diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,35042 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 160.25641025641025,
+  "eval_steps": 500,
+  "global_step": 50000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03205128205128205,
+      "grad_norm": 145.42877197265625,
+      "learning_rate": 0.0001,
+      "loss": 20.0804,
+      "step": 10
+    },
+    {
+      "epoch": 0.0641025641025641,
+      "grad_norm": 83.6687240600586,
+      "learning_rate": 0.0001,
+      "loss": 4.8561,
+      "step": 20
+    },
+    {
+      "epoch": 0.09615384615384616,
+      "grad_norm": 66.00541687011719,
+      "learning_rate": 0.0001,
+      "loss": 3.7012,
+      "step": 30
+    },
+    {
+      "epoch": 0.1282051282051282,
+      "grad_norm": 62.172115325927734,
+      "learning_rate": 0.0001,
+      "loss": 3.1671,
+      "step": 40
+    },
+    {
+      "epoch": 0.16025641025641027,
+      "grad_norm": 53.95061492919922,
+      "learning_rate": 0.0001,
+      "loss": 2.9338,
+      "step": 50
+    },
+    {
+      "epoch": 0.19230769230769232,
+      "grad_norm": 45.865421295166016,
+      "learning_rate": 0.0001,
+      "loss": 2.7134,
+      "step": 60
+    },
+    {
+      "epoch": 0.22435897435897437,
+      "grad_norm": 49.08586502075195,
+      "learning_rate": 0.0001,
+      "loss": 2.5253,
+      "step": 70
+    },
+    {
+      "epoch": 0.2564102564102564,
+      "grad_norm": 40.58842468261719,
+      "learning_rate": 0.0001,
+      "loss": 2.3174,
+      "step": 80
+    },
+    {
+      "epoch": 0.28846153846153844,
+      "grad_norm": 42.8148193359375,
+      "learning_rate": 0.0001,
+      "loss": 2.1433,
+      "step": 90
+    },
+    {
+      "epoch": 0.32051282051282054,
+      "grad_norm": 43.857879638671875,
+      "learning_rate": 0.0001,
+      "loss": 2.0303,
+      "step": 100
+    },
+    {
+      "epoch": 0.3525641025641026,
+      "grad_norm": 37.79813766479492,
+      "learning_rate": 0.0001,
+      "loss": 1.8997,
+      "step": 110
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 42.826358795166016,
+      "learning_rate": 0.0001,
+      "loss": 1.8136,
+      "step": 120
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 45.04273223876953,
+      "learning_rate": 0.0001,
+      "loss": 1.705,
+      "step": 130
+    },
+    {
+      "epoch": 0.44871794871794873,
+      "grad_norm": 36.918312072753906,
+      "learning_rate": 0.0001,
+      "loss": 1.6156,
+      "step": 140
+    },
+    {
+      "epoch": 0.4807692307692308,
+      "grad_norm": 45.2752685546875,
+      "learning_rate": 0.0001,
+      "loss": 1.513,
+      "step": 150
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 32.438270568847656,
+      "learning_rate": 0.0001,
+      "loss": 1.4138,
+      "step": 160
+    },
+    {
+      "epoch": 0.5448717948717948,
+      "grad_norm": 40.694786071777344,
+      "learning_rate": 0.0001,
+      "loss": 1.3299,
+      "step": 170
+    },
+    {
+      "epoch": 0.5769230769230769,
+      "grad_norm": 32.529083251953125,
+      "learning_rate": 0.0001,
+      "loss": 1.2605,
+      "step": 180
+    },
+    {
+      "epoch": 0.6089743589743589,
+      "grad_norm": 33.108646392822266,
+      "learning_rate": 0.0001,
+      "loss": 1.1649,
+      "step": 190
+    },
+    {
+      "epoch": 0.6410256410256411,
+      "grad_norm": 38.705482482910156,
+      "learning_rate": 0.0001,
+      "loss": 1.1184,
+      "step": 200
+    },
+    {
+      "epoch": 0.6730769230769231,
+      "grad_norm": 37.113067626953125,
+      "learning_rate": 0.0001,
+      "loss": 1.0611,
+      "step": 210
+    },
+    {
+      "epoch": 0.7051282051282052,
+      "grad_norm": 36.26287841796875,
+      "learning_rate": 0.0001,
+      "loss": 1.0048,
+      "step": 220
+    },
+    {
+      "epoch": 0.7371794871794872,
+      "grad_norm": 37.88580322265625,
+      "learning_rate": 0.0001,
+      "loss": 0.939,
+      "step": 230
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 38.3851432800293,
+      "learning_rate": 0.0001,
+      "loss": 0.8938,
+      "step": 240
+    },
+    {
+      "epoch": 0.8012820512820513,
+      "grad_norm": 33.174652099609375,
+      "learning_rate": 0.0001,
+      "loss": 0.8694,
+      "step": 250
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 27.991905212402344,
+      "learning_rate": 0.0001,
+      "loss": 0.8073,
+      "step": 260
+    },
+    {
+      "epoch": 0.8653846153846154,
+      "grad_norm": 31.897321701049805,
+      "learning_rate": 0.0001,
+      "loss": 0.7375,
+      "step": 270
+    },
+    {
+      "epoch": 0.8974358974358975,
+      "grad_norm": 30.297447204589844,
+      "learning_rate": 0.0001,
+      "loss": 0.6981,
+      "step": 280
+    },
+    {
+      "epoch": 0.9294871794871795,
+      "grad_norm": 26.0345458984375,
+      "learning_rate": 0.0001,
+      "loss": 0.6543,
+      "step": 290
+    },
+    {
+      "epoch": 0.9615384615384616,
+      "grad_norm": 26.216842651367188,
+      "learning_rate": 0.0001,
+      "loss": 0.6314,
+      "step": 300
+    },
+    {
+      "epoch": 0.9935897435897436,
+      "grad_norm": 24.59775733947754,
+      "learning_rate": 0.0001,
+      "loss": 0.598,
+      "step": 310
+    },
+    {
+      "epoch": 1.0256410256410255,
+      "grad_norm": 28.747257232666016,
+      "learning_rate": 0.0001,
+      "loss": 0.5461,
+      "step": 320
+    },
+    {
+      "epoch": 1.0576923076923077,
+      "grad_norm": 24.38837432861328,
+      "learning_rate": 0.0001,
+      "loss": 0.5168,
+      "step": 330
+    },
+    {
+      "epoch": 1.0897435897435896,
+      "grad_norm": 22.864219665527344,
+      "learning_rate": 0.0001,
+      "loss": 0.5058,
+      "step": 340
+    },
+    {
+      "epoch": 1.1217948717948718,
+      "grad_norm": 21.214086532592773,
+      "learning_rate": 0.0001,
+      "loss": 0.4755,
+      "step": 350
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 22.62112808227539,
+      "learning_rate": 0.0001,
+      "loss": 0.4667,
+      "step": 360
+    },
+    {
+      "epoch": 1.185897435897436,
+      "grad_norm": 30.92898941040039,
+      "learning_rate": 0.0001,
+      "loss": 0.4388,
+      "step": 370
+    },
+    {
+      "epoch": 1.217948717948718,
+      "grad_norm": 23.30482292175293,
+      "learning_rate": 0.0001,
+      "loss": 0.4496,
+      "step": 380
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 22.879722595214844,
+      "learning_rate": 0.0001,
+      "loss": 0.4176,
+      "step": 390
+    },
+    {
+      "epoch": 1.282051282051282,
+      "grad_norm": 27.631187438964844,
+      "learning_rate": 0.0001,
+      "loss": 0.4063,
+      "step": 400
+    },
+    {
+      "epoch": 1.314102564102564,
+      "grad_norm": 23.220075607299805,
+      "learning_rate": 0.0001,
+      "loss": 0.3862,
+      "step": 410
+    },
+    {
+      "epoch": 1.3461538461538463,
+      "grad_norm": 19.1767635345459,
+      "learning_rate": 0.0001,
+      "loss": 0.3558,
+      "step": 420
+    },
+    {
+      "epoch": 1.3782051282051282,
+      "grad_norm": 18.207799911499023,
+      "learning_rate": 0.0001,
+      "loss": 0.3334,
+      "step": 430
+    },
+    {
+      "epoch": 1.4102564102564101,
+      "grad_norm": 20.599430084228516,
+      "learning_rate": 0.0001,
+      "loss": 0.3336,
+      "step": 440
+    },
+    {
+      "epoch": 1.4423076923076923,
+      "grad_norm": 21.528234481811523,
+      "learning_rate": 0.0001,
+      "loss": 0.3188,
+      "step": 450
+    },
+    {
+      "epoch": 1.4743589743589745,
+      "grad_norm": 20.507978439331055,
+      "learning_rate": 0.0001,
+      "loss": 0.3045,
+      "step": 460
+    },
+    {
+      "epoch": 1.5064102564102564,
+      "grad_norm": 18.63322639465332,
+      "learning_rate": 0.0001,
+      "loss": 0.2926,
+      "step": 470
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 21.291637420654297,
+      "learning_rate": 0.0001,
+      "loss": 0.2838,
+      "step": 480
+    },
+    {
+      "epoch": 1.5705128205128205,
+      "grad_norm": 19.38383674621582,
+      "learning_rate": 0.0001,
+      "loss": 0.2746,
+      "step": 490
+    },
+    {
+      "epoch": 1.6025641025641026,
+      "grad_norm": 18.03575897216797,
+      "learning_rate": 0.0001,
+      "loss": 0.2578,
+      "step": 500
+    },
+    {
+      "epoch": 1.6346153846153846,
+      "grad_norm": 17.893205642700195,
+      "learning_rate": 0.0001,
+      "loss": 0.2569,
+      "step": 510
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 17.73887062072754,
+      "learning_rate": 0.0001,
+      "loss": 0.2436,
+      "step": 520
+    },
+    {
+      "epoch": 1.6987179487179487,
+      "grad_norm": 20.73174285888672,
+      "learning_rate": 0.0001,
+      "loss": 0.2445,
+      "step": 530
+    },
+    {
+      "epoch": 1.7307692307692308,
+      "grad_norm": 18.894405364990234,
+      "learning_rate": 0.0001,
+      "loss": 0.242,
+      "step": 540
+    },
+    {
+      "epoch": 1.7628205128205128,
+      "grad_norm": 19.611589431762695,
+      "learning_rate": 0.0001,
+      "loss": 0.2338,
+      "step": 550
+    },
+    {
+      "epoch": 1.7948717948717947,
+      "grad_norm": 16.299564361572266,
+      "learning_rate": 0.0001,
+      "loss": 0.2196,
+      "step": 560
+    },
+    {
+      "epoch": 1.8269230769230769,
+      "grad_norm": 17.933652877807617,
+      "learning_rate": 0.0001,
+      "loss": 0.2199,
+      "step": 570
+    },
+    {
+      "epoch": 1.858974358974359,
+      "grad_norm": 17.967666625976562,
+      "learning_rate": 0.0001,
+      "loss": 0.2129,
+      "step": 580
+    },
+    {
+      "epoch": 1.891025641025641,
+      "grad_norm": 14.8726224899292,
+      "learning_rate": 0.0001,
+      "loss": 0.2031,
+      "step": 590
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 16.91478157043457,
+      "learning_rate": 0.0001,
+      "loss": 0.2053,
+      "step": 600
+    },
+    {
+      "epoch": 1.9551282051282053,
+      "grad_norm": 16.59592628479004,
+      "learning_rate": 0.0001,
+      "loss": 0.2133,
+      "step": 610
+    },
+    {
+      "epoch": 1.9871794871794872,
+      "grad_norm": 15.259805679321289,
+      "learning_rate": 0.0001,
+      "loss": 0.2066,
+      "step": 620
+    },
+    {
+      "epoch": 2.019230769230769,
+      "grad_norm": 18.76848602294922,
+      "learning_rate": 0.0001,
+      "loss": 0.1966,
+      "step": 630
+    },
+    {
+      "epoch": 2.051282051282051,
+      "grad_norm": 14.207231521606445,
+      "learning_rate": 0.0001,
+      "loss": 0.1979,
+      "step": 640
+    },
+    {
+      "epoch": 2.0833333333333335,
+      "grad_norm": 14.777447700500488,
+      "learning_rate": 0.0001,
+      "loss": 0.1997,
+      "step": 650
+    },
+    {
+      "epoch": 2.1153846153846154,
+      "grad_norm": 15.572965621948242,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 660
+    },
+    {
+      "epoch": 2.1474358974358974,
+      "grad_norm": 13.691197395324707,
+      "learning_rate": 0.0001,
+      "loss": 0.1809,
+      "step": 670
+    },
+    {
+      "epoch": 2.1794871794871793,
+      "grad_norm": 13.38075065612793,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 680
+    },
+    {
+      "epoch": 2.2115384615384617,
+      "grad_norm": 13.9527587890625,
+      "learning_rate": 0.0001,
+      "loss": 0.1786,
+      "step": 690
+    },
+    {
+      "epoch": 2.2435897435897436,
+      "grad_norm": 12.5586519241333,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 700
+    },
+    {
+      "epoch": 2.2756410256410255,
+      "grad_norm": 16.166288375854492,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 710
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 17.70206642150879,
+      "learning_rate": 0.0001,
+      "loss": 0.1746,
+      "step": 720
+    },
+    {
+      "epoch": 2.33974358974359,
+      "grad_norm": 15.4883394241333,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 730
+    },
+    {
+      "epoch": 2.371794871794872,
+      "grad_norm": 15.548429489135742,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 740
+    },
+    {
+      "epoch": 2.4038461538461537,
+      "grad_norm": 17.554248809814453,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 750
+    },
+    {
+      "epoch": 2.435897435897436,
+      "grad_norm": 13.252490043640137,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 760
+    },
+    {
+      "epoch": 2.467948717948718,
+      "grad_norm": 13.457218170166016,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 770
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 14.840357780456543,
+      "learning_rate": 0.0001,
+      "loss": 0.1573,
+      "step": 780
+    },
+    {
+      "epoch": 2.532051282051282,
+      "grad_norm": 12.185482025146484,
+      "learning_rate": 0.0001,
+      "loss": 0.1559,
+      "step": 790
+    },
+    {
+      "epoch": 2.564102564102564,
+      "grad_norm": 11.851032257080078,
+      "learning_rate": 0.0001,
+      "loss": 0.1501,
+      "step": 800
+    },
+    {
+      "epoch": 2.5961538461538463,
+      "grad_norm": 12.129788398742676,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 810
+    },
+    {
+      "epoch": 2.628205128205128,
+      "grad_norm": 14.25014591217041,
+      "learning_rate": 0.0001,
+      "loss": 0.1509,
+      "step": 820
+    },
+    {
+      "epoch": 2.66025641025641,
+      "grad_norm": 15.993083953857422,
+      "learning_rate": 0.0001,
+      "loss": 0.1509,
+      "step": 830
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 15.221981048583984,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 840
+    },
+    {
+      "epoch": 2.7243589743589745,
+      "grad_norm": 13.071995735168457,
+      "learning_rate": 0.0001,
+      "loss": 0.1444,
+      "step": 850
+    },
+    {
+      "epoch": 2.7564102564102564,
+      "grad_norm": 15.202935218811035,
+      "learning_rate": 0.0001,
+      "loss": 0.1407,
+      "step": 860
+    },
+    {
+      "epoch": 2.7884615384615383,
+      "grad_norm": 12.28674030303955,
+      "learning_rate": 0.0001,
+      "loss": 0.1355,
+      "step": 870
+    },
+    {
+      "epoch": 2.8205128205128203,
+      "grad_norm": 12.3435640335083,
+      "learning_rate": 0.0001,
+      "loss": 0.1356,
+      "step": 880
+    },
+    {
+      "epoch": 2.8525641025641026,
+      "grad_norm": 13.236529350280762,
+      "learning_rate": 0.0001,
+      "loss": 0.1323,
+      "step": 890
+    },
+    {
+      "epoch": 2.8846153846153846,
+      "grad_norm": 9.97934341430664,
+      "learning_rate": 0.0001,
+      "loss": 0.1348,
+      "step": 900
+    },
+    {
+      "epoch": 2.9166666666666665,
+      "grad_norm": 13.574583053588867,
+      "learning_rate": 0.0001,
+      "loss": 0.1292,
+      "step": 910
+    },
+    {
+      "epoch": 2.948717948717949,
+      "grad_norm": 11.287897109985352,
+      "learning_rate": 0.0001,
+      "loss": 0.1252,
+      "step": 920
+    },
+    {
+      "epoch": 2.980769230769231,
+      "grad_norm": 11.3424711227417,
+      "learning_rate": 0.0001,
+      "loss": 0.1243,
+      "step": 930
+    },
+    {
+      "epoch": 3.0128205128205128,
+      "grad_norm": 12.857080459594727,
+      "learning_rate": 0.0001,
+      "loss": 0.1273,
+      "step": 940
+    },
+    {
+      "epoch": 3.0448717948717947,
+      "grad_norm": 12.469817161560059,
+      "learning_rate": 0.0001,
+      "loss": 0.1242,
+      "step": 950
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 13.245798110961914,
+      "learning_rate": 0.0001,
+      "loss": 0.1288,
+      "step": 960
+    },
+    {
+      "epoch": 3.108974358974359,
+      "grad_norm": 12.69701099395752,
+      "learning_rate": 0.0001,
+      "loss": 0.1228,
+      "step": 970
+    },
+    {
+      "epoch": 3.141025641025641,
+      "grad_norm": 11.858522415161133,
+      "learning_rate": 0.0001,
+      "loss": 0.1182,
+      "step": 980
+    },
+    {
+      "epoch": 3.173076923076923,
+      "grad_norm": 11.331292152404785,
+      "learning_rate": 0.0001,
+      "loss": 0.1193,
+      "step": 990
+    },
+    {
+      "epoch": 3.2051282051282053,
+      "grad_norm": 9.510218620300293,
+      "learning_rate": 0.0001,
+      "loss": 0.1149,
+      "step": 1000
+    },
+    {
+      "epoch": 3.2371794871794872,
+      "grad_norm": 11.03514289855957,
+      "learning_rate": 0.0001,
+      "loss": 0.1147,
+      "step": 1010
+    },
+    {
+      "epoch": 3.269230769230769,
+      "grad_norm": 11.662793159484863,
+      "learning_rate": 0.0001,
+      "loss": 0.1129,
+      "step": 1020
+    },
+    {
+      "epoch": 3.301282051282051,
+      "grad_norm": 11.718436241149902,
+      "learning_rate": 0.0001,
+      "loss": 0.1148,
+      "step": 1030
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 11.538015365600586,
+      "learning_rate": 0.0001,
+      "loss": 0.1115,
+      "step": 1040
+    },
+    {
+      "epoch": 3.3653846153846154,
+      "grad_norm": 12.65234088897705,
+      "learning_rate": 0.0001,
+      "loss": 0.1158,
+      "step": 1050
+    },
+    {
+      "epoch": 3.3974358974358974,
+      "grad_norm": 9.606489181518555,
+      "learning_rate": 0.0001,
+      "loss": 0.1131,
+      "step": 1060
+    },
+    {
+      "epoch": 3.4294871794871793,
+      "grad_norm": 11.021942138671875,
+      "learning_rate": 0.0001,
+      "loss": 0.1108,
+      "step": 1070
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 11.315251350402832,
+      "learning_rate": 0.0001,
+      "loss": 0.1141,
+      "step": 1080
+    },
+    {
+      "epoch": 3.4935897435897436,
+      "grad_norm": 13.794599533081055,
+      "learning_rate": 0.0001,
+      "loss": 0.1177,
+      "step": 1090
+    },
+    {
+      "epoch": 3.5256410256410255,
+      "grad_norm": 13.967779159545898,
+      "learning_rate": 0.0001,
+      "loss": 0.1138,
+      "step": 1100
+    },
+    {
+      "epoch": 3.5576923076923075,
+      "grad_norm": 10.828083992004395,
+      "learning_rate": 0.0001,
+      "loss": 0.115,
+      "step": 1110
+    },
+    {
+      "epoch": 3.58974358974359,
+      "grad_norm": 12.068327903747559,
+      "learning_rate": 0.0001,
+      "loss": 0.1122,
+      "step": 1120
+    },
+    {
+      "epoch": 3.621794871794872,
+      "grad_norm": 10.905609130859375,
+      "learning_rate": 0.0001,
+      "loss": 0.1071,
+      "step": 1130
+    },
+    {
+      "epoch": 3.6538461538461537,
+      "grad_norm": 10.249520301818848,
+      "learning_rate": 0.0001,
+      "loss": 0.1115,
+      "step": 1140
+    },
+    {
+      "epoch": 3.685897435897436,
+      "grad_norm": 11.281134605407715,
+      "learning_rate": 0.0001,
+      "loss": 0.1138,
+      "step": 1150
+    },
+    {
+      "epoch": 3.717948717948718,
+      "grad_norm": 10.699358940124512,
+      "learning_rate": 0.0001,
+      "loss": 0.1127,
+      "step": 1160
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 12.13260555267334,
+      "learning_rate": 0.0001,
+      "loss": 0.1111,
+      "step": 1170
+    },
+    {
+      "epoch": 3.782051282051282,
+      "grad_norm": 13.125444412231445,
+      "learning_rate": 0.0001,
+      "loss": 0.1063,
+      "step": 1180
+    },
+    {
+      "epoch": 3.814102564102564,
+      "grad_norm": 11.16601276397705,
+      "learning_rate": 0.0001,
+      "loss": 0.1035,
+      "step": 1190
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 11.007014274597168,
+      "learning_rate": 0.0001,
+      "loss": 0.1013,
+      "step": 1200
+    },
+    {
+      "epoch": 3.878205128205128,
+      "grad_norm": 11.488813400268555,
+      "learning_rate": 0.0001,
+      "loss": 0.0995,
+      "step": 1210
+    },
+    {
+      "epoch": 3.91025641025641,
+      "grad_norm": 8.965542793273926,
+      "learning_rate": 0.0001,
+      "loss": 0.0995,
+      "step": 1220
+    },
+    {
+      "epoch": 3.9423076923076925,
+      "grad_norm": 12.162210464477539,
+      "learning_rate": 0.0001,
+      "loss": 0.0984,
+      "step": 1230
+    },
+    {
+      "epoch": 3.9743589743589745,
+      "grad_norm": 10.48455810546875,
+      "learning_rate": 0.0001,
+      "loss": 0.0969,
+      "step": 1240
+    },
+    {
+      "epoch": 4.006410256410256,
+      "grad_norm": 8.865561485290527,
+      "learning_rate": 0.0001,
+      "loss": 0.0997,
+      "step": 1250
+    },
+    {
+      "epoch": 4.038461538461538,
+      "grad_norm": 10.782670974731445,
+      "learning_rate": 0.0001,
+      "loss": 0.0977,
+      "step": 1260
+    },
+    {
+      "epoch": 4.07051282051282,
+      "grad_norm": 10.074785232543945,
+      "learning_rate": 0.0001,
+      "loss": 0.098,
+      "step": 1270
+    },
+    {
+      "epoch": 4.102564102564102,
+      "grad_norm": 9.265982627868652,
+      "learning_rate": 0.0001,
+      "loss": 0.0935,
+      "step": 1280
+    },
+    {
+      "epoch": 4.134615384615385,
+      "grad_norm": 9.139842987060547,
+      "learning_rate": 0.0001,
+      "loss": 0.0949,
+      "step": 1290
+    },
+    {
+      "epoch": 4.166666666666667,
+      "grad_norm": 11.679908752441406,
+      "learning_rate": 0.0001,
+      "loss": 0.0921,
+      "step": 1300
+    },
+    {
+      "epoch": 4.198717948717949,
+      "grad_norm": 9.957094192504883,
+      "learning_rate": 0.0001,
+      "loss": 0.0936,
+      "step": 1310
+    },
+    {
+      "epoch": 4.230769230769231,
+      "grad_norm": 10.352608680725098,
+      "learning_rate": 0.0001,
+      "loss": 0.0912,
+      "step": 1320
+    },
+    {
+      "epoch": 4.262820512820513,
+      "grad_norm": 9.618124961853027,
+      "learning_rate": 0.0001,
+      "loss": 0.0942,
+      "step": 1330
+    },
+    {
+      "epoch": 4.294871794871795,
+      "grad_norm": 8.931426048278809,
+      "learning_rate": 0.0001,
+      "loss": 0.0983,
+      "step": 1340
+    },
+    {
+      "epoch": 4.326923076923077,
+      "grad_norm": 8.72685718536377,
+      "learning_rate": 0.0001,
+      "loss": 0.0942,
+      "step": 1350
+    },
+    {
+      "epoch": 4.358974358974359,
+      "grad_norm": 12.073081016540527,
+      "learning_rate": 0.0001,
+      "loss": 0.0921,
+      "step": 1360
+    },
+    {
+      "epoch": 4.391025641025641,
+      "grad_norm": 10.850428581237793,
+      "learning_rate": 0.0001,
+      "loss": 0.0927,
+      "step": 1370
+    },
+    {
+      "epoch": 4.423076923076923,
+      "grad_norm": 13.887635231018066,
+      "learning_rate": 0.0001,
+      "loss": 0.095,
+      "step": 1380
+    },
+    {
+      "epoch": 4.455128205128205,
+      "grad_norm": 8.08206844329834,
+      "learning_rate": 0.0001,
+      "loss": 0.0925,
+      "step": 1390
+    },
+    {
+      "epoch": 4.487179487179487,
+      "grad_norm": 8.544827461242676,
+      "learning_rate": 0.0001,
+      "loss": 0.0881,
+      "step": 1400
+    },
+    {
+      "epoch": 4.519230769230769,
+      "grad_norm": 7.460143089294434,
+      "learning_rate": 0.0001,
+      "loss": 0.0865,
+      "step": 1410
+    },
+    {
+      "epoch": 4.551282051282051,
+      "grad_norm": 9.715693473815918,
+      "learning_rate": 0.0001,
+      "loss": 0.0858,
+      "step": 1420
+    },
+    {
+      "epoch": 4.583333333333333,
+      "grad_norm": 8.187448501586914,
+      "learning_rate": 0.0001,
+      "loss": 0.0853,
+      "step": 1430
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 8.072632789611816,
+      "learning_rate": 0.0001,
+      "loss": 0.0851,
+      "step": 1440
+    },
+    {
+      "epoch": 4.647435897435898,
+      "grad_norm": 9.01579761505127,
+      "learning_rate": 0.0001,
+      "loss": 0.0822,
+      "step": 1450
+    },
+    {
+      "epoch": 4.67948717948718,
+      "grad_norm": 8.434136390686035,
+      "learning_rate": 0.0001,
+      "loss": 0.0849,
+      "step": 1460
+    },
+    {
+      "epoch": 4.711538461538462,
+      "grad_norm": 8.921215057373047,
+      "learning_rate": 0.0001,
+      "loss": 0.085,
+      "step": 1470
+    },
+    {
+      "epoch": 4.743589743589744,
+      "grad_norm": 9.078075408935547,
+      "learning_rate": 0.0001,
+      "loss": 0.0826,
+      "step": 1480
+    },
+    {
+      "epoch": 4.7756410256410255,
+      "grad_norm": 8.944608688354492,
+      "learning_rate": 0.0001,
+      "loss": 0.0786,
+      "step": 1490
+    },
+    {
+      "epoch": 4.8076923076923075,
+      "grad_norm": 10.6521577835083,
+      "learning_rate": 0.0001,
+      "loss": 0.0798,
+      "step": 1500
+    },
+    {
+      "epoch": 4.839743589743589,
+      "grad_norm": 9.344939231872559,
+      "learning_rate": 0.0001,
+      "loss": 0.0831,
+      "step": 1510
+    },
+    {
+      "epoch": 4.871794871794872,
+      "grad_norm": 9.095298767089844,
+      "learning_rate": 0.0001,
+      "loss": 0.0824,
+      "step": 1520
+    },
+    {
+      "epoch": 4.903846153846154,
+      "grad_norm": 9.012145042419434,
+      "learning_rate": 0.0001,
+      "loss": 0.0793,
+      "step": 1530
+    },
+    {
+      "epoch": 4.935897435897436,
+      "grad_norm": 8.63984489440918,
+      "learning_rate": 0.0001,
+      "loss": 0.0789,
+      "step": 1540
+    },
+    {
+      "epoch": 4.967948717948718,
+      "grad_norm": 9.23857593536377,
+      "learning_rate": 0.0001,
+      "loss": 0.0779,
+      "step": 1550
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 10.09559154510498,
+      "learning_rate": 0.0001,
+      "loss": 0.0776,
+      "step": 1560
+    },
+    {
+      "epoch": 5.032051282051282,
+      "grad_norm": 8.263296127319336,
+      "learning_rate": 0.0001,
+      "loss": 0.0766,
+      "step": 1570
+    },
+    {
+      "epoch": 5.064102564102564,
+      "grad_norm": 9.643431663513184,
+      "learning_rate": 0.0001,
+      "loss": 0.0759,
+      "step": 1580
+    },
+    {
+      "epoch": 5.096153846153846,
+      "grad_norm": 9.427698135375977,
+      "learning_rate": 0.0001,
+      "loss": 0.0793,
+      "step": 1590
+    },
+    {
+      "epoch": 5.128205128205128,
+      "grad_norm": 8.024136543273926,
+      "learning_rate": 0.0001,
+      "loss": 0.0755,
+      "step": 1600
+    },
+    {
+      "epoch": 5.160256410256411,
+      "grad_norm": 8.440216064453125,
+      "learning_rate": 0.0001,
+      "loss": 0.0737,
+      "step": 1610
+    },
+    {
+      "epoch": 5.1923076923076925,
+      "grad_norm": 9.354905128479004,
+      "learning_rate": 0.0001,
+      "loss": 0.073,
+      "step": 1620
+    },
+    {
+      "epoch": 5.2243589743589745,
+      "grad_norm": 8.812016487121582,
+      "learning_rate": 0.0001,
+      "loss": 0.0714,
+      "step": 1630
+    },
+    {
+      "epoch": 5.256410256410256,
+      "grad_norm": 9.55749225616455,
+      "learning_rate": 0.0001,
+      "loss": 0.0711,
+      "step": 1640
+    },
+    {
+      "epoch": 5.288461538461538,
+      "grad_norm": 8.780061721801758,
+      "learning_rate": 0.0001,
+      "loss": 0.0726,
+      "step": 1650
+    },
+    {
+      "epoch": 5.32051282051282,
+      "grad_norm": 7.259555816650391,
+      "learning_rate": 0.0001,
+      "loss": 0.0712,
+      "step": 1660
+    },
+    {
+      "epoch": 5.352564102564102,
+      "grad_norm": 8.978029251098633,
+      "learning_rate": 0.0001,
+      "loss": 0.0715,
+      "step": 1670
+    },
+    {
+      "epoch": 5.384615384615385,
+      "grad_norm": 9.483819007873535,
+      "learning_rate": 0.0001,
+      "loss": 0.0715,
+      "step": 1680
+    },
+    {
+      "epoch": 5.416666666666667,
+      "grad_norm": 8.16633415222168,
+      "learning_rate": 0.0001,
+      "loss": 0.071,
+      "step": 1690
+    },
+    {
+      "epoch": 5.448717948717949,
+      "grad_norm": 8.834074974060059,
+      "learning_rate": 0.0001,
+      "loss": 0.0707,
+      "step": 1700
+    },
+    {
+      "epoch": 5.480769230769231,
+      "grad_norm": 8.87706470489502,
+      "learning_rate": 0.0001,
+      "loss": 0.0712,
+      "step": 1710
+    },
+    {
+      "epoch": 5.512820512820513,
+      "grad_norm": 8.785717964172363,
+      "learning_rate": 0.0001,
+      "loss": 0.0688,
+      "step": 1720
+    },
+    {
+      "epoch": 5.544871794871795,
+      "grad_norm": 9.460667610168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0712,
+      "step": 1730
+    },
+    {
+      "epoch": 5.576923076923077,
+      "grad_norm": 7.309500217437744,
+      "learning_rate": 0.0001,
+      "loss": 0.0712,
+      "step": 1740
+    },
+    {
+      "epoch": 5.608974358974359,
+      "grad_norm": 8.23578929901123,
+      "learning_rate": 0.0001,
+      "loss": 0.0709,
+      "step": 1750
+    },
+    {
+      "epoch": 5.641025641025641,
+      "grad_norm": 8.063543319702148,
+      "learning_rate": 0.0001,
+      "loss": 0.0692,
+      "step": 1760
+    },
+    {
+      "epoch": 5.673076923076923,
+      "grad_norm": 6.716485500335693,
+      "learning_rate": 0.0001,
+      "loss": 0.0674,
+      "step": 1770
+    },
+    {
+      "epoch": 5.705128205128205,
+      "grad_norm": 8.969748497009277,
+      "learning_rate": 0.0001,
+      "loss": 0.0686,
+      "step": 1780
+    },
+    {
+      "epoch": 5.737179487179487,
+      "grad_norm": 7.19386625289917,
+      "learning_rate": 0.0001,
+      "loss": 0.0681,
+      "step": 1790
+    },
+    {
+      "epoch": 5.769230769230769,
+      "grad_norm": 10.67116641998291,
+      "learning_rate": 0.0001,
+      "loss": 0.07,
+      "step": 1800
+    },
+    {
+      "epoch": 5.801282051282051,
+      "grad_norm": 7.234352111816406,
+      "learning_rate": 0.0001,
+      "loss": 0.0689,
+      "step": 1810
+    },
+    {
+      "epoch": 5.833333333333333,
+      "grad_norm": 8.62362289428711,
+      "learning_rate": 0.0001,
+      "loss": 0.066,
+      "step": 1820
+    },
+    {
+      "epoch": 5.865384615384615,
+      "grad_norm": 10.142123222351074,
+      "learning_rate": 0.0001,
+      "loss": 0.0709,
+      "step": 1830
+    },
+    {
+      "epoch": 5.897435897435898,
+      "grad_norm": 9.105326652526855,
+      "learning_rate": 0.0001,
+      "loss": 0.069,
+      "step": 1840
+    },
+    {
+      "epoch": 5.92948717948718,
+      "grad_norm": 9.1032133102417,
+      "learning_rate": 0.0001,
+      "loss": 0.0664,
+      "step": 1850
+    },
+    {
+      "epoch": 5.961538461538462,
+      "grad_norm": 7.680888652801514,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 1860
+    },
+    {
+      "epoch": 5.993589743589744,
+      "grad_norm": 7.140402317047119,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 1870
+    },
+    {
+      "epoch": 6.0256410256410255,
+      "grad_norm": 9.224740028381348,
+      "learning_rate": 0.0001,
+      "loss": 0.0656,
+      "step": 1880
+    },
+    {
+      "epoch": 6.0576923076923075,
+      "grad_norm": 8.018192291259766,
+      "learning_rate": 0.0001,
+      "loss": 0.0623,
+      "step": 1890
+    },
+    {
+      "epoch": 6.089743589743589,
+      "grad_norm": 7.111850261688232,
+      "learning_rate": 0.0001,
+      "loss": 0.064,
+      "step": 1900
+    },
+    {
+      "epoch": 6.121794871794871,
+      "grad_norm": 6.5859270095825195,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 1910
+    },
+    {
+      "epoch": 6.153846153846154,
+      "grad_norm": 7.601027011871338,
+      "learning_rate": 0.0001,
+      "loss": 0.0633,
+      "step": 1920
+    },
+    {
+      "epoch": 6.185897435897436,
+      "grad_norm": 7.0342512130737305,
+      "learning_rate": 0.0001,
+      "loss": 0.0626,
+      "step": 1930
+    },
+    {
+      "epoch": 6.217948717948718,
+      "grad_norm": 6.453479290008545,
+      "learning_rate": 0.0001,
+      "loss": 0.0623,
+      "step": 1940
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 7.58675479888916,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "step": 1950
+    },
+    {
+      "epoch": 6.282051282051282,
+      "grad_norm": 8.58928394317627,
+      "learning_rate": 0.0001,
+      "loss": 0.0675,
+      "step": 1960
+    },
+    {
+      "epoch": 6.314102564102564,
+      "grad_norm": 7.057661533355713,
+      "learning_rate": 0.0001,
+      "loss": 0.0622,
+      "step": 1970
+    },
+    {
+      "epoch": 6.346153846153846,
+      "grad_norm": 6.5727925300598145,
+      "learning_rate": 0.0001,
+      "loss": 0.0612,
+      "step": 1980
+    },
+    {
+      "epoch": 6.378205128205128,
+      "grad_norm": 7.9517903327941895,
+      "learning_rate": 0.0001,
+      "loss": 0.0633,
+      "step": 1990
+    },
+    {
+      "epoch": 6.410256410256411,
+      "grad_norm": 9.338974952697754,
+      "learning_rate": 0.0001,
+      "loss": 0.0672,
+      "step": 2000
+    },
+    {
+      "epoch": 6.4423076923076925,
+      "grad_norm": 9.999692916870117,
+      "learning_rate": 0.0001,
+      "loss": 0.0661,
+      "step": 2010
+    },
+    {
+      "epoch": 6.4743589743589745,
+      "grad_norm": 10.284932136535645,
+      "learning_rate": 0.0001,
+      "loss": 0.0608,
+      "step": 2020
+    },
+    {
+      "epoch": 6.506410256410256,
+      "grad_norm": 6.788944244384766,
+      "learning_rate": 0.0001,
+      "loss": 0.0567,
+      "step": 2030
+    },
+    {
+      "epoch": 6.538461538461538,
+      "grad_norm": 7.548011302947998,
+      "learning_rate": 0.0001,
+      "loss": 0.0555,
+      "step": 2040
+    },
+    {
+      "epoch": 6.57051282051282,
+      "grad_norm": 6.80021858215332,
+      "learning_rate": 0.0001,
+      "loss": 0.0546,
+      "step": 2050
+    },
+    {
+      "epoch": 6.602564102564102,
+      "grad_norm": 6.221144199371338,
+      "learning_rate": 0.0001,
+      "loss": 0.0574,
+      "step": 2060
+    },
+    {
+      "epoch": 6.634615384615385,
+      "grad_norm": 6.181380748748779,
+      "learning_rate": 0.0001,
+      "loss": 0.0565,
+      "step": 2070
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 7.520815849304199,
+      "learning_rate": 0.0001,
+      "loss": 0.0585,
+      "step": 2080
+    },
+    {
+      "epoch": 6.698717948717949,
+      "grad_norm": 7.462428569793701,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "step": 2090
+    },
+    {
+      "epoch": 6.730769230769231,
+      "grad_norm": 7.517280578613281,
+      "learning_rate": 0.0001,
+      "loss": 0.0552,
+      "step": 2100
+    },
+    {
+      "epoch": 6.762820512820513,
+      "grad_norm": 7.333137035369873,
+      "learning_rate": 0.0001,
+      "loss": 0.0556,
+      "step": 2110
+    },
+    {
+      "epoch": 6.794871794871795,
+      "grad_norm": 7.370513916015625,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "step": 2120
+    },
+    {
+      "epoch": 6.826923076923077,
+      "grad_norm": 6.427040100097656,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 2130
+    },
+    {
+      "epoch": 6.858974358974359,
+      "grad_norm": 6.2469658851623535,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 2140
+    },
+    {
+      "epoch": 6.891025641025641,
+      "grad_norm": 6.883134365081787,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 2150
+    },
+    {
+      "epoch": 6.923076923076923,
+      "grad_norm": 6.6050262451171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 2160
+    },
+    {
+      "epoch": 6.955128205128205,
+      "grad_norm": 6.507558822631836,
+      "learning_rate": 0.0001,
+      "loss": 0.0557,
+      "step": 2170
+    },
+    {
+      "epoch": 6.987179487179487,
+      "grad_norm": 5.983650207519531,
+      "learning_rate": 0.0001,
+      "loss": 0.0571,
+      "step": 2180
+    },
+    {
+      "epoch": 7.019230769230769,
+      "grad_norm": 8.619466781616211,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 2190
+    },
+    {
+      "epoch": 7.051282051282051,
+      "grad_norm": 6.656548976898193,
+      "learning_rate": 0.0001,
+      "loss": 0.0592,
+      "step": 2200
+    },
+    {
+      "epoch": 7.083333333333333,
+      "grad_norm": 6.790521621704102,
+      "learning_rate": 0.0001,
+      "loss": 0.0593,
+      "step": 2210
+    },
+    {
+      "epoch": 7.115384615384615,
+      "grad_norm": 7.247766017913818,
+      "learning_rate": 0.0001,
+      "loss": 0.0601,
+      "step": 2220
+    },
+    {
+      "epoch": 7.147435897435898,
+      "grad_norm": 8.490907669067383,
+      "learning_rate": 0.0001,
+      "loss": 0.0602,
+      "step": 2230
+    },
+    {
+      "epoch": 7.17948717948718,
+      "grad_norm": 6.294097423553467,
+      "learning_rate": 0.0001,
+      "loss": 0.0579,
+      "step": 2240
+    },
+    {
+      "epoch": 7.211538461538462,
+      "grad_norm": 6.9430251121521,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "step": 2250
+    },
+    {
+      "epoch": 7.243589743589744,
+      "grad_norm": 6.313124179840088,
+      "learning_rate": 0.0001,
+      "loss": 0.0549,
+      "step": 2260
+    },
+    {
+      "epoch": 7.2756410256410255,
+      "grad_norm": 5.5781569480896,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 2270
+    },
+    {
+      "epoch": 7.3076923076923075,
+      "grad_norm": 5.950911521911621,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 2280
+    },
+    {
+      "epoch": 7.339743589743589,
+      "grad_norm": 6.755985736846924,
+      "learning_rate": 0.0001,
+      "loss": 0.0542,
+      "step": 2290
+    },
+    {
+      "epoch": 7.371794871794872,
+      "grad_norm": 7.540384769439697,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 2300
+    },
+    {
+      "epoch": 7.403846153846154,
+      "grad_norm": 8.423117637634277,
+      "learning_rate": 0.0001,
+      "loss": 0.0541,
+      "step": 2310
+    },
+    {
+      "epoch": 7.435897435897436,
+      "grad_norm": 7.655261516571045,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 2320
+    },
+    {
+      "epoch": 7.467948717948718,
+      "grad_norm": 6.582742691040039,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 2330
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 10.306877136230469,
+      "learning_rate": 0.0001,
+      "loss": 0.0597,
+      "step": 2340
+    },
+    {
+      "epoch": 7.532051282051282,
+      "grad_norm": 7.554524898529053,
+      "learning_rate": 0.0001,
+      "loss": 0.0572,
+      "step": 2350
+    },
+    {
+      "epoch": 7.564102564102564,
+      "grad_norm": 6.894124984741211,
+      "learning_rate": 0.0001,
+      "loss": 0.0544,
+      "step": 2360
+    },
+    {
+      "epoch": 7.596153846153846,
+      "grad_norm": 5.91453742980957,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "step": 2370
+    },
+    {
+      "epoch": 7.628205128205128,
+      "grad_norm": 7.348963737487793,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 2380
+    },
+    {
+      "epoch": 7.660256410256411,
+      "grad_norm": 6.308079719543457,
+      "learning_rate": 0.0001,
+      "loss": 0.0568,
+      "step": 2390
+    },
+    {
+      "epoch": 7.6923076923076925,
+      "grad_norm": 6.994787216186523,
+      "learning_rate": 0.0001,
+      "loss": 0.0557,
+      "step": 2400
+    },
+    {
+      "epoch": 7.7243589743589745,
+      "grad_norm": 6.730224609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0561,
+      "step": 2410
+    },
+    {
+      "epoch": 7.756410256410256,
+      "grad_norm": 7.677487373352051,
+      "learning_rate": 0.0001,
+      "loss": 0.0529,
+      "step": 2420
+    },
+    {
+      "epoch": 7.788461538461538,
+      "grad_norm": 6.131002902984619,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 2430
+    },
+    {
+      "epoch": 7.82051282051282,
+      "grad_norm": 5.967788219451904,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 2440
+    },
+    {
+      "epoch": 7.852564102564102,
+      "grad_norm": 9.216577529907227,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "step": 2450
+    },
+    {
+      "epoch": 7.884615384615385,
+      "grad_norm": 7.197536945343018,
+      "learning_rate": 0.0001,
+      "loss": 0.0511,
+      "step": 2460
+    },
+    {
