241,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 196450
+    },
+    {
+      "epoch": 1292.5,
+      "grad_norm": 1.0718955993652344,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 196460
+    },
+    {
+      "epoch": 1292.5657894736842,
+      "grad_norm": 1.0958354473114014,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 196470
+    },
+    {
+      "epoch": 1292.6315789473683,
+      "grad_norm": 1.5984266996383667,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 196480
+    },
+    {
+      "epoch": 1292.6973684210527,
+      "grad_norm": 1.1815001964569092,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 196490
+    },
+    {
+      "epoch": 1292.7631578947369,
+      "grad_norm": 1.5802987813949585,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 196500
+    },
+    {
+      "epoch": 1292.828947368421,
+      "grad_norm": 1.259151816368103,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 196510
+    },
+    {
+      "epoch": 1292.8947368421052,
+      "grad_norm": 1.411533236503601,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 196520
+    },
+    {
+      "epoch": 1292.9605263157894,
+      "grad_norm": 0.8868615031242371,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 196530
+    },
+    {
+      "epoch": 1293.0263157894738,
+      "grad_norm": 0.9721224904060364,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 196540
+    },
+    {
+      "epoch": 1293.092105263158,
+      "grad_norm": 1.0972973108291626,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 196550
+    },
+    {
+      "epoch": 1293.157894736842,
+      "grad_norm": 0.9356346130371094,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 196560
+    },
+    {
+      "epoch": 1293.2236842105262,
+      "grad_norm": 1.2877929210662842,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 196570
+    },
+    {
+      "epoch": 1293.2894736842106,
+      "grad_norm": 0.9061616659164429,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 196580
+    },
+    {
+      "epoch": 1293.3552631578948,
+      "grad_norm": 1.1051673889160156,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 196590
+    },
+    {
+      "epoch": 1293.421052631579,
+      "grad_norm": 1.0051326751708984,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 196600
+    },
+    {
+      "epoch": 1293.4868421052631,
+      "grad_norm": 1.1201657056808472,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 196610
+    },
+    {
+      "epoch": 1293.5526315789473,
+      "grad_norm": 0.9499081969261169,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 196620
+    },
+    {
+      "epoch": 1293.6184210526317,
+      "grad_norm": 0.87836092710495,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 196630
+    },
+    {
+      "epoch": 1293.6842105263158,
+      "grad_norm": 0.5928868651390076,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 196640
+    },
+    {
+      "epoch": 1293.75,
+      "grad_norm": 0.7514234781265259,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 196650
+    },
+    {
+      "epoch": 1293.8157894736842,
+      "grad_norm": 0.8907245397567749,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 196660
+    },
+    {
+      "epoch": 1293.8815789473683,
+      "grad_norm": 0.7311668395996094,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 196670
+    },
+    {
+      "epoch": 1293.9473684210527,
+      "grad_norm": 1.2829216718673706,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 196680
+    },
+    {
+      "epoch": 1294.0131578947369,
+      "grad_norm": 0.797347366809845,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 196690
+    },
+    {
+      "epoch": 1294.078947368421,
+      "grad_norm": 1.1499863862991333,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 196700
+    },
+    {
+      "epoch": 1294.1447368421052,
+      "grad_norm": 1.033013939857483,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 196710
+    },
+    {
+      "epoch": 1294.2105263157894,
+      "grad_norm": 0.9540650248527527,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 196720
+    },
+    {
+      "epoch": 1294.2763157894738,
+      "grad_norm": 1.2450796365737915,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 196730
+    },
+    {
+      "epoch": 1294.342105263158,
+      "grad_norm": 0.8691481947898865,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 196740
+    },
+    {
+      "epoch": 1294.407894736842,
+      "grad_norm": 1.0715320110321045,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 196750
+    },
+    {
+      "epoch": 1294.4736842105262,
+      "grad_norm": 1.0476354360580444,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 196760
+    },
+    {
+      "epoch": 1294.5394736842106,
+      "grad_norm": 0.9503270983695984,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 196770
+    },
+    {
+      "epoch": 1294.6052631578948,
+      "grad_norm": 1.1776961088180542,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 196780
+    },
+    {
+      "epoch": 1294.671052631579,
+      "grad_norm": 0.6950992345809937,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 196790
+    },
+    {
+      "epoch": 1294.7368421052631,
+      "grad_norm": 1.3217885494232178,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 196800
+    },
+    {
+      "epoch": 1294.8026315789473,
+      "grad_norm": 0.9721072912216187,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 196810
+    },
+    {
+      "epoch": 1294.8684210526317,
+      "grad_norm": 0.996015191078186,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 196820
+    },
+    {
+      "epoch": 1294.9342105263158,
+      "grad_norm": 1.1851248741149902,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 196830
+    },
+    {
+      "epoch": 1295.0,
+      "grad_norm": 1.0879162549972534,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 196840
+    },
+    {
+      "epoch": 1295.0657894736842,
+      "grad_norm": 0.9492610692977905,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 196850
+    },
+    {
+      "epoch": 1295.1315789473683,
+      "grad_norm": 1.0667566061019897,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 196860
+    },
+    {
+      "epoch": 1295.1973684210527,
+      "grad_norm": 0.8629001975059509,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 196870
+    },
+    {
+      "epoch": 1295.2631578947369,
+      "grad_norm": 0.9111774563789368,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 196880
+    },
+    {
+      "epoch": 1295.328947368421,
+      "grad_norm": 0.9569632411003113,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 196890
+    },
+    {
+      "epoch": 1295.3947368421052,