+      "epoch": 7.916666666666667,
+      "grad_norm": 5.935634613037109,
+      "learning_rate": 0.0001,
+      "loss": 0.0478,
+      "step": 2470
+    },
+    {
+      "epoch": 7.948717948717949,
+      "grad_norm": 5.8652424812316895,
+      "learning_rate": 0.0001,
+      "loss": 0.0483,
+      "step": 2480
+    },
+    {
+      "epoch": 7.980769230769231,
+      "grad_norm": 6.469454288482666,
+      "learning_rate": 0.0001,
+      "loss": 0.0478,
+      "step": 2490
+    },
+    {
+      "epoch": 8.012820512820513,
+      "grad_norm": 5.61533260345459,
+      "learning_rate": 0.0001,
+      "loss": 0.0493,
+      "step": 2500
+    },
+    {
+      "epoch": 8.044871794871796,
+      "grad_norm": 7.056205749511719,
+      "learning_rate": 0.0001,
+      "loss": 0.0513,
+      "step": 2510
+    },
+    {
+      "epoch": 8.076923076923077,
+      "grad_norm": 6.137714385986328,
+      "learning_rate": 0.0001,
+      "loss": 0.0507,
+      "step": 2520
+    },
+    {
+      "epoch": 8.10897435897436,
+      "grad_norm": 5.819190502166748,
+      "learning_rate": 0.0001,
+      "loss": 0.0501,
+      "step": 2530
+    },
+    {
+      "epoch": 8.14102564102564,
+      "grad_norm": 5.246364116668701,
+      "learning_rate": 0.0001,
+      "loss": 0.0507,
+      "step": 2540
+    },
+    {
+      "epoch": 8.173076923076923,
+      "grad_norm": 6.629395008087158,
+      "learning_rate": 0.0001,
+      "loss": 0.0489,
+      "step": 2550
+    },
+    {
+      "epoch": 8.205128205128204,
+      "grad_norm": 6.244475364685059,
+      "learning_rate": 0.0001,
+      "loss": 0.0474,
+      "step": 2560
+    },
+    {
+      "epoch": 8.237179487179487,
+      "grad_norm": 5.888215065002441,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 2570
+    },
+    {
+      "epoch": 8.26923076923077,
+      "grad_norm": 6.043148994445801,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 2580
+    },
+    {
+      "epoch": 8.301282051282051,
+      "grad_norm": 7.372574329376221,
+      "learning_rate": 0.0001,
+      "loss": 0.0473,
+      "step": 2590
+    },
+    {
+      "epoch": 8.333333333333334,
+      "grad_norm": 6.209153652191162,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 2600
+    },
+    {
+      "epoch": 8.365384615384615,
+      "grad_norm": 5.507105350494385,
+      "learning_rate": 0.0001,
+      "loss": 0.0472,
+      "step": 2610
+    },
+    {
+      "epoch": 8.397435897435898,
+      "grad_norm": 5.01945686340332,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "step": 2620
+    },
+    {
+      "epoch": 8.429487179487179,
+      "grad_norm": 5.078667163848877,
+      "learning_rate": 0.0001,
+      "loss": 0.0472,
+      "step": 2630
+    },
+    {
+      "epoch": 8.461538461538462,
+      "grad_norm": 4.403445243835449,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 2640
+    },
+    {
+      "epoch": 8.493589743589745,
+      "grad_norm": 5.570326805114746,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 2650
+    },
+    {
+      "epoch": 8.525641025641026,
+      "grad_norm": 4.945068359375,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 2660
+    },
+    {
+      "epoch": 8.557692307692308,
+      "grad_norm": 5.301838397979736,
+      "learning_rate": 0.0001,
+      "loss": 0.0467,
+      "step": 2670
+    },
+    {
+      "epoch": 8.58974358974359,
+      "grad_norm": 6.728024482727051,
+      "learning_rate": 0.0001,
+      "loss": 0.0483,
+      "step": 2680
+    },
+    {
+      "epoch": 8.621794871794872,
+      "grad_norm": 8.002473831176758,
+      "learning_rate": 0.0001,
+      "loss": 0.0482,
+      "step": 2690
+    },
+    {
+      "epoch": 8.653846153846153,
+      "grad_norm": 6.8310866355896,
+      "learning_rate": 0.0001,
+      "loss": 0.0501,
+      "step": 2700
+    },
+    {
+      "epoch": 8.685897435897436,
+      "grad_norm": 6.069472789764404,
+      "learning_rate": 0.0001,
+      "loss": 0.0489,
+      "step": 2710
+    },
+    {
+      "epoch": 8.717948717948717,
+      "grad_norm": 5.730343818664551,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 2720
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 5.661256313323975,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 2730
+    },
+    {
+      "epoch": 8.782051282051283,
+      "grad_norm": 5.085428714752197,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 2740
+    },
+    {
+      "epoch": 8.814102564102564,
+      "grad_norm": 4.996331214904785,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 2750
+    },
+    {
+      "epoch": 8.846153846153847,
+      "grad_norm": 5.410341739654541,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 2760
+    },
+    {
+      "epoch": 8.878205128205128,
+      "grad_norm": 7.1424560546875,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 2770
+    },
+    {
+      "epoch": 8.91025641025641,
+      "grad_norm": 6.48975133895874,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 2780
+    },
+    {
+      "epoch": 8.942307692307692,
+      "grad_norm": 5.6859941482543945,
+      "learning_rate": 0.0001,
+      "loss": 0.0488,
+      "step": 2790
+    },
+    {
+      "epoch": 8.974358974358974,
+      "grad_norm": 5.954822063446045,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 2800
+    },
+    {
+      "epoch": 9.006410256410257,
+      "grad_norm": 5.6773529052734375,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 2810
+    },
+    {
+      "epoch": 9.038461538461538,
+      "grad_norm": 5.297764778137207,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 2820
+    },
+    {
+      "epoch": 9.070512820512821,
+      "grad_norm": 6.272767543792725,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 2830
+    },
+    {
+      "epoch": 9.102564102564102,
+      "grad_norm": 6.466505527496338,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 2840
+    },
+    {
+      "epoch": 9.134615384615385,
+      "grad_norm": 5.490339279174805,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 2850
+    },
+    {
+      "epoch": 9.166666666666666,
+      "grad_norm": 5.731348514556885,
+      "learning_rate": 0.0001,
+      "loss": 0.0466,
+      "step": 2860
+    },
+    {
+      "epoch": 9.198717948717949,
+      "grad_norm": 4.985128879547119,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 2870
+    },
+    {
+      "epoch": 9.23076923076923,
+      "grad_norm": 7.122063636779785,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 2880
+    },
+    {
+      "epoch": 9.262820512820513,
+      "grad_norm": 7.768131256103516,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 2890
+    },
+    {
+      "epoch": 9.294871794871796,
+      "grad_norm": 5.471693992614746,
+      "learning_rate": 0.0001,
+      "loss": 0.042,
+      "step": 2900
+    },
+    {
+      "epoch": 9.326923076923077,
+      "grad_norm": 5.0532684326171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0424,
+      "step": 2910
+    },
+    {
+      "epoch": 9.35897435897436,
+      "grad_norm": 5.2089948654174805,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 2920
+    },
+    {
+      "epoch": 9.39102564102564,
+      "grad_norm": 5.9514031410217285,
+      "learning_rate": 0.0001,
+      "loss": 0.0412,
+      "step": 2930
+    },
+    {
+      "epoch": 9.423076923076923,
+      "grad_norm": 5.216250419616699,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 2940
+    },
+    {
+      "epoch": 9.455128205128204,
+      "grad_norm": 4.857729911804199,
+      "learning_rate": 0.0001,
+      "loss": 0.0424,
+      "step": 2950
+    },
+    {
+      "epoch": 9.487179487179487,
+      "grad_norm": 7.547190189361572,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 2960
+    },
+    {
+      "epoch": 9.51923076923077,
+      "grad_norm": 5.238833427429199,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 2970
+    },
+    {
+      "epoch": 9.551282051282051,
+      "grad_norm": 5.245550632476807,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 2980
+    },
+    {
+      "epoch": 9.583333333333334,
+      "grad_norm": 4.865394115447998,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "step": 2990
+    },
+    {
+      "epoch": 9.615384615384615,
+      "grad_norm": 6.217046737670898,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 3000
+    },
+    {
+      "epoch": 9.647435897435898,
+      "grad_norm": 5.559901714324951,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 3010
+    },
+    {
+      "epoch": 9.679487179487179,
+      "grad_norm": 5.55620813369751,
+      "learning_rate": 0.0001,
+      "loss": 0.0444,
+      "step": 3020
+    },
+    {
+      "epoch": 9.711538461538462,
+      "grad_norm": 4.596827030181885,
+      "learning_rate": 0.0001,
+      "loss": 0.0422,
+      "step": 3030
+    },
+    {
+      "epoch": 9.743589743589745,
+      "grad_norm": 5.213090896606445,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 3040
+    },
+    {
+      "epoch": 9.775641025641026,
+      "grad_norm": 4.659507751464844,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 3050
+    },
+    {
+      "epoch": 9.807692307692308,
+      "grad_norm": 4.362442493438721,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 3060
+    },
+    {
+      "epoch": 9.83974358974359,
+      "grad_norm": 4.87863302230835,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 3070
+    },
+    {
+      "epoch": 9.871794871794872,
+      "grad_norm": 5.917473316192627,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 3080
+    },
+    {
+      "epoch": 9.903846153846153,
+      "grad_norm": 3.8518383502960205,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 3090
+    },
+    {
+      "epoch": 9.935897435897436,
+      "grad_norm": 3.961597442626953,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 3100
+    },
+    {
+      "epoch": 9.967948717948717,
+      "grad_norm": 4.7606072425842285,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 3110
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 5.570981025695801,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 3120
+    },
+    {
+      "epoch": 10.032051282051283,
+      "grad_norm": 5.384578227996826,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 3130
+    },
+    {
+      "epoch": 10.064102564102564,
+      "grad_norm": 5.301525592803955,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 3140
+    },
+    {
+      "epoch": 10.096153846153847,
+      "grad_norm": 4.737154960632324,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 3150
+    },
+    {
+      "epoch": 10.128205128205128,
+      "grad_norm": 5.188713550567627,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 3160
+    },
+    {
+      "epoch": 10.16025641025641,
+      "grad_norm": 4.968835830688477,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 3170
+    },
+    {
+      "epoch": 10.192307692307692,
+      "grad_norm": 5.069928169250488,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 3180
+    },
+    {
+      "epoch": 10.224358974358974,
+      "grad_norm": 4.344878673553467,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 3190
+    },
+    {
+      "epoch": 10.256410256410255,
+      "grad_norm": 4.91372013092041,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 3200
+    },
+    {
+      "epoch": 10.288461538461538,
+      "grad_norm": 4.5926384925842285,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 3210
+    },
+    {
+      "epoch": 10.320512820512821,
+      "grad_norm": 5.0196452140808105,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 3220
+    },
+    {
+      "epoch": 10.352564102564102,
+      "grad_norm": 5.456750869750977,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 3230
+    },
+    {
+      "epoch": 10.384615384615385,
+      "grad_norm": 4.43285608291626,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 3240
+    },
+    {
+      "epoch": 10.416666666666666,
+      "grad_norm": 4.3033294677734375,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 3250
+    },
+    {
+      "epoch": 10.448717948717949,
+      "grad_norm": 5.545897483825684,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 3260
+    },
+    {
+      "epoch": 10.48076923076923,
+      "grad_norm": 4.494649410247803,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 3270
+    },
+    {
+      "epoch": 10.512820512820513,
+      "grad_norm": 4.719374656677246,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 3280
+    },
+    {
+      "epoch": 10.544871794871796,
+      "grad_norm": 4.451051712036133,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 3290
+    },
+    {
+      "epoch": 10.576923076923077,
+      "grad_norm": 3.9088149070739746,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 3300
+    },
+    {
+      "epoch": 10.60897435897436,
+      "grad_norm": 4.624056816101074,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 3310
+    },
+    {
+      "epoch": 10.64102564102564,
+      "grad_norm": 5.079074382781982,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 3320
+    },
+    {
+      "epoch": 10.673076923076923,
+      "grad_norm": 4.921101093292236,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 3330
+    },
+    {
+      "epoch": 10.705128205128204,
+      "grad_norm": 5.064759254455566,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 3340
+    },
+    {
+      "epoch": 10.737179487179487,
+      "grad_norm": 5.043102264404297,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 3350
+    },
+    {
+      "epoch": 10.76923076923077,
+      "grad_norm": 4.193841934204102,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 3360
+    },
+    {
+      "epoch": 10.801282051282051,
+      "grad_norm": 3.7491838932037354,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 3370
+    },
+    {
+      "epoch": 10.833333333333334,
+      "grad_norm": 4.258098602294922,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 3380
+    },
+    {
+      "epoch": 10.865384615384615,
+      "grad_norm": 4.254199504852295,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 3390
+    },
+    {
+      "epoch": 10.897435897435898,
+      "grad_norm": 4.406967639923096,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 3400
+    },
+    {
+      "epoch": 10.929487179487179,
+      "grad_norm": 4.336716651916504,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 3410
+    },
+    {
+      "epoch": 10.961538461538462,
+      "grad_norm": 4.996073246002197,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 3420
+    },
+    {
+      "epoch": 10.993589743589745,
+      "grad_norm": 4.6920013427734375,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 3430
+    },
+    {
+      "epoch": 11.025641025641026,
+      "grad_norm": 4.517393589019775,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 3440
+    },
+    {
+      "epoch": 11.057692307692308,
+      "grad_norm": 3.6874966621398926,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 3450
+    },
+    {
+      "epoch": 11.08974358974359,
+      "grad_norm": 4.2535552978515625,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 3460
+    },
+    {
+      "epoch": 11.121794871794872,
+      "grad_norm": 4.914954662322998,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 3470
+    },
+    {
+      "epoch": 11.153846153846153,
+      "grad_norm": 4.995485305786133,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 3480
+    },
+    {
+      "epoch": 11.185897435897436,
+      "grad_norm": 4.81497049331665,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 3490
+    },
+    {
+      "epoch": 11.217948717948717,
+      "grad_norm": 4.135605335235596,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 3500
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 3.687518835067749,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 3510
+    },
+    {
+      "epoch": 11.282051282051283,
+      "grad_norm": 4.5489630699157715,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 3520
+    },
+    {
+      "epoch": 11.314102564102564,
+      "grad_norm": 3.8071272373199463,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 3530
+    },
+    {
+      "epoch": 11.346153846153847,
+      "grad_norm": 4.854300022125244,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 3540
+    },
+    {
+      "epoch": 11.378205128205128,
+      "grad_norm": 4.61160135269165,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 3550
+    },
+    {
+      "epoch": 11.41025641025641,
+      "grad_norm": 4.875280380249023,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 3560
+    },
+    {
+      "epoch": 11.442307692307692,
+      "grad_norm": 4.7223286628723145,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 3570
+    },
+    {
+      "epoch": 11.474358974358974,
+      "grad_norm": 4.143892765045166,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 3580
+    },
+    {
+      "epoch": 11.506410256410255,
+      "grad_norm": 5.398232936859131,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 3590
+    },
+    {
+      "epoch": 11.538461538461538,
+      "grad_norm": 4.407661437988281,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 3600
+    },
+    {
+      "epoch": 11.570512820512821,
+      "grad_norm": 3.741880416870117,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 3610
+    },
+    {
+      "epoch": 11.602564102564102,
+      "grad_norm": 4.240623950958252,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 3620
+    },
+    {
+      "epoch": 11.634615384615385,
+      "grad_norm": 3.367664098739624,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 3630
+    },
+    {
+      "epoch": 11.666666666666666,
+      "grad_norm": 4.1250457763671875,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 3640
+    },
+    {
+      "epoch": 11.698717948717949,
+      "grad_norm": 4.0287604331970215,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 3650
+    },
+    {
+      "epoch": 11.73076923076923,
+      "grad_norm": 4.443467140197754,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 3660
+    },
+    {
+      "epoch": 11.762820512820513,
+      "grad_norm": 3.6539993286132812,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 3670
+    },
+    {
+      "epoch": 11.794871794871796,
+      "grad_norm": 4.227948188781738,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 3680
+    },
+    {
+      "epoch": 11.826923076923077,
+      "grad_norm": 3.68633770942688,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 3690
+    },
+    {
+      "epoch": 11.85897435897436,
+      "grad_norm": 3.656583547592163,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 3700
+    },
+    {
+      "epoch": 11.89102564102564,
+      "grad_norm": 4.235436916351318,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 3710
+    },
+    {
+      "epoch": 11.923076923076923,
+      "grad_norm": 4.543222427368164,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 3720
+    },
+    {
+      "epoch": 11.955128205128204,
+      "grad_norm": 4.055953025817871,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 3730
+    },
+    {
+      "epoch": 11.987179487179487,
+      "grad_norm": 4.113011360168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 3740
+    },
+    {
+      "epoch": 12.01923076923077,
+      "grad_norm": 4.730022430419922,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 3750
+    },
+    {
+      "epoch": 12.051282051282051,
+      "grad_norm": 4.484661102294922,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 3760
+    },
+    {
+      "epoch": 12.083333333333334,
+      "grad_norm": 3.8935561180114746,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 3770
+    },
+    {
+      "epoch": 12.115384615384615,
+      "grad_norm": 3.980353593826294,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 3780
+    },
+    {
+      "epoch": 12.147435897435898,
+      "grad_norm": 3.7835426330566406,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 3790
+    },
+    {
+      "epoch": 12.179487179487179,
+      "grad_norm": 4.0831379890441895,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 3800
+    },
+    {
+      "epoch": 12.211538461538462,
+      "grad_norm": 4.162652015686035,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 3810
+    },
+    {
+      "epoch": 12.243589743589743,
+      "grad_norm": 4.762477397918701,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 3820
+    },
+    {
+      "epoch": 12.275641025641026,
+      "grad_norm": 4.4887003898620605,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 3830
+    },
+    {
+      "epoch": 12.307692307692308,
+      "grad_norm": 3.653463840484619,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 3840
+    },
+    {
+      "epoch": 12.33974358974359,
+      "grad_norm": 4.109398365020752,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 3850
+    },
+    {
+      "epoch": 12.371794871794872,
+      "grad_norm": 4.510799407958984,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 3860
+    },
+    {
+      "epoch": 12.403846153846153,
+      "grad_norm": 3.45520281791687,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 3870
+    },
+    {
+      "epoch": 12.435897435897436,
+      "grad_norm": 3.957491874694824,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 3880
+    },
+    {
+      "epoch": 12.467948717948717,
+      "grad_norm": 4.278698444366455,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 3890
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 4.236378192901611,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 3900
+    },
+    {
+      "epoch": 12.532051282051283,
+      "grad_norm": 4.181147575378418,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 3910
+    },
+    {
+      "epoch": 12.564102564102564,
+      "grad_norm": 3.346686601638794,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 3920
+    },
+    {
+      "epoch": 12.596153846153847,
+      "grad_norm": 4.464163303375244,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 3930
+    },
+    {
+      "epoch": 12.628205128205128,
+      "grad_norm": 4.485145092010498,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 3940
+    },
+    {
+      "epoch": 12.66025641025641,
+      "grad_norm": 4.550953388214111,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 3950
+    },
+    {
+      "epoch": 12.692307692307692,
+      "grad_norm": 3.587400436401367,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 3960
+    },
+    {
+      "epoch": 12.724358974358974,
+      "grad_norm": 3.720364570617676,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 3970
+    },
+    {
+      "epoch": 12.756410256410255,
+      "grad_norm": 3.298011302947998,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 3980
+    },
+    {
+      "epoch": 12.788461538461538,
+      "grad_norm": 3.621622323989868,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 3990
+    },
+    {
+      "epoch": 12.820512820512821,
+      "grad_norm": 3.8080344200134277,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 4000
+    },
+    {
+      "epoch": 12.852564102564102,
+      "grad_norm": 4.31474494934082,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 4010
+    },
+    {
+      "epoch": 12.884615384615385,
+      "grad_norm": 3.8488495349884033,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 4020
+    },
+    {
+      "epoch": 12.916666666666666,
+      "grad_norm": 3.866185188293457,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 4030
+    },
+    {
+      "epoch": 12.948717948717949,
+      "grad_norm": 3.8579750061035156,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 4040
+    },
+    {
+      "epoch": 12.98076923076923,
+      "grad_norm": 2.9426093101501465,
+      "learning_rate": 0.0001,
+      "loss": 0.0296,
+      "step": 4050
+    },
+    {
+      "epoch": 13.012820512820513,
+      "grad_norm": 3.6344223022460938,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 4060
+    },
+    {
+      "epoch": 13.044871794871796,
+      "grad_norm": 3.9723308086395264,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 4070
+    },
+    {
+      "epoch": 13.076923076923077,
+      "grad_norm": 2.9895920753479004,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 4080
+    },
+    {
+      "epoch": 13.10897435897436,
+      "grad_norm": 4.082196235656738,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 4090
+    },
+    {
+      "epoch": 13.14102564102564,
+      "grad_norm": 3.8368873596191406,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 4100
+    },
+    {
+      "epoch": 13.173076923076923,
+      "grad_norm": 3.2625792026519775,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 4110
+    },
+    {
+      "epoch": 13.205128205128204,
+      "grad_norm": 3.4162421226501465,
+      "learning_rate": 0.0001,
+      "loss": 0.0308,
+      "step": 4120
+    },
+    {
+      "epoch": 13.237179487179487,
+      "grad_norm": 3.7453558444976807,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 4130
+    },
+    {
+      "epoch": 13.26923076923077,
+      "grad_norm": 3.457040548324585,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 4140
+    },
+    {
+      "epoch": 13.301282051282051,
+      "grad_norm": 3.1488914489746094,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 4150
+    },
+    {
+      "epoch": 13.333333333333334,
+      "grad_norm": 3.7592713832855225,
+      "learning_rate": 0.0001,
+      "loss": 0.0304,
+      "step": 4160
+    },
+    {
+      "epoch": 13.365384615384615,
+      "grad_norm": 4.262688636779785,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 4170
+    },
+    {
+      "epoch": 13.397435897435898,
+      "grad_norm": 3.418375015258789,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 4180
+    },
+    {
+      "epoch": 13.429487179487179,
+      "grad_norm": 3.6605513095855713,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 4190
+    },
+    {
+      "epoch": 13.461538461538462,
+      "grad_norm": 3.208761692047119,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 4200
+    },
+    {
+      "epoch": 13.493589743589745,
+      "grad_norm": 4.166501998901367,
+      "learning_rate": 0.0001,
+      "loss": 0.0299,
+      "step": 4210
+    },
+    {
+      "epoch": 13.525641025641026,
+      "grad_norm": 3.794900417327881,
+      "learning_rate": 0.0001,
+      "loss": 0.029,
+      "step": 4220
+    },
+    {
+      "epoch": 13.557692307692308,
+      "grad_norm": 3.735645294189453,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 4230
+    },
+    {
+      "epoch": 13.58974358974359,
+      "grad_norm": 3.2368414402008057,
+      "learning_rate": 0.0001,
+      "loss": 0.0275,
+      "step": 4240
+    },
+    {
+      "epoch": 13.621794871794872,
+      "grad_norm": 3.424647808074951,
+      "learning_rate": 0.0001,
+      "loss": 0.0291,
+      "step": 4250
+    },
+    {
+      "epoch": 13.653846153846153,
+      "grad_norm": 3.579198122024536,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 4260
+    },
+    {
+      "epoch": 13.685897435897436,
+      "grad_norm": 4.197644233703613,
+      "learning_rate": 0.0001,
+      "loss": 0.029,
+      "step": 4270
+    },
+    {
+      "epoch": 13.717948717948717,
+      "grad_norm": 3.4035584926605225,
+      "learning_rate": 0.0001,
+      "loss": 0.0294,
+      "step": 4280
+    },
+    {
+      "epoch": 13.75,
+      "grad_norm": 3.8743956089019775,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 4290
+    },
+    {
+      "epoch": 13.782051282051283,
+      "grad_norm": 3.4530327320098877,
+      "learning_rate": 0.0001,
+      "loss": 0.0294,
+      "step": 4300
+    },
+    {
+      "epoch": 13.814102564102564,
+      "grad_norm": 2.9512808322906494,
+      "learning_rate": 0.0001,
+      "loss": 0.0281,
+      "step": 4310
+    },
+    {
+      "epoch": 13.846153846153847,
+      "grad_norm": 3.4321718215942383,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 4320
+    },
+    {
+      "epoch": 13.878205128205128,
+      "grad_norm": 3.928579807281494,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 4330
+    },
+    {
+      "epoch": 13.91025641025641,
+      "grad_norm": 3.5125350952148438,
+      "learning_rate": 0.0001,
+      "loss": 0.0287,
+      "step": 4340
+    },
+    {
+      "epoch": 13.942307692307692,
+      "grad_norm": 3.907015323638916,
+      "learning_rate": 0.0001,
+      "loss": 0.0304,
+      "step": 4350
+    },
+    {
+      "epoch": 13.974358974358974,
+      "grad_norm": 4.098099231719971,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 4360
+    },
+    {
+      "epoch": 14.006410256410257,
+      "grad_norm": 4.517259120941162,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 4370
+    },
+    {
+      "epoch": 14.038461538461538,
+      "grad_norm": 3.4022507667541504,
+      "learning_rate": 0.0001,
+      "loss": 0.0293,
+      "step": 4380
+    },
+    {
+      "epoch": 14.070512820512821,
+      "grad_norm": 3.059760093688965,
+      "learning_rate": 0.0001,
+      "loss": 0.0304,
+      "step": 4390
+    },
+    {
+      "epoch": 14.102564102564102,
+      "grad_norm": 3.1346914768218994,
+      "learning_rate": 0.0001,
+      "loss": 0.0288,
+      "step": 4400
+    },
+    {
+      "epoch": 14.134615384615385,
+      "grad_norm": 3.325901985168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 4410
+    },
+    {
+      "epoch": 14.166666666666666,
+      "grad_norm": 3.421664237976074,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 4420
+    },
+    {
+      "epoch": 14.198717948717949,
+      "grad_norm": 4.67044734954834,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 4430
+    },
+    {
+      "epoch": 14.23076923076923,
+      "grad_norm": 2.964694023132324,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 4440
+    },
+    {
+      "epoch": 14.262820512820513,
+      "grad_norm": 3.6508755683898926,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 4450
+    },
+    {
+      "epoch": 14.294871794871796,
+      "grad_norm": 2.9293110370635986,
+      "learning_rate": 0.0001,
+      "loss": 0.0306,
+      "step": 4460
+    },
+    {
+      "epoch": 14.326923076923077,
+      "grad_norm": 3.4285552501678467,
+      "learning_rate": 0.0001,
+      "loss": 0.0297,
+      "step": 4470
+    },
+    {
+      "epoch": 14.35897435897436,
+      "grad_norm": 3.3838372230529785,
+      "learning_rate": 0.0001,
+      "loss": 0.0289,
+      "step": 4480
+    },
+    {
+      "epoch": 14.39102564102564,
+      "grad_norm": 3.4895076751708984,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 4490
+    },
+    {
+      "epoch": 14.423076923076923,
+      "grad_norm": 3.67922306060791,
+      "learning_rate": 0.0001,
+      "loss": 0.0287,
+      "step": 4500
+    },
+    {
+      "epoch": 14.455128205128204,
+      "grad_norm": 3.141451835632324,
+      "learning_rate": 0.0001,
+      "loss": 0.0302,
+      "step": 4510
+    },
+    {
+      "epoch": 14.487179487179487,
+      "grad_norm": 3.8157565593719482,
+      "learning_rate": 0.0001,
+      "loss": 0.0308,
+      "step": 4520
+    },
+    {
+      "epoch": 14.51923076923077,
+      "grad_norm": 3.1633265018463135,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 4530
+    },
+    {
+      "epoch": 14.551282051282051,
+      "grad_norm": 2.991029739379883,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 4540
+    },
+    {
+      "epoch": 14.583333333333334,
+      "grad_norm": 3.257967710494995,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 4550
+    },
+    {
+      "epoch": 14.615384615384615,
+      "grad_norm": 3.183598041534424,
+      "learning_rate": 0.0001,
+      "loss": 0.0294,
+      "step": 4560
+    },
+    {
+      "epoch": 14.647435897435898,
+      "grad_norm": 3.12715744972229,
+      "learning_rate": 0.0001,
+      "loss": 0.0273,
+      "step": 4570
+    },
+    {
+      "epoch": 14.679487179487179,
+      "grad_norm": 3.050215482711792,
+      "learning_rate": 0.0001,
+      "loss": 0.0281,
+      "step": 4580
+    },
+    {
+      "epoch": 14.711538461538462,
+      "grad_norm": 2.788644790649414,
+      "learning_rate": 0.0001,
+      "loss": 0.0291,
+      "step": 4590
+    },
+    {
+      "epoch": 14.743589743589745,
+      "grad_norm": 2.7550079822540283,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 4600
+    },
+    {
+      "epoch": 14.775641025641026,
+      "grad_norm": 3.036320686340332,
+      "learning_rate": 0.0001,
+      "loss": 0.0283,
+      "step": 4610
+    },
+    {
+      "epoch": 14.807692307692308,
+      "grad_norm": 3.14253830909729,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 4620
+    },
+    {
+      "epoch": 14.83974358974359,
+      "grad_norm": 3.3128609657287598,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 4630
+    },
+    {
+      "epoch": 14.871794871794872,
+      "grad_norm": 3.703613519668579,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "step": 4640
+    },
+    {
+      "epoch": 14.903846153846153,
+      "grad_norm": 3.200819492340088,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 4650
+    },
+    {
+      "epoch": 14.935897435897436,
+      "grad_norm": 3.1905035972595215,
+      "learning_rate": 0.0001,
+      "loss": 0.0279,
+      "step": 4660
+    },
+    {
+      "epoch": 14.967948717948717,
+      "grad_norm": 3.5679359436035156,
+      "learning_rate": 0.0001,
+      "loss": 0.0285,
+      "step": 4670
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 3.186164379119873,
+      "learning_rate": 0.0001,
+      "loss": 0.0275,
+      "step": 4680
+    },
+    {
+      "epoch": 15.032051282051283,
+      "grad_norm": 3.4353549480438232,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "step": 4690
+    },
+    {
+      "epoch": 15.064102564102564,
+      "grad_norm": 4.012332916259766,
+      "learning_rate": 0.0001,
+      "loss": 0.027,
+      "step": 4700
+    },
+    {
+      "epoch": 15.096153846153847,
+      "grad_norm": 3.876591205596924,
+      "learning_rate": 0.0001,
+      "loss": 0.0277,
+      "step": 4710
+    },
+    {
+      "epoch": 15.128205128205128,
+      "grad_norm": 3.5996615886688232,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 4720
+    },
+    {
+      "epoch": 15.16025641025641,
+      "grad_norm": 3.262385845184326,
+      "learning_rate": 0.0001,
+      "loss": 0.0289,
+      "step": 4730
+    },
+    {
+      "epoch": 15.192307692307692,
+      "grad_norm": 3.4985899925231934,
+      "learning_rate": 0.0001,
+      "loss": 0.0272,
+      "step": 4740
+    },
+    {
+      "epoch": 15.224358974358974,
+      "grad_norm": 3.459765672683716,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 4750
+    },
+    {
+      "epoch": 15.256410256410255,
+      "grad_norm": 3.9838860034942627,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 4760
+    },
+    {
+      "epoch": 15.288461538461538,
+      "grad_norm": 3.4298784732818604,
+      "learning_rate": 0.0001,
+      "loss": 0.0291,
+      "step": 4770
+    },
+    {
+      "epoch": 15.320512820512821,
+      "grad_norm": 3.4209749698638916,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 4780
+    },
+    {
+      "epoch": 15.352564102564102,
+      "grad_norm": 3.0022754669189453,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 4790
+    },
+    {
+      "epoch": 15.384615384615385,
+      "grad_norm": 2.818871259689331,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 4800
+    },
+    {
+      "epoch": 15.416666666666666,
+      "grad_norm": 2.450993537902832,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 4810
+    },
+    {
+      "epoch": 15.448717948717949,
+      "grad_norm": 3.0150392055511475,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 4820
+    },
+    {
+      "epoch": 15.48076923076923,
+      "grad_norm": 2.8579487800598145,
+      "learning_rate": 0.0001,
+      "loss": 0.0303,
+      "step": 4830
+    },
+    {
+      "epoch": 15.512820512820513,
+      "grad_norm": 2.9444453716278076,
+      "learning_rate": 0.0001,
+      "loss": 0.0287,
+      "step": 4840
+    },
+    {
+      "epoch": 15.544871794871796,
+      "grad_norm": 3.794285535812378,
+      "learning_rate": 0.0001,
+      "loss": 0.0285,
+      "step": 4850
+    },
+    {
+      "epoch": 15.576923076923077,
+      "grad_norm": 2.889913320541382,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 4860
+    },
+    {
+      "epoch": 15.60897435897436,
+      "grad_norm": 2.4692916870117188,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 4870
+    },
+    {
+      "epoch": 15.64102564102564,
+      "grad_norm": 3.302612781524658,
+      "learning_rate": 0.0001,
+      "loss": 0.0275,
+      "step": 4880
+    },
+    {
+      "epoch": 15.673076923076923,
+      "grad_norm": 2.7743074893951416,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 4890
+    },
+    {
+      "epoch": 15.705128205128204,
+      "grad_norm": 2.996398448944092,
+      "learning_rate": 0.0001,
+      "loss": 0.0273,
+      "step": 4900
+    },
+    {
+      "epoch": 15.737179487179487,
+      "grad_norm": 3.6971888542175293,
+      "learning_rate": 0.0001,
+      "loss": 0.0267,
+      "step": 4910
+    },
+    {
+      "epoch": 15.76923076923077,
+      "grad_norm": 2.7379348278045654,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 4920
+    },
+    {
+      "epoch": 15.801282051282051,
+      "grad_norm": 3.1581263542175293,
+      "learning_rate": 0.0001,
+      "loss": 0.0267,
+      "step": 4930
+    },
+    {
+      "epoch": 15.833333333333334,
+      "grad_norm": 2.776794910430908,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 4940
+    },
+    {
+      "epoch": 15.865384615384615,
+      "grad_norm": 3.4279465675354004,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 4950
+    },
+    {
+      "epoch": 15.897435897435898,
+      "grad_norm": 2.716580867767334,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 4960
+    },
+    {
+      "epoch": 15.929487179487179,
+      "grad_norm": 2.619068145751953,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 4970
+    },
+    {
+      "epoch": 15.961538461538462,
+      "grad_norm": 3.163851022720337,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 4980
+    },
+    {
+      "epoch": 15.993589743589745,
+      "grad_norm": 3.260380744934082,
+      "learning_rate": 0.0001,
+      "loss": 0.0271,
+      "step": 4990
+    },
+    {
+      "epoch": 16.025641025641026,
+      "grad_norm": 2.7413055896759033,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 5000
+    },
+    {
+      "epoch": 16.057692307692307,
+      "grad_norm": 2.8356895446777344,
+      "learning_rate": 0.0001,
+      "loss": 0.0264,
+      "step": 5010
+    },
+    {
+      "epoch": 16.08974358974359,
+      "grad_norm": 2.7685391902923584,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 5020
+    },
+    {
+      "epoch": 16.121794871794872,
+      "grad_norm": 2.8314268589019775,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 5030
+    },
+    {
+      "epoch": 16.153846153846153,
+      "grad_norm": 2.75520658493042,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 5040
+    },
+    {
+      "epoch": 16.185897435897434,
+      "grad_norm": 2.4560556411743164,
+      "learning_rate": 0.0001,
+      "loss": 0.0269,
+      "step": 5050
+    },
+    {
+      "epoch": 16.21794871794872,
+      "grad_norm": 3.321467399597168,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 5060
+    },
+    {
+      "epoch": 16.25,
+      "grad_norm": 2.636418104171753,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 5070
+    },