+      "grad_norm": 0.8886310458183289,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 196900
+    },
+    {
+      "epoch": 1295.4605263157894,
+      "grad_norm": 1.119640588760376,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 196910
+    },
+    {
+      "epoch": 1295.5263157894738,
+      "grad_norm": 0.7103597521781921,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 196920
+    },
+    {
+      "epoch": 1295.592105263158,
+      "grad_norm": 1.0032562017440796,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 196930
+    },
+    {
+      "epoch": 1295.657894736842,
+      "grad_norm": 0.6882836818695068,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "step": 196940
+    },
+    {
+      "epoch": 1295.7236842105262,
+      "grad_norm": 1.1084208488464355,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 196950
+    },
+    {
+      "epoch": 1295.7894736842106,
+      "grad_norm": 1.101213812828064,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 196960
+    },
+    {
+      "epoch": 1295.8552631578948,
+      "grad_norm": 1.2237820625305176,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 196970
+    },
+    {
+      "epoch": 1295.921052631579,
+      "grad_norm": 1.164002537727356,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 196980
+    },
+    {
+      "epoch": 1295.9868421052631,
+      "grad_norm": 0.5948657989501953,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 196990
+    },
+    {
+      "epoch": 1296.0526315789473,
+      "grad_norm": 0.8001142144203186,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 197000
+    },
+    {
+      "epoch": 1296.1184210526317,
+      "grad_norm": 0.9679588675498962,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 197010
+    },
+    {
+      "epoch": 1296.1842105263158,
+      "grad_norm": 0.7849892377853394,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 197020
+    },
+    {
+      "epoch": 1296.25,
+      "grad_norm": 0.9673242568969727,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 197030
+    },
+    {
+      "epoch": 1296.3157894736842,
+      "grad_norm": 0.968097984790802,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 197040
+    },
+    {
+      "epoch": 1296.3815789473683,
+      "grad_norm": 0.8999632596969604,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 197050
+    },
+    {
+      "epoch": 1296.4473684210527,
+      "grad_norm": 0.7191337943077087,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 197060
+    },
+    {
+      "epoch": 1296.5131578947369,
+      "grad_norm": 1.1006282567977905,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 197070
+    },
+    {
+      "epoch": 1296.578947368421,
+      "grad_norm": 1.0084878206253052,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 197080
+    },
+    {
+      "epoch": 1296.6447368421052,
+      "grad_norm": 0.9073280096054077,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 197090
+    },
+    {
+      "epoch": 1296.7105263157894,
+      "grad_norm": 0.8795638680458069,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 197100
+    },
+    {
+      "epoch": 1296.7763157894738,
+      "grad_norm": 1.0686129331588745,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 197110
+    },
+    {
+      "epoch": 1296.842105263158,
+      "grad_norm": 0.8052754402160645,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 197120
+    },
+    {
+      "epoch": 1296.907894736842,
+      "grad_norm": 0.8181116580963135,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 197130
+    },
+    {
+      "epoch": 1296.9736842105262,
+      "grad_norm": 1.0895122289657593,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 197140
+    },
+    {
+      "epoch": 1297.0394736842106,
+      "grad_norm": 0.7319552898406982,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 197150
+    },
+    {
+      "epoch": 1297.1052631578948,
+      "grad_norm": 1.2623764276504517,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 197160
+    },
+    {
+      "epoch": 1297.171052631579,
+      "grad_norm": 1.238411784172058,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 197170
+    },
+    {
+      "epoch": 1297.2368421052631,
+      "grad_norm": 1.0991114377975464,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 197180
+    },
+    {
+      "epoch": 1297.3026315789473,
+      "grad_norm": 0.8547118306159973,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 197190
+    },
+    {
+      "epoch": 1297.3684210526317,
+      "grad_norm": 1.1389788389205933,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 197200
+    },
+    {
+      "epoch": 1297.4342105263158,
+      "grad_norm": 1.2709906101226807,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 197210
+    },
+    {
+      "epoch": 1297.5,
+      "grad_norm": 0.9706898331642151,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 197220
+    },
+    {
+      "epoch": 1297.5657894736842,
+      "grad_norm": 0.8353281021118164,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 197230
+    },
+    {
+      "epoch": 1297.6315789473683,
+      "grad_norm": 1.0063377618789673,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 197240
+    },
+    {
+      "epoch": 1297.6973684210527,
+      "grad_norm": 1.0630806684494019,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 197250
+    },
+    {
+      "epoch": 1297.7631578947369,
+      "grad_norm": 0.7940020561218262,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 197260
+    },
+    {
+      "epoch": 1297.828947368421,
+      "grad_norm": 1.0338921546936035,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 197270
+    },
+    {
+      "epoch": 1297.8947368421052,
+      "grad_norm": 1.1591367721557617,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 197280
+    },
+    {
+      "epoch": 1297.9605263157894,
+      "grad_norm": 0.9339560270309448,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 197290
+    },
+    {
+      "epoch": 1298.0263157894738,
+      "grad_norm": 1.0255118608474731,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 197300
+    },
+    {
+      "epoch": 1298.092105263158,
+      "grad_norm": 1.0539302825927734,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 197310
+    },
+    {
+      "epoch": 1298.157894736842,
+      "grad_norm": 0.9863162636756897,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 197320
+    },
+    {
+      "epoch": 1298.2236842105262,
+      "grad_norm": 1.1610512733459473,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 197330
+    },
+    {
+      "epoch": 1298.2894736842106,
+      "grad_norm": 0.9269651174545288,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 197340
+    },
+    {
+      "epoch": 1298.3552631578948,
+      "grad_norm": 0.8783165812492371,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 197350
+    },
+    {
+      "epoch": 1298.421052631579,
+      "grad_norm": 0.9454626441001892,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 197360
+    },
+    {
+      "epoch": 1298.4868421052631,
+      "grad_norm": 0.9759495854377747,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 197370
+    },
+    {
+      "epoch": 1298.5526315789473,
+      "grad_norm": 1.3680849075317383,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 197380
+    },
+    {
+      "epoch": 1298.6184210526317,
+      "grad_norm": 1.3781981468200684,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 197390
+    },
+    {
+      "epoch": 1298.6842105263158,
+      "grad_norm": 1.1003533601760864,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 197400
+    },
+    {
+      "epoch": 1298.75,
+      "grad_norm": 1.1373398303985596,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 197410
+    },
+    {
+      "epoch": 1298.8157894736842,
+      "grad_norm": 0.7512350082397461,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 197420
+    },
+    {
+      "epoch": 1298.8815789473683,
+      "grad_norm": 1.035928726196289,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 197430
+    },
+    {
+      "epoch": 1298.9473684210527,
+      "grad_norm": 1.0015102624893188,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 197440
+    },
+    {
+      "epoch": 1299.0131578947369,
+      "grad_norm": 1.1075206995010376,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 197450
+    },
+    {
+      "epoch": 1299.078947368421,
+      "grad_norm": 1.2172952890396118,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 197460
+    },
+    {
+      "epoch": 1299.1447368421052,
+      "grad_norm": 1.0585336685180664,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 197470
+    },
+    {
+      "epoch": 1299.2105263157894,
+      "grad_norm": 0.9066409468650818,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 197480
+    },
+    {
+      "epoch": 1299.2763157894738,
+      "grad_norm": 1.2567695379257202,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 197490
+    },
+    {
+      "epoch": 1299.342105263158,
+      "grad_norm": 1.0604040622711182,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 197500
+    },
+    {
+      "epoch": 1299.407894736842,
+      "grad_norm": 1.0578597784042358,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 197510
+    },
+    {
+      "epoch": 1299.4736842105262,
+      "grad_norm": 0.9691057205200195,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 197520
+    },
+    {
+      "epoch": 1299.5394736842106,
+      "grad_norm": 1.079813003540039,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 197530
+    },
+    {
+      "epoch": 1299.6052631578948,
+      "grad_norm": 1.141811490058899,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 197540
+    },
+    {
+      "epoch": 1299.671052631579,
+      "grad_norm": 0.8607233166694641,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 197550
+    },
+    {
+      "epoch": 1299.7368421052631,
+      "grad_norm": 0.815772533416748,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 197560
+    },
+    {
+      "epoch": 1299.8026315789473,
+      "grad_norm": 0.8933233022689819,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 197570
+    },
+    {
+      "epoch": 1299.8684210526317,
+      "grad_norm": 0.9021427035331726,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 197580
+    },
+    {
+      "epoch": 1299.9342105263158,
+      "grad_norm": 0.763812780380249,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 197590
+    },
+    {
+      "epoch": 1300.0,
+      "grad_norm": 0.9004136323928833,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 197600
+    },
+    {
+      "epoch": 1300.0657894736842,
+      "grad_norm": 0.7930411100387573,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 197610
+    },
+    {
+      "epoch": 1300.1315789473683,
+      "grad_norm": 1.1747939586639404,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 197620
+    },
+    {
+      "epoch": 1300.1973684210527,
+      "grad_norm": 0.7587085962295532,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 197630
+    },
+    {
+      "epoch": 1300.2631578947369,
+      "grad_norm": 0.8571013808250427,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 197640
+    },
+    {
+      "epoch": 1300.328947368421,
+      "grad_norm": 1.2042564153671265,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 197650
+    },
+    {
+      "epoch": 1300.3947368421052,
+      "grad_norm": 1.0559279918670654,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 197660
+    },
+    {
+      "epoch": 1300.4605263157894,
+      "grad_norm": 1.0643962621688843,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 197670
+    },
+    {
+      "epoch": 1300.5263157894738,
+      "grad_norm": 1.2402312755584717,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 197680
+    },
+    {
+      "epoch": 1300.592105263158,
+      "grad_norm": 1.1355198621749878,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 197690
+    },
+    {
+      "epoch": 1300.657894736842,
+      "grad_norm": 0.8521161675453186,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 197700
+    },
+    {
+      "epoch": 1300.7236842105262,
+      "grad_norm": 0.7998043894767761,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 197710
+    },
+    {
+      "epoch": 1300.7894736842106,
+      "grad_norm": 1.1688848733901978,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 197720
+    },
+    {
+      "epoch": 1300.8552631578948,
+      "grad_norm": 1.0704905986785889,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 197730
+    },
+    {
+      "epoch": 1300.921052631579,
+      "grad_norm": 0.9947075247764587,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 197740
+    },
+    {
+      "epoch": 1300.9868421052631,
+      "grad_norm": 1.168898582458496,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 197750
+    },
+    {
+      "epoch": 1301.0526315789473,
+      "grad_norm": 1.1258999109268188,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 197760
+    },
+    {
+      "epoch": 1301.1184210526317,
+      "grad_norm": 0.9728697538375854,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 197770
+    },
+    {
+      "epoch": 1301.1842105263158,
+      "grad_norm": 0.923896849155426,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 197780
+    },
+    {
+      "epoch": 1301.25,
+      "grad_norm": 1.1840898990631104,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 197790