+    {
+      "epoch": 16.28205128205128,
+      "grad_norm": 2.7601397037506104,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "step": 5080
+    },
+    {
+      "epoch": 16.314102564102566,
+      "grad_norm": 2.817948579788208,
+      "learning_rate": 0.0001,
+      "loss": 0.0273,
+      "step": 5090
+    },
+    {
+      "epoch": 16.346153846153847,
+      "grad_norm": 2.772094249725342,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 5100
+    },
+    {
+      "epoch": 16.378205128205128,
+      "grad_norm": 3.1215574741363525,
+      "learning_rate": 0.0001,
+      "loss": 0.026,
+      "step": 5110
+    },
+    {
+      "epoch": 16.41025641025641,
+      "grad_norm": 3.4062461853027344,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 5120
+    },
+    {
+      "epoch": 16.442307692307693,
+      "grad_norm": 3.18473219871521,
+      "learning_rate": 0.0001,
+      "loss": 0.0276,
+      "step": 5130
+    },
+    {
+      "epoch": 16.474358974358974,
+      "grad_norm": 3.0861685276031494,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 5140
+    },
+    {
+      "epoch": 16.506410256410255,
+      "grad_norm": 2.9566004276275635,
+      "learning_rate": 0.0001,
+      "loss": 0.0266,
+      "step": 5150
+    },
+    {
+      "epoch": 16.53846153846154,
+      "grad_norm": 3.822904348373413,
+      "learning_rate": 0.0001,
+      "loss": 0.0264,
+      "step": 5160
+    },
+    {
+      "epoch": 16.57051282051282,
+      "grad_norm": 3.1629083156585693,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 5170
+    },
+    {
+      "epoch": 16.602564102564102,
+      "grad_norm": 3.647480010986328,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 5180
+    },
+    {
+      "epoch": 16.634615384615383,
+      "grad_norm": 2.755328893661499,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 5190
+    },
+    {
+      "epoch": 16.666666666666668,
+      "grad_norm": 3.1172730922698975,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 5200
+    },
+    {
+      "epoch": 16.69871794871795,
+      "grad_norm": 3.389695644378662,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 5210
+    },
+    {
+      "epoch": 16.73076923076923,
+      "grad_norm": 2.943272113800049,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 5220
+    },
+    {
+      "epoch": 16.76282051282051,
+      "grad_norm": 3.0306015014648438,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 5230
+    },
+    {
+      "epoch": 16.794871794871796,
+      "grad_norm": 2.903876304626465,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 5240
+    },
+    {
+      "epoch": 16.826923076923077,
+      "grad_norm": 2.816786050796509,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 5250
+    },
+    {
+      "epoch": 16.858974358974358,
+      "grad_norm": 2.6454901695251465,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 5260
+    },
+    {
+      "epoch": 16.891025641025642,
+      "grad_norm": 3.7566497325897217,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 5270
+    },
+    {
+      "epoch": 16.923076923076923,
+      "grad_norm": 2.796703815460205,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 5280
+    },
+    {
+      "epoch": 16.955128205128204,
+      "grad_norm": 2.937497615814209,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 5290
+    },
+    {
+      "epoch": 16.98717948717949,
+      "grad_norm": 2.9849588871002197,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 5300
+    },
+    {
+      "epoch": 17.01923076923077,
+      "grad_norm": 2.407482147216797,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 5310
+    },
+    {
+      "epoch": 17.05128205128205,
+      "grad_norm": 3.0295231342315674,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 5320
+    },
+    {
+      "epoch": 17.083333333333332,
+      "grad_norm": 2.06109881401062,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 5330
+    },
+    {
+      "epoch": 17.115384615384617,
+      "grad_norm": 2.759565591812134,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 5340
+    },
+    {
+      "epoch": 17.147435897435898,
+      "grad_norm": 2.688222646713257,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 5350
+    },
+    {
+      "epoch": 17.17948717948718,
+      "grad_norm": 3.021909236907959,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 5360
+    },
+    {
+      "epoch": 17.21153846153846,
+      "grad_norm": 3.3294031620025635,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 5370
+    },
+    {
+      "epoch": 17.243589743589745,
+      "grad_norm": 3.160064458847046,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 5380
+    },
+    {
+      "epoch": 17.275641025641026,
+      "grad_norm": 2.675417423248291,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 5390
+    },
+    {
+      "epoch": 17.307692307692307,
+      "grad_norm": 2.710097551345825,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 5400
+    },
+    {
+      "epoch": 17.33974358974359,
+      "grad_norm": 2.692667007446289,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 5410
+    },
+    {
+      "epoch": 17.371794871794872,
+      "grad_norm": 2.859224557876587,
+      "learning_rate": 0.0001,
+      "loss": 0.0268,
+      "step": 5420
+    },
+    {
+      "epoch": 17.403846153846153,
+      "grad_norm": 2.5094034671783447,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 5430
+    },
+    {
+      "epoch": 17.435897435897434,
+      "grad_norm": 2.7888588905334473,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 5440
+    },
+    {
+      "epoch": 17.46794871794872,
+      "grad_norm": 2.8299038410186768,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 5450
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 2.4854736328125,
+      "learning_rate": 0.0001,
+      "loss": 0.0267,
+      "step": 5460
+    },
+    {
+      "epoch": 17.53205128205128,
+      "grad_norm": 3.0461745262145996,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 5470
+    },
+    {
+      "epoch": 17.564102564102566,
+      "grad_norm": 2.777853012084961,
+      "learning_rate": 0.0001,
+      "loss": 0.0284,
+      "step": 5480
+    },
+    {
+      "epoch": 17.596153846153847,
+      "grad_norm": 3.1325180530548096,
+      "learning_rate": 0.0001,
+      "loss": 0.0263,
+      "step": 5490
+    },
+    {
+      "epoch": 17.628205128205128,
+      "grad_norm": 2.5857458114624023,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 5500
+    },
+    {
+      "epoch": 17.66025641025641,
+      "grad_norm": 2.790210485458374,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 5510
+    },
+    {
+      "epoch": 17.692307692307693,
+      "grad_norm": 2.7265243530273438,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 5520
+    },
+    {
+      "epoch": 17.724358974358974,
+      "grad_norm": 2.411297559738159,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 5530
+    },
+    {
+      "epoch": 17.756410256410255,
+      "grad_norm": 2.676563262939453,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 5540
+    },
+    {
+      "epoch": 17.78846153846154,
+      "grad_norm": 2.641489267349243,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 5550
+    },
+    {
+      "epoch": 17.82051282051282,
+      "grad_norm": 2.326695203781128,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 5560
+    },
+    {
+      "epoch": 17.852564102564102,
+      "grad_norm": 2.580791711807251,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 5570
+    },
+    {
+      "epoch": 17.884615384615383,
+      "grad_norm": 2.747269630432129,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 5580
+    },
+    {
+      "epoch": 17.916666666666668,
+      "grad_norm": 2.7453222274780273,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 5590
+    },
+    {
+      "epoch": 17.94871794871795,
+      "grad_norm": 3.111733913421631,
+      "learning_rate": 0.0001,
+      "loss": 0.0261,
+      "step": 5600
+    },
+    {
+      "epoch": 17.98076923076923,
+      "grad_norm": 2.4817349910736084,
+      "learning_rate": 0.0001,
+      "loss": 0.0246,
+      "step": 5610
+    },
+    {
+      "epoch": 18.012820512820515,
+      "grad_norm": 2.717355251312256,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 5620
+    },
+    {
+      "epoch": 18.044871794871796,
+      "grad_norm": 2.6020970344543457,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 5630
+    },
+    {
+      "epoch": 18.076923076923077,
+      "grad_norm": 2.3232879638671875,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 5640
+    },
+    {
+      "epoch": 18.108974358974358,
+      "grad_norm": 3.297537326812744,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 5650
+    },
+    {
+      "epoch": 18.141025641025642,
+      "grad_norm": 2.599656343460083,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 5660
+    },
+    {
+      "epoch": 18.173076923076923,
+      "grad_norm": 2.395692825317383,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 5670
+    },
+    {
+      "epoch": 18.205128205128204,
+      "grad_norm": 3.4894423484802246,
+      "learning_rate": 0.0001,
+      "loss": 0.0254,
+      "step": 5680
+    },
+    {
+      "epoch": 18.237179487179485,
+      "grad_norm": 2.5700626373291016,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 5690
+    },
+    {
+      "epoch": 18.26923076923077,
+      "grad_norm": 2.467526912689209,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 5700
+    },
+    {
+      "epoch": 18.30128205128205,
+      "grad_norm": 2.463859796524048,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 5710
+    },
+    {
+      "epoch": 18.333333333333332,
+      "grad_norm": 2.331685781478882,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 5720
+    },
+    {
+      "epoch": 18.365384615384617,
+      "grad_norm": 2.687025785446167,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 5730
+    },
+    {
+      "epoch": 18.397435897435898,
+      "grad_norm": 2.6946585178375244,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 5740
+    },
+    {
+      "epoch": 18.42948717948718,
+      "grad_norm": 4.065126895904541,
+      "learning_rate": 0.0001,
+      "loss": 0.0249,
+      "step": 5750
+    },
+    {
+      "epoch": 18.46153846153846,
+      "grad_norm": 3.0193850994110107,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 5760
+    },
+    {
+      "epoch": 18.493589743589745,
+      "grad_norm": 2.3813676834106445,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 5770
+    },
+    {
+      "epoch": 18.525641025641026,
+      "grad_norm": 2.905806064605713,
+      "learning_rate": 0.0001,
+      "loss": 0.0242,
+      "step": 5780
+    },
+    {
+      "epoch": 18.557692307692307,
+      "grad_norm": 2.765190601348877,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 5790
+    },
+    {
+      "epoch": 18.58974358974359,
+      "grad_norm": 3.3380815982818604,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 5800
+    },
+    {
+      "epoch": 18.621794871794872,
+      "grad_norm": 2.718698501586914,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 5810
+    },
+    {
+      "epoch": 18.653846153846153,
+      "grad_norm": 2.614901304244995,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 5820
+    },
+    {
+      "epoch": 18.685897435897434,
+      "grad_norm": 2.574202537536621,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 5830
+    },
+    {
+      "epoch": 18.71794871794872,
+      "grad_norm": 2.6516714096069336,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 5840
+    },
+    {
+      "epoch": 18.75,
+      "grad_norm": 2.7536203861236572,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 5850
+    },
+    {
+      "epoch": 18.78205128205128,
+      "grad_norm": 2.491481304168701,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 5860
+    },
+    {
+      "epoch": 18.814102564102566,
+      "grad_norm": 2.470709800720215,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 5870
+    },
+    {
+      "epoch": 18.846153846153847,
+      "grad_norm": 3.192702293395996,
+      "learning_rate": 0.0001,
+      "loss": 0.0245,
+      "step": 5880
+    },
+    {
+      "epoch": 18.878205128205128,
+      "grad_norm": 2.8426668643951416,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 5890
+    },
+    {
+      "epoch": 18.91025641025641,
+      "grad_norm": 2.8473124504089355,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 5900
+    },
+    {
+      "epoch": 18.942307692307693,
+      "grad_norm": 2.7011146545410156,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 5910
+    },
+    {
+      "epoch": 18.974358974358974,
+      "grad_norm": 3.1229372024536133,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 5920
+    },
+    {
+      "epoch": 19.006410256410255,
+      "grad_norm": 2.228893280029297,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 5930
+    },
+    {
+      "epoch": 19.03846153846154,
+      "grad_norm": 2.7579166889190674,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 5940
+    },
+    {
+      "epoch": 19.07051282051282,
+      "grad_norm": 2.7614972591400146,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 5950
+    },
+    {
+      "epoch": 19.102564102564102,
+      "grad_norm": 2.413581132888794,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 5960
+    },
+    {
+      "epoch": 19.134615384615383,
+      "grad_norm": 2.4113729000091553,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 5970
+    },
+    {
+      "epoch": 19.166666666666668,
+      "grad_norm": 2.2972846031188965,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 5980
+    },
+    {
+      "epoch": 19.19871794871795,
+      "grad_norm": 2.6480324268341064,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 5990
+    },
+    {
+      "epoch": 19.23076923076923,
+      "grad_norm": 2.4940879344940186,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 6000
+    },
+    {
+      "epoch": 19.262820512820515,
+      "grad_norm": 2.253333330154419,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 6010
+    },
+    {
+      "epoch": 19.294871794871796,
+      "grad_norm": 2.3195667266845703,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 6020
+    },
+    {
+      "epoch": 19.326923076923077,
+      "grad_norm": 2.736644983291626,
+      "learning_rate": 0.0001,
+      "loss": 0.0255,
+      "step": 6030
+    },
+    {
+      "epoch": 19.358974358974358,
+      "grad_norm": 2.726344347000122,
+      "learning_rate": 0.0001,
+      "loss": 0.025,
+      "step": 6040
+    },
+    {
+      "epoch": 19.391025641025642,
+      "grad_norm": 2.4727766513824463,
+      "learning_rate": 0.0001,
+      "loss": 0.0257,
+      "step": 6050
+    },
+    {
+      "epoch": 19.423076923076923,
+      "grad_norm": 2.180943250656128,
+      "learning_rate": 0.0001,
+      "loss": 0.0251,
+      "step": 6060
+    },
+    {
+      "epoch": 19.455128205128204,
+      "grad_norm": 2.8678736686706543,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 6070
+    },
+    {
+      "epoch": 19.487179487179485,
+      "grad_norm": 2.2936999797821045,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 6080
+    },
+    {
+      "epoch": 19.51923076923077,
+      "grad_norm": 2.4637374877929688,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 6090
+    },
+    {
+      "epoch": 19.55128205128205,
+      "grad_norm": 2.506727933883667,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 6100
+    },
+    {
+      "epoch": 19.583333333333332,
+      "grad_norm": 2.695742607116699,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 6110
+    },
+    {
+      "epoch": 19.615384615384617,
+      "grad_norm": 1.895630955696106,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 6120
+    },
+    {
+      "epoch": 19.647435897435898,
+      "grad_norm": 2.3296878337860107,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 6130
+    },
+    {
+      "epoch": 19.67948717948718,
+      "grad_norm": 2.370215654373169,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 6140
+    },
+    {
+      "epoch": 19.71153846153846,
+      "grad_norm": 2.5559608936309814,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 6150
+    },
+    {
+      "epoch": 19.743589743589745,
+      "grad_norm": 2.5297179222106934,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 6160
+    },
+    {
+      "epoch": 19.775641025641026,
+      "grad_norm": 2.324643611907959,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 6170
+    },
+    {
+      "epoch": 19.807692307692307,
+      "grad_norm": 2.2911105155944824,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 6180
+    },
+    {
+      "epoch": 19.83974358974359,
+      "grad_norm": 2.9543862342834473,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 6190
+    },
+    {
+      "epoch": 19.871794871794872,
+      "grad_norm": 2.917512893676758,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 6200
+    },
+    {
+      "epoch": 19.903846153846153,
+      "grad_norm": 2.4353842735290527,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 6210
+    },
+    {
+      "epoch": 19.935897435897434,
+      "grad_norm": 2.6304476261138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 6220
+    },
+    {
+      "epoch": 19.96794871794872,
+      "grad_norm": 2.483510732650757,
+      "learning_rate": 0.0001,
+      "loss": 0.0233,
+      "step": 6230
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 2.3691775798797607,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 6240
+    },
+    {
+      "epoch": 20.03205128205128,
+      "grad_norm": 2.09486722946167,
+      "learning_rate": 0.0001,
+      "loss": 0.0236,
+      "step": 6250
+    },
+    {
+      "epoch": 20.064102564102566,
+      "grad_norm": 2.1885087490081787,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 6260
+    },
+    {
+      "epoch": 20.096153846153847,
+      "grad_norm": 2.0216422080993652,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 6270
+    },
+    {
+      "epoch": 20.128205128205128,
+      "grad_norm": 2.1108345985412598,
+      "learning_rate": 0.0001,
+      "loss": 0.0238,
+      "step": 6280
+    },
+    {
+      "epoch": 20.16025641025641,
+      "grad_norm": 2.2598941326141357,
+      "learning_rate": 0.0001,
+      "loss": 0.024,
+      "step": 6290
+    },
+    {
+      "epoch": 20.192307692307693,
+      "grad_norm": 2.2350850105285645,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 6300
+    },
+    {
+      "epoch": 20.224358974358974,
+      "grad_norm": 3.0093796253204346,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 6310
+    },
+    {
+      "epoch": 20.256410256410255,
+      "grad_norm": 2.567873001098633,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 6320
+    },
+    {
+      "epoch": 20.28846153846154,
+      "grad_norm": 2.135219097137451,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 6330
+    },
+    {
+      "epoch": 20.32051282051282,
+      "grad_norm": 2.4754984378814697,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 6340
+    },
+    {
+      "epoch": 20.352564102564102,
+      "grad_norm": 2.2614574432373047,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 6350
+    },
+    {
+      "epoch": 20.384615384615383,
+      "grad_norm": 2.169696807861328,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 6360
+    },
+    {
+      "epoch": 20.416666666666668,
+      "grad_norm": 2.1439125537872314,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 6370
+    },
+    {
+      "epoch": 20.44871794871795,
+      "grad_norm": 2.07303786277771,
+      "learning_rate": 0.0001,
+      "loss": 0.0244,
+      "step": 6380
+    },
+    {
+      "epoch": 20.48076923076923,
+      "grad_norm": 2.453007698059082,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 6390
+    },
+    {
+      "epoch": 20.51282051282051,
+      "grad_norm": 2.4704322814941406,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 6400
+    },
+    {
+      "epoch": 20.544871794871796,
+      "grad_norm": 2.5562000274658203,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 6410
+    },
+    {
+      "epoch": 20.576923076923077,
+      "grad_norm": 2.3574604988098145,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 6420
+    },
+    {
+      "epoch": 20.608974358974358,
+      "grad_norm": 3.05255126953125,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 6430
+    },
+    {
+      "epoch": 20.641025641025642,
+      "grad_norm": 2.613553762435913,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 6440
+    },
+    {
+      "epoch": 20.673076923076923,
+      "grad_norm": 2.202878713607788,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 6450
+    },
+    {
+      "epoch": 20.705128205128204,
+      "grad_norm": 2.5271248817443848,
+      "learning_rate": 0.0001,
+      "loss": 0.0237,
+      "step": 6460
+    },
+    {
+      "epoch": 20.73717948717949,
+      "grad_norm": 2.105153799057007,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 6470
+    },
+    {
+      "epoch": 20.76923076923077,
+      "grad_norm": 2.071165084838867,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 6480
+    },
+    {
+      "epoch": 20.80128205128205,
+      "grad_norm": 2.869441270828247,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 6490
+    },
+    {
+      "epoch": 20.833333333333332,
+      "grad_norm": 2.190702438354492,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 6500
+    },
+    {
+      "epoch": 20.865384615384617,
+      "grad_norm": 2.2532246112823486,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 6510
+    },
+    {
+      "epoch": 20.897435897435898,
+      "grad_norm": 2.661111354827881,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 6520
+    },
+    {
+      "epoch": 20.92948717948718,
+      "grad_norm": 2.6565158367156982,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 6530
+    },
+    {
+      "epoch": 20.96153846153846,
+      "grad_norm": 2.3633830547332764,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 6540
+    },
+    {
+      "epoch": 20.993589743589745,
+      "grad_norm": 2.6252074241638184,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 6550
+    },
+    {
+      "epoch": 21.025641025641026,
+      "grad_norm": 2.4006121158599854,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 6560
+    },
+    {
+      "epoch": 21.057692307692307,
+      "grad_norm": 2.4913341999053955,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 6570
+    },
+    {
+      "epoch": 21.08974358974359,
+      "grad_norm": 2.3227920532226562,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 6580
+    },
+    {
+      "epoch": 21.121794871794872,
+      "grad_norm": 2.0752267837524414,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 6590
+    },
+    {
+      "epoch": 21.153846153846153,
+      "grad_norm": 2.50242018699646,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 6600
+    },
+    {
+      "epoch": 21.185897435897434,
+      "grad_norm": 1.8786053657531738,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 6610
+    },
+    {
+      "epoch": 21.21794871794872,
+      "grad_norm": 2.378067970275879,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 6620
+    },
+    {
+      "epoch": 21.25,
+      "grad_norm": 2.319430351257324,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 6630
+    },
+    {
+      "epoch": 21.28205128205128,
+      "grad_norm": 1.8938727378845215,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 6640
+    },
+    {
+      "epoch": 21.314102564102566,
+      "grad_norm": 2.1133806705474854,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 6650
+    },
+    {
+      "epoch": 21.346153846153847,
+      "grad_norm": 2.332225799560547,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 6660
+    },
+    {
+      "epoch": 21.378205128205128,
+      "grad_norm": 2.2579565048217773,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 6670
+    },
+    {
+      "epoch": 21.41025641025641,
+      "grad_norm": 2.5183629989624023,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 6680
+    },
+    {
+      "epoch": 21.442307692307693,
+      "grad_norm": 2.5661392211914062,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 6690
+    },
+    {
+      "epoch": 21.474358974358974,
+      "grad_norm": 2.442814588546753,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 6700
+    },
+    {
+      "epoch": 21.506410256410255,
+      "grad_norm": 1.843664288520813,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 6710
+    },
+    {
+      "epoch": 21.53846153846154,
+      "grad_norm": 2.2522521018981934,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 6720
+    },
+    {
+      "epoch": 21.57051282051282,
+      "grad_norm": 1.9240140914916992,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 6730
+    },
+    {
+      "epoch": 21.602564102564102,
+      "grad_norm": 1.9302911758422852,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 6740
+    },
+    {
+      "epoch": 21.634615384615383,
+      "grad_norm": 2.5263028144836426,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 6750
+    },
+    {
+      "epoch": 21.666666666666668,
+      "grad_norm": 2.1400835514068604,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 6760
+    },
+    {
+      "epoch": 21.69871794871795,
+      "grad_norm": 2.32651948928833,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 6770
+    },
+    {
+      "epoch": 21.73076923076923,
+      "grad_norm": 2.0497686862945557,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 6780
+    },
+    {
+      "epoch": 21.76282051282051,
+      "grad_norm": 2.4226438999176025,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 6790
+    },
+    {
+      "epoch": 21.794871794871796,
+      "grad_norm": 2.4037322998046875,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 6800
+    },
+    {
+      "epoch": 21.826923076923077,
+      "grad_norm": 2.570404052734375,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 6810
+    },
+    {
+      "epoch": 21.858974358974358,
+      "grad_norm": 1.6865414381027222,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 6820
+    },
+    {
+      "epoch": 21.891025641025642,
+      "grad_norm": 2.3927488327026367,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 6830
+    },
+    {
+      "epoch": 21.923076923076923,
+      "grad_norm": 2.3751089572906494,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 6840
+    },
+    {
+      "epoch": 21.955128205128204,
+      "grad_norm": 2.0440986156463623,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 6850
+    },
+    {
+      "epoch": 21.98717948717949,
+      "grad_norm": 2.1035609245300293,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 6860
+    },
+    {
+      "epoch": 22.01923076923077,
+      "grad_norm": 2.1280577182769775,
+      "learning_rate": 0.0001,
+      "loss": 0.0224,
+      "step": 6870
+    },
+    {
+      "epoch": 22.05128205128205,
+      "grad_norm": 2.1338977813720703,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 6880
+    },
+    {
+      "epoch": 22.083333333333332,
+      "grad_norm": 2.0862832069396973,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 6890
+    },
+    {
+      "epoch": 22.115384615384617,
+      "grad_norm": 2.1095523834228516,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 6900
+    },
+    {
+      "epoch": 22.147435897435898,
+      "grad_norm": 2.227233409881592,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 6910
+    },
+    {
+      "epoch": 22.17948717948718,
+      "grad_norm": 2.2682812213897705,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 6920
+    },
+    {
+      "epoch": 22.21153846153846,
+      "grad_norm": 2.2864134311676025,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 6930
+    },
+    {
+      "epoch": 22.243589743589745,
+      "grad_norm": 2.0292115211486816,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 6940
+    },
+    {
+      "epoch": 22.275641025641026,
+      "grad_norm": 2.436767101287842,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 6950
+    },
+    {
+      "epoch": 22.307692307692307,
+      "grad_norm": 1.95342218875885,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 6960
+    },
+    {
+      "epoch": 22.33974358974359,
+      "grad_norm": 2.2495529651641846,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 6970
+    },
+    {
+      "epoch": 22.371794871794872,
+      "grad_norm": 1.9991117715835571,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 6980
+    },
+    {
+      "epoch": 22.403846153846153,
+      "grad_norm": 2.1541218757629395,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 6990
+    },
+    {
+      "epoch": 22.435897435897434,
+      "grad_norm": 1.9657899141311646,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 7000
+    },
+    {
+      "epoch": 22.46794871794872,
+      "grad_norm": 1.965909719467163,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 7010
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 1.9067782163619995,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 7020
+    },
+    {
+      "epoch": 22.53205128205128,
+      "grad_norm": 2.555567502975464,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 7030
+    },
+    {
+      "epoch": 22.564102564102566,
+      "grad_norm": 2.5767838954925537,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 7040
+    },
+    {
+      "epoch": 22.596153846153847,
+      "grad_norm": 2.3919405937194824,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 7050
+    },
+    {
+      "epoch": 22.628205128205128,
+      "grad_norm": 1.9437181949615479,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 7060
+    },
+    {
+      "epoch": 22.66025641025641,
+      "grad_norm": 2.065418243408203,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 7070
+    },
+    {
+      "epoch": 22.692307692307693,
+      "grad_norm": 2.129448652267456,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 7080
+    },
+    {
+      "epoch": 22.724358974358974,
+      "grad_norm": 2.1343557834625244,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 7090
+    },
+    {
+      "epoch": 22.756410256410255,
+      "grad_norm": 2.4218642711639404,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 7100
+    },
+    {
+      "epoch": 22.78846153846154,
+      "grad_norm": 1.9601199626922607,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 7110
+    },
+    {
+      "epoch": 22.82051282051282,
+      "grad_norm": 1.7631748914718628,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 7120
+    },
+    {
+      "epoch": 22.852564102564102,
+      "grad_norm": 1.897560715675354,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 7130
+    },
+    {
+      "epoch": 22.884615384615383,
+      "grad_norm": 2.1872565746307373,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 7140
+    },
+    {
+      "epoch": 22.916666666666668,
+      "grad_norm": 2.1789636611938477,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 7150
+    },
+    {
+      "epoch": 22.94871794871795,
+      "grad_norm": 1.9641361236572266,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 7160
+    },
+    {
+      "epoch": 22.98076923076923,
+      "grad_norm": 2.1320912837982178,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 7170
+    },
+    {
+      "epoch": 23.012820512820515,
+      "grad_norm": 2.203303098678589,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 7180
+    },
+    {
+      "epoch": 23.044871794871796,
+      "grad_norm": 2.392852783203125,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 7190
+    },
+    {
+      "epoch": 23.076923076923077,
+      "grad_norm": 1.8540951013565063,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 7200
+    },
+    {
+      "epoch": 23.108974358974358,
+      "grad_norm": 2.3745718002319336,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 7210
+    },
+    {
+      "epoch": 23.141025641025642,
+      "grad_norm": 2.336437463760376,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 7220
+    },
+    {
+      "epoch": 23.173076923076923,
+      "grad_norm": 1.7846848964691162,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 7230
+    },
+    {
+      "epoch": 23.205128205128204,
+      "grad_norm": 2.301988363265991,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 7240
+    },
+    {
+      "epoch": 23.237179487179485,
+      "grad_norm": 2.0795865058898926,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 7250
+    },
+    {
+      "epoch": 23.26923076923077,
+      "grad_norm": 2.4783356189727783,
+      "learning_rate": 0.0001,
+      "loss": 0.0212,
+      "step": 7260
+    },
+    {
+      "epoch": 23.30128205128205,
+      "grad_norm": 2.104689598083496,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 7270
+    },
+    {
+      "epoch": 23.333333333333332,
+      "grad_norm": 2.4220950603485107,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 7280
+    },
+    {
+      "epoch": 23.365384615384617,
+      "grad_norm": 2.6191937923431396,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 7290
+    },
+    {
+      "epoch": 23.397435897435898,
+      "grad_norm": 1.9180433750152588,
+      "learning_rate": 0.0001,
+      "loss": 0.0227,
+      "step": 7300
+    },
+    {
+      "epoch": 23.42948717948718,
+      "grad_norm": 2.486530303955078,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 7310
+    },
+    {
+      "epoch": 23.46153846153846,
+      "grad_norm": 2.07006573677063,
+      "learning_rate": 0.0001,
+      "loss": 0.0223,
+      "step": 7320
+    },
+    {
+      "epoch": 23.493589743589745,
+      "grad_norm": 2.626356840133667,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 7330
+    },
+    {
+      "epoch": 23.525641025641026,
+      "grad_norm": 2.134913444519043,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 7340
+    },
+    {
+      "epoch": 23.557692307692307,
+      "grad_norm": 1.9528292417526245,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 7350
+    },
+    {
+      "epoch": 23.58974358974359,
+      "grad_norm": 2.161001205444336,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 7360
+    },
+    {
+      "epoch": 23.621794871794872,
+      "grad_norm": 1.8617204427719116,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 7370
+    },
+    {
+      "epoch": 23.653846153846153,
+      "grad_norm": 2.1380679607391357,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 7380
+    },
+    {
+      "epoch": 23.685897435897434,
+      "grad_norm": 2.2671968936920166,
+      "learning_rate": 0.0001,
+      "loss": 0.0221,
+      "step": 7390
+    },
+    {
+      "epoch": 23.71794871794872,
+      "grad_norm": 2.1539597511291504,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 7400
+    },
+    {
+      "epoch": 23.75,
+      "grad_norm": 2.167410373687744,
+      "learning_rate": 0.0001,
+      "loss": 0.0217,
+      "step": 7410
+    },
+    {
+      "epoch": 23.78205128205128,
+      "grad_norm": 2.0001449584960938,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 7420
+    },
+    {
+      "epoch": 23.814102564102566,
+      "grad_norm": 2.5041747093200684,
+      "learning_rate": 0.0001,
+      "loss": 0.022,
+      "step": 7430
+    },
+    {
+      "epoch": 23.846153846153847,
+      "grad_norm": 1.6780283451080322,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 7440
+    },
+    {
+      "epoch": 23.878205128205128,
+      "grad_norm": 2.3446335792541504,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 7450
+    },
+    {
+      "epoch": 23.91025641025641,
+      "grad_norm": 1.895558476448059,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 7460
+    },
+    {
+      "epoch": 23.942307692307693,
+      "grad_norm": 2.1128745079040527,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 7470
+    },
+    {
+      "epoch": 23.974358974358974,
+      "grad_norm": 2.3684892654418945,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 7480
+    },
+    {
+      "epoch": 24.006410256410255,
+      "grad_norm": 2.323725938796997,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 7490
+    },
+    {
+      "epoch": 24.03846153846154,
+      "grad_norm": 2.288409471511841,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 7500
+    },
+    {
+      "epoch": 24.07051282051282,
+      "grad_norm": 2.137967824935913,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 7510
+    },
+    {
+      "epoch": 24.102564102564102,
+      "grad_norm": 1.858307957649231,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 7520
+    },
+    {
+      "epoch": 24.134615384615383,
+      "grad_norm": 2.170950412750244,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 7530
+    },
+    {
+      "epoch": 24.166666666666668,
+      "grad_norm": 1.8133512735366821,
+      "learning_rate": 0.0001,
+      "loss": 0.0222,
+      "step": 7540
+    },
+    {
+      "epoch": 24.19871794871795,
+      "grad_norm": 2.5387940406799316,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 7550
+    },
+    {
+      "epoch": 24.23076923076923,
+      "grad_norm": 2.1800637245178223,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 7560
+    },
+    {
+      "epoch": 24.262820512820515,
+      "grad_norm": 2.244605779647827,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 7570
+    },
+    {
+      "epoch": 24.294871794871796,
+      "grad_norm": 2.0990140438079834,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 7580
+    },
+    {
+      "epoch": 24.326923076923077,
+      "grad_norm": 2.0584089756011963,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 7590
+    },
+    {
+      "epoch": 24.358974358974358,
+      "grad_norm": 2.062410354614258,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 7600
+    },
+    {
+      "epoch": 24.391025641025642,
+      "grad_norm": 2.0045151710510254,
+      "learning_rate": 0.0001,
+      "loss": 0.0197,
+      "step": 7610
+    },
+    {
+      "epoch": 24.423076923076923,
+      "grad_norm": 1.9695228338241577,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 7620
+    },
+    {
+      "epoch": 24.455128205128204,
+      "grad_norm": 2.3742501735687256,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 7630
+    },
+    {
+      "epoch": 24.487179487179485,
+      "grad_norm": 2.856541872024536,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 7640
+    },
+    {
+      "epoch": 24.51923076923077,
+      "grad_norm": 2.219289541244507,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 7650
+    },
+    {
+      "epoch": 24.55128205128205,
+      "grad_norm": 1.9282513856887817,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 7660
+    },
+    {
+      "epoch": 24.583333333333332,
+      "grad_norm": 1.867789387702942,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 7670
+    },
+    {
+      "epoch": 24.615384615384617,
+      "grad_norm": 2.2784132957458496,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 7680
+    },
+    {
+      "epoch": 24.647435897435898,
+      "grad_norm": 2.4913110733032227,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 7690
+    },
+    {
+      "epoch": 24.67948717948718,
+      "grad_norm": 2.1106958389282227,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 7700
+    },
+    {
+      "epoch": 24.71153846153846,
+      "grad_norm": 2.3220162391662598,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 7710
+    },
+    {
+      "epoch": 24.743589743589745,
+      "grad_norm": 1.8951692581176758,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 7720
+    },
+    {
+      "epoch": 24.775641025641026,
+      "grad_norm": 2.1384031772613525,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 7730
+    },
+    {
+      "epoch": 24.807692307692307,
+      "grad_norm": 1.9938547611236572,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 7740
+    },
+    {
+      "epoch": 24.83974358974359,
+      "grad_norm": 1.9380080699920654,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 7750
+    },
+    {
+      "epoch": 24.871794871794872,
+      "grad_norm": 1.5975866317749023,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 7760
+    },
+    {
+      "epoch": 24.903846153846153,
+      "grad_norm": 2.239264488220215,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 7770
+    },
+    {
+      "epoch": 24.935897435897434,
+      "grad_norm": 2.4920594692230225,
+      "learning_rate": 0.0001,
+      "loss": 0.0239,
+      "step": 7780
+    },
+    {
+      "epoch": 24.96794871794872,
+      "grad_norm": 2.7871365547180176,
+      "learning_rate": 0.0001,
+      "loss": 0.0234,
+      "step": 7790
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 2.4696574211120605,
+      "learning_rate": 0.0001,
+      "loss": 0.0229,
+      "step": 7800
+    },
+    {
+      "epoch": 25.03205128205128,
+      "grad_norm": 2.469994306564331,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 7810
+    },
+    {
+      "epoch": 25.064102564102566,
+      "grad_norm": 2.5168166160583496,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 7820
+    },
+    {
+      "epoch": 25.096153846153847,
+      "grad_norm": 2.444796085357666,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 7830
+    },
+    {
+      "epoch": 25.128205128205128,
+      "grad_norm": 2.2356746196746826,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 7840
+    },
+    {
+      "epoch": 25.16025641025641,
+      "grad_norm": 2.0409486293792725,
+      "learning_rate": 0.0001,
+      "loss": 0.0211,
+      "step": 7850
+    },
+    {
+      "epoch": 25.192307692307693,
+      "grad_norm": 1.7760604619979858,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 7860
+    },
+    {
+      "epoch": 25.224358974358974,
+      "grad_norm": 1.8342646360397339,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 7870
+    },
+    {
+      "epoch": 25.256410256410255,
+      "grad_norm": 2.3203277587890625,
+      "learning_rate": 0.0001,
+      "loss": 0.0225,
+      "step": 7880
+    },
+    {
+      "epoch": 25.28846153846154,
+      "grad_norm": 1.9366159439086914,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 7890
+    },
+    {
+      "epoch": 25.32051282051282,
+      "grad_norm": 1.7144862413406372,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 7900
+    },
+    {
+      "epoch": 25.352564102564102,
+      "grad_norm": 2.2737598419189453,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 7910
+    },
+    {
+      "epoch": 25.384615384615383,
+      "grad_norm": 1.9341888427734375,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 7920
+    },
+    {
+      "epoch": 25.416666666666668,
+      "grad_norm": 1.7581288814544678,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 7930
+    },
+    {
+      "epoch": 25.44871794871795,
+      "grad_norm": 1.869982123374939,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 7940
+    },
+    {
+      "epoch": 25.48076923076923,
+      "grad_norm": 1.8544893264770508,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 7950
+    },
+    {
+      "epoch": 25.51282051282051,
+      "grad_norm": 1.7364909648895264,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 7960
+    },
+    {
+      "epoch": 25.544871794871796,
+      "grad_norm": 1.7695960998535156,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 7970
+    },
+    {
+      "epoch": 25.576923076923077,
+      "grad_norm": 2.172821283340454,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 7980
+    },
+    {
+      "epoch": 25.608974358974358,
+      "grad_norm": 2.1765735149383545,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 7990
+    },
+    {
+      "epoch": 25.641025641025642,
+      "grad_norm": 2.0085885524749756,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 8000
+    },
+    {
+      "epoch": 25.673076923076923,
+      "grad_norm": 2.3065953254699707,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 8010
+    },
+    {
+      "epoch": 25.705128205128204,
+      "grad_norm": 2.410230875015259,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 8020
+    },
+    {
+      "epoch": 25.73717948717949,
+      "grad_norm": 2.310319662094116,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 8030
+    },
+    {
+      "epoch": 25.76923076923077,
+      "grad_norm": 2.0291545391082764,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 8040
+    },
+    {
+      "epoch": 25.80128205128205,
+      "grad_norm": 2.006711959838867,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 8050
+    },
+    {
+      "epoch": 25.833333333333332,
+      "grad_norm": 1.8809735774993896,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 8060
+    },
+    {
+      "epoch": 25.865384615384617,
+      "grad_norm": 2.013092517852783,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 8070
+    },
+    {
+      "epoch": 25.897435897435898,
+      "grad_norm": 2.210293769836426,
+      "learning_rate": 0.0001,
+      "loss": 0.0219,
+      "step": 8080
+    },
+    {
+      "epoch": 25.92948717948718,
+      "grad_norm": 2.7968313694000244,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 8090
+    },
+    {
+      "epoch": 25.96153846153846,
+      "grad_norm": 2.4631497859954834,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 8100
+    },
+    {
+      "epoch": 25.993589743589745,
+      "grad_norm": 1.9022027254104614,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 8110
+    },
+    {
+      "epoch": 26.025641025641026,
+      "grad_norm": 2.1356441974639893,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 8120
+    },
+    {
+      "epoch": 26.057692307692307,
+      "grad_norm": 2.04964017868042,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 8130
+    },
+    {
+      "epoch": 26.08974358974359,
+      "grad_norm": 2.1263701915740967,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 8140
+    },
+    {
+      "epoch": 26.121794871794872,
+      "grad_norm": 1.9381874799728394,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 8150
+    },
+    {
+      "epoch": 26.153846153846153,
+      "grad_norm": 1.7397258281707764,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 8160
+    },
+    {
+      "epoch": 26.185897435897434,
+      "grad_norm": 1.730484127998352,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 8170
+    },
+    {
+      "epoch": 26.21794871794872,
+      "grad_norm": 1.938844084739685,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 8180
+    },
+    {
+      "epoch": 26.25,
+      "grad_norm": 1.7573728561401367,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 8190
+    },
+    {
+      "epoch": 26.28205128205128,
+      "grad_norm": 1.718780279159546,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 8200
+    },
+    {
+      "epoch": 26.314102564102566,
+      "grad_norm": 1.935135841369629,
+      "learning_rate": 0.0001,
+      "loss": 0.0203,
+      "step": 8210
+    },
+    {
+      "epoch": 26.346153846153847,
+      "grad_norm": 2.130481719970703,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 8220
+    },
+    {
+      "epoch": 26.378205128205128,
+      "grad_norm": 1.9044467210769653,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 8230
+    },
+    {
+      "epoch": 26.41025641025641,
+      "grad_norm": 1.978486180305481,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 8240
+    },
+    {
+      "epoch": 26.442307692307693,
+      "grad_norm": 2.023418664932251,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 8250
+    },
+    {
+      "epoch": 26.474358974358974,
+      "grad_norm": 2.100163221359253,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 8260
+    },
+    {
+      "epoch": 26.506410256410255,
+      "grad_norm": 2.157423734664917,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 8270
+    },
+    {
+      "epoch": 26.53846153846154,
+      "grad_norm": 1.7843403816223145,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 8280
+    },
+    {
+      "epoch": 26.57051282051282,
+      "grad_norm": 1.842915415763855,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 8290
+    },
+    {
+      "epoch": 26.602564102564102,
+      "grad_norm": 1.9403526782989502,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 8300
+    },
+    {
+      "epoch": 26.634615384615383,
+      "grad_norm": 2.2746317386627197,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 8310
+    },
+    {
+      "epoch": 26.666666666666668,
+      "grad_norm": 1.571892499923706,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 8320
+    },
+    {
+      "epoch": 26.69871794871795,
+      "grad_norm": 2.25111722946167,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 8330
+    },
+    {
+      "epoch": 26.73076923076923,
+      "grad_norm": 2.0354979038238525,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 8340
+    },
+    {
+      "epoch": 26.76282051282051,
+      "grad_norm": 1.6025217771530151,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 8350
+    },
+    {
+      "epoch": 26.794871794871796,
+      "grad_norm": 1.6221263408660889,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 8360
+    },
+    {
+      "epoch": 26.826923076923077,
+      "grad_norm": 1.7288620471954346,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 8370
+    },
+    {
+      "epoch": 26.858974358974358,
+      "grad_norm": 1.5191155672073364,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 8380
+    },
+    {
+      "epoch": 26.891025641025642,
+      "grad_norm": 1.906168818473816,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 8390
+    },
+    {
+      "epoch": 26.923076923076923,
+      "grad_norm": 1.7442294359207153,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 8400
+    },
+    {
+      "epoch": 26.955128205128204,
+      "grad_norm": 2.054586410522461,
+      "learning_rate": 0.0001,
+      "loss": 0.0204,
+      "step": 8410
+    },
+    {
+      "epoch": 26.98717948717949,
+      "grad_norm": 2.085461378097534,
+      "learning_rate": 0.0001,
+      "loss": 0.0218,
+      "step": 8420
+    },
+    {
+      "epoch": 27.01923076923077,
+      "grad_norm": 1.8055580854415894,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 8430
+    },
+    {
+      "epoch": 27.05128205128205,
+      "grad_norm": 1.9273357391357422,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 8440
+    },
+    {
+      "epoch": 27.083333333333332,
+      "grad_norm": 2.007023811340332,
+      "learning_rate": 0.0001,
+      "loss": 0.0215,
+      "step": 8450
+    },
+    {
+      "epoch": 27.115384615384617,
+      "grad_norm": 2.073061466217041,
+      "learning_rate": 0.0001,
+      "loss": 0.0226,
+      "step": 8460
+    },
+    {
+      "epoch": 27.147435897435898,
+      "grad_norm": 1.737353801727295,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 8470
+    },
+    {
+      "epoch": 27.17948717948718,
+      "grad_norm": 1.8155568838119507,
+      "learning_rate": 0.0001,
+      "loss": 0.0214,
+      "step": 8480
+    },
+    {
+      "epoch": 27.21153846153846,
+      "grad_norm": 1.986191749572754,
+      "learning_rate": 0.0001,
+      "loss": 0.0213,
+      "step": 8490
+    },
+    {
+      "epoch": 27.243589743589745,
+      "grad_norm": 2.0215954780578613,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 8500
+    },
+    {
+      "epoch": 27.275641025641026,
+      "grad_norm": 1.8612140417099,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 8510
+    },
+    {
+      "epoch": 27.307692307692307,
+      "grad_norm": 2.0668447017669678,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 8520
+    },
+    {
+      "epoch": 27.33974358974359,
+      "grad_norm": 2.5615179538726807,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 8530
+    },
+    {
+      "epoch": 27.371794871794872,
+      "grad_norm": 2.1426122188568115,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 8540
+    },
+    {
+      "epoch": 27.403846153846153,
+      "grad_norm": 2.039597511291504,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 8550
+    },
+    {
+      "epoch": 27.435897435897434,
+      "grad_norm": 1.9490858316421509,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 8560
+    },
+    {
+      "epoch": 27.46794871794872,
+      "grad_norm": 1.8246827125549316,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 8570
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 2.178185224533081,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 8580
+    },
+    {
+      "epoch": 27.53205128205128,
+      "grad_norm": 2.253363847732544,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 8590
+    },
+    {
+      "epoch": 27.564102564102566,
+      "grad_norm": 2.3686952590942383,
+      "learning_rate": 0.0001,
+      "loss": 0.0208,
+      "step": 8600
+    },
+    {
+      "epoch": 27.596153846153847,
+      "grad_norm": 2.4660863876342773,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 8610
+    },
+    {
+      "epoch": 27.628205128205128,
+      "grad_norm": 1.959808588027954,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 8620
+    },
+    {
+      "epoch": 27.66025641025641,
+      "grad_norm": 2.0045642852783203,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 8630
+    },
+    {
+      "epoch": 27.692307692307693,
+      "grad_norm": 1.8085780143737793,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 8640
+    },
+    {
+      "epoch": 27.724358974358974,
+      "grad_norm": 1.910142183303833,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 8650
+    },
+    {
+      "epoch": 27.756410256410255,
+      "grad_norm": 1.9513264894485474,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 8660
+    },
+    {
+      "epoch": 27.78846153846154,
+      "grad_norm": 2.377110242843628,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 8670
+    },
+    {
+      "epoch": 27.82051282051282,
+      "grad_norm": 1.6742669343948364,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 8680
+    },
+    {
+      "epoch": 27.852564102564102,
+      "grad_norm": 1.699354648590088,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 8690
+    },
+    {
+      "epoch": 27.884615384615383,
+      "grad_norm": 1.4424294233322144,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 8700
+    },
+    {
+      "epoch": 27.916666666666668,
+      "grad_norm": 1.6105619668960571,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 8710
+    },
+    {
+      "epoch": 27.94871794871795,
+      "grad_norm": 1.8652199506759644,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 8720
+    },
+    {
+      "epoch": 27.98076923076923,
+      "grad_norm": 2.243945360183716,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 8730
+    },
+    {
+      "epoch": 28.012820512820515,
+      "grad_norm": 1.7259777784347534,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 8740
+    },
+    {
+      "epoch": 28.044871794871796,
+      "grad_norm": 2.385288715362549,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 8750
+    },
+    {
+      "epoch": 28.076923076923077,
+      "grad_norm": 1.9848723411560059,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 8760
+    },
+    {
+      "epoch": 28.108974358974358,
+      "grad_norm": 1.5904815196990967,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 8770
+    },
+    {
+      "epoch": 28.141025641025642,
+      "grad_norm": 1.6414109468460083,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 8780
+    },
+    {
+      "epoch": 28.173076923076923,
+      "grad_norm": 1.714098334312439,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 8790
+    },
+    {
+      "epoch": 28.205128205128204,
+      "grad_norm": 1.8831086158752441,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 8800
+    },
+    {
+      "epoch": 28.237179487179485,
+      "grad_norm": 1.8963987827301025,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 8810
+    },
+    {
+      "epoch": 28.26923076923077,
+      "grad_norm": 1.717739462852478,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 8820
+    },
+    {
+      "epoch": 28.30128205128205,
+      "grad_norm": 1.7032802104949951,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 8830
+    },
+    {
+      "epoch": 28.333333333333332,
+      "grad_norm": 1.729918360710144,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 8840
+    },
+    {
+      "epoch": 28.365384615384617,
+      "grad_norm": 1.716692566871643,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 8850
+    },
+    {
+      "epoch": 28.397435897435898,
+      "grad_norm": 1.7888103723526,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 8860
+    },
+    {
+      "epoch": 28.42948717948718,
+      "grad_norm": 1.797007441520691,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 8870
+    },
+    {
+      "epoch": 28.46153846153846,
+      "grad_norm": 1.801880121231079,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 8880
+    },
+    {
+      "epoch": 28.493589743589745,
+      "grad_norm": 1.8519458770751953,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 8890
+    },
+    {
+      "epoch": 28.525641025641026,
+      "grad_norm": 1.8267154693603516,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 8900
+    },
+    {
+      "epoch": 28.557692307692307,
+      "grad_norm": 2.3088431358337402,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 8910
+    },
+    {
+      "epoch": 28.58974358974359,
+      "grad_norm": 1.8550670146942139,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 8920
+    },
+    {
+      "epoch": 28.621794871794872,
+      "grad_norm": 1.854724645614624,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 8930
+    },
+    {
+      "epoch": 28.653846153846153,
+      "grad_norm": 2.2919209003448486,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 8940
+    },
+    {
+      "epoch": 28.685897435897434,
+      "grad_norm": 2.234786033630371,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 8950
+    },
+    {
+      "epoch": 28.71794871794872,
+      "grad_norm": 2.169624090194702,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 8960
+    },
+    {
+      "epoch": 28.75,
+      "grad_norm": 1.8134227991104126,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 8970
+    },
+    {
+      "epoch": 28.78205128205128,
+      "grad_norm": 1.7343730926513672,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 8980
+    },
+    {
+      "epoch": 28.814102564102566,
+      "grad_norm": 1.6826586723327637,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 8990
+    },
+    {
+      "epoch": 28.846153846153847,
+      "grad_norm": 1.915418267250061,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 9000
+    },
+    {
+      "epoch": 28.878205128205128,
+      "grad_norm": 2.0777266025543213,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 9010
+    },
+    {
+      "epoch": 28.91025641025641,
+      "grad_norm": 2.2467477321624756,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 9020
+    },
+    {
+      "epoch": 28.942307692307693,
+      "grad_norm": 2.112637996673584,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 9030
+    },
+    {
+      "epoch": 28.974358974358974,
+      "grad_norm": 1.714940071105957,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 9040
+    },
+    {
+      "epoch": 29.006410256410255,
+      "grad_norm": 2.274172782897949,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 9050
+    },
+    {
+      "epoch": 29.03846153846154,
+      "grad_norm": 1.8130053281784058,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 9060
+    },
+    {
+      "epoch": 29.07051282051282,
+      "grad_norm": 1.8634759187698364,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 9070
+    },
+    {
+      "epoch": 29.102564102564102,
+      "grad_norm": 1.8549264669418335,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 9080
+    },
+    {
+      "epoch": 29.134615384615383,
+      "grad_norm": 2.0329017639160156,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 9090
+    },
+    {
+      "epoch": 29.166666666666668,
+      "grad_norm": 1.8701411485671997,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 9100
+    },
+    {
+      "epoch": 29.19871794871795,
+      "grad_norm": 1.9479246139526367,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 9110
+    },
+    {
+      "epoch": 29.23076923076923,
+      "grad_norm": 2.0467023849487305,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 9120
+    },
+    {
+      "epoch": 29.262820512820515,
+      "grad_norm": 1.904484510421753,
+      "learning_rate": 0.0001,
+      "loss": 0.0196,
+      "step": 9130
+    },
+    {
+      "epoch": 29.294871794871796,
+      "grad_norm": 2.2637813091278076,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 9140
+    },
+    {
+      "epoch": 29.326923076923077,
+      "grad_norm": 1.3883939981460571,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 9150
+    },
+    {
+      "epoch": 29.358974358974358,
+      "grad_norm": 1.9284652471542358,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 9160
+    },
+    {
+      "epoch": 29.391025641025642,
+      "grad_norm": 1.8267935514450073,
+      "learning_rate": 0.0001,
+      "loss": 0.0201,
+      "step": 9170
+    },
+    {
+      "epoch": 29.423076923076923,
+      "grad_norm": 2.002413272857666,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 9180
+    },
+    {
+      "epoch": 29.455128205128204,
+      "grad_norm": 1.834088683128357,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 9190
+    },
+    {
+      "epoch": 29.487179487179485,
+      "grad_norm": 2.17870831489563,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 9200
+    },
+    {
+      "epoch": 29.51923076923077,
+      "grad_norm": 1.7619277238845825,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 9210
+    },
+    {
+      "epoch": 29.55128205128205,
+      "grad_norm": 1.777395248413086,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 9220
+    },
+    {
+      "epoch": 29.583333333333332,
+      "grad_norm": 2.167571783065796,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 9230
+    },
+    {
+      "epoch": 29.615384615384617,
+      "grad_norm": 1.8097535371780396,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 9240
+    },
+    {
+      "epoch": 29.647435897435898,
+      "grad_norm": 2.1970696449279785,
+      "learning_rate": 0.0001,
+      "loss": 0.0198,
+      "step": 9250
+    },
+    {
+      "epoch": 29.67948717948718,
+      "grad_norm": 1.8905006647109985,
+      "learning_rate": 0.0001,
+      "loss": 0.0199,
+      "step": 9260
+    },
+    {
+      "epoch": 29.71153846153846,
+      "grad_norm": 1.7193009853363037,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 9270
+    },
+    {
+      "epoch": 29.743589743589745,
+      "grad_norm": 1.4801058769226074,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 9280
+    },
+    {
+      "epoch": 29.775641025641026,
+      "grad_norm": 1.6776012182235718,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 9290
+    },
+    {
+      "epoch": 29.807692307692307,
+      "grad_norm": 1.4904963970184326,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 9300
+    },
+    {
+      "epoch": 29.83974358974359,
+      "grad_norm": 1.8040814399719238,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 9310
+    },
+    {
+      "epoch": 29.871794871794872,
+      "grad_norm": 1.8322135210037231,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 9320
+    },
+    {
+      "epoch": 29.903846153846153,
+      "grad_norm": 1.747743010520935,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 9330
+    },
+    {
+      "epoch": 29.935897435897434,
+      "grad_norm": 1.8544098138809204,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 9340
+    },
+    {
+      "epoch": 29.96794871794872,
+      "grad_norm": 1.9972087144851685,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 9350
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 1.7024741172790527,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 9360
+    },
+    {
+      "epoch": 30.03205128205128,
+      "grad_norm": 1.8836469650268555,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 9370
+    },
+    {
+      "epoch": 30.064102564102566,
+      "grad_norm": 2.031193494796753,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 9380
+    },
+    {
+      "epoch": 30.096153846153847,
+      "grad_norm": 1.626508116722107,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 9390
+    },
+    {
+      "epoch": 30.128205128205128,
+      "grad_norm": 1.6179336309432983,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 9400
+    },
+    {
+      "epoch": 30.16025641025641,
+      "grad_norm": 1.4257943630218506,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 9410
+    },
+    {
+      "epoch": 30.192307692307693,
+      "grad_norm": 1.6536033153533936,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 9420
+    },
+    {
+      "epoch": 30.224358974358974,
+      "grad_norm": 2.018357515335083,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 9430
+    },
+    {
+      "epoch": 30.256410256410255,
+      "grad_norm": 1.7419980764389038,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 9440
+    },
+    {
+      "epoch": 30.28846153846154,
+      "grad_norm": 1.6756658554077148,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 9450
+    },
+    {
+      "epoch": 30.32051282051282,
+      "grad_norm": 1.7000274658203125,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 9460
+    },
+    {
+      "epoch": 30.352564102564102,
+      "grad_norm": 1.5291976928710938,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 9470
+    },
+    {
+      "epoch": 30.384615384615383,
+      "grad_norm": 1.6135364770889282,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 9480
+    },
+    {
+      "epoch": 30.416666666666668,
+      "grad_norm": 2.024808883666992,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 9490
+    },
+    {
+      "epoch": 30.44871794871795,
+      "grad_norm": 1.7810136079788208,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 9500
+    },
+    {
+      "epoch": 30.48076923076923,
+      "grad_norm": 1.8307726383209229,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 9510
+    },
+    {
+      "epoch": 30.51282051282051,
+      "grad_norm": 2.151576042175293,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 9520
+    },
+    {
+      "epoch": 30.544871794871796,
+      "grad_norm": 1.9137002229690552,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 9530
+    },
+    {
+      "epoch": 30.576923076923077,
+      "grad_norm": 1.5770434141159058,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 9540
+    },
+    {
+      "epoch": 30.608974358974358,
+      "grad_norm": 1.8552402257919312,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 9550
+    },
+    {
+      "epoch": 30.641025641025642,
+      "grad_norm": 1.8927452564239502,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 9560
+    },
+    {
+      "epoch": 30.673076923076923,
+      "grad_norm": 1.97144615650177,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 9570
+    },
+    {
+      "epoch": 30.705128205128204,
+      "grad_norm": 1.8376692533493042,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 9580
+    },
+    {
+      "epoch": 30.73717948717949,
+      "grad_norm": 2.3407609462738037,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 9590
+    },
+    {
+      "epoch": 30.76923076923077,
+      "grad_norm": 2.0139622688293457,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 9600
+    },
+    {
+      "epoch": 30.80128205128205,
+      "grad_norm": 1.859351396560669,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 9610
+    },
+    {
+      "epoch": 30.833333333333332,
+      "grad_norm": 1.705877661705017,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 9620
+    },
+    {
+      "epoch": 30.865384615384617,
+      "grad_norm": 1.5252801179885864,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 9630
+    },
+    {
+      "epoch": 30.897435897435898,
+      "grad_norm": 1.6541659832000732,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 9640
+    },
+    {
+      "epoch": 30.92948717948718,
+      "grad_norm": 2.085771083831787,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 9650
+    },
+    {
+      "epoch": 30.96153846153846,
+      "grad_norm": 1.8677423000335693,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 9660
+    },
+    {
+      "epoch": 30.993589743589745,
+      "grad_norm": 1.5798794031143188,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 9670
+    },
+    {
+      "epoch": 31.025641025641026,
+      "grad_norm": 1.619641900062561,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 9680
+    },
+    {
+      "epoch": 31.057692307692307,
+      "grad_norm": 1.5838899612426758,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 9690
+    },
+    {
+      "epoch": 31.08974358974359,
+      "grad_norm": 1.7778080701828003,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 9700
+    },
+    {
+      "epoch": 31.121794871794872,
+      "grad_norm": 1.7065622806549072,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 9710
+    },
+    {
+      "epoch": 31.153846153846153,
+      "grad_norm": 1.6062777042388916,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 9720
+    },
+    {
+      "epoch": 31.185897435897434,
+      "grad_norm": 1.7777336835861206,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 9730
+    },
+    {
+      "epoch": 31.21794871794872,
+      "grad_norm": 1.8336225748062134,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 9740
+    },
+    {
+      "epoch": 31.25,
+      "grad_norm": 2.0107240676879883,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 9750
+    },
+    {
+      "epoch": 31.28205128205128,
+      "grad_norm": 1.8503575325012207,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 9760
+    },
+    {
+      "epoch": 31.314102564102566,
+      "grad_norm": 1.6372531652450562,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 9770
+    },
+    {
+      "epoch": 31.346153846153847,
+      "grad_norm": 2.07255482673645,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 9780
+    },
+    {
+      "epoch": 31.378205128205128,
+      "grad_norm": 2.7348620891571045,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 9790
+    },
+    {
+      "epoch": 31.41025641025641,
+      "grad_norm": 2.7878835201263428,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 9800
+    },
+    {
+      "epoch": 31.442307692307693,
+      "grad_norm": 1.8688615560531616,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 9810
+    },
+    {
+      "epoch": 31.474358974358974,
+      "grad_norm": 2.1157665252685547,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 9820
+    },
+    {
+      "epoch": 31.506410256410255,
+      "grad_norm": 1.6749210357666016,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 9830
+    },
+    {
+      "epoch": 31.53846153846154,
+      "grad_norm": 2.4320120811462402,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 9840
+    },
+    {
+      "epoch": 31.57051282051282,
+      "grad_norm": 2.5270416736602783,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 9850
+    },
+    {
+      "epoch": 31.602564102564102,
+      "grad_norm": 2.167774200439453,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 9860
+    },
+    {
+      "epoch": 31.634615384615383,
+      "grad_norm": 1.7432974576950073,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 9870
+    },
+    {
+      "epoch": 31.666666666666668,
+      "grad_norm": 2.2192223072052,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 9880
+    },
+    {
+      "epoch": 31.69871794871795,
+      "grad_norm": 1.979580283164978,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 9890
+    },
+    {
+      "epoch": 31.73076923076923,
+      "grad_norm": 2.3416621685028076,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 9900
+    },
+    {
+      "epoch": 31.76282051282051,
+      "grad_norm": 2.3209855556488037,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 9910
+    },
+    {
+      "epoch": 31.794871794871796,
+      "grad_norm": 2.5318620204925537,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 9920
+    },
+    {
+      "epoch": 31.826923076923077,
+      "grad_norm": 2.025780200958252,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 9930
+    },
+    {
+      "epoch": 31.858974358974358,
+      "grad_norm": 1.636176347732544,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 9940
+    },
+    {
+      "epoch": 31.891025641025642,
+      "grad_norm": 1.7482041120529175,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 9950
+    },
+    {
+      "epoch": 31.923076923076923,
+      "grad_norm": 1.5328348875045776,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 9960
+    },
+    {
+      "epoch": 31.955128205128204,
+      "grad_norm": 2.1234805583953857,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 9970
+    },
+    {
+      "epoch": 31.98717948717949,
+      "grad_norm": 1.9471029043197632,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 9980
+    },
+    {
+      "epoch": 32.01923076923077,
+      "grad_norm": 1.508803129196167,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 9990
+    },
+    {
+      "epoch": 32.05128205128205,
+      "grad_norm": 1.7476258277893066,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 10000
+    },
+    {
+      "epoch": 32.083333333333336,
+      "grad_norm": 1.660592794418335,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 10010
+    },
+    {
+      "epoch": 32.11538461538461,
+      "grad_norm": 1.4389138221740723,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 10020
+    },
+    {
+      "epoch": 32.1474358974359,
+      "grad_norm": 1.6087933778762817,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 10030
+    },
+    {
+      "epoch": 32.17948717948718,
+      "grad_norm": 1.7865159511566162,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 10040
+    },
+    {
+      "epoch": 32.21153846153846,
+      "grad_norm": 1.6194140911102295,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 10050
+    },
+    {
+      "epoch": 32.243589743589745,
+      "grad_norm": 1.791205883026123,
+      "learning_rate": 0.0001,
+      "loss": 0.0183,
+      "step": 10060
+    },
+    {
+      "epoch": 32.27564102564103,
+      "grad_norm": 2.012035846710205,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 10070
+    },
+    {
+      "epoch": 32.30769230769231,
+      "grad_norm": 1.525893211364746,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 10080
+    },
+    {
+      "epoch": 32.33974358974359,
+      "grad_norm": 1.4312409162521362,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 10090
+    },
+    {
+      "epoch": 32.37179487179487,
+      "grad_norm": 1.6360130310058594,
+      "learning_rate": 0.0001,
+      "loss": 0.0186,
+      "step": 10100
+    },
+    {
+      "epoch": 32.40384615384615,
+      "grad_norm": 1.7542210817337036,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 10110
+    },
+    {
+      "epoch": 32.43589743589744,
+      "grad_norm": 2.096468448638916,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 10120
+    },
+    {
+      "epoch": 32.467948717948715,
+      "grad_norm": 1.9179474115371704,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 10130
+    },
+    {
+      "epoch": 32.5,
+      "grad_norm": 1.7056092023849487,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 10140
+    },
+    {
+      "epoch": 32.532051282051285,
+      "grad_norm": 1.769641637802124,
+      "learning_rate": 0.0001,
+      "loss": 0.0182,
+      "step": 10150
+    },
+    {
+      "epoch": 32.56410256410256,
+      "grad_norm": 1.9853566884994507,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 10160
+    },
+    {
+      "epoch": 32.59615384615385,
+      "grad_norm": 1.4187015295028687,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 10170
+    },
+    {
+      "epoch": 32.62820512820513,
+      "grad_norm": 1.4672508239746094,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 10180
+    },
+    {
+      "epoch": 32.66025641025641,
+      "grad_norm": 1.6329678297042847,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 10190
+    },
+    {
+      "epoch": 32.69230769230769,
+      "grad_norm": 1.4605880975723267,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 10200
+    },
+    {
+      "epoch": 32.72435897435897,
+      "grad_norm": 1.4270951747894287,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 10210
+    },
+    {
+      "epoch": 32.756410256410255,
+      "grad_norm": 1.8195741176605225,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 10220
+    },
+    {
+      "epoch": 32.78846153846154,
+      "grad_norm": 1.4219516515731812,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 10230
+    },
+    {
+      "epoch": 32.82051282051282,
+      "grad_norm": 1.5619579553604126,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 10240
+    },
+    {
+      "epoch": 32.8525641025641,
+      "grad_norm": 1.213438868522644,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 10250
+    },
+    {
+      "epoch": 32.88461538461539,
+      "grad_norm": 1.8593034744262695,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 10260
+    },
+    {
+      "epoch": 32.916666666666664,
+      "grad_norm": 1.7562487125396729,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 10270
+    },
+    {
+      "epoch": 32.94871794871795,
+      "grad_norm": 1.6841872930526733,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 10280
+    },
+    {
+      "epoch": 32.98076923076923,
+      "grad_norm": 1.8166817426681519,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 10290
+    },
+    {
+      "epoch": 33.01282051282051,
+      "grad_norm": 1.7628390789031982,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 10300
+    },
+    {
+      "epoch": 33.044871794871796,
+      "grad_norm": 2.1147663593292236,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 10310
+    },
+    {
+      "epoch": 33.07692307692308,
+      "grad_norm": 1.9083815813064575,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 10320
+    },
+    {
+      "epoch": 33.10897435897436,
+      "grad_norm": 1.72088623046875,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 10330
+    },
+    {
+      "epoch": 33.14102564102564,
+      "grad_norm": 1.5126211643218994,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 10340
+    },
+    {
+      "epoch": 33.17307692307692,
+      "grad_norm": 1.5763148069381714,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 10350
+    },
+    {
+      "epoch": 33.205128205128204,
+      "grad_norm": 1.7654956579208374,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 10360
+    },
+    {
+      "epoch": 33.23717948717949,
+      "grad_norm": 1.7382022142410278,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 10370
+    },
+    {
+      "epoch": 33.26923076923077,