+    },
+    {
+      "epoch": 1301.3157894736842,
+      "grad_norm": 1.1185284852981567,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 197800
+    },
+    {
+      "epoch": 1301.3815789473683,
+      "grad_norm": 0.7468376159667969,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 197810
+    },
+    {
+      "epoch": 1301.4473684210527,
+      "grad_norm": 0.6839302778244019,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 197820
+    },
+    {
+      "epoch": 1301.5131578947369,
+      "grad_norm": 0.9034473299980164,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 197830
+    },
+    {
+      "epoch": 1301.578947368421,
+      "grad_norm": 1.357205867767334,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 197840
+    },
+    {
+      "epoch": 1301.6447368421052,
+      "grad_norm": 1.2651640176773071,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 197850
+    },
+    {
+      "epoch": 1301.7105263157894,
+      "grad_norm": 0.9837512969970703,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 197860
+    },
+    {
+      "epoch": 1301.7763157894738,
+      "grad_norm": 1.1584852933883667,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 197870
+    },
+    {
+      "epoch": 1301.842105263158,
+      "grad_norm": 0.8707510232925415,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 197880
+    },
+    {
+      "epoch": 1301.907894736842,
+      "grad_norm": 1.073400855064392,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 197890
+    },
+    {
+      "epoch": 1301.9736842105262,
+      "grad_norm": 1.0655863285064697,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 197900
+    },
+    {
+      "epoch": 1302.0394736842106,
+      "grad_norm": 0.9237596988677979,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 197910
+    },
+    {
+      "epoch": 1302.1052631578948,
+      "grad_norm": 1.0519744157791138,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 197920
+    },
+    {
+      "epoch": 1302.171052631579,
+      "grad_norm": 0.7526938319206238,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 197930
+    },
+    {
+      "epoch": 1302.2368421052631,
+      "grad_norm": 0.9896842837333679,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 197940
+    },
+    {
+      "epoch": 1302.3026315789473,
+      "grad_norm": 0.7161365151405334,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 197950
+    },
+    {
+      "epoch": 1302.3684210526317,
+      "grad_norm": 1.2991057634353638,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 197960
+    },
+    {
+      "epoch": 1302.4342105263158,
+      "grad_norm": 1.0487098693847656,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 197970
+    },
+    {
+      "epoch": 1302.5,
+      "grad_norm": 1.1503612995147705,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 197980
+    },
+    {
+      "epoch": 1302.5657894736842,
+      "grad_norm": 1.0851579904556274,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 197990
+    },
+    {
+      "epoch": 1302.6315789473683,
+      "grad_norm": 1.2855541706085205,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "step": 198000
+    },
+    {
+      "epoch": 1302.6973684210527,
+      "grad_norm": 1.1550692319869995,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 198010
+    },
+    {
+      "epoch": 1302.7631578947369,
+      "grad_norm": 0.8146882653236389,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 198020
+    },
+    {
+      "epoch": 1302.828947368421,
+      "grad_norm": 0.6463797092437744,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 198030
+    },
+    {
+      "epoch": 1302.8947368421052,
+      "grad_norm": 1.289481520652771,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 198040
+    },
+    {
+      "epoch": 1302.9605263157894,
+      "grad_norm": 0.8075095415115356,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 198050
+    },
+    {
+      "epoch": 1303.0263157894738,
+      "grad_norm": 1.2249948978424072,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 198060
+    },
+    {
+      "epoch": 1303.092105263158,
+      "grad_norm": 0.8465560674667358,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 198070
+    },
+    {
+      "epoch": 1303.157894736842,
+      "grad_norm": 0.8450498580932617,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 198080
+    },
+    {
+      "epoch": 1303.2236842105262,
+      "grad_norm": 1.2542957067489624,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 198090
+    },
+    {
+      "epoch": 1303.2894736842106,
+      "grad_norm": 0.9559205770492554,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 198100
+    },
+    {
+      "epoch": 1303.3552631578948,
+      "grad_norm": 0.6290387511253357,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 198110
+    },
+    {
+      "epoch": 1303.421052631579,
+      "grad_norm": 0.8307482600212097,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 198120
+    },
+    {
+      "epoch": 1303.4868421052631,
+      "grad_norm": 0.7818741202354431,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 198130
+    },
+    {
+      "epoch": 1303.5526315789473,
+      "grad_norm": 0.7035477757453918,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 198140
+    },
+    {
+      "epoch": 1303.6184210526317,
+      "grad_norm": 0.9868258833885193,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 198150
+    },
+    {
+      "epoch": 1303.6842105263158,
+      "grad_norm": 1.5310142040252686,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 198160
+    },
+    {
+      "epoch": 1303.75,
+      "grad_norm": 1.3995954990386963,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 198170
+    },
+    {
+      "epoch": 1303.8157894736842,
+      "grad_norm": 1.1260607242584229,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 198180
+    },
+    {
+      "epoch": 1303.8815789473683,
+      "grad_norm": 1.8549559116363525,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 198190
+    },
+    {
+      "epoch": 1303.9473684210527,
+      "grad_norm": 1.5865100622177124,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 198200
+    },
+    {
+      "epoch": 1304.0131578947369,
+      "grad_norm": 1.6978919506072998,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 198210
+    },
+    {
+      "epoch": 1304.078947368421,
+      "grad_norm": 1.3336944580078125,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 198220
+    },
+    {
+      "epoch": 1304.1447368421052,
+      "grad_norm": 1.2469390630722046,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 198230
+    },
+    {
+      "epoch": 1304.2105263157894,