+      "grad_norm": 1.39749014377594,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 10380
+    },
+    {
+      "epoch": 33.30128205128205,
+      "grad_norm": 1.5260436534881592,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 10390
+    },
+    {
+      "epoch": 33.333333333333336,
+      "grad_norm": 1.7159744501113892,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 10400
+    },
+    {
+      "epoch": 33.36538461538461,
+      "grad_norm": 1.4590882062911987,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 10410
+    },
+    {
+      "epoch": 33.3974358974359,
+      "grad_norm": 1.6430726051330566,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 10420
+    },
+    {
+      "epoch": 33.42948717948718,
+      "grad_norm": 1.4286346435546875,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 10430
+    },
+    {
+      "epoch": 33.46153846153846,
+      "grad_norm": 1.4771382808685303,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 10440
+    },
+    {
+      "epoch": 33.493589743589745,
+      "grad_norm": 1.593119740486145,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 10450
+    },
+    {
+      "epoch": 33.52564102564103,
+      "grad_norm": 1.5224659442901611,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 10460
+    },
+    {
+      "epoch": 33.55769230769231,
+      "grad_norm": 1.524033546447754,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 10470
+    },
+    {
+      "epoch": 33.58974358974359,
+      "grad_norm": 1.5558611154556274,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 10480
+    },
+    {
+      "epoch": 33.62179487179487,
+      "grad_norm": 1.7433397769927979,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 10490
+    },
+    {
+      "epoch": 33.65384615384615,
+      "grad_norm": 1.615350365638733,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 10500
+    },
+    {
+      "epoch": 33.68589743589744,
+      "grad_norm": 1.7088472843170166,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 10510
+    },
+    {
+      "epoch": 33.717948717948715,
+      "grad_norm": 1.7948963642120361,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 10520
+    },
+    {
+      "epoch": 33.75,
+      "grad_norm": 1.5258243083953857,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 10530
+    },
+    {
+      "epoch": 33.782051282051285,
+      "grad_norm": 1.4905561208724976,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 10540
+    },
+    {
+      "epoch": 33.81410256410256,
+      "grad_norm": 1.451151967048645,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 10550
+    },
+    {
+      "epoch": 33.84615384615385,
+      "grad_norm": 1.9629911184310913,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 10560
+    },
+    {
+      "epoch": 33.87820512820513,
+      "grad_norm": 1.856055736541748,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 10570
+    },
+    {
+      "epoch": 33.91025641025641,
+      "grad_norm": 1.7479232549667358,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 10580
+    },
+    {
+      "epoch": 33.94230769230769,
+      "grad_norm": 1.484237551689148,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 10590
+    },
+    {
+      "epoch": 33.97435897435897,
+      "grad_norm": 1.5893609523773193,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 10600
+    },
+    {
+      "epoch": 34.006410256410255,
+      "grad_norm": 1.5691338777542114,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 10610
+    },
+    {
+      "epoch": 34.03846153846154,
+      "grad_norm": 1.7397927045822144,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 10620
+    },
+    {
+      "epoch": 34.07051282051282,
+      "grad_norm": 1.4689606428146362,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 10630
+    },
+    {
+      "epoch": 34.1025641025641,
+      "grad_norm": 1.6532474756240845,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 10640
+    },
+    {
+      "epoch": 34.13461538461539,
+      "grad_norm": 1.8410770893096924,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 10650
+    },
+    {
+      "epoch": 34.166666666666664,
+      "grad_norm": 1.8348032236099243,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 10660
+    },
+    {
+      "epoch": 34.19871794871795,
+      "grad_norm": 1.4887001514434814,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 10670
+    },
+    {
+      "epoch": 34.23076923076923,
+      "grad_norm": 1.4936854839324951,
+      "learning_rate": 0.0001,
+      "loss": 0.0191,
+      "step": 10680
+    },
+    {
+      "epoch": 34.26282051282051,
+      "grad_norm": 1.789055347442627,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 10690
+    },
+    {
+      "epoch": 34.294871794871796,
+      "grad_norm": 1.883852243423462,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 10700
+    },
+    {
+      "epoch": 34.32692307692308,
+      "grad_norm": 2.233919858932495,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 10710
+    },
+    {
+      "epoch": 34.35897435897436,
+      "grad_norm": 1.7235773801803589,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 10720
+    },
+    {
+      "epoch": 34.39102564102564,
+      "grad_norm": 2.107471466064453,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 10730
+    },
+    {
+      "epoch": 34.42307692307692,
+      "grad_norm": 2.2029969692230225,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 10740
+    },
+    {
+      "epoch": 34.455128205128204,
+      "grad_norm": 1.9910861253738403,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 10750
+    },
+    {
+      "epoch": 34.48717948717949,
+      "grad_norm": 1.961442232131958,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 10760
+    },
+    {
+      "epoch": 34.51923076923077,
+      "grad_norm": 1.7527135610580444,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 10770
+    },
+    {
+      "epoch": 34.55128205128205,
+      "grad_norm": 1.7521926164627075,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 10780
+    },
+    {
+      "epoch": 34.583333333333336,
+      "grad_norm": 1.80336332321167,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 10790
+    },
+    {
+      "epoch": 34.61538461538461,
+      "grad_norm": 2.0010595321655273,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 10800
+    },
+    {
+      "epoch": 34.6474358974359,
+      "grad_norm": 1.8658488988876343,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 10810
+    },
+    {
+      "epoch": 34.67948717948718,
+      "grad_norm": 1.960755467414856,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 10820
+    },
+    {
+      "epoch": 34.71153846153846,
+      "grad_norm": 1.6390820741653442,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 10830
+    },
+    {
+      "epoch": 34.743589743589745,
+      "grad_norm": 1.6425048112869263,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 10840
+    },
+    {
+      "epoch": 34.77564102564103,
+      "grad_norm": 1.676430344581604,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 10850
+    },
+    {
+      "epoch": 34.80769230769231,
+      "grad_norm": 1.679273009300232,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 10860
+    },
+    {
+      "epoch": 34.83974358974359,
+      "grad_norm": 1.7746658325195312,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 10870
+    },
+    {
+      "epoch": 34.87179487179487,
+      "grad_norm": 1.9075653553009033,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 10880
+    },
+    {
+      "epoch": 34.90384615384615,
+      "grad_norm": 1.8361091613769531,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 10890
+    },
+    {
+      "epoch": 34.93589743589744,
+      "grad_norm": 1.5293519496917725,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 10900
+    },
+    {
+      "epoch": 34.967948717948715,
+      "grad_norm": 1.6580746173858643,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 10910
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 1.6816896200180054,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 10920
+    },
+    {
+      "epoch": 35.032051282051285,
+      "grad_norm": 1.620327115058899,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 10930
+    },
+    {
+      "epoch": 35.06410256410256,
+      "grad_norm": 1.5595970153808594,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 10940
+    },
+    {
+      "epoch": 35.09615384615385,
+      "grad_norm": 1.4124690294265747,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 10950
+    },
+    {
+      "epoch": 35.12820512820513,
+      "grad_norm": 1.757730484008789,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 10960
+    },
+    {
+      "epoch": 35.16025641025641,
+      "grad_norm": 1.5117120742797852,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 10970
+    },
+    {
+      "epoch": 35.19230769230769,
+      "grad_norm": 2.1199913024902344,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 10980
+    },
+    {
+      "epoch": 35.22435897435897,
+      "grad_norm": 1.3876138925552368,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 10990
+    },
+    {
+      "epoch": 35.256410256410255,
+      "grad_norm": 1.8702820539474487,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 11000
+    },
+    {
+      "epoch": 35.28846153846154,
+      "grad_norm": 1.59559166431427,
+      "learning_rate": 0.0001,
+      "loss": 0.0168,
+      "step": 11010
+    },
+    {
+      "epoch": 35.32051282051282,
+      "grad_norm": 1.6516940593719482,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 11020
+    },
+    {
+      "epoch": 35.3525641025641,
+      "grad_norm": 1.7315785884857178,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 11030
+    },
+    {
+      "epoch": 35.38461538461539,
+      "grad_norm": 1.5910173654556274,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 11040
+    },
+    {
+      "epoch": 35.416666666666664,
+      "grad_norm": 1.4917683601379395,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 11050
+    },
+    {
+      "epoch": 35.44871794871795,
+      "grad_norm": 1.8072000741958618,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 11060
+    },
+    {
+      "epoch": 35.48076923076923,
+      "grad_norm": 1.5946332216262817,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 11070
+    },
+    {
+      "epoch": 35.51282051282051,
+      "grad_norm": 1.7303773164749146,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 11080
+    },
+    {
+      "epoch": 35.544871794871796,
+      "grad_norm": 1.653912901878357,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 11090
+    },
+    {
+      "epoch": 35.57692307692308,
+      "grad_norm": 1.6298012733459473,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 11100
+    },
+    {
+      "epoch": 35.60897435897436,
+      "grad_norm": 1.5533034801483154,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 11110
+    },
+    {
+      "epoch": 35.64102564102564,
+      "grad_norm": 1.4732062816619873,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 11120
+    },
+    {
+      "epoch": 35.67307692307692,
+      "grad_norm": 1.2958409786224365,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 11130
+    },
+    {
+      "epoch": 35.705128205128204,
+      "grad_norm": 1.5336447954177856,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 11140
+    },
+    {
+      "epoch": 35.73717948717949,
+      "grad_norm": 1.837249517440796,
+      "learning_rate": 0.0001,
+      "loss": 0.0247,
+      "step": 11150
+    },
+    {
+      "epoch": 35.76923076923077,
+      "grad_norm": 2.6507222652435303,
+      "learning_rate": 0.0001,
+      "loss": 0.0189,
+      "step": 11160
+    },
+    {
+      "epoch": 35.80128205128205,
+      "grad_norm": 2.099679708480835,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 11170
+    },
+    {
+      "epoch": 35.833333333333336,
+      "grad_norm": 2.455796241760254,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 11180
+    },
+    {
+      "epoch": 35.86538461538461,
+      "grad_norm": 2.0012834072113037,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 11190
+    },
+    {
+      "epoch": 35.8974358974359,
+      "grad_norm": 1.5275911092758179,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 11200
+    },
+    {
+      "epoch": 35.92948717948718,
+      "grad_norm": 1.6477304697036743,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 11210
+    },
+    {
+      "epoch": 35.96153846153846,
+      "grad_norm": 1.7257068157196045,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 11220
+    },
+    {
+      "epoch": 35.993589743589745,
+      "grad_norm": 1.8624606132507324,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 11230
+    },
+    {
+      "epoch": 36.02564102564103,
+      "grad_norm": 1.8021501302719116,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 11240
+    },
+    {
+      "epoch": 36.05769230769231,
+      "grad_norm": 1.873957633972168,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 11250
+    },
+    {
+      "epoch": 36.08974358974359,
+      "grad_norm": 1.6910736560821533,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 11260
+    },
+    {
+      "epoch": 36.12179487179487,
+      "grad_norm": 1.9445109367370605,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 11270
+    },
+    {
+      "epoch": 36.15384615384615,
+      "grad_norm": 1.7874106168746948,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 11280
+    },
+    {
+      "epoch": 36.18589743589744,
+      "grad_norm": 2.35673189163208,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 11290
+    },
+    {
+      "epoch": 36.217948717948715,
+      "grad_norm": 1.9054241180419922,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 11300
+    },
+    {
+      "epoch": 36.25,
+      "grad_norm": 1.7486035823822021,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 11310
+    },
+    {
+      "epoch": 36.282051282051285,
+      "grad_norm": 1.7607545852661133,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 11320
+    },
+    {
+      "epoch": 36.31410256410256,
+      "grad_norm": 1.4948348999023438,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 11330
+    },
+    {
+      "epoch": 36.34615384615385,
+      "grad_norm": 1.6036207675933838,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 11340
+    },
+    {
+      "epoch": 36.37820512820513,
+      "grad_norm": 1.7479506731033325,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 11350
+    },
+    {
+      "epoch": 36.41025641025641,
+      "grad_norm": 1.9252262115478516,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 11360
+    },
+    {
+      "epoch": 36.44230769230769,
+      "grad_norm": 1.8868968486785889,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 11370
+    },
+    {
+      "epoch": 36.47435897435897,
+      "grad_norm": 1.7382066249847412,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 11380
+    },
+    {
+      "epoch": 36.506410256410255,
+      "grad_norm": 1.7188570499420166,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 11390
+    },
+    {
+      "epoch": 36.53846153846154,
+      "grad_norm": 1.6367896795272827,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 11400
+    },
+    {
+      "epoch": 36.57051282051282,
+      "grad_norm": 1.3695664405822754,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 11410
+    },
+    {
+      "epoch": 36.6025641025641,
+      "grad_norm": 1.7137398719787598,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 11420
+    },
+    {
+      "epoch": 36.63461538461539,
+      "grad_norm": 1.5719554424285889,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 11430
+    },
+    {
+      "epoch": 36.666666666666664,
+      "grad_norm": 1.4912947416305542,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 11440
+    },
+    {
+      "epoch": 36.69871794871795,
+      "grad_norm": 1.134800910949707,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 11450
+    },
+    {
+      "epoch": 36.73076923076923,
+      "grad_norm": 1.1075400114059448,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 11460
+    },
+    {
+      "epoch": 36.76282051282051,
+      "grad_norm": 1.5913994312286377,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 11470
+    },
+    {
+      "epoch": 36.794871794871796,
+      "grad_norm": 1.5238467454910278,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 11480
+    },
+    {
+      "epoch": 36.82692307692308,
+      "grad_norm": 1.5352369546890259,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 11490
+    },
+    {
+      "epoch": 36.85897435897436,
+      "grad_norm": 1.3048542737960815,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 11500
+    },
+    {
+      "epoch": 36.89102564102564,
+      "grad_norm": 1.4833670854568481,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 11510
+    },
+    {
+      "epoch": 36.92307692307692,
+      "grad_norm": 1.3695210218429565,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 11520
+    },
+    {
+      "epoch": 36.955128205128204,
+      "grad_norm": 1.5082851648330688,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 11530
+    },
+    {
+      "epoch": 36.98717948717949,
+      "grad_norm": 1.8328081369400024,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 11540
+    },
+    {
+      "epoch": 37.01923076923077,
+      "grad_norm": 1.6446616649627686,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 11550
+    },
+    {
+      "epoch": 37.05128205128205,
+      "grad_norm": 1.431779384613037,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 11560
+    },
+    {
+      "epoch": 37.083333333333336,
+      "grad_norm": 1.397458553314209,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 11570
+    },
+    {
+      "epoch": 37.11538461538461,
+      "grad_norm": 1.353140950202942,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 11580
+    },
+    {
+      "epoch": 37.1474358974359,
+      "grad_norm": 1.465466022491455,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 11590
+    },
+    {
+      "epoch": 37.17948717948718,
+      "grad_norm": 1.4127053022384644,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 11600
+    },
+    {
+      "epoch": 37.21153846153846,
+      "grad_norm": 1.5576893091201782,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 11610
+    },
+    {
+      "epoch": 37.243589743589745,
+      "grad_norm": 1.4888213872909546,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 11620
+    },
+    {
+      "epoch": 37.27564102564103,
+      "grad_norm": 1.484317660331726,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 11630
+    },
+    {
+      "epoch": 37.30769230769231,
+      "grad_norm": 1.5621172189712524,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 11640
+    },
+    {
+      "epoch": 37.33974358974359,
+      "grad_norm": 1.4635461568832397,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 11650
+    },
+    {
+      "epoch": 37.37179487179487,
+      "grad_norm": 1.5379810333251953,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 11660
+    },
+    {
+      "epoch": 37.40384615384615,
+      "grad_norm": 1.707191824913025,
+      "learning_rate": 0.0001,
+      "loss": 0.018,
+      "step": 11670
+    },
+    {
+      "epoch": 37.43589743589744,
+      "grad_norm": 1.6384049654006958,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 11680
+    },
+    {
+      "epoch": 37.467948717948715,
+      "grad_norm": 1.946465015411377,
+      "learning_rate": 0.0001,
+      "loss": 0.0177,
+      "step": 11690
+    },
+    {
+      "epoch": 37.5,
+      "grad_norm": 1.6456594467163086,
+      "learning_rate": 0.0001,
+      "loss": 0.0192,
+      "step": 11700
+    },
+    {
+      "epoch": 37.532051282051285,
+      "grad_norm": 1.7535289525985718,
+      "learning_rate": 0.0001,
+      "loss": 0.0202,
+      "step": 11710
+    },
+    {
+      "epoch": 37.56410256410256,
+      "grad_norm": 1.3954745531082153,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 11720
+    },
+    {
+      "epoch": 37.59615384615385,
+      "grad_norm": 1.6023287773132324,
+      "learning_rate": 0.0001,
+      "loss": 0.0172,
+      "step": 11730
+    },
+    {
+      "epoch": 37.62820512820513,
+      "grad_norm": 1.53830885887146,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 11740
+    },
+    {
+      "epoch": 37.66025641025641,
+      "grad_norm": 1.7311429977416992,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 11750
+    },
+    {
+      "epoch": 37.69230769230769,
+      "grad_norm": 1.7503981590270996,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 11760
+    },
+    {
+      "epoch": 37.72435897435897,
+      "grad_norm": 1.5489487648010254,
+      "learning_rate": 0.0001,
+      "loss": 0.0173,
+      "step": 11770
+    },
+    {
+      "epoch": 37.756410256410255,
+      "grad_norm": 1.6182324886322021,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 11780
+    },
+    {
+      "epoch": 37.78846153846154,
+      "grad_norm": 1.662867546081543,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 11790
+    },
+    {
+      "epoch": 37.82051282051282,
+      "grad_norm": 1.6210496425628662,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 11800
+    },
+    {
+      "epoch": 37.8525641025641,
+      "grad_norm": 1.318446159362793,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 11810
+    },
+    {
+      "epoch": 37.88461538461539,
+      "grad_norm": 1.5441850423812866,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 11820
+    },
+    {
+      "epoch": 37.916666666666664,
+      "grad_norm": 1.622084140777588,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 11830
+    },
+    {
+      "epoch": 37.94871794871795,
+      "grad_norm": 1.637273907661438,
+      "learning_rate": 0.0001,
+      "loss": 0.0179,
+      "step": 11840
+    },
+    {
+      "epoch": 37.98076923076923,
+      "grad_norm": 1.4368484020233154,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 11850
+    },
+    {
+      "epoch": 38.01282051282051,
+      "grad_norm": 1.7037521600723267,
+      "learning_rate": 0.0001,
+      "loss": 0.0181,
+      "step": 11860
+    },
+    {
+      "epoch": 38.044871794871796,
+      "grad_norm": 1.7442257404327393,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 11870
+    },
+    {
+      "epoch": 38.07692307692308,
+      "grad_norm": 1.4991378784179688,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 11880
+    },
+    {
+      "epoch": 38.10897435897436,
+      "grad_norm": 1.3807876110076904,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 11890
+    },
+    {
+      "epoch": 38.14102564102564,
+      "grad_norm": 1.401655912399292,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 11900
+    },
+    {
+      "epoch": 38.17307692307692,
+      "grad_norm": 1.5714393854141235,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 11910
+    },
+    {
+      "epoch": 38.205128205128204,
+      "grad_norm": 1.3692444562911987,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 11920
+    },
+    {
+      "epoch": 38.23717948717949,
+      "grad_norm": 1.443835973739624,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 11930
+    },
+    {
+      "epoch": 38.26923076923077,
+      "grad_norm": 1.58891761302948,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 11940
+    },
+    {
+      "epoch": 38.30128205128205,
+      "grad_norm": 1.58780038356781,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 11950
+    },
+    {
+      "epoch": 38.333333333333336,
+      "grad_norm": 1.5330605506896973,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 11960
+    },
+    {
+      "epoch": 38.36538461538461,
+      "grad_norm": 1.8901503086090088,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 11970
+    },
+    {
+      "epoch": 38.3974358974359,
+      "grad_norm": 1.5068856477737427,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 11980
+    },
+    {
+      "epoch": 38.42948717948718,
+      "grad_norm": 1.5701942443847656,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 11990
+    },
+    {
+      "epoch": 38.46153846153846,
+      "grad_norm": 1.1868116855621338,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 12000
+    },
+    {
+      "epoch": 38.493589743589745,
+      "grad_norm": 1.7219698429107666,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 12010
+    },
+    {
+      "epoch": 38.52564102564103,
+      "grad_norm": 1.4672821760177612,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 12020
+    },
+    {
+      "epoch": 38.55769230769231,
+      "grad_norm": 1.7000876665115356,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 12030
+    },
+    {
+      "epoch": 38.58974358974359,
+      "grad_norm": 1.2536815404891968,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 12040
+    },
+    {
+      "epoch": 38.62179487179487,
+      "grad_norm": 1.7941112518310547,
+      "learning_rate": 0.0001,
+      "loss": 0.0175,
+      "step": 12050
+    },
+    {
+      "epoch": 38.65384615384615,
+      "grad_norm": 1.2826061248779297,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 12060
+    },
+    {
+      "epoch": 38.68589743589744,
+      "grad_norm": 1.3117648363113403,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 12070
+    },
+    {
+      "epoch": 38.717948717948715,
+      "grad_norm": 1.3213527202606201,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 12080
+    },
+    {
+      "epoch": 38.75,
+      "grad_norm": 1.3278855085372925,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 12090
+    },
+    {
+      "epoch": 38.782051282051285,
+      "grad_norm": 1.3082314729690552,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 12100
+    },
+    {
+      "epoch": 38.81410256410256,
+      "grad_norm": 1.3913761377334595,
+      "learning_rate": 0.0001,
+      "loss": 0.0163,
+      "step": 12110
+    },
+    {
+      "epoch": 38.84615384615385,
+      "grad_norm": 1.5182521343231201,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 12120
+    },
+    {
+      "epoch": 38.87820512820513,
+      "grad_norm": 1.5669957399368286,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 12130
+    },
+    {
+      "epoch": 38.91025641025641,
+      "grad_norm": 1.4659616947174072,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 12140
+    },
+    {
+      "epoch": 38.94230769230769,
+      "grad_norm": 1.2350541353225708,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 12150
+    },
+    {
+      "epoch": 38.97435897435897,
+      "grad_norm": 1.331242561340332,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 12160
+    },
+    {
+      "epoch": 39.006410256410255,
+      "grad_norm": 1.556071162223816,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 12170
+    },
+    {
+      "epoch": 39.03846153846154,
+      "grad_norm": 1.533072829246521,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 12180
+    },
+    {
+      "epoch": 39.07051282051282,
+      "grad_norm": 2.0657548904418945,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 12190
+    },
+    {
+      "epoch": 39.1025641025641,
+      "grad_norm": 1.8604674339294434,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 12200
+    },
+    {
+      "epoch": 39.13461538461539,
+      "grad_norm": 1.7973824739456177,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 12210
+    },
+    {
+      "epoch": 39.166666666666664,
+      "grad_norm": 1.541810154914856,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 12220
+    },
+    {
+      "epoch": 39.19871794871795,
+      "grad_norm": 1.5012316703796387,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 12230
+    },
+    {
+      "epoch": 39.23076923076923,
+      "grad_norm": 1.777346134185791,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 12240
+    },
+    {
+      "epoch": 39.26282051282051,
+      "grad_norm": 1.8042408227920532,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 12250
+    },
+    {
+      "epoch": 39.294871794871796,
+      "grad_norm": 1.703172206878662,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 12260
+    },
+    {
+      "epoch": 39.32692307692308,
+      "grad_norm": 1.238478660583496,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 12270
+    },
+    {
+      "epoch": 39.35897435897436,
+      "grad_norm": 1.5187033414840698,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 12280
+    },
+    {
+      "epoch": 39.39102564102564,
+      "grad_norm": 1.4580196142196655,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 12290
+    },
+    {
+      "epoch": 39.42307692307692,
+      "grad_norm": 1.401454210281372,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 12300
+    },
+    {
+      "epoch": 39.455128205128204,
+      "grad_norm": 1.2959507703781128,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 12310
+    },
+    {
+      "epoch": 39.48717948717949,
+      "grad_norm": 1.4067796468734741,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 12320
+    },
+    {
+      "epoch": 39.51923076923077,
+      "grad_norm": 1.2092689275741577,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 12330
+    },
+    {
+      "epoch": 39.55128205128205,
+      "grad_norm": 1.3149386644363403,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 12340
+    },
+    {
+      "epoch": 39.583333333333336,
+      "grad_norm": 1.13405442237854,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 12350
+    },
+    {
+      "epoch": 39.61538461538461,
+      "grad_norm": 1.6851937770843506,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 12360
+    },
+    {
+      "epoch": 39.6474358974359,
+      "grad_norm": 1.260632038116455,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 12370
+    },
+    {
+      "epoch": 39.67948717948718,
+      "grad_norm": 1.4260509014129639,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 12380
+    },
+    {
+      "epoch": 39.71153846153846,
+      "grad_norm": 1.2130883932113647,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 12390
+    },
+    {
+      "epoch": 39.743589743589745,
+      "grad_norm": 1.6260617971420288,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 12400
+    },
+    {
+      "epoch": 39.77564102564103,
+      "grad_norm": 1.7585242986679077,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 12410
+    },
+    {
+      "epoch": 39.80769230769231,
+      "grad_norm": 1.3568482398986816,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 12420
+    },
+    {
+      "epoch": 39.83974358974359,
+      "grad_norm": 1.4918653964996338,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 12430
+    },
+    {
+      "epoch": 39.87179487179487,
+      "grad_norm": 1.4998669624328613,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 12440
+    },
+    {
+      "epoch": 39.90384615384615,
+      "grad_norm": 1.6185706853866577,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 12450
+    },
+    {
+      "epoch": 39.93589743589744,
+      "grad_norm": 1.634658932685852,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 12460
+    },
+    {
+      "epoch": 39.967948717948715,
+      "grad_norm": 1.342952013015747,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 12470
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 1.2517924308776855,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 12480
+    },
+    {
+      "epoch": 40.032051282051285,
+      "grad_norm": 1.3209261894226074,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 12490
+    },
+    {
+      "epoch": 40.06410256410256,
+      "grad_norm": 1.3419283628463745,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 12500
+    },
+    {
+      "epoch": 40.09615384615385,
+      "grad_norm": 1.5633323192596436,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 12510
+    },
+    {
+      "epoch": 40.12820512820513,
+      "grad_norm": 1.7954598665237427,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 12520
+    },
+    {
+      "epoch": 40.16025641025641,
+      "grad_norm": 1.719743251800537,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 12530
+    },
+    {
+      "epoch": 40.19230769230769,
+      "grad_norm": 1.7671940326690674,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 12540
+    },
+    {
+      "epoch": 40.22435897435897,
+      "grad_norm": 1.5557199716567993,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 12550
+    },
+    {
+      "epoch": 40.256410256410255,
+      "grad_norm": 1.4633153676986694,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 12560
+    },
+    {
+      "epoch": 40.28846153846154,
+      "grad_norm": 1.4462223052978516,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 12570
+    },
+    {
+      "epoch": 40.32051282051282,
+      "grad_norm": 1.4049561023712158,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 12580
+    },
+    {
+      "epoch": 40.3525641025641,
+      "grad_norm": 1.6395725011825562,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 12590
+    },
+    {
+      "epoch": 40.38461538461539,
+      "grad_norm": 1.4848397970199585,
+      "learning_rate": 0.0001,
+      "loss": 0.0169,
+      "step": 12600
+    },
+    {
+      "epoch": 40.416666666666664,
+      "grad_norm": 1.537334680557251,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 12610
+    },
+    {
+      "epoch": 40.44871794871795,
+      "grad_norm": 1.7367362976074219,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 12620
+    },
+    {
+      "epoch": 40.48076923076923,
+      "grad_norm": 1.3062987327575684,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 12630
+    },
+    {
+      "epoch": 40.51282051282051,
+      "grad_norm": 1.4020715951919556,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 12640
+    },
+    {
+      "epoch": 40.544871794871796,
+      "grad_norm": 1.4328738451004028,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 12650
+    },
+    {
+      "epoch": 40.57692307692308,
+      "grad_norm": 1.7578388452529907,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 12660
+    },
+    {
+      "epoch": 40.60897435897436,
+      "grad_norm": 1.6971560716629028,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 12670
+    },
+    {
+      "epoch": 40.64102564102564,
+      "grad_norm": 1.2680612802505493,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 12680
+    },
+    {
+      "epoch": 40.67307692307692,
+      "grad_norm": 1.5615694522857666,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 12690
+    },
+    {
+      "epoch": 40.705128205128204,
+      "grad_norm": 1.6551638841629028,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 12700
+    },
+    {
+      "epoch": 40.73717948717949,
+      "grad_norm": 1.3193349838256836,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 12710
+    },
+    {
+      "epoch": 40.76923076923077,
+      "grad_norm": 1.321285367012024,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 12720
+    },
+    {
+      "epoch": 40.80128205128205,
+      "grad_norm": 1.5391532182693481,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 12730
+    },
+    {
+      "epoch": 40.833333333333336,
+      "grad_norm": 1.4566560983657837,
+      "learning_rate": 0.0001,
+      "loss": 0.0174,
+      "step": 12740
+    },
+    {
+      "epoch": 40.86538461538461,
+      "grad_norm": 1.3870965242385864,
+      "learning_rate": 0.0001,
+      "loss": 0.0185,
+      "step": 12750
+    },
+    {
+      "epoch": 40.8974358974359,
+      "grad_norm": 1.7458410263061523,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 12760
+    },
+    {
+      "epoch": 40.92948717948718,
+      "grad_norm": 1.3945047855377197,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 12770
+    },
+    {
+      "epoch": 40.96153846153846,
+      "grad_norm": 1.3938902616500854,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 12780
+    },
+    {
+      "epoch": 40.993589743589745,
+      "grad_norm": 1.6878256797790527,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 12790
+    },
+    {
+      "epoch": 41.02564102564103,
+      "grad_norm": 1.1318292617797852,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 12800
+    },
+    {
+      "epoch": 41.05769230769231,
+      "grad_norm": 1.2203469276428223,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 12810
+    },
+    {
+      "epoch": 41.08974358974359,
+      "grad_norm": 1.328615665435791,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 12820
+    },
+    {
+      "epoch": 41.12179487179487,
+      "grad_norm": 1.3646776676177979,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 12830
+    },
+    {
+      "epoch": 41.15384615384615,
+      "grad_norm": 1.4550570249557495,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 12840
+    },
+    {
+      "epoch": 41.18589743589744,
+      "grad_norm": 1.5695714950561523,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 12850
+    },
+    {
+      "epoch": 41.217948717948715,
+      "grad_norm": 1.6237770318984985,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 12860
+    },
+    {
+      "epoch": 41.25,
+      "grad_norm": 1.4848991632461548,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 12870
+    },
+    {
+      "epoch": 41.282051282051285,
+      "grad_norm": 1.5800526142120361,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 12880
+    },
+    {
+      "epoch": 41.31410256410256,
+      "grad_norm": 1.51021146774292,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 12890
+    },
+    {
+      "epoch": 41.34615384615385,
+      "grad_norm": 1.8164012432098389,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 12900
+    },
+    {
+      "epoch": 41.37820512820513,
+      "grad_norm": 1.4754937887191772,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 12910
+    },
+    {
+      "epoch": 41.41025641025641,
+      "grad_norm": 1.1023662090301514,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 12920
+    },
+    {
+      "epoch": 41.44230769230769,
+      "grad_norm": 1.262595772743225,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 12930
+    },
+    {
+      "epoch": 41.47435897435897,
+      "grad_norm": 1.2925360202789307,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 12940
+    },
+    {
+      "epoch": 41.506410256410255,
+      "grad_norm": 1.1823949813842773,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 12950
+    },
+    {
+      "epoch": 41.53846153846154,
+      "grad_norm": 1.623037576675415,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 12960
+    },
+    {
+      "epoch": 41.57051282051282,
+      "grad_norm": 1.3906683921813965,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 12970
+    },
+    {
+      "epoch": 41.6025641025641,
+      "grad_norm": 1.5441356897354126,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 12980
+    },
+    {
+      "epoch": 41.63461538461539,
+      "grad_norm": 1.2952722311019897,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 12990
+    },
+    {
+      "epoch": 41.666666666666664,
+      "grad_norm": 2.0250637531280518,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 13000
+    },
+    {
+      "epoch": 41.69871794871795,
+      "grad_norm": 1.3920475244522095,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 13010
+    },
+    {
+      "epoch": 41.73076923076923,
+      "grad_norm": 1.2069278955459595,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 13020
+    },
+    {
+      "epoch": 41.76282051282051,