+      "grad_norm": 1.321698546409607,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 198240
+    },
+    {
+      "epoch": 1304.2763157894738,
+      "grad_norm": 0.9702907800674438,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 198250
+    },
+    {
+      "epoch": 1304.342105263158,
+      "grad_norm": 1.213561773300171,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 198260
+    },
+    {
+      "epoch": 1304.407894736842,
+      "grad_norm": 1.2231154441833496,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 198270
+    },
+    {
+      "epoch": 1304.4736842105262,
+      "grad_norm": 1.1985797882080078,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 198280
+    },
+    {
+      "epoch": 1304.5394736842106,
+      "grad_norm": 0.9450637102127075,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 198290
+    },
+    {
+      "epoch": 1304.6052631578948,
+      "grad_norm": 0.9720757007598877,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 198300
+    },
+    {
+      "epoch": 1304.671052631579,
+      "grad_norm": 0.849722683429718,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 198310
+    },
+    {
+      "epoch": 1304.7368421052631,
+      "grad_norm": 1.1238338947296143,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 198320
+    },
+    {
+      "epoch": 1304.8026315789473,
+      "grad_norm": 1.1874395608901978,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 198330
+    },
+    {
+      "epoch": 1304.8684210526317,
+      "grad_norm": 1.0838104486465454,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 198340
+    },
+    {
+      "epoch": 1304.9342105263158,
+      "grad_norm": 1.154099464416504,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 198350
+    },
+    {
+      "epoch": 1305.0,
+      "grad_norm": 1.3620731830596924,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 198360
+    },
+    {
+      "epoch": 1305.0657894736842,
+      "grad_norm": 1.2841746807098389,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 198370
+    },
+    {
+      "epoch": 1305.1315789473683,
+      "grad_norm": 0.7634298801422119,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 198380
+    },
+    {
+      "epoch": 1305.1973684210527,
+      "grad_norm": 0.82984858751297,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 198390
+    },
+    {
+      "epoch": 1305.2631578947369,
+      "grad_norm": 1.093189001083374,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 198400
+    },
+    {
+      "epoch": 1305.328947368421,
+      "grad_norm": 0.9606587290763855,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 198410
+    },
+    {
+      "epoch": 1305.3947368421052,
+      "grad_norm": 0.9093089699745178,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 198420
+    },
+    {
+      "epoch": 1305.4605263157894,
+      "grad_norm": 0.9972700476646423,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 198430
+    },
+    {
+      "epoch": 1305.5263157894738,
+      "grad_norm": 0.724269688129425,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 198440
+    },
+    {
+      "epoch": 1305.592105263158,
+      "grad_norm": 0.6498937606811523,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 198450
+    },
+    {
+      "epoch": 1305.657894736842,
+      "grad_norm": 0.9374182224273682,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 198460
+    },
+    {
+      "epoch": 1305.7236842105262,
+      "grad_norm": 1.0840044021606445,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 198470
+    },
+    {
+      "epoch": 1305.7894736842106,
+      "grad_norm": 0.8806322813034058,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 198480
+    },
+    {
+      "epoch": 1305.8552631578948,
+      "grad_norm": 1.1491776704788208,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 198490
+    },
+    {
+      "epoch": 1305.921052631579,
+      "grad_norm": 1.1558951139450073,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 198500
+    },
+    {
+      "epoch": 1305.9868421052631,
+      "grad_norm": 0.8697763085365295,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 198510
+    },
+    {
+      "epoch": 1306.0526315789473,
+      "grad_norm": 0.7972436547279358,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 198520
+    },
+    {
+      "epoch": 1306.1184210526317,
+      "grad_norm": 0.7652742266654968,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 198530
+    },
+    {
+      "epoch": 1306.1842105263158,
+      "grad_norm": 1.0677361488342285,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 198540
+    },
+    {
+      "epoch": 1306.25,
+      "grad_norm": 0.9967530965805054,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 198550
+    },
+    {
+      "epoch": 1306.3157894736842,
+      "grad_norm": 0.8501853942871094,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 198560
+    },
+    {
+      "epoch": 1306.3815789473683,
+      "grad_norm": 1.3013569116592407,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 198570
+    },
+    {
+      "epoch": 1306.4473684210527,
+      "grad_norm": 0.9409443140029907,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 198580
+    },
+    {
+      "epoch": 1306.5131578947369,
+      "grad_norm": 0.9747039675712585,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 198590
+    },
+    {
+      "epoch": 1306.578947368421,
+      "grad_norm": 1.0571503639221191,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 198600
+    },
+    {
+      "epoch": 1306.6447368421052,
+      "grad_norm": 0.7892196178436279,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 198610
+    },
+    {
+      "epoch": 1306.7105263157894,
+      "grad_norm": 1.2711453437805176,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 198620
+    },
+    {
+      "epoch": 1306.7763157894738,
+      "grad_norm": 1.0646253824234009,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 198630
+    },
+    {
+      "epoch": 1306.842105263158,
+      "grad_norm": 0.9739760160446167,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 198640
+    },
+    {
+      "epoch": 1306.907894736842,
+      "grad_norm": 1.054294466972351,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 198650
+    },
+    {
+      "epoch": 1306.9736842105262,
+      "grad_norm": 0.9815489649772644,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 198660
+    },
+    {
+      "epoch": 1307.0394736842106,
+      "grad_norm": 1.2176196575164795,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 198670
+    },
+    {
+      "epoch": 1307.1052631578948,
+      "grad_norm": 0.6826998591423035,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 198680
+    },
+    {
+      "epoch": 1307.171052631579,
+      "grad_norm": 0.6777370572090149,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 198690
+    },
+    {
+      "epoch": 1307.2368421052631,
+      "grad_norm": 0.7783268690109253,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 198700
+    },
+    {
+      "epoch": 1307.3026315789473,
+      "grad_norm": 0.9428658485412598,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 198710
+    },
+    {
+      "epoch": 1307.3684210526317,
+      "grad_norm": 0.6998884081840515,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 198720
+    },
+    {
+      "epoch": 1307.4342105263158,
+      "grad_norm": 0.6625298261642456,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 198730
+    },
+    {
+      "epoch": 1307.5,
+      "grad_norm": 0.868622362613678,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 198740
+    },
+    {
+      "epoch": 1307.5657894736842,
+      "grad_norm": 0.9482408165931702,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 198750
+    },
+    {
+      "epoch": 1307.6315789473683,
+      "grad_norm": 0.9562262296676636,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 198760
+    },
+    {
+      "epoch": 1307.6973684210527,
+      "grad_norm": 1.1496707201004028,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 198770
+    },
+    {
+      "epoch": 1307.7631578947369,
+      "grad_norm": 1.1873894929885864,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 198780
+    },
+    {
+      "epoch": 1307.828947368421,
+      "grad_norm": 1.321275234222412,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 198790
+    },
+    {
+      "epoch": 1307.8947368421052,
+      "grad_norm": 1.3467177152633667,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 198800
+    },
+    {
+      "epoch": 1307.9605263157894,
+      "grad_norm": 1.229738473892212,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 198810
+    },
+    {
+      "epoch": 1308.0263157894738,
+      "grad_norm": 1.2855587005615234,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 198820
+    },
+    {
+      "epoch": 1308.092105263158,
+      "grad_norm": 0.9853659272193909,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 198830
+    },
+    {
+      "epoch": 1308.157894736842,
+      "grad_norm": 1.0061135292053223,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 198840
+    },
+    {
+      "epoch": 1308.2236842105262,
+      "grad_norm": 0.6921624541282654,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 198850
+    },
+    {
+      "epoch": 1308.2894736842106,
+      "grad_norm": 0.9409551024436951,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 198860
+    },
+    {
+      "epoch": 1308.3552631578948,
+      "grad_norm": 1.0803308486938477,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 198870
+    },
+    {
+      "epoch": 1308.421052631579,
+      "grad_norm": 1.107574224472046,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 198880
+    },
+    {
+      "epoch": 1308.4868421052631,
+      "grad_norm": 1.182572603225708,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 198890
+    },
+    {
+      "epoch": 1308.5526315789473,
+      "grad_norm": 1.107548713684082,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 198900
+    },
+    {
+      "epoch": 1308.6184210526317,
+      "grad_norm": 1.1584941148757935,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 198910
+    },
+    {
+      "epoch": 1308.6842105263158,
+      "grad_norm": 1.4747897386550903,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 198920
+    },
+    {
+      "epoch": 1308.75,
+      "grad_norm": 1.3047585487365723,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 198930
+    },
+    {
+      "epoch": 1308.8157894736842,
+      "grad_norm": 1.0019290447235107,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 198940
+    },
+    {
+      "epoch": 1308.8815789473683,
+      "grad_norm": 0.7977175712585449,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 198950
+    },
+    {
+      "epoch": 1308.9473684210527,
+      "grad_norm": 1.2986831665039062,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 198960
+    },
+    {
+      "epoch": 1309.0131578947369,
+      "grad_norm": 1.226833701133728,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 198970
+    },
+    {
+      "epoch": 1309.078947368421,
+      "grad_norm": 0.9073089957237244,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 198980
+    },
+    {
+      "epoch": 1309.1447368421052,
+      "grad_norm": 0.9509008526802063,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 198990
+    },
+    {
+      "epoch": 1309.2105263157894,
+      "grad_norm": 1.083253026008606,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 199000
+    },
+    {
+      "epoch": 1309.2763157894738,
+      "grad_norm": 0.731686532497406,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 199010
+    },
+    {
+      "epoch": 1309.342105263158,
+      "grad_norm": 1.5306459665298462,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 199020
+    },
+    {
+      "epoch": 1309.407894736842,
+      "grad_norm": 1.169585108757019,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 199030
+    },
+    {
+      "epoch": 1309.4736842105262,
+      "grad_norm": 1.225286841392517,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 199040
+    },
+    {
+      "epoch": 1309.5394736842106,
+      "grad_norm": 1.1703866720199585,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 199050
+    },
+    {
+      "epoch": 1309.6052631578948,
+      "grad_norm": 1.381737470626831,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 199060
+    },
+    {
+      "epoch": 1309.671052631579,
+      "grad_norm": 0.9777848720550537,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 199070
+    },
+    {
+      "epoch": 1309.7368421052631,
+      "grad_norm": 1.1087908744812012,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 199080
+    },
+    {
+      "epoch": 1309.8026315789473,
+      "grad_norm": 1.5419447422027588,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 199090
+    },
+    {
+      "epoch": 1309.8684210526317,
+      "grad_norm": 1.1349118947982788,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 199100
+    },
+    {
+      "epoch": 1309.9342105263158,
+      "grad_norm": 1.1746926307678223,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 199110
+    },
+    {
+      "epoch": 1310.0,
+      "grad_norm": 0.8852264881134033,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 199120
+    },
+    {
+      "epoch": 1310.0657894736842,
+      "grad_norm": 1.0382320880889893,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 199130
+    },
+    {
+      "epoch": 1310.1315789473683,