+      "grad_norm": 1.4182066917419434,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 13030
+    },
+    {
+      "epoch": 41.794871794871796,
+      "grad_norm": 1.4021706581115723,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 13040
+    },
+    {
+      "epoch": 41.82692307692308,
+      "grad_norm": 1.264643669128418,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 13050
+    },
+    {
+      "epoch": 41.85897435897436,
+      "grad_norm": 1.7241528034210205,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 13060
+    },
+    {
+      "epoch": 41.89102564102564,
+      "grad_norm": 1.5512839555740356,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 13070
+    },
+    {
+      "epoch": 41.92307692307692,
+      "grad_norm": 1.5593472719192505,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 13080
+    },
+    {
+      "epoch": 41.955128205128204,
+      "grad_norm": 1.3050546646118164,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 13090
+    },
+    {
+      "epoch": 41.98717948717949,
+      "grad_norm": 1.4949992895126343,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 13100
+    },
+    {
+      "epoch": 42.01923076923077,
+      "grad_norm": 1.367732048034668,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 13110
+    },
+    {
+      "epoch": 42.05128205128205,
+      "grad_norm": 1.5874683856964111,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 13120
+    },
+    {
+      "epoch": 42.083333333333336,
+      "grad_norm": 1.2341395616531372,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 13130
+    },
+    {
+      "epoch": 42.11538461538461,
+      "grad_norm": 1.3235291242599487,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 13140
+    },
+    {
+      "epoch": 42.1474358974359,
+      "grad_norm": 1.10971200466156,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 13150
+    },
+    {
+      "epoch": 42.17948717948718,
+      "grad_norm": 1.1652225255966187,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 13160
+    },
+    {
+      "epoch": 42.21153846153846,
+      "grad_norm": 1.4924333095550537,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 13170
+    },
+    {
+      "epoch": 42.243589743589745,
+      "grad_norm": 1.535416841506958,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 13180
+    },
+    {
+      "epoch": 42.27564102564103,
+      "grad_norm": 1.278149127960205,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 13190
+    },
+    {
+      "epoch": 42.30769230769231,
+      "grad_norm": 1.4892760515213013,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 13200
+    },
+    {
+      "epoch": 42.33974358974359,
+      "grad_norm": 1.361525297164917,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 13210
+    },
+    {
+      "epoch": 42.37179487179487,
+      "grad_norm": 1.4055627584457397,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 13220
+    },
+    {
+      "epoch": 42.40384615384615,
+      "grad_norm": 1.5213630199432373,
+      "learning_rate": 0.0001,
+      "loss": 0.0155,
+      "step": 13230
+    },
+    {
+      "epoch": 42.43589743589744,
+      "grad_norm": 1.263697624206543,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 13240
+    },
+    {
+      "epoch": 42.467948717948715,
+      "grad_norm": 1.5194586515426636,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 13250
+    },
+    {
+      "epoch": 42.5,
+      "grad_norm": 1.1274211406707764,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 13260
+    },
+    {
+      "epoch": 42.532051282051285,
+      "grad_norm": 1.4212441444396973,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 13270
+    },
+    {
+      "epoch": 42.56410256410256,
+      "grad_norm": 1.2597031593322754,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 13280
+    },
+    {
+      "epoch": 42.59615384615385,
+      "grad_norm": 1.4647955894470215,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 13290
+    },
+    {
+      "epoch": 42.62820512820513,
+      "grad_norm": 1.4848425388336182,
+      "learning_rate": 0.0001,
+      "loss": 0.0156,
+      "step": 13300
+    },
+    {
+      "epoch": 42.66025641025641,
+      "grad_norm": 1.567398190498352,
+      "learning_rate": 0.0001,
+      "loss": 0.0164,
+      "step": 13310
+    },
+    {
+      "epoch": 42.69230769230769,
+      "grad_norm": 1.3729661703109741,
+      "learning_rate": 0.0001,
+      "loss": 0.0158,
+      "step": 13320
+    },
+    {
+      "epoch": 42.72435897435897,
+      "grad_norm": 1.560835361480713,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 13330
+    },
+    {
+      "epoch": 42.756410256410255,
+      "grad_norm": 1.4814984798431396,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 13340
+    },
+    {
+      "epoch": 42.78846153846154,
+      "grad_norm": 1.7237474918365479,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 13350
+    },
+    {
+      "epoch": 42.82051282051282,
+      "grad_norm": 1.1336373090744019,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 13360
+    },
+    {
+      "epoch": 42.8525641025641,
+      "grad_norm": 1.4071494340896606,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 13370
+    },
+    {
+      "epoch": 42.88461538461539,
+      "grad_norm": 1.1990396976470947,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 13380
+    },
+    {
+      "epoch": 42.916666666666664,
+      "grad_norm": 1.4490772485733032,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 13390
+    },
+    {
+      "epoch": 42.94871794871795,
+      "grad_norm": 1.7106400728225708,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 13400
+    },
+    {
+      "epoch": 42.98076923076923,
+      "grad_norm": 1.6688423156738281,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 13410
+    },
+    {
+      "epoch": 43.01282051282051,
+      "grad_norm": 1.2650384902954102,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 13420
+    },
+    {
+      "epoch": 43.044871794871796,
+      "grad_norm": 1.3470165729522705,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 13430
+    },
+    {
+      "epoch": 43.07692307692308,
+      "grad_norm": 1.5844476222991943,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 13440
+    },
+    {
+      "epoch": 43.10897435897436,
+      "grad_norm": 1.3076056241989136,
+      "learning_rate": 0.0001,
+      "loss": 0.0161,
+      "step": 13450
+    },
+    {
+      "epoch": 43.14102564102564,
+      "grad_norm": 1.4139776229858398,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 13460
+    },
+    {
+      "epoch": 43.17307692307692,
+      "grad_norm": 1.3998370170593262,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 13470
+    },
+    {
+      "epoch": 43.205128205128204,
+      "grad_norm": 1.4907950162887573,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 13480
+    },
+    {
+      "epoch": 43.23717948717949,
+      "grad_norm": 1.423680067062378,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 13490
+    },
+    {
+      "epoch": 43.26923076923077,
+      "grad_norm": 1.1949522495269775,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 13500
+    },
+    {
+      "epoch": 43.30128205128205,
+      "grad_norm": 1.57328200340271,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 13510
+    },
+    {
+      "epoch": 43.333333333333336,
+      "grad_norm": 1.3706156015396118,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 13520
+    },
+    {
+      "epoch": 43.36538461538461,
+      "grad_norm": 1.5411814451217651,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 13530
+    },
+    {
+      "epoch": 43.3974358974359,
+      "grad_norm": 1.3211857080459595,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 13540
+    },
+    {
+      "epoch": 43.42948717948718,
+      "grad_norm": 1.1519310474395752,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 13550
+    },
+    {
+      "epoch": 43.46153846153846,
+      "grad_norm": 1.5038219690322876,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 13560
+    },
+    {
+      "epoch": 43.493589743589745,
+      "grad_norm": 1.1524192094802856,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 13570
+    },
+    {
+      "epoch": 43.52564102564103,
+      "grad_norm": 1.4315845966339111,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 13580
+    },
+    {
+      "epoch": 43.55769230769231,
+      "grad_norm": 1.416010856628418,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 13590
+    },
+    {
+      "epoch": 43.58974358974359,
+      "grad_norm": 1.250651478767395,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 13600
+    },
+    {
+      "epoch": 43.62179487179487,
+      "grad_norm": 1.3116458654403687,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 13610
+    },
+    {
+      "epoch": 43.65384615384615,
+      "grad_norm": 1.2933439016342163,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 13620
+    },
+    {
+      "epoch": 43.68589743589744,
+      "grad_norm": 1.271322250366211,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 13630
+    },
+    {
+      "epoch": 43.717948717948715,
+      "grad_norm": 1.8542050123214722,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 13640
+    },
+    {
+      "epoch": 43.75,
+      "grad_norm": 1.4845478534698486,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 13650
+    },
+    {
+      "epoch": 43.782051282051285,
+      "grad_norm": 1.6054221391677856,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 13660
+    },
+    {
+      "epoch": 43.81410256410256,
+      "grad_norm": 1.2341477870941162,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 13670
+    },
+    {
+      "epoch": 43.84615384615385,
+      "grad_norm": 1.5661543607711792,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 13680
+    },
+    {
+      "epoch": 43.87820512820513,
+      "grad_norm": 1.5912723541259766,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 13690
+    },
+    {
+      "epoch": 43.91025641025641,
+      "grad_norm": 1.1883914470672607,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 13700
+    },
+    {
+      "epoch": 43.94230769230769,
+      "grad_norm": 1.873012661933899,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 13710
+    },
+    {
+      "epoch": 43.97435897435897,
+      "grad_norm": 1.815854787826538,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 13720
+    },
+    {
+      "epoch": 44.006410256410255,
+      "grad_norm": 1.9235273599624634,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 13730
+    },
+    {
+      "epoch": 44.03846153846154,
+      "grad_norm": 1.6557371616363525,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 13740
+    },
+    {
+      "epoch": 44.07051282051282,
+      "grad_norm": 1.6052653789520264,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 13750
+    },
+    {
+      "epoch": 44.1025641025641,
+      "grad_norm": 1.5591837167739868,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 13760
+    },
+    {
+      "epoch": 44.13461538461539,
+      "grad_norm": 1.6697862148284912,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 13770
+    },
+    {
+      "epoch": 44.166666666666664,
+      "grad_norm": 1.4162569046020508,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 13780
+    },
+    {
+      "epoch": 44.19871794871795,
+      "grad_norm": 1.3389195203781128,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 13790
+    },
+    {
+      "epoch": 44.23076923076923,
+      "grad_norm": 1.5349199771881104,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 13800
+    },
+    {
+      "epoch": 44.26282051282051,
+      "grad_norm": 1.4865868091583252,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 13810
+    },
+    {
+      "epoch": 44.294871794871796,
+      "grad_norm": 1.282641887664795,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 13820
+    },
+    {
+      "epoch": 44.32692307692308,
+      "grad_norm": 1.8219002485275269,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 13830
+    },
+    {
+      "epoch": 44.35897435897436,
+      "grad_norm": 1.8339409828186035,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 13840
+    },
+    {
+      "epoch": 44.39102564102564,
+      "grad_norm": 1.3240617513656616,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 13850
+    },
+    {
+      "epoch": 44.42307692307692,
+      "grad_norm": 1.5358703136444092,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 13860
+    },
+    {
+      "epoch": 44.455128205128204,
+      "grad_norm": 1.8734097480773926,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 13870
+    },
+    {
+      "epoch": 44.48717948717949,
+      "grad_norm": 1.7670713663101196,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 13880
+    },
+    {
+      "epoch": 44.51923076923077,
+      "grad_norm": 2.054593563079834,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 13890
+    },
+    {
+      "epoch": 44.55128205128205,
+      "grad_norm": 2.046670436859131,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 13900
+    },
+    {
+      "epoch": 44.583333333333336,
+      "grad_norm": 2.0899605751037598,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 13910
+    },
+    {
+      "epoch": 44.61538461538461,
+      "grad_norm": 1.6416637897491455,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 13920
+    },
+    {
+      "epoch": 44.6474358974359,
+      "grad_norm": 1.4238673448562622,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 13930
+    },
+    {
+      "epoch": 44.67948717948718,
+      "grad_norm": 1.4182482957839966,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 13940
+    },
+    {
+      "epoch": 44.71153846153846,
+      "grad_norm": 1.5908417701721191,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 13950
+    },
+    {
+      "epoch": 44.743589743589745,
+      "grad_norm": 1.6526551246643066,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 13960
+    },
+    {
+      "epoch": 44.77564102564103,
+      "grad_norm": 1.6272114515304565,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 13970
+    },
+    {
+      "epoch": 44.80769230769231,
+      "grad_norm": 1.2589800357818604,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 13980
+    },
+    {
+      "epoch": 44.83974358974359,
+      "grad_norm": 1.5461876392364502,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 13990
+    },
+    {
+      "epoch": 44.87179487179487,
+      "grad_norm": 1.1033819913864136,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 14000
+    },
+    {
+      "epoch": 44.90384615384615,
+      "grad_norm": 1.5191293954849243,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 14010
+    },
+    {
+      "epoch": 44.93589743589744,
+      "grad_norm": 1.4580175876617432,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 14020
+    },
+    {
+      "epoch": 44.967948717948715,
+      "grad_norm": 1.3626344203948975,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 14030
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 0.9937858581542969,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 14040
+    },
+    {
+      "epoch": 45.032051282051285,
+      "grad_norm": 1.299930214881897,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 14050
+    },
+    {
+      "epoch": 45.06410256410256,
+      "grad_norm": 0.9954428672790527,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 14060
+    },
+    {
+      "epoch": 45.09615384615385,
+      "grad_norm": 1.4350980520248413,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 14070
+    },
+    {
+      "epoch": 45.12820512820513,
+      "grad_norm": 0.9185075759887695,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 14080
+    },
+    {
+      "epoch": 45.16025641025641,
+      "grad_norm": 0.9538195729255676,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 14090
+    },
+    {
+      "epoch": 45.19230769230769,
+      "grad_norm": 1.2636340856552124,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 14100
+    },
+    {
+      "epoch": 45.22435897435897,
+      "grad_norm": 1.201856255531311,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 14110
+    },
+    {
+      "epoch": 45.256410256410255,
+      "grad_norm": 1.4358103275299072,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 14120
+    },
+    {
+      "epoch": 45.28846153846154,
+      "grad_norm": 1.2462881803512573,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 14130
+    },
+    {
+      "epoch": 45.32051282051282,
+      "grad_norm": 1.4640687704086304,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 14140
+    },
+    {
+      "epoch": 45.3525641025641,
+      "grad_norm": 1.1967284679412842,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 14150
+    },
+    {
+      "epoch": 45.38461538461539,
+      "grad_norm": 2.1066782474517822,
+      "learning_rate": 0.0001,
+      "loss": 0.0157,
+      "step": 14160
+    },
+    {
+      "epoch": 45.416666666666664,
+      "grad_norm": 1.9150457382202148,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 14170
+    },
+    {
+      "epoch": 45.44871794871795,
+      "grad_norm": 1.1601903438568115,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 14180
+    },
+    {
+      "epoch": 45.48076923076923,
+      "grad_norm": 0.9451940655708313,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 14190
+    },
+    {
+      "epoch": 45.51282051282051,
+      "grad_norm": 1.444657802581787,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 14200
+    },
+    {
+      "epoch": 45.544871794871796,
+      "grad_norm": 1.4755773544311523,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 14210
+    },
+    {
+      "epoch": 45.57692307692308,
+      "grad_norm": 1.7286694049835205,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 14220
+    },
+    {
+      "epoch": 45.60897435897436,
+      "grad_norm": 1.3827892541885376,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 14230
+    },
+    {
+      "epoch": 45.64102564102564,
+      "grad_norm": 1.382319688796997,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 14240
+    },
+    {
+      "epoch": 45.67307692307692,
+      "grad_norm": 1.4534835815429688,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 14250
+    },
+    {
+      "epoch": 45.705128205128204,
+      "grad_norm": 1.5834344625473022,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 14260
+    },
+    {
+      "epoch": 45.73717948717949,
+      "grad_norm": 1.742770791053772,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 14270
+    },
+    {
+      "epoch": 45.76923076923077,
+      "grad_norm": 1.3905388116836548,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 14280
+    },
+    {
+      "epoch": 45.80128205128205,
+      "grad_norm": 1.4791237115859985,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 14290
+    },
+    {
+      "epoch": 45.833333333333336,
+      "grad_norm": 1.6211235523223877,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 14300
+    },
+    {
+      "epoch": 45.86538461538461,
+      "grad_norm": 1.5651072263717651,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 14310
+    },
+    {
+      "epoch": 45.8974358974359,
+      "grad_norm": 1.717585563659668,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 14320
+    },
+    {
+      "epoch": 45.92948717948718,
+      "grad_norm": 1.3768366575241089,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 14330
+    },
+    {
+      "epoch": 45.96153846153846,
+      "grad_norm": 1.1047064065933228,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 14340
+    },
+    {
+      "epoch": 45.993589743589745,
+      "grad_norm": 1.1818536520004272,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 14350
+    },
+    {
+      "epoch": 46.02564102564103,
+      "grad_norm": 1.3353221416473389,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 14360
+    },
+    {
+      "epoch": 46.05769230769231,
+      "grad_norm": 2.10288143157959,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 14370
+    },
+    {
+      "epoch": 46.08974358974359,
+      "grad_norm": 1.3731688261032104,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 14380
+    },
+    {
+      "epoch": 46.12179487179487,
+      "grad_norm": 1.853304386138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 14390
+    },
+    {
+      "epoch": 46.15384615384615,
+      "grad_norm": 1.4864933490753174,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 14400
+    },
+    {
+      "epoch": 46.18589743589744,
+      "grad_norm": 1.3976699113845825,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 14410
+    },
+    {
+      "epoch": 46.217948717948715,
+      "grad_norm": 1.3165374994277954,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 14420
+    },
+    {
+      "epoch": 46.25,
+      "grad_norm": 1.0162922143936157,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 14430
+    },
+    {
+      "epoch": 46.282051282051285,
+      "grad_norm": 1.495338797569275,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 14440
+    },
+    {
+      "epoch": 46.31410256410256,
+      "grad_norm": 1.980352759361267,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 14450
+    },
+    {
+      "epoch": 46.34615384615385,
+      "grad_norm": 1.8832917213439941,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 14460
+    },
+    {
+      "epoch": 46.37820512820513,
+      "grad_norm": 1.7287254333496094,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 14470
+    },
+    {
+      "epoch": 46.41025641025641,
+      "grad_norm": 1.3833401203155518,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 14480
+    },
+    {
+      "epoch": 46.44230769230769,
+      "grad_norm": 1.3457419872283936,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 14490
+    },
+    {
+      "epoch": 46.47435897435897,
+      "grad_norm": 1.2328802347183228,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 14500
+    },
+    {
+      "epoch": 46.506410256410255,
+      "grad_norm": 1.7252110242843628,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 14510
+    },
+    {
+      "epoch": 46.53846153846154,
+      "grad_norm": 1.5339007377624512,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 14520
+    },
+    {
+      "epoch": 46.57051282051282,
+      "grad_norm": 1.675951361656189,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 14530
+    },
+    {
+      "epoch": 46.6025641025641,
+      "grad_norm": 1.3617160320281982,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 14540
+    },
+    {
+      "epoch": 46.63461538461539,
+      "grad_norm": 1.5837761163711548,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 14550
+    },
+    {
+      "epoch": 46.666666666666664,
+      "grad_norm": 1.472544550895691,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 14560
+    },
+    {
+      "epoch": 46.69871794871795,
+      "grad_norm": 1.2550029754638672,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 14570
+    },
+    {
+      "epoch": 46.73076923076923,
+      "grad_norm": 1.8888616561889648,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 14580
+    },
+    {
+      "epoch": 46.76282051282051,
+      "grad_norm": 1.5199072360992432,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 14590
+    },
+    {
+      "epoch": 46.794871794871796,
+      "grad_norm": 1.4733220338821411,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 14600
+    },
+    {
+      "epoch": 46.82692307692308,
+      "grad_norm": 1.464176058769226,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 14610
+    },
+    {
+      "epoch": 46.85897435897436,
+      "grad_norm": 1.2077198028564453,
+      "learning_rate": 0.0001,
+      "loss": 0.0151,
+      "step": 14620
+    },
+    {
+      "epoch": 46.89102564102564,
+      "grad_norm": 1.2515724897384644,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 14630
+    },
+    {
+      "epoch": 46.92307692307692,
+      "grad_norm": 1.5269984006881714,
+      "learning_rate": 0.0001,
+      "loss": 0.0154,
+      "step": 14640
+    },
+    {
+      "epoch": 46.955128205128204,
+      "grad_norm": 1.2782231569290161,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 14650
+    },
+    {
+      "epoch": 46.98717948717949,
+      "grad_norm": 1.1790506839752197,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 14660
+    },
+    {
+      "epoch": 47.01923076923077,
+      "grad_norm": 1.2892917394638062,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 14670
+    },
+    {
+      "epoch": 47.05128205128205,
+      "grad_norm": 1.1887890100479126,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 14680
+    },
+    {
+      "epoch": 47.083333333333336,
+      "grad_norm": 1.4662361145019531,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 14690
+    },
+    {
+      "epoch": 47.11538461538461,
+      "grad_norm": 1.1511508226394653,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 14700
+    },
+    {
+      "epoch": 47.1474358974359,
+      "grad_norm": 1.2079713344573975,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 14710
+    },
+    {
+      "epoch": 47.17948717948718,
+      "grad_norm": 1.3370943069458008,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 14720
+    },
+    {
+      "epoch": 47.21153846153846,
+      "grad_norm": 1.2838523387908936,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 14730
+    },
+    {
+      "epoch": 47.243589743589745,
+      "grad_norm": 2.080634593963623,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 14740
+    },
+    {
+      "epoch": 47.27564102564103,
+      "grad_norm": 1.3096469640731812,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 14750
+    },
+    {
+      "epoch": 47.30769230769231,
+      "grad_norm": 2.1603307723999023,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 14760
+    },
+    {
+      "epoch": 47.33974358974359,
+      "grad_norm": 1.3911470174789429,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 14770
+    },
+    {
+      "epoch": 47.37179487179487,
+      "grad_norm": 1.4585803747177124,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 14780
+    },
+    {
+      "epoch": 47.40384615384615,
+      "grad_norm": 2.2072370052337646,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 14790
+    },
+    {
+      "epoch": 47.43589743589744,
+      "grad_norm": 1.5878092050552368,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 14800
+    },
+    {
+      "epoch": 47.467948717948715,
+      "grad_norm": 1.564808964729309,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 14810
+    },
+    {
+      "epoch": 47.5,
+      "grad_norm": 1.6466211080551147,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 14820
+    },
+    {
+      "epoch": 47.532051282051285,
+      "grad_norm": 1.4056334495544434,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 14830
+    },
+    {
+      "epoch": 47.56410256410256,
+      "grad_norm": 1.1757656335830688,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 14840
+    },
+    {
+      "epoch": 47.59615384615385,
+      "grad_norm": 1.5556154251098633,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 14850
+    },
+    {
+      "epoch": 47.62820512820513,
+      "grad_norm": 1.8548475503921509,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 14860
+    },
+    {
+      "epoch": 47.66025641025641,
+      "grad_norm": 1.4188292026519775,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 14870
+    },
+    {
+      "epoch": 47.69230769230769,
+      "grad_norm": 1.4770623445510864,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 14880
+    },
+    {
+      "epoch": 47.72435897435897,
+      "grad_norm": 1.2625296115875244,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 14890
+    },
+    {
+      "epoch": 47.756410256410255,
+      "grad_norm": 1.3725394010543823,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 14900
+    },
+    {
+      "epoch": 47.78846153846154,
+      "grad_norm": 1.379301905632019,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 14910
+    },
+    {
+      "epoch": 47.82051282051282,
+      "grad_norm": 1.2022149562835693,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 14920
+    },
+    {
+      "epoch": 47.8525641025641,
+      "grad_norm": 1.4325090646743774,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 14930
+    },
+    {
+      "epoch": 47.88461538461539,
+      "grad_norm": 1.4309990406036377,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 14940
+    },
+    {
+      "epoch": 47.916666666666664,
+      "grad_norm": 1.496914267539978,
+      "learning_rate": 0.0001,
+      "loss": 0.0153,
+      "step": 14950
+    },
+    {
+      "epoch": 47.94871794871795,
+      "grad_norm": 1.5934255123138428,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 14960
+    },
+    {
+      "epoch": 47.98076923076923,
+      "grad_norm": 1.8460350036621094,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 14970
+    },
+    {
+      "epoch": 48.01282051282051,
+      "grad_norm": 1.3358546495437622,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 14980
+    },
+    {
+      "epoch": 48.044871794871796,
+      "grad_norm": 1.4186465740203857,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 14990
+    },
+    {
+      "epoch": 48.07692307692308,
+      "grad_norm": 1.5329601764678955,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 15000
+    },
+    {
+      "epoch": 48.10897435897436,
+      "grad_norm": 1.4672949314117432,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 15010
+    },
+    {
+      "epoch": 48.14102564102564,
+      "grad_norm": 1.1525704860687256,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 15020
+    },
+    {
+      "epoch": 48.17307692307692,
+      "grad_norm": 1.2373465299606323,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 15030
+    },
+    {
+      "epoch": 48.205128205128204,
+      "grad_norm": 0.8530870676040649,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15040
+    },
+    {
+      "epoch": 48.23717948717949,
+      "grad_norm": 1.5703604221343994,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15050
+    },
+    {
+      "epoch": 48.26923076923077,
+      "grad_norm": 1.7103209495544434,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 15060
+    },
+    {
+      "epoch": 48.30128205128205,
+      "grad_norm": 1.3758243322372437,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 15070
+    },
+    {
+      "epoch": 48.333333333333336,
+      "grad_norm": 1.7681550979614258,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 15080
+    },
+    {
+      "epoch": 48.36538461538461,
+      "grad_norm": 1.36439847946167,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 15090
+    },
+    {
+      "epoch": 48.3974358974359,
+      "grad_norm": 1.3048949241638184,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15100
+    },
+    {
+      "epoch": 48.42948717948718,
+      "grad_norm": 1.0822452306747437,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 15110
+    },
+    {
+      "epoch": 48.46153846153846,
+      "grad_norm": 1.7394616603851318,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 15120
+    },
+    {
+      "epoch": 48.493589743589745,
+      "grad_norm": 1.700535535812378,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 15130
+    },
+    {
+      "epoch": 48.52564102564103,
+      "grad_norm": 1.6045435667037964,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 15140
+    },
+    {
+      "epoch": 48.55769230769231,
+      "grad_norm": 1.4822567701339722,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 15150
+    },
+    {
+      "epoch": 48.58974358974359,
+      "grad_norm": 1.2139326333999634,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 15160
+    },
+    {
+      "epoch": 48.62179487179487,
+      "grad_norm": 1.6960335969924927,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 15170
+    },
+    {
+      "epoch": 48.65384615384615,
+      "grad_norm": 1.217214822769165,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 15180
+    },
+    {
+      "epoch": 48.68589743589744,
+      "grad_norm": 1.4866142272949219,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 15190
+    },
+    {
+      "epoch": 48.717948717948715,
+      "grad_norm": 1.3999232053756714,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 15200
+    },
+    {
+      "epoch": 48.75,
+      "grad_norm": 1.4303278923034668,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 15210
+    },
+    {
+      "epoch": 48.782051282051285,
+      "grad_norm": 1.454598307609558,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 15220
+    },
+    {
+      "epoch": 48.81410256410256,
+      "grad_norm": 1.4881043434143066,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 15230
+    },
+    {
+      "epoch": 48.84615384615385,
+      "grad_norm": 1.9722723960876465,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 15240
+    },
+    {
+      "epoch": 48.87820512820513,
+      "grad_norm": 1.7739686965942383,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 15250
+    },
+    {
+      "epoch": 48.91025641025641,
+      "grad_norm": 2.4034087657928467,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 15260
+    },
+    {
+      "epoch": 48.94230769230769,
+      "grad_norm": 2.0294106006622314,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 15270
+    },
+    {
+      "epoch": 48.97435897435897,
+      "grad_norm": 2.035243272781372,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 15280
+    },
+    {
+      "epoch": 49.006410256410255,
+      "grad_norm": 1.8231216669082642,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 15290
+    },
+    {
+      "epoch": 49.03846153846154,
+      "grad_norm": 1.627069354057312,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 15300
+    },
+    {
+      "epoch": 49.07051282051282,
+      "grad_norm": 1.7875434160232544,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 15310
+    },
+    {
+      "epoch": 49.1025641025641,
+      "grad_norm": 1.87114679813385,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 15320
+    },
+    {
+      "epoch": 49.13461538461539,
+      "grad_norm": 1.3154364824295044,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 15330
+    },
+    {
+      "epoch": 49.166666666666664,
+      "grad_norm": 1.6026724576950073,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 15340
+    },
+    {
+      "epoch": 49.19871794871795,
+      "grad_norm": 1.2864066362380981,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 15350
+    },
+    {
+      "epoch": 49.23076923076923,
+      "grad_norm": 1.415216088294983,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 15360
+    },
+    {
+      "epoch": 49.26282051282051,
+      "grad_norm": 1.5178031921386719,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 15370
+    },
+    {
+      "epoch": 49.294871794871796,
+      "grad_norm": 1.6735670566558838,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 15380
+    },
+    {
+      "epoch": 49.32692307692308,
+      "grad_norm": 1.3305575847625732,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15390
+    },
+    {
+      "epoch": 49.35897435897436,
+      "grad_norm": 1.471673846244812,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 15400
+    },
+    {
+      "epoch": 49.39102564102564,
+      "grad_norm": 2.4644076824188232,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 15410
+    },
+    {
+      "epoch": 49.42307692307692,
+      "grad_norm": 2.0089967250823975,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 15420
+    },
+    {
+      "epoch": 49.455128205128204,
+      "grad_norm": 2.086679697036743,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 15430
+    },
+    {
+      "epoch": 49.48717948717949,
+      "grad_norm": 1.5340145826339722,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15440
+    },
+    {
+      "epoch": 49.51923076923077,
+      "grad_norm": 1.5413615703582764,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15450
+    },
+    {
+      "epoch": 49.55128205128205,
+      "grad_norm": 1.4605519771575928,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 15460
+    },
+    {
+      "epoch": 49.583333333333336,
+      "grad_norm": 1.5865014791488647,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 15470
+    },
+    {
+      "epoch": 49.61538461538461,
+      "grad_norm": 1.566868543624878,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 15480
+    },
+    {
+      "epoch": 49.6474358974359,
+      "grad_norm": 1.5282906293869019,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 15490
+    },
+    {
+      "epoch": 49.67948717948718,
+      "grad_norm": 1.2353575229644775,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 15500
+    },
+    {
+      "epoch": 49.71153846153846,
+      "grad_norm": 1.5142247676849365,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 15510
+    },
+    {
+      "epoch": 49.743589743589745,
+      "grad_norm": 1.7747505903244019,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 15520
+    },
+    {
+      "epoch": 49.77564102564103,
+      "grad_norm": 1.2723251581192017,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 15530
+    },
+    {
+      "epoch": 49.80769230769231,
+      "grad_norm": 1.5743162631988525,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 15540
+    },
+    {
+      "epoch": 49.83974358974359,
+      "grad_norm": 1.619694709777832,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15550
+    },
+    {
+      "epoch": 49.87179487179487,
+      "grad_norm": 2.8937857151031494,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 15560
+    },
+    {
+      "epoch": 49.90384615384615,
+      "grad_norm": 1.915523886680603,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 15570
+    },
+    {
+      "epoch": 49.93589743589744,
+      "grad_norm": 2.460894823074341,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 15580
+    },
+    {
+      "epoch": 49.967948717948715,
+      "grad_norm": 1.7079333066940308,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 15590
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 1.414375901222229,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 15600
+    },
+    {
+      "epoch": 50.032051282051285,
+      "grad_norm": 1.677878499031067,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 15610
+    },
+    {
+      "epoch": 50.06410256410256,
+      "grad_norm": 1.5026888847351074,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 15620
+    },
+    {
+      "epoch": 50.09615384615385,
+      "grad_norm": 1.309737205505371,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 15630
+    },
+    {
+      "epoch": 50.12820512820513,
+      "grad_norm": 1.4199328422546387,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 15640
+    },
+    {
+      "epoch": 50.16025641025641,
+      "grad_norm": 1.4536559581756592,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 15650
+    },
+    {
+      "epoch": 50.19230769230769,
+      "grad_norm": 1.3872767686843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 15660
+    },
+    {
+      "epoch": 50.22435897435897,
+      "grad_norm": 1.3499122858047485,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 15670
+    },
+    {
+      "epoch": 50.256410256410255,
+      "grad_norm": 1.5733795166015625,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 15680
+    },
+    {
+      "epoch": 50.28846153846154,
+      "grad_norm": 1.5587791204452515,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 15690