+      "grad_norm": 1.3453278541564941,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 199140
+    },
+    {
+      "epoch": 1310.1973684210527,
+      "grad_norm": 1.1044265031814575,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 199150
+    },
+    {
+      "epoch": 1310.2631578947369,
+      "grad_norm": 0.6985461711883545,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 199160
+    },
+    {
+      "epoch": 1310.328947368421,
+      "grad_norm": 1.2022053003311157,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 199170
+    },
+    {
+      "epoch": 1310.3947368421052,
+      "grad_norm": 1.3073654174804688,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 199180
+    },
+    {
+      "epoch": 1310.4605263157894,
+      "grad_norm": 0.9767325520515442,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 199190
+    },
+    {
+      "epoch": 1310.5263157894738,
+      "grad_norm": 0.7886372208595276,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 199200
+    },
+    {
+      "epoch": 1310.592105263158,
+      "grad_norm": 1.1879262924194336,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 199210
+    },
+    {
+      "epoch": 1310.657894736842,
+      "grad_norm": 1.3642725944519043,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 199220
+    },
+    {
+      "epoch": 1310.7236842105262,
+      "grad_norm": 0.7498663663864136,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 199230
+    },
+    {
+      "epoch": 1310.7894736842106,
+      "grad_norm": 0.6963070034980774,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 199240
+    },
+    {
+      "epoch": 1310.8552631578948,
+      "grad_norm": 1.091447114944458,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 199250
+    },
+    {
+      "epoch": 1310.921052631579,
+      "grad_norm": 0.7097912430763245,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 199260
+    },
+    {
+      "epoch": 1310.9868421052631,
+      "grad_norm": 0.9089406728744507,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 199270
+    },
+    {
+      "epoch": 1311.0526315789473,
+      "grad_norm": 1.0399571657180786,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 199280
+    },
+    {
+      "epoch": 1311.1184210526317,
+      "grad_norm": 0.8863736987113953,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 199290
+    },
+    {
+      "epoch": 1311.1842105263158,
+      "grad_norm": 0.7154498100280762,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 199300
+    },
+    {
+      "epoch": 1311.25,
+      "grad_norm": 0.644045352935791,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 199310
+    },
+    {
+      "epoch": 1311.3157894736842,
+      "grad_norm": 1.04582941532135,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 199320
+    },
+    {
+      "epoch": 1311.3815789473683,
+      "grad_norm": 0.9396390318870544,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 199330
+    },
+    {
+      "epoch": 1311.4473684210527,
+      "grad_norm": 0.959603488445282,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 199340
+    },
+    {
+      "epoch": 1311.5131578947369,
+      "grad_norm": 1.1740236282348633,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 199350
+    },
+    {
+      "epoch": 1311.578947368421,
+      "grad_norm": 0.7795823216438293,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 199360
+    },
+    {
+      "epoch": 1311.6447368421052,
+      "grad_norm": 1.1692177057266235,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 199370
+    },
+    {
+      "epoch": 1311.7105263157894,
+      "grad_norm": 1.053812026977539,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 199380
+    },
+    {
+      "epoch": 1311.7763157894738,
+      "grad_norm": 1.0118703842163086,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 199390
+    },
+    {
+      "epoch": 1311.842105263158,
+      "grad_norm": 0.7939363718032837,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 199400
+    },
+    {
+      "epoch": 1311.907894736842,
+      "grad_norm": 1.0312461853027344,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 199410
+    },
+    {
+      "epoch": 1311.9736842105262,
+      "grad_norm": 1.237866759300232,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 199420
+    },
+    {
+      "epoch": 1312.0394736842106,
+      "grad_norm": 1.2162584066390991,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 199430
+    },
+    {
+      "epoch": 1312.1052631578948,
+      "grad_norm": 0.9375420212745667,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 199440
+    },
+    {
+      "epoch": 1312.171052631579,
+      "grad_norm": 1.0901834964752197,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 199450
+    },
+    {
+      "epoch": 1312.2368421052631,
+      "grad_norm": 1.2144420146942139,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 199460
+    },
+    {
+      "epoch": 1312.3026315789473,
+      "grad_norm": 1.3519539833068848,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 199470
+    },
+    {
+      "epoch": 1312.3684210526317,
+      "grad_norm": 0.9426065683364868,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 199480
+    },
+    {
+      "epoch": 1312.4342105263158,
+      "grad_norm": 0.7949636578559875,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 199490
+    },
+    {
+      "epoch": 1312.5,
+      "grad_norm": 1.197507619857788,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 199500
+    },
+    {
+      "epoch": 1312.5657894736842,
+      "grad_norm": 1.0171011686325073,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 199510
+    },
+    {
+      "epoch": 1312.6315789473683,
+      "grad_norm": 0.9687016010284424,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 199520
+    },
+    {
+      "epoch": 1312.6973684210527,
+      "grad_norm": 1.1884241104125977,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 199530
+    },
+    {
+      "epoch": 1312.7631578947369,
+      "grad_norm": 1.3759465217590332,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 199540
+    },
+    {
+      "epoch": 1312.828947368421,
+      "grad_norm": 1.368860125541687,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 199550
+    },
+    {
+      "epoch": 1312.8947368421052,
+      "grad_norm": 1.2088021039962769,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 199560
+    },
+    {
+      "epoch": 1312.9605263157894,
+      "grad_norm": 0.524134635925293,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 199570
+    },
+    {
+      "epoch": 1313.0263157894738,
+      "grad_norm": 1.3162397146224976,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 199580
+    },
+    {