+    },
+    {
+      "epoch": 50.32051282051282,
+      "grad_norm": 1.684955358505249,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 15700
+    },
+    {
+      "epoch": 50.3525641025641,
+      "grad_norm": 1.6490137577056885,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 15710
+    },
+    {
+      "epoch": 50.38461538461539,
+      "grad_norm": 1.592980146408081,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 15720
+    },
+    {
+      "epoch": 50.416666666666664,
+      "grad_norm": 1.7272244691848755,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 15730
+    },
+    {
+      "epoch": 50.44871794871795,
+      "grad_norm": 1.6149604320526123,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 15740
+    },
+    {
+      "epoch": 50.48076923076923,
+      "grad_norm": 1.1569535732269287,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 15750
+    },
+    {
+      "epoch": 50.51282051282051,
+      "grad_norm": 1.817934513092041,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 15760
+    },
+    {
+      "epoch": 50.544871794871796,
+      "grad_norm": 1.9634262323379517,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 15770
+    },
+    {
+      "epoch": 50.57692307692308,
+      "grad_norm": 1.9494473934173584,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 15780
+    },
+    {
+      "epoch": 50.60897435897436,
+      "grad_norm": 1.576435923576355,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 15790
+    },
+    {
+      "epoch": 50.64102564102564,
+      "grad_norm": 1.5815507173538208,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 15800
+    },
+    {
+      "epoch": 50.67307692307692,
+      "grad_norm": 1.339893102645874,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 15810
+    },
+    {
+      "epoch": 50.705128205128204,
+      "grad_norm": 1.421189546585083,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 15820
+    },
+    {
+      "epoch": 50.73717948717949,
+      "grad_norm": 1.5153155326843262,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 15830
+    },
+    {
+      "epoch": 50.76923076923077,
+      "grad_norm": 1.6086560487747192,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 15840
+    },
+    {
+      "epoch": 50.80128205128205,
+      "grad_norm": 1.3780909776687622,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15850
+    },
+    {
+      "epoch": 50.833333333333336,
+      "grad_norm": 1.4494799375534058,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 15860
+    },
+    {
+      "epoch": 50.86538461538461,
+      "grad_norm": 1.34677255153656,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 15870
+    },
+    {
+      "epoch": 50.8974358974359,
+      "grad_norm": 1.7435455322265625,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 15880
+    },
+    {
+      "epoch": 50.92948717948718,
+      "grad_norm": 1.2408010959625244,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 15890
+    },
+    {
+      "epoch": 50.96153846153846,
+      "grad_norm": 1.194413185119629,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 15900
+    },
+    {
+      "epoch": 50.993589743589745,
+      "grad_norm": 1.4560867547988892,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 15910
+    },
+    {
+      "epoch": 51.02564102564103,
+      "grad_norm": 1.3255025148391724,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 15920
+    },
+    {
+      "epoch": 51.05769230769231,
+      "grad_norm": 1.2940136194229126,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 15930
+    },
+    {
+      "epoch": 51.08974358974359,
+      "grad_norm": 1.206061840057373,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 15940
+    },
+    {
+      "epoch": 51.12179487179487,
+      "grad_norm": 1.2382256984710693,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 15950
+    },
+    {
+      "epoch": 51.15384615384615,
+      "grad_norm": 1.208896517753601,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15960
+    },
+    {
+      "epoch": 51.18589743589744,
+      "grad_norm": 1.2731941938400269,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 15970
+    },
+    {
+      "epoch": 51.217948717948715,
+      "grad_norm": 1.4141924381256104,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 15980
+    },
+    {
+      "epoch": 51.25,
+      "grad_norm": 1.7986109256744385,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 15990
+    },
+    {
+      "epoch": 51.282051282051285,
+      "grad_norm": 1.767656922340393,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 16000
+    },
+    {
+      "epoch": 51.31410256410256,
+      "grad_norm": 1.4901806116104126,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 16010
+    },
+    {
+      "epoch": 51.34615384615385,
+      "grad_norm": 1.3905203342437744,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 16020
+    },
+    {
+      "epoch": 51.37820512820513,
+      "grad_norm": 1.4642879962921143,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 16030
+    },
+    {
+      "epoch": 51.41025641025641,
+      "grad_norm": 1.2400258779525757,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 16040
+    },
+    {
+      "epoch": 51.44230769230769,
+      "grad_norm": 1.4077702760696411,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 16050
+    },
+    {
+      "epoch": 51.47435897435897,
+      "grad_norm": 1.5742915868759155,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 16060
+    },
+    {
+      "epoch": 51.506410256410255,
+      "grad_norm": 1.3528639078140259,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 16070
+    },
+    {
+      "epoch": 51.53846153846154,
+      "grad_norm": 1.4054526090621948,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 16080
+    },
+    {
+      "epoch": 51.57051282051282,
+      "grad_norm": 1.1153438091278076,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 16090
+    },
+    {
+      "epoch": 51.6025641025641,
+      "grad_norm": 1.7053669691085815,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 16100
+    },
+    {
+      "epoch": 51.63461538461539,
+      "grad_norm": 1.8781402111053467,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 16110
+    },
+    {
+      "epoch": 51.666666666666664,
+      "grad_norm": 1.5646321773529053,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 16120
+    },
+    {
+      "epoch": 51.69871794871795,
+      "grad_norm": 1.667620301246643,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 16130
+    },
+    {
+      "epoch": 51.73076923076923,
+      "grad_norm": 1.4889291524887085,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16140
+    },
+    {
+      "epoch": 51.76282051282051,
+      "grad_norm": 1.3238680362701416,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 16150
+    },
+    {
+      "epoch": 51.794871794871796,
+      "grad_norm": 1.476738691329956,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 16160
+    },
+    {
+      "epoch": 51.82692307692308,
+      "grad_norm": 1.5132322311401367,
+      "learning_rate": 0.0001,
+      "loss": 0.0146,
+      "step": 16170
+    },
+    {
+      "epoch": 51.85897435897436,
+      "grad_norm": 1.4690226316452026,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 16180
+    },
+    {
+      "epoch": 51.89102564102564,
+      "grad_norm": 1.4132080078125,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 16190
+    },
+    {
+      "epoch": 51.92307692307692,
+      "grad_norm": 1.7080174684524536,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16200
+    },
+    {
+      "epoch": 51.955128205128204,
+      "grad_norm": 1.121627688407898,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 16210
+    },
+    {
+      "epoch": 51.98717948717949,
+      "grad_norm": 1.4819918870925903,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 16220
+    },
+    {
+      "epoch": 52.01923076923077,
+      "grad_norm": 1.7715235948562622,
+      "learning_rate": 0.0001,
+      "loss": 0.0159,
+      "step": 16230
+    },
+    {
+      "epoch": 52.05128205128205,
+      "grad_norm": 1.2185920476913452,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 16240
+    },
+    {
+      "epoch": 52.083333333333336,
+      "grad_norm": 1.7912441492080688,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16250
+    },
+    {
+      "epoch": 52.11538461538461,
+      "grad_norm": 1.2204526662826538,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 16260
+    },
+    {
+      "epoch": 52.1474358974359,
+      "grad_norm": 1.5983049869537354,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 16270
+    },
+    {
+      "epoch": 52.17948717948718,
+      "grad_norm": 1.3988988399505615,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 16280
+    },
+    {
+      "epoch": 52.21153846153846,
+      "grad_norm": 1.0410239696502686,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 16290
+    },
+    {
+      "epoch": 52.243589743589745,
+      "grad_norm": 1.3750203847885132,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 16300
+    },
+    {
+      "epoch": 52.27564102564103,
+      "grad_norm": 1.2083162069320679,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 16310
+    },
+    {
+      "epoch": 52.30769230769231,
+      "grad_norm": 1.0154967308044434,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 16320
+    },
+    {
+      "epoch": 52.33974358974359,
+      "grad_norm": 1.7153292894363403,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 16330
+    },
+    {
+      "epoch": 52.37179487179487,
+      "grad_norm": 1.5804824829101562,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 16340
+    },
+    {
+      "epoch": 52.40384615384615,
+      "grad_norm": 1.5865135192871094,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 16350
+    },
+    {
+      "epoch": 52.43589743589744,
+      "grad_norm": 1.3534780740737915,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 16360
+    },
+    {
+      "epoch": 52.467948717948715,
+      "grad_norm": 1.5527431964874268,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 16370
+    },
+    {
+      "epoch": 52.5,
+      "grad_norm": 1.3315883874893188,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 16380
+    },
+    {
+      "epoch": 52.532051282051285,
+      "grad_norm": 1.3130961656570435,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 16390
+    },
+    {
+      "epoch": 52.56410256410256,
+      "grad_norm": 1.1426150798797607,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 16400
+    },
+    {
+      "epoch": 52.59615384615385,
+      "grad_norm": 1.1611061096191406,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 16410
+    },
+    {
+      "epoch": 52.62820512820513,
+      "grad_norm": 1.1703641414642334,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 16420
+    },
+    {
+      "epoch": 52.66025641025641,
+      "grad_norm": 1.3240578174591064,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 16430
+    },
+    {
+      "epoch": 52.69230769230769,
+      "grad_norm": 1.2958043813705444,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 16440
+    },
+    {
+      "epoch": 52.72435897435897,
+      "grad_norm": 1.0470552444458008,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 16450
+    },
+    {
+      "epoch": 52.756410256410255,
+      "grad_norm": 1.0053988695144653,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 16460
+    },
+    {
+      "epoch": 52.78846153846154,
+      "grad_norm": 1.26897394657135,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 16470
+    },
+    {
+      "epoch": 52.82051282051282,
+      "grad_norm": 1.4356075525283813,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 16480
+    },
+    {
+      "epoch": 52.8525641025641,
+      "grad_norm": 1.3930684328079224,
+      "learning_rate": 0.0001,
+      "loss": 0.0149,
+      "step": 16490
+    },
+    {
+      "epoch": 52.88461538461539,
+      "grad_norm": 1.1448893547058105,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16500
+    },
+    {
+      "epoch": 52.916666666666664,
+      "grad_norm": 1.0874011516571045,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 16510
+    },
+    {
+      "epoch": 52.94871794871795,
+      "grad_norm": 1.3930158615112305,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 16520
+    },
+    {
+      "epoch": 52.98076923076923,
+      "grad_norm": 1.257588505744934,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16530
+    },
+    {
+      "epoch": 53.01282051282051,
+      "grad_norm": 1.493470549583435,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 16540
+    },
+    {
+      "epoch": 53.044871794871796,
+      "grad_norm": 1.4423413276672363,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16550
+    },
+    {
+      "epoch": 53.07692307692308,
+      "grad_norm": 1.3001172542572021,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 16560
+    },
+    {
+      "epoch": 53.10897435897436,
+      "grad_norm": 1.1401937007904053,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16570
+    },
+    {
+      "epoch": 53.14102564102564,
+      "grad_norm": 1.6418911218643188,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 16580
+    },
+    {
+      "epoch": 53.17307692307692,
+      "grad_norm": 1.2128185033798218,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16590
+    },
+    {
+      "epoch": 53.205128205128204,
+      "grad_norm": 1.384463906288147,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 16600
+    },
+    {
+      "epoch": 53.23717948717949,
+      "grad_norm": 2.2022547721862793,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 16610
+    },
+    {
+      "epoch": 53.26923076923077,
+      "grad_norm": 1.6769269704818726,
+      "learning_rate": 0.0001,
+      "loss": 0.0143,
+      "step": 16620
+    },
+    {
+      "epoch": 53.30128205128205,
+      "grad_norm": 1.3624937534332275,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 16630
+    },
+    {
+      "epoch": 53.333333333333336,
+      "grad_norm": 1.2216691970825195,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16640
+    },
+    {
+      "epoch": 53.36538461538461,
+      "grad_norm": 1.396540880203247,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 16650
+    },
+    {
+      "epoch": 53.3974358974359,
+      "grad_norm": 1.7450000047683716,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 16660
+    },
+    {
+      "epoch": 53.42948717948718,
+      "grad_norm": 1.2713321447372437,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 16670
+    },
+    {
+      "epoch": 53.46153846153846,
+      "grad_norm": 1.5428752899169922,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 16680
+    },
+    {
+      "epoch": 53.493589743589745,
+      "grad_norm": 1.6088272333145142,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 16690
+    },
+    {
+      "epoch": 53.52564102564103,
+      "grad_norm": 1.6688463687896729,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 16700
+    },
+    {
+      "epoch": 53.55769230769231,
+      "grad_norm": 1.5507452487945557,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 16710
+    },
+    {
+      "epoch": 53.58974358974359,
+      "grad_norm": 1.099798560142517,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 16720
+    },
+    {
+      "epoch": 53.62179487179487,
+      "grad_norm": 1.262628436088562,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 16730
+    },
+    {
+      "epoch": 53.65384615384615,
+      "grad_norm": 1.3971819877624512,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 16740
+    },
+    {
+      "epoch": 53.68589743589744,
+      "grad_norm": 1.418124794960022,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 16750
+    },
+    {
+      "epoch": 53.717948717948715,
+      "grad_norm": 1.3373891115188599,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 16760
+    },
+    {
+      "epoch": 53.75,
+      "grad_norm": 1.3503834009170532,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 16770
+    },
+    {
+      "epoch": 53.782051282051285,
+      "grad_norm": 1.4083828926086426,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 16780
+    },
+    {
+      "epoch": 53.81410256410256,
+      "grad_norm": 1.4798964262008667,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 16790
+    },
+    {
+      "epoch": 53.84615384615385,
+      "grad_norm": 1.1556235551834106,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 16800
+    },
+    {
+      "epoch": 53.87820512820513,
+      "grad_norm": 1.4434691667556763,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 16810
+    },
+    {
+      "epoch": 53.91025641025641,
+      "grad_norm": 1.2701916694641113,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 16820
+    },
+    {
+      "epoch": 53.94230769230769,
+      "grad_norm": 1.3718668222427368,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 16830
+    },
+    {
+      "epoch": 53.97435897435897,
+      "grad_norm": 1.0122467279434204,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 16840
+    },
+    {
+      "epoch": 54.006410256410255,
+      "grad_norm": 1.0677884817123413,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 16850
+    },
+    {
+      "epoch": 54.03846153846154,
+      "grad_norm": 1.191892385482788,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 16860
+    },
+    {
+      "epoch": 54.07051282051282,
+      "grad_norm": 1.4344733953475952,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 16870
+    },
+    {
+      "epoch": 54.1025641025641,
+      "grad_norm": 1.3972355127334595,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 16880
+    },
+    {
+      "epoch": 54.13461538461539,
+      "grad_norm": 1.0642157793045044,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 16890
+    },
+    {
+      "epoch": 54.166666666666664,
+      "grad_norm": 1.6169337034225464,
+      "learning_rate": 0.0001,
+      "loss": 0.0145,
+      "step": 16900
+    },
+    {
+      "epoch": 54.19871794871795,
+      "grad_norm": 1.4977760314941406,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 16910
+    },
+    {
+      "epoch": 54.23076923076923,
+      "grad_norm": 1.2846943140029907,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 16920
+    },
+    {
+      "epoch": 54.26282051282051,
+      "grad_norm": 1.1140047311782837,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 16930
+    },
+    {
+      "epoch": 54.294871794871796,
+      "grad_norm": 1.3244255781173706,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 16940
+    },
+    {
+      "epoch": 54.32692307692308,
+      "grad_norm": 1.1772371530532837,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 16950
+    },
+    {
+      "epoch": 54.35897435897436,
+      "grad_norm": 1.100095272064209,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 16960
+    },
+    {
+      "epoch": 54.39102564102564,
+      "grad_norm": 1.2272884845733643,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 16970
+    },
+    {
+      "epoch": 54.42307692307692,
+      "grad_norm": 1.0108572244644165,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 16980
+    },
+    {
+      "epoch": 54.455128205128204,
+      "grad_norm": 1.7262169122695923,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 16990
+    },
+    {
+      "epoch": 54.48717948717949,
+      "grad_norm": 1.3549970388412476,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 17000
+    },
+    {
+      "epoch": 54.51923076923077,
+      "grad_norm": 1.181462287902832,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 17010
+    },
+    {
+      "epoch": 54.55128205128205,
+      "grad_norm": 0.784734845161438,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 17020
+    },
+    {
+      "epoch": 54.583333333333336,
+      "grad_norm": 0.9180074334144592,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 17030
+    },
+    {
+      "epoch": 54.61538461538461,
+      "grad_norm": 1.3891832828521729,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 17040
+    },
+    {
+      "epoch": 54.6474358974359,
+      "grad_norm": 1.348059058189392,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 17050
+    },
+    {
+      "epoch": 54.67948717948718,
+      "grad_norm": 1.1906883716583252,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 17060
+    },
+    {
+      "epoch": 54.71153846153846,
+      "grad_norm": 1.175998330116272,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 17070
+    },
+    {
+      "epoch": 54.743589743589745,
+      "grad_norm": 1.4609676599502563,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 17080
+    },
+    {
+      "epoch": 54.77564102564103,
+      "grad_norm": 1.48897123336792,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 17090
+    },
+    {
+      "epoch": 54.80769230769231,
+      "grad_norm": 1.7713348865509033,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 17100
+    },
+    {
+      "epoch": 54.83974358974359,
+      "grad_norm": 1.4105178117752075,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 17110
+    },
+    {
+      "epoch": 54.87179487179487,
+      "grad_norm": 1.7611223459243774,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 17120
+    },
+    {
+      "epoch": 54.90384615384615,
+      "grad_norm": 1.8028101921081543,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17130
+    },
+    {
+      "epoch": 54.93589743589744,
+      "grad_norm": 1.420149564743042,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 17140
+    },
+    {
+      "epoch": 54.967948717948715,
+      "grad_norm": 1.3608647584915161,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 17150
+    },
+    {
+      "epoch": 55.0,
+      "grad_norm": 1.6324608325958252,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 17160
+    },
+    {
+      "epoch": 55.032051282051285,
+      "grad_norm": 1.538195252418518,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 17170
+    },
+    {
+      "epoch": 55.06410256410256,
+      "grad_norm": 1.3910542726516724,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 17180
+    },
+    {
+      "epoch": 55.09615384615385,
+      "grad_norm": 1.004798173904419,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 17190
+    },
+    {
+      "epoch": 55.12820512820513,
+      "grad_norm": 1.3086049556732178,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 17200
+    },
+    {
+      "epoch": 55.16025641025641,
+      "grad_norm": 1.1735886335372925,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 17210
+    },
+    {
+      "epoch": 55.19230769230769,
+      "grad_norm": 1.5379364490509033,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 17220
+    },
+    {
+      "epoch": 55.22435897435897,
+      "grad_norm": 1.0317988395690918,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 17230
+    },
+    {
+      "epoch": 55.256410256410255,
+      "grad_norm": 1.3773512840270996,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 17240
+    },
+    {
+      "epoch": 55.28846153846154,
+      "grad_norm": 1.0974445343017578,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 17250
+    },
+    {
+      "epoch": 55.32051282051282,
+      "grad_norm": 1.4704886674880981,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 17260
+    },
+    {
+      "epoch": 55.3525641025641,
+      "grad_norm": 1.003848910331726,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 17270
+    },
+    {
+      "epoch": 55.38461538461539,
+      "grad_norm": 1.3684247732162476,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 17280
+    },
+    {
+      "epoch": 55.416666666666664,
+      "grad_norm": 1.1051337718963623,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 17290
+    },
+    {
+      "epoch": 55.44871794871795,
+      "grad_norm": 1.2800270318984985,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17300
+    },
+    {
+      "epoch": 55.48076923076923,
+      "grad_norm": 0.9806950092315674,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 17310
+    },
+    {
+      "epoch": 55.51282051282051,
+      "grad_norm": 1.0376447439193726,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 17320
+    },
+    {
+      "epoch": 55.544871794871796,
+      "grad_norm": 1.1198196411132812,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 17330
+    },
+    {
+      "epoch": 55.57692307692308,
+      "grad_norm": 0.9544689059257507,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 17340
+    },
+    {
+      "epoch": 55.60897435897436,
+      "grad_norm": 1.247563362121582,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 17350
+    },
+    {
+      "epoch": 55.64102564102564,
+      "grad_norm": 1.4982805252075195,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17360
+    },
+    {
+      "epoch": 55.67307692307692,
+      "grad_norm": 0.9972005486488342,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17370
+    },
+    {
+      "epoch": 55.705128205128204,
+      "grad_norm": 1.0328195095062256,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 17380
+    },
+    {
+      "epoch": 55.73717948717949,
+      "grad_norm": 1.0262706279754639,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 17390
+    },
+    {
+      "epoch": 55.76923076923077,
+      "grad_norm": 1.0356111526489258,
+      "learning_rate": 0.0001,
+      "loss": 0.014,
+      "step": 17400
+    },
+    {
+      "epoch": 55.80128205128205,
+      "grad_norm": 1.4231936931610107,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 17410
+    },
+    {
+      "epoch": 55.833333333333336,
+      "grad_norm": 1.4223966598510742,
+      "learning_rate": 0.0001,
+      "loss": 0.015,
+      "step": 17420
+    },
+    {
+      "epoch": 55.86538461538461,
+      "grad_norm": 0.9809154272079468,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 17430
+    },
+    {
+      "epoch": 55.8974358974359,
+      "grad_norm": 1.0889447927474976,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 17440
+    },
+    {
+      "epoch": 55.92948717948718,
+      "grad_norm": 0.9776862263679504,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 17450
+    },
+    {
+      "epoch": 55.96153846153846,
+      "grad_norm": 1.0052626132965088,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 17460
+    },
+    {
+      "epoch": 55.993589743589745,
+      "grad_norm": 1.011157512664795,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 17470
+    },
+    {
+      "epoch": 56.02564102564103,
+      "grad_norm": 1.0476146936416626,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17480
+    },
+    {
+      "epoch": 56.05769230769231,
+      "grad_norm": 1.0499918460845947,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 17490
+    },
+    {
+      "epoch": 56.08974358974359,
+      "grad_norm": 1.5909006595611572,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17500
+    },
+    {
+      "epoch": 56.12179487179487,
+      "grad_norm": 1.2326154708862305,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 17510
+    },
+    {
+      "epoch": 56.15384615384615,
+      "grad_norm": 0.8537552952766418,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 17520
+    },
+    {
+      "epoch": 56.18589743589744,
+      "grad_norm": 0.9533272385597229,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 17530
+    },
+    {
+      "epoch": 56.217948717948715,
+      "grad_norm": 0.7692682147026062,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 17540
+    },
+    {
+      "epoch": 56.25,
+      "grad_norm": 1.152483582496643,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 17550
+    },
+    {
+      "epoch": 56.282051282051285,
+      "grad_norm": 0.732848048210144,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17560
+    },
+    {
+      "epoch": 56.31410256410256,
+      "grad_norm": 1.1604119539260864,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 17570
+    },
+    {
+      "epoch": 56.34615384615385,
+      "grad_norm": 1.0696848630905151,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 17580
+    },
+    {
+      "epoch": 56.37820512820513,
+      "grad_norm": 1.5233752727508545,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 17590
+    },
+    {
+      "epoch": 56.41025641025641,
+      "grad_norm": 1.022600531578064,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 17600
+    },
+    {
+      "epoch": 56.44230769230769,
+      "grad_norm": 1.4820618629455566,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 17610
+    },
+    {
+      "epoch": 56.47435897435897,
+      "grad_norm": 1.242737889289856,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 17620
+    },
+    {
+      "epoch": 56.506410256410255,
+      "grad_norm": 1.3280558586120605,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 17630
+    },
+    {
+      "epoch": 56.53846153846154,
+      "grad_norm": 0.9943426847457886,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 17640
+    },
+    {
+      "epoch": 56.57051282051282,
+      "grad_norm": 0.8471247553825378,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 17650
+    },
+    {
+      "epoch": 56.6025641025641,
+      "grad_norm": 1.0678421258926392,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 17660
+    },
+    {
+      "epoch": 56.63461538461539,
+      "grad_norm": 1.2157783508300781,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 17670
+    },
+    {
+      "epoch": 56.666666666666664,
+      "grad_norm": 1.4285845756530762,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 17680
+    },
+    {
+      "epoch": 56.69871794871795,
+      "grad_norm": 1.1493422985076904,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 17690
+    },
+    {
+      "epoch": 56.73076923076923,
+      "grad_norm": 1.5110819339752197,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 17700
+    },
+    {
+      "epoch": 56.76282051282051,
+      "grad_norm": 1.4330930709838867,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 17710
+    },
+    {
+      "epoch": 56.794871794871796,
+      "grad_norm": 1.0147254467010498,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17720
+    },
+    {
+      "epoch": 56.82692307692308,
+      "grad_norm": 1.3442928791046143,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 17730
+    },
+    {
+      "epoch": 56.85897435897436,
+      "grad_norm": 1.207861304283142,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 17740
+    },
+    {
+      "epoch": 56.89102564102564,
+      "grad_norm": 1.399232029914856,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17750
+    },
+    {
+      "epoch": 56.92307692307692,
+      "grad_norm": 1.5031102895736694,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 17760
+    },
+    {
+      "epoch": 56.955128205128204,
+      "grad_norm": 1.0924499034881592,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 17770
+    },
+    {
+      "epoch": 56.98717948717949,
+      "grad_norm": 1.1358128786087036,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 17780
+    },
+    {
+      "epoch": 57.01923076923077,
+      "grad_norm": 1.4541473388671875,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 17790
+    },
+    {
+      "epoch": 57.05128205128205,
+      "grad_norm": 1.565310001373291,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 17800
+    },
+    {
+      "epoch": 57.083333333333336,
+      "grad_norm": 1.8956371545791626,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 17810
+    },
+    {
+      "epoch": 57.11538461538461,
+      "grad_norm": 1.6977638006210327,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 17820
+    },
+    {
+      "epoch": 57.1474358974359,
+      "grad_norm": 1.7681266069412231,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 17830
+    },
+    {
+      "epoch": 57.17948717948718,
+      "grad_norm": 1.9807020425796509,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 17840
+    },
+    {
+      "epoch": 57.21153846153846,
+      "grad_norm": 1.6330432891845703,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 17850
+    },
+    {
+      "epoch": 57.243589743589745,
+      "grad_norm": 3.074298858642578,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 17860
+    },
+    {
+      "epoch": 57.27564102564103,
+      "grad_norm": 2.341331720352173,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 17870
+    },
+    {
+      "epoch": 57.30769230769231,
+      "grad_norm": 1.4445675611495972,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 17880
+    },
+    {
+      "epoch": 57.33974358974359,
+      "grad_norm": 1.4310181140899658,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 17890
+    },
+    {
+      "epoch": 57.37179487179487,
+      "grad_norm": 1.2766107320785522,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 17900
+    },
+    {
+      "epoch": 57.40384615384615,
+      "grad_norm": 1.3263412714004517,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 17910
+    },
+    {
+      "epoch": 57.43589743589744,
+      "grad_norm": 1.4371927976608276,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 17920
+    },
+    {
+      "epoch": 57.467948717948715,
+      "grad_norm": 1.2367323637008667,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 17930
+    },
+    {
+      "epoch": 57.5,
+      "grad_norm": 1.1817916631698608,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 17940
+    },
+    {
+      "epoch": 57.532051282051285,
+      "grad_norm": 1.1999475955963135,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 17950
+    },
+    {
+      "epoch": 57.56410256410256,
+      "grad_norm": 1.7190736532211304,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 17960
+    },
+    {
+      "epoch": 57.59615384615385,
+      "grad_norm": 1.5168575048446655,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 17970
+    },
+    {
+      "epoch": 57.62820512820513,
+      "grad_norm": 1.1874288320541382,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 17980
+    },
+    {
+      "epoch": 57.66025641025641,
+      "grad_norm": 1.251605749130249,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 17990
+    },
+    {
+      "epoch": 57.69230769230769,
+      "grad_norm": 1.1939078569412231,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 18000
+    },
+    {
+      "epoch": 57.72435897435897,
+      "grad_norm": 1.6976265907287598,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 18010
+    },
+    {
+      "epoch": 57.756410256410255,
+      "grad_norm": 1.1873260736465454,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 18020
+    },
+    {
+      "epoch": 57.78846153846154,
+      "grad_norm": 1.255031704902649,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 18030
+    },
+    {
+      "epoch": 57.82051282051282,
+      "grad_norm": 1.4751715660095215,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 18040
+    },
+    {
+      "epoch": 57.8525641025641,
+      "grad_norm": 1.510852575302124,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 18050
+    },
+    {
+      "epoch": 57.88461538461539,
+      "grad_norm": 1.4847655296325684,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 18060
+    },
+    {
+      "epoch": 57.916666666666664,
+      "grad_norm": 1.2553046941757202,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 18070
+    },
+    {
+      "epoch": 57.94871794871795,
+      "grad_norm": 1.595548152923584,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 18080
+    },
+    {
+      "epoch": 57.98076923076923,
+      "grad_norm": 1.5961867570877075,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 18090
+    },
+    {
+      "epoch": 58.01282051282051,
+      "grad_norm": 1.6778208017349243,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 18100
+    },
+    {
+      "epoch": 58.044871794871796,
+      "grad_norm": 1.5833665132522583,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 18110
+    },
+    {
+      "epoch": 58.07692307692308,
+      "grad_norm": 1.8187263011932373,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 18120
+    },
+    {
+      "epoch": 58.10897435897436,
+      "grad_norm": 1.1798661947250366,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 18130
+    },
+    {
+      "epoch": 58.14102564102564,
+      "grad_norm": 1.3269481658935547,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 18140
+    },
+    {
+      "epoch": 58.17307692307692,
+      "grad_norm": 1.4189085960388184,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 18150
+    },
+    {
+      "epoch": 58.205128205128204,
+      "grad_norm": 0.9958335161209106,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 18160
+    },
+    {
+      "epoch": 58.23717948717949,
+      "grad_norm": 1.2420506477355957,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 18170
+    },
+    {
+      "epoch": 58.26923076923077,
+      "grad_norm": 1.6102923154830933,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 18180
+    },
+    {
+      "epoch": 58.30128205128205,
+      "grad_norm": 1.5268480777740479,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 18190
+    },
+    {
+      "epoch": 58.333333333333336,
+      "grad_norm": 2.1934449672698975,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 18200
+    },
+    {
+      "epoch": 58.36538461538461,
+      "grad_norm": 1.552107572555542,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 18210
+    },
+    {
+      "epoch": 58.3974358974359,
+      "grad_norm": 1.2479557991027832,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 18220
+    },
+    {
+      "epoch": 58.42948717948718,
+      "grad_norm": 1.6763089895248413,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 18230
+    },
+    {
+      "epoch": 58.46153846153846,
+      "grad_norm": 1.7659544944763184,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 18240
+    },
+    {
+      "epoch": 58.493589743589745,
+      "grad_norm": 1.3076142072677612,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 18250
+    },
+    {
+      "epoch": 58.52564102564103,
+      "grad_norm": 1.6671817302703857,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 18260
+    },
+    {
+      "epoch": 58.55769230769231,
+      "grad_norm": 1.5456042289733887,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 18270
+    },
+    {
+      "epoch": 58.58974358974359,
+      "grad_norm": 1.1391674280166626,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 18280
+    },
+    {
+      "epoch": 58.62179487179487,
+      "grad_norm": 1.442305088043213,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 18290
+    },
+    {
+      "epoch": 58.65384615384615,
+      "grad_norm": 1.2266080379486084,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 18300
+    },
+    {
+      "epoch": 58.68589743589744,
+      "grad_norm": 1.049283504486084,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 18310
+    },
+    {
+      "epoch": 58.717948717948715,
+      "grad_norm": 1.3582065105438232,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 18320
+    },
+    {
+      "epoch": 58.75,
+      "grad_norm": 1.298092007637024,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 18330
+    },
+    {
+      "epoch": 58.782051282051285,
+      "grad_norm": 1.468706727027893,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 18340
+    },
+    {
+      "epoch": 58.81410256410256,
+      "grad_norm": 1.7458161115646362,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 18350
+    },
+    {