+      "epoch": 1313.092105263158,
+      "grad_norm": 1.575825810432434,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 199590
+    },
+    {
+      "epoch": 1313.157894736842,
+      "grad_norm": 1.1224581003189087,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 199600
+    },
+    {
+      "epoch": 1313.2236842105262,
+      "grad_norm": 1.0737751722335815,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 199610
+    },
+    {
+      "epoch": 1313.2894736842106,
+      "grad_norm": 0.9804463982582092,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 199620
+    },
+    {
+      "epoch": 1313.3552631578948,
+      "grad_norm": 0.9724069833755493,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 199630
+    },
+    {
+      "epoch": 1313.421052631579,
+      "grad_norm": 1.1374396085739136,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 199640
+    },
+    {
+      "epoch": 1313.4868421052631,
+      "grad_norm": 0.8667737245559692,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 199650
+    },
+    {
+      "epoch": 1313.5526315789473,
+      "grad_norm": 0.9983107447624207,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 199660
+    },
+    {
+      "epoch": 1313.6184210526317,
+      "grad_norm": 1.2259620428085327,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 199670
+    },
+    {
+      "epoch": 1313.6842105263158,
+      "grad_norm": 0.9851988554000854,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 199680
+    },
+    {
+      "epoch": 1313.75,
+      "grad_norm": 0.5888131260871887,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 199690
+    },
+    {
+      "epoch": 1313.8157894736842,
+      "grad_norm": 0.9664670825004578,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 199700
+    },
+    {
+      "epoch": 1313.8815789473683,
+      "grad_norm": 0.9594868421554565,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 199710
+    },
+    {
+      "epoch": 1313.9473684210527,
+      "grad_norm": 0.9215731620788574,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 199720
+    },
+    {
+      "epoch": 1314.0131578947369,
+      "grad_norm": 1.1164031028747559,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 199730
+    },
+    {
+      "epoch": 1314.078947368421,
+      "grad_norm": 0.9203608632087708,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 199740
+    },
+    {
+      "epoch": 1314.1447368421052,
+      "grad_norm": 0.9199415445327759,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 199750
+    },
+    {
+      "epoch": 1314.2105263157894,
+      "grad_norm": 1.0740537643432617,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 199760
+    },
+    {
+      "epoch": 1314.2763157894738,
+      "grad_norm": 0.9334034323692322,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 199770
+    },
+    {
+      "epoch": 1314.342105263158,
+      "grad_norm": 1.0269443988800049,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 199780
+    },
+    {
+      "epoch": 1314.407894736842,
+      "grad_norm": 0.8074368834495544,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 199790
+    },
+    {
+      "epoch": 1314.4736842105262,
+      "grad_norm": 0.929581880569458,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 199800
+    },
+    {
+      "epoch": 1314.5394736842106,
+      "grad_norm": 0.8619239926338196,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 199810
+    },
+    {
+      "epoch": 1314.6052631578948,
+      "grad_norm": 1.4143906831741333,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 199820
+    },
+    {
+      "epoch": 1314.671052631579,
+      "grad_norm": 1.1911793947219849,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 199830
+    },
+    {
+      "epoch": 1314.7368421052631,
+      "grad_norm": 1.1507445573806763,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 199840
+    },
+    {
+      "epoch": 1314.8026315789473,
+      "grad_norm": 1.1112452745437622,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 199850
+    },
+    {
+      "epoch": 1314.8684210526317,
+      "grad_norm": 1.0850579738616943,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 199860
+    },
+    {
+      "epoch": 1314.9342105263158,
+      "grad_norm": 1.0612223148345947,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 199870
+    },
+    {
+      "epoch": 1315.0,
+      "grad_norm": 0.6841676831245422,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 199880
+    },
+    {
+      "epoch": 1315.0657894736842,
+      "grad_norm": 0.7603421211242676,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 199890
+    },
+    {
+      "epoch": 1315.1315789473683,
+      "grad_norm": 0.862991988658905,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 199900
+    },
+    {
+      "epoch": 1315.1973684210527,
+      "grad_norm": 0.9214925169944763,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 199910
+    },
+    {
+      "epoch": 1315.2631578947369,
+      "grad_norm": 1.2615870237350464,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 199920
+    },
+    {
+      "epoch": 1315.328947368421,
+      "grad_norm": 1.203710675239563,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 199930
+    },
+    {
+      "epoch": 1315.3947368421052,
+      "grad_norm": 1.0401464700698853,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 199940
+    },
+    {
+      "epoch": 1315.4605263157894,
+      "grad_norm": 0.9347955584526062,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 199950
+    },
+    {
+      "epoch": 1315.5263157894738,
+      "grad_norm": 0.8496779799461365,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 199960
+    },
+    {
+      "epoch": 1315.592105263158,
+      "grad_norm": 1.1234538555145264,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 199970
+    },
+    {
+      "epoch": 1315.657894736842,
+      "grad_norm": 1.2898207902908325,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 199980
+    },
+    {
+      "epoch": 1315.7236842105262,
+      "grad_norm": 0.773065984249115,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 199990
+    },
+    {
+      "epoch": 1315.7894736842106,
+      "grad_norm": 1.2071137428283691,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 200000
+    },
+    {
+      "epoch": 1315.7894736842106,
+      "step": 200000,
+      "total_flos": 0.0,
+      "train_loss": 0.01848614209484309,
+      "train_runtime": 93663.6535,
+      "train_samples_per_second": 136.659,
+      "train_steps_per_second": 2.135
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 200000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1316,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}