+      "epoch": 58.84615384615385,
+      "grad_norm": 1.4176757335662842,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 18360
+    },
+    {
+      "epoch": 58.87820512820513,
+      "grad_norm": 1.0672798156738281,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 18370
+    },
+    {
+      "epoch": 58.91025641025641,
+      "grad_norm": 1.208035945892334,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 18380
+    },
+    {
+      "epoch": 58.94230769230769,
+      "grad_norm": 1.2418867349624634,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 18390
+    },
+    {
+      "epoch": 58.97435897435897,
+      "grad_norm": 0.9402397871017456,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 18400
+    },
+    {
+      "epoch": 59.006410256410255,
+      "grad_norm": 1.4007312059402466,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 18410
+    },
+    {
+      "epoch": 59.03846153846154,
+      "grad_norm": 1.6821624040603638,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 18420
+    },
+    {
+      "epoch": 59.07051282051282,
+      "grad_norm": 1.3144562244415283,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 18430
+    },
+    {
+      "epoch": 59.1025641025641,
+      "grad_norm": 1.2391189336776733,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 18440
+    },
+    {
+      "epoch": 59.13461538461539,
+      "grad_norm": 1.0244916677474976,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 18450
+    },
+    {
+      "epoch": 59.166666666666664,
+      "grad_norm": 1.2896842956542969,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 18460
+    },
+    {
+      "epoch": 59.19871794871795,
+      "grad_norm": 1.1247986555099487,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 18470
+    },
+    {
+      "epoch": 59.23076923076923,
+      "grad_norm": 1.2049024105072021,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 18480
+    },
+    {
+      "epoch": 59.26282051282051,
+      "grad_norm": 1.321616291999817,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 18490
+    },
+    {
+      "epoch": 59.294871794871796,
+      "grad_norm": 1.093476414680481,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 18500
+    },
+    {
+      "epoch": 59.32692307692308,
+      "grad_norm": 0.9919162392616272,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 18510
+    },
+    {
+      "epoch": 59.35897435897436,
+      "grad_norm": 1.6398780345916748,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 18520
+    },
+    {
+      "epoch": 59.39102564102564,
+      "grad_norm": 1.0882028341293335,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 18530
+    },
+    {
+      "epoch": 59.42307692307692,
+      "grad_norm": 1.1184825897216797,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 18540
+    },
+    {
+      "epoch": 59.455128205128204,
+      "grad_norm": 1.7411212921142578,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 18550
+    },
+    {
+      "epoch": 59.48717948717949,
+      "grad_norm": 1.5743950605392456,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 18560
+    },
+    {
+      "epoch": 59.51923076923077,
+      "grad_norm": 1.2143372297286987,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 18570
+    },
+    {
+      "epoch": 59.55128205128205,
+      "grad_norm": 1.2221040725708008,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 18580
+    },
+    {
+      "epoch": 59.583333333333336,
+      "grad_norm": 1.4637328386306763,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 18590
+    },
+    {
+      "epoch": 59.61538461538461,
+      "grad_norm": 1.65518057346344,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 18600
+    },
+    {
+      "epoch": 59.6474358974359,
+      "grad_norm": 1.4632313251495361,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 18610
+    },
+    {
+      "epoch": 59.67948717948718,
+      "grad_norm": 2.0206124782562256,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 18620
+    },
+    {
+      "epoch": 59.71153846153846,
+      "grad_norm": 1.2958875894546509,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 18630
+    },
+    {
+      "epoch": 59.743589743589745,
+      "grad_norm": 1.42147696018219,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 18640
+    },
+    {
+      "epoch": 59.77564102564103,
+      "grad_norm": 1.6636754274368286,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 18650
+    },
+    {
+      "epoch": 59.80769230769231,
+      "grad_norm": 1.5987577438354492,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 18660
+    },
+    {
+      "epoch": 59.83974358974359,
+      "grad_norm": 1.810102105140686,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 18670
+    },
+    {
+      "epoch": 59.87179487179487,
+      "grad_norm": 1.0205459594726562,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 18680
+    },
+    {
+      "epoch": 59.90384615384615,
+      "grad_norm": 1.0555462837219238,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 18690
+    },
+    {
+      "epoch": 59.93589743589744,
+      "grad_norm": 1.6258447170257568,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 18700
+    },
+    {
+      "epoch": 59.967948717948715,
+      "grad_norm": 1.0191829204559326,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 18710
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 1.1081922054290771,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 18720
+    },
+    {
+      "epoch": 60.032051282051285,
+      "grad_norm": 1.0206409692764282,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 18730
+    },
+    {
+      "epoch": 60.06410256410256,
+      "grad_norm": 1.5927419662475586,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 18740
+    },
+    {
+      "epoch": 60.09615384615385,
+      "grad_norm": 0.9318953156471252,
+      "learning_rate": 0.0001,
+      "loss": 0.0142,
+      "step": 18750
+    },
+    {
+      "epoch": 60.12820512820513,
+      "grad_norm": 1.1129295825958252,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 18760
+    },
+    {
+      "epoch": 60.16025641025641,
+      "grad_norm": 0.8539644479751587,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 18770
+    },
+    {
+      "epoch": 60.19230769230769,
+      "grad_norm": 0.9165988564491272,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 18780
+    },
+    {
+      "epoch": 60.22435897435897,
+      "grad_norm": 1.1204606294631958,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 18790
+    },
+    {
+      "epoch": 60.256410256410255,
+      "grad_norm": 0.9787343740463257,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 18800
+    },
+    {
+      "epoch": 60.28846153846154,
+      "grad_norm": 1.319193959236145,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 18810
+    },
+    {
+      "epoch": 60.32051282051282,
+      "grad_norm": 1.4229881763458252,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 18820
+    },
+    {
+      "epoch": 60.3525641025641,
+      "grad_norm": 1.4905132055282593,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 18830
+    },
+    {
+      "epoch": 60.38461538461539,
+      "grad_norm": 1.0357842445373535,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 18840
+    },
+    {
+      "epoch": 60.416666666666664,
+      "grad_norm": 0.8399243950843811,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 18850
+    },
+    {
+      "epoch": 60.44871794871795,
+      "grad_norm": 1.0657216310501099,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 18860
+    },
+    {
+      "epoch": 60.48076923076923,
+      "grad_norm": 1.175038456916809,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 18870
+    },
+    {
+      "epoch": 60.51282051282051,
+      "grad_norm": 0.9690041542053223,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 18880
+    },
+    {
+      "epoch": 60.544871794871796,
+      "grad_norm": 1.1231399774551392,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 18890
+    },
+    {
+      "epoch": 60.57692307692308,
+      "grad_norm": 1.2021162509918213,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 18900
+    },
+    {
+      "epoch": 60.60897435897436,
+      "grad_norm": 1.5186773538589478,
+      "learning_rate": 0.0001,
+      "loss": 0.0135,
+      "step": 18910
+    },
+    {
+      "epoch": 60.64102564102564,
+      "grad_norm": 1.2103148698806763,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 18920
+    },
+    {
+      "epoch": 60.67307692307692,
+      "grad_norm": 1.2799115180969238,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 18930
+    },
+    {
+      "epoch": 60.705128205128204,
+      "grad_norm": 1.4718652963638306,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 18940
+    },
+    {
+      "epoch": 60.73717948717949,
+      "grad_norm": 1.9685345888137817,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 18950
+    },
+    {
+      "epoch": 60.76923076923077,
+      "grad_norm": 1.8907387256622314,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 18960
+    },
+    {
+      "epoch": 60.80128205128205,
+      "grad_norm": 1.9886386394500732,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 18970
+    },
+    {
+      "epoch": 60.833333333333336,
+      "grad_norm": 1.61800217628479,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 18980
+    },
+    {
+      "epoch": 60.86538461538461,
+      "grad_norm": 1.545074701309204,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 18990
+    },
+    {
+      "epoch": 60.8974358974359,
+      "grad_norm": 1.6831417083740234,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 19000
+    },
+    {
+      "epoch": 60.92948717948718,
+      "grad_norm": 1.5398935079574585,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19010
+    },
+    {
+      "epoch": 60.96153846153846,
+      "grad_norm": 1.467044711112976,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 19020
+    },
+    {
+      "epoch": 60.993589743589745,
+      "grad_norm": 1.1924155950546265,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 19030
+    },
+    {
+      "epoch": 61.02564102564103,
+      "grad_norm": 1.148616075515747,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 19040
+    },
+    {
+      "epoch": 61.05769230769231,
+      "grad_norm": 1.0978047847747803,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 19050
+    },
+    {
+      "epoch": 61.08974358974359,
+      "grad_norm": 1.060099482536316,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 19060
+    },
+    {
+      "epoch": 61.12179487179487,
+      "grad_norm": 0.8691592216491699,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 19070
+    },
+    {
+      "epoch": 61.15384615384615,
+      "grad_norm": 0.9859428405761719,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 19080
+    },
+    {
+      "epoch": 61.18589743589744,
+      "grad_norm": 0.8158553838729858,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 19090
+    },
+    {
+      "epoch": 61.217948717948715,
+      "grad_norm": 0.9494746327400208,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 19100
+    },
+    {
+      "epoch": 61.25,
+      "grad_norm": 1.2810505628585815,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 19110
+    },
+    {
+      "epoch": 61.282051282051285,
+      "grad_norm": 1.437111735343933,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 19120
+    },
+    {
+      "epoch": 61.31410256410256,
+      "grad_norm": 1.0480616092681885,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19130
+    },
+    {
+      "epoch": 61.34615384615385,
+      "grad_norm": 1.4759416580200195,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 19140
+    },
+    {
+      "epoch": 61.37820512820513,
+      "grad_norm": 1.593888521194458,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 19150
+    },
+    {
+      "epoch": 61.41025641025641,
+      "grad_norm": 1.183132529258728,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 19160
+    },
+    {
+      "epoch": 61.44230769230769,
+      "grad_norm": 1.3196659088134766,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 19170
+    },
+    {
+      "epoch": 61.47435897435897,
+      "grad_norm": 1.4504307508468628,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 19180
+    },
+    {
+      "epoch": 61.506410256410255,
+      "grad_norm": 1.3405909538269043,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 19190
+    },
+    {
+      "epoch": 61.53846153846154,
+      "grad_norm": 1.3851195573806763,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 19200
+    },
+    {
+      "epoch": 61.57051282051282,
+      "grad_norm": 1.3768770694732666,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 19210
+    },
+    {
+      "epoch": 61.6025641025641,
+      "grad_norm": 0.9944049119949341,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 19220
+    },
+    {
+      "epoch": 61.63461538461539,
+      "grad_norm": 1.5631253719329834,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 19230
+    },
+    {
+      "epoch": 61.666666666666664,
+      "grad_norm": 1.3342106342315674,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 19240
+    },
+    {
+      "epoch": 61.69871794871795,
+      "grad_norm": 1.2315316200256348,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 19250
+    },
+    {
+      "epoch": 61.73076923076923,
+      "grad_norm": 1.0863651037216187,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 19260
+    },
+    {
+      "epoch": 61.76282051282051,
+      "grad_norm": 1.221527099609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 19270
+    },
+    {
+      "epoch": 61.794871794871796,
+      "grad_norm": 1.2538185119628906,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 19280
+    },
+    {
+      "epoch": 61.82692307692308,
+      "grad_norm": 1.2475394010543823,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 19290
+    },
+    {
+      "epoch": 61.85897435897436,
+      "grad_norm": 0.9422110319137573,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19300
+    },
+    {
+      "epoch": 61.89102564102564,
+      "grad_norm": 0.9777013659477234,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19310
+    },
+    {
+      "epoch": 61.92307692307692,
+      "grad_norm": 1.240251898765564,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 19320
+    },
+    {
+      "epoch": 61.955128205128204,
+      "grad_norm": 1.0625427961349487,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19330
+    },
+    {
+      "epoch": 61.98717948717949,
+      "grad_norm": 1.52616548538208,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 19340
+    },
+    {
+      "epoch": 62.01923076923077,
+      "grad_norm": 0.906545102596283,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 19350
+    },
+    {
+      "epoch": 62.05128205128205,
+      "grad_norm": 0.9048050045967102,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19360
+    },
+    {
+      "epoch": 62.083333333333336,
+      "grad_norm": 1.0792332887649536,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 19370
+    },
+    {
+      "epoch": 62.11538461538461,
+      "grad_norm": 0.90876305103302,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 19380
+    },
+    {
+      "epoch": 62.1474358974359,
+      "grad_norm": 1.1188457012176514,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 19390
+    },
+    {
+      "epoch": 62.17948717948718,
+      "grad_norm": 1.391676425933838,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19400
+    },
+    {
+      "epoch": 62.21153846153846,
+      "grad_norm": 1.2929073572158813,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 19410
+    },
+    {
+      "epoch": 62.243589743589745,
+      "grad_norm": 1.2122009992599487,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 19420
+    },
+    {
+      "epoch": 62.27564102564103,
+      "grad_norm": 1.6493210792541504,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 19430
+    },
+    {
+      "epoch": 62.30769230769231,
+      "grad_norm": 1.1758599281311035,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 19440
+    },
+    {
+      "epoch": 62.33974358974359,
+      "grad_norm": 1.2287060022354126,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 19450
+    },
+    {
+      "epoch": 62.37179487179487,
+      "grad_norm": 1.0224422216415405,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 19460
+    },
+    {
+      "epoch": 62.40384615384615,
+      "grad_norm": 1.140117883682251,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 19470
+    },
+    {
+      "epoch": 62.43589743589744,
+      "grad_norm": 1.2577097415924072,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 19480
+    },
+    {
+      "epoch": 62.467948717948715,
+      "grad_norm": 1.4739961624145508,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 19490
+    },
+    {
+      "epoch": 62.5,
+      "grad_norm": 1.3620439767837524,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 19500
+    },
+    {
+      "epoch": 62.532051282051285,
+      "grad_norm": 1.255001187324524,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 19510
+    },
+    {
+      "epoch": 62.56410256410256,
+      "grad_norm": 1.3384207487106323,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19520
+    },
+    {
+      "epoch": 62.59615384615385,
+      "grad_norm": 0.9095895886421204,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 19530
+    },
+    {
+      "epoch": 62.62820512820513,
+      "grad_norm": 0.8337371945381165,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 19540
+    },
+    {
+      "epoch": 62.66025641025641,
+      "grad_norm": 4.446610450744629,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 19550
+    },
+    {
+      "epoch": 62.69230769230769,
+      "grad_norm": 2.3526084423065186,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 19560
+    },
+    {
+      "epoch": 62.72435897435897,
+      "grad_norm": 2.019357681274414,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 19570
+    },
+    {
+      "epoch": 62.756410256410255,
+      "grad_norm": 1.6962350606918335,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 19580
+    },
+    {
+      "epoch": 62.78846153846154,
+      "grad_norm": 1.3606287240982056,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 19590
+    },
+    {
+      "epoch": 62.82051282051282,
+      "grad_norm": 1.261122465133667,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 19600
+    },
+    {
+      "epoch": 62.8525641025641,
+      "grad_norm": 1.4540374279022217,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 19610
+    },
+    {
+      "epoch": 62.88461538461539,
+      "grad_norm": 1.1713305711746216,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 19620
+    },
+    {
+      "epoch": 62.916666666666664,
+      "grad_norm": 1.161605715751648,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 19630
+    },
+    {
+      "epoch": 62.94871794871795,
+      "grad_norm": 1.4071753025054932,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 19640
+    },
+    {
+      "epoch": 62.98076923076923,
+      "grad_norm": 1.5455074310302734,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 19650
+    },
+    {
+      "epoch": 63.01282051282051,
+      "grad_norm": 1.438223123550415,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 19660
+    },
+    {
+      "epoch": 63.044871794871796,
+      "grad_norm": 1.266689658164978,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 19670
+    },
+    {
+      "epoch": 63.07692307692308,
+      "grad_norm": 1.5929737091064453,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 19680
+    },
+    {
+      "epoch": 63.10897435897436,
+      "grad_norm": 0.9853650331497192,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 19690
+    },
+    {
+      "epoch": 63.14102564102564,
+      "grad_norm": 1.3139735460281372,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 19700
+    },
+    {
+      "epoch": 63.17307692307692,
+      "grad_norm": 1.2268078327178955,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 19710
+    },
+    {
+      "epoch": 63.205128205128204,
+      "grad_norm": 1.4885138273239136,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 19720
+    },
+    {
+      "epoch": 63.23717948717949,
+      "grad_norm": 1.233221411705017,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 19730
+    },
+    {
+      "epoch": 63.26923076923077,
+      "grad_norm": 0.9820259213447571,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 19740
+    },
+    {
+      "epoch": 63.30128205128205,
+      "grad_norm": 1.2478140592575073,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 19750
+    },
+    {
+      "epoch": 63.333333333333336,
+      "grad_norm": 1.2932506799697876,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 19760
+    },
+    {
+      "epoch": 63.36538461538461,
+      "grad_norm": 1.431780457496643,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 19770
+    },
+    {
+      "epoch": 63.3974358974359,
+      "grad_norm": 1.431242823600769,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 19780
+    },
+    {
+      "epoch": 63.42948717948718,
+      "grad_norm": 1.3104459047317505,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 19790
+    },
+    {
+      "epoch": 63.46153846153846,
+      "grad_norm": 1.546678066253662,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 19800
+    },
+    {
+      "epoch": 63.493589743589745,
+      "grad_norm": 1.4960306882858276,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 19810
+    },
+    {
+      "epoch": 63.52564102564103,
+      "grad_norm": 1.6159216165542603,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 19820
+    },
+    {
+      "epoch": 63.55769230769231,
+      "grad_norm": 1.3992942571640015,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 19830
+    },
+    {
+      "epoch": 63.58974358974359,
+      "grad_norm": 1.0829657316207886,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 19840
+    },
+    {
+      "epoch": 63.62179487179487,
+      "grad_norm": 1.140849232673645,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 19850
+    },
+    {
+      "epoch": 63.65384615384615,
+      "grad_norm": 1.2526427507400513,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 19860
+    },
+    {
+      "epoch": 63.68589743589744,
+      "grad_norm": 1.3013288974761963,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 19870
+    },
+    {
+      "epoch": 63.717948717948715,
+      "grad_norm": 1.3629698753356934,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 19880
+    },
+    {
+      "epoch": 63.75,
+      "grad_norm": 1.1920071840286255,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 19890
+    },
+    {
+      "epoch": 63.782051282051285,
+      "grad_norm": 0.9606302976608276,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 19900
+    },
+    {
+      "epoch": 63.81410256410256,
+      "grad_norm": 1.1831989288330078,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 19910
+    },
+    {
+      "epoch": 63.84615384615385,
+      "grad_norm": 1.3321900367736816,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19920
+    },
+    {
+      "epoch": 63.87820512820513,
+      "grad_norm": 1.6233710050582886,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 19930
+    },
+    {
+      "epoch": 63.91025641025641,
+      "grad_norm": 1.4060050249099731,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 19940
+    },
+    {
+      "epoch": 63.94230769230769,
+      "grad_norm": 1.0801221132278442,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 19950
+    },
+    {
+      "epoch": 63.97435897435897,
+      "grad_norm": 1.1020352840423584,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 19960
+    },
+    {
+      "epoch": 64.00641025641026,
+      "grad_norm": 1.4181499481201172,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 19970
+    },
+    {
+      "epoch": 64.03846153846153,
+      "grad_norm": 1.4832642078399658,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 19980
+    },
+    {
+      "epoch": 64.07051282051282,
+      "grad_norm": 1.1693286895751953,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 19990
+    },
+    {
+      "epoch": 64.1025641025641,
+      "grad_norm": 1.0700246095657349,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 20000
+    },
+    {
+      "epoch": 64.13461538461539,
+      "grad_norm": 0.8735447525978088,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 20010
+    },
+    {
+      "epoch": 64.16666666666667,
+      "grad_norm": 0.9818481802940369,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 20020
+    },
+    {
+      "epoch": 64.19871794871794,
+      "grad_norm": 1.858477234840393,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 20030
+    },
+    {
+      "epoch": 64.23076923076923,
+      "grad_norm": 2.1235194206237793,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 20040
+    },
+    {
+      "epoch": 64.26282051282051,
+      "grad_norm": 1.8485760688781738,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 20050
+    },
+    {
+      "epoch": 64.2948717948718,
+      "grad_norm": 2.1517722606658936,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 20060
+    },
+    {
+      "epoch": 64.32692307692308,
+      "grad_norm": 1.3691585063934326,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 20070
+    },
+    {
+      "epoch": 64.35897435897436,
+      "grad_norm": 1.1294678449630737,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 20080
+    },
+    {
+      "epoch": 64.39102564102564,
+      "grad_norm": 1.23639976978302,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 20090
+    },
+    {
+      "epoch": 64.42307692307692,
+      "grad_norm": 1.1775705814361572,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 20100
+    },
+    {
+      "epoch": 64.4551282051282,
+      "grad_norm": 1.2619948387145996,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 20110
+    },
+    {
+      "epoch": 64.48717948717949,
+      "grad_norm": 1.2052122354507446,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 20120
+    },
+    {
+      "epoch": 64.51923076923077,
+      "grad_norm": 0.9637640118598938,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 20130
+    },
+    {
+      "epoch": 64.55128205128206,
+      "grad_norm": 1.70613431930542,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 20140
+    },
+    {
+      "epoch": 64.58333333333333,
+      "grad_norm": 1.0535590648651123,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 20150
+    },
+    {
+      "epoch": 64.61538461538461,
+      "grad_norm": 1.1551158428192139,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 20160
+    },
+    {
+      "epoch": 64.6474358974359,
+      "grad_norm": 1.1508761644363403,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 20170
+    },
+    {
+      "epoch": 64.67948717948718,
+      "grad_norm": 1.0730911493301392,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 20180
+    },
+    {
+      "epoch": 64.71153846153847,
+      "grad_norm": 1.5915906429290771,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 20190
+    },
+    {
+      "epoch": 64.74358974358974,
+      "grad_norm": 0.9380902051925659,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 20200
+    },
+    {
+      "epoch": 64.77564102564102,
+      "grad_norm": 1.6945871114730835,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 20210
+    },
+    {
+      "epoch": 64.8076923076923,
+      "grad_norm": 1.536543607711792,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 20220
+    },
+    {
+      "epoch": 64.83974358974359,
+      "grad_norm": 1.4294445514678955,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 20230
+    },
+    {
+      "epoch": 64.87179487179488,
+      "grad_norm": 1.1754438877105713,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 20240
+    },
+    {
+      "epoch": 64.90384615384616,
+      "grad_norm": 1.0873045921325684,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 20250
+    },
+    {
+      "epoch": 64.93589743589743,
+      "grad_norm": 1.188880205154419,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 20260
+    },
+    {
+      "epoch": 64.96794871794872,
+      "grad_norm": 1.2947502136230469,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 20270
+    },
+    {
+      "epoch": 65.0,
+      "grad_norm": 1.0750346183776855,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 20280
+    },
+    {
+      "epoch": 65.03205128205128,
+      "grad_norm": 1.1244451999664307,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 20290
+    },
+    {
+      "epoch": 65.06410256410257,
+      "grad_norm": 1.2410163879394531,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 20300
+    },
+    {
+      "epoch": 65.09615384615384,
+      "grad_norm": 1.228567123413086,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 20310
+    },
+    {
+      "epoch": 65.12820512820512,
+      "grad_norm": 1.3367619514465332,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 20320
+    },
+    {
+      "epoch": 65.16025641025641,
+      "grad_norm": 1.3260774612426758,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 20330
+    },
+    {
+      "epoch": 65.1923076923077,
+      "grad_norm": 1.1298267841339111,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 20340
+    },
+    {
+      "epoch": 65.22435897435898,
+      "grad_norm": 1.2249096632003784,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 20350
+    },
+    {
+      "epoch": 65.25641025641026,
+      "grad_norm": 1.246633768081665,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 20360
+    },
+    {
+      "epoch": 65.28846153846153,
+      "grad_norm": 1.514626145362854,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 20370
+    },
+    {
+      "epoch": 65.32051282051282,
+      "grad_norm": 1.4261901378631592,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 20380
+    },
+    {
+      "epoch": 65.3525641025641,
+      "grad_norm": 1.2556589841842651,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 20390
+    },
+    {
+      "epoch": 65.38461538461539,
+      "grad_norm": 1.1882492303848267,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 20400
+    },
+    {
+      "epoch": 65.41666666666667,
+      "grad_norm": 1.1384873390197754,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 20410
+    },
+    {
+      "epoch": 65.44871794871794,
+      "grad_norm": 1.100624680519104,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 20420
+    },
+    {
+      "epoch": 65.48076923076923,
+      "grad_norm": 1.486050009727478,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 20430
+    },
+    {
+      "epoch": 65.51282051282051,
+      "grad_norm": 1.3560627698898315,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 20440
+    },
+    {
+      "epoch": 65.5448717948718,
+      "grad_norm": 1.4271970987319946,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 20450
+    },
+    {
+      "epoch": 65.57692307692308,
+      "grad_norm": 1.1529395580291748,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 20460
+    },
+    {
+      "epoch": 65.60897435897436,
+      "grad_norm": 1.2171262502670288,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 20470
+    },
+    {
+      "epoch": 65.64102564102564,
+      "grad_norm": 1.0476348400115967,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 20480
+    },
+    {
+      "epoch": 65.67307692307692,
+      "grad_norm": 1.1601155996322632,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 20490
+    },
+    {
+      "epoch": 65.7051282051282,
+      "grad_norm": 1.1055949926376343,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 20500
+    },
+    {
+      "epoch": 65.73717948717949,
+      "grad_norm": 1.442237138748169,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 20510
+    },
+    {
+      "epoch": 65.76923076923077,
+      "grad_norm": 1.5051391124725342,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 20520
+    },
+    {
+      "epoch": 65.80128205128206,
+      "grad_norm": 1.8782075643539429,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 20530
+    },
+    {
+      "epoch": 65.83333333333333,
+      "grad_norm": 1.6831891536712646,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 20540
+    },
+    {
+      "epoch": 65.86538461538461,
+      "grad_norm": 1.5795146226882935,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 20550
+    },
+    {
+      "epoch": 65.8974358974359,
+      "grad_norm": 1.2715002298355103,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 20560
+    },
+    {
+      "epoch": 65.92948717948718,
+      "grad_norm": 1.4611449241638184,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 20570
+    },
+    {
+      "epoch": 65.96153846153847,
+      "grad_norm": 1.4521058797836304,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 20580
+    },
+    {
+      "epoch": 65.99358974358974,
+      "grad_norm": 1.1715470552444458,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 20590
+    },
+    {
+      "epoch": 66.02564102564102,
+      "grad_norm": 1.3753546476364136,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 20600
+    },
+    {
+      "epoch": 66.0576923076923,
+      "grad_norm": 1.1590269804000854,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 20610
+    },
+    {
+      "epoch": 66.08974358974359,
+      "grad_norm": 1.101024866104126,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 20620
+    },
+    {
+      "epoch": 66.12179487179488,
+      "grad_norm": 1.4786295890808105,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 20630
+    },
+    {
+      "epoch": 66.15384615384616,
+      "grad_norm": 1.432580828666687,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 20640
+    },
+    {
+      "epoch": 66.18589743589743,
+      "grad_norm": 1.0329145193099976,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 20650
+    },
+    {
+      "epoch": 66.21794871794872,
+      "grad_norm": 1.321162223815918,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 20660
+    },
+    {
+      "epoch": 66.25,
+      "grad_norm": 1.3834888935089111,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 20670
+    },
+    {
+      "epoch": 66.28205128205128,
+      "grad_norm": 1.2590265274047852,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 20680
+    },
+    {
+      "epoch": 66.31410256410257,
+      "grad_norm": 1.0004030466079712,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 20690
+    },
+    {
+      "epoch": 66.34615384615384,
+      "grad_norm": 1.2772572040557861,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 20700
+    },
+    {
+      "epoch": 66.37820512820512,
+      "grad_norm": 1.3465126752853394,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 20710
+    },
+    {
+      "epoch": 66.41025641025641,
+      "grad_norm": 1.5117549896240234,
+      "learning_rate": 0.0001,
+      "loss": 0.0138,
+      "step": 20720
+    },
+    {
+      "epoch": 66.4423076923077,
+      "grad_norm": 0.9366006851196289,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 20730
+    },
+    {
+      "epoch": 66.47435897435898,
+      "grad_norm": 1.3556883335113525,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 20740
+    },
+    {
+      "epoch": 66.50641025641026,
+      "grad_norm": 1.050779938697815,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 20750
+    },
+    {
+      "epoch": 66.53846153846153,
+      "grad_norm": 1.0957973003387451,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 20760
+    },
+    {
+      "epoch": 66.57051282051282,
+      "grad_norm": 1.4670192003250122,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 20770
+    },
+    {
+      "epoch": 66.6025641025641,
+      "grad_norm": 1.1430222988128662,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 20780
+    },
+    {
+      "epoch": 66.63461538461539,
+      "grad_norm": 1.2880457639694214,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 20790
+    },
+    {
+      "epoch": 66.66666666666667,
+      "grad_norm": 1.1442592144012451,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 20800
+    },
+    {
+      "epoch": 66.69871794871794,
+      "grad_norm": 1.167825698852539,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 20810
+    },
+    {
+      "epoch": 66.73076923076923,
+      "grad_norm": 1.302177906036377,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 20820
+    },
+    {
+      "epoch": 66.76282051282051,
+      "grad_norm": 1.2192535400390625,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 20830
+    },
+    {
+      "epoch": 66.7948717948718,
+      "grad_norm": 1.2615104913711548,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 20840
+    },
+    {
+      "epoch": 66.82692307692308,
+      "grad_norm": 1.2252488136291504,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 20850
+    },
+    {
+      "epoch": 66.85897435897436,
+      "grad_norm": 0.9252914190292358,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 20860
+    },
+    {
+      "epoch": 66.89102564102564,
+      "grad_norm": 0.8750860095024109,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 20870
+    },
+    {
+      "epoch": 66.92307692307692,
+      "grad_norm": 1.106040358543396,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 20880
+    },
+    {
+      "epoch": 66.9551282051282,
+      "grad_norm": 1.0484557151794434,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 20890
+    },
+    {
+      "epoch": 66.98717948717949,
+      "grad_norm": 0.9983659982681274,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 20900
+    },
+    {
+      "epoch": 67.01923076923077,
+      "grad_norm": 1.1089123487472534,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 20910
+    },
+    {
+      "epoch": 67.05128205128206,
+      "grad_norm": 0.9915053844451904,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 20920
+    },
+    {
+      "epoch": 67.08333333333333,
+      "grad_norm": 1.6002578735351562,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 20930
+    },
+    {
+      "epoch": 67.11538461538461,
+      "grad_norm": 1.6830337047576904,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 20940
+    },
+    {
+      "epoch": 67.1474358974359,
+      "grad_norm": 1.900041937828064,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 20950
+    },
+    {
+      "epoch": 67.17948717948718,
+      "grad_norm": 1.5794618129730225,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 20960
+    },
+    {
+      "epoch": 67.21153846153847,
+      "grad_norm": 1.244606375694275,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 20970
+    },
+    {
+      "epoch": 67.24358974358974,
+      "grad_norm": 1.1785281896591187,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 20980
+    },
+    {
+      "epoch": 67.27564102564102,
+      "grad_norm": 1.1645562648773193,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 20990
+    },
+    {
+      "epoch": 67.3076923076923,
+      "grad_norm": 1.395159363746643,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 21000
+    },
+    {
+      "epoch": 67.33974358974359,
+      "grad_norm": 1.209384799003601,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 21010
+    },
+    {
+      "epoch": 67.37179487179488,
+      "grad_norm": 1.4062445163726807,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 21020
+    },
+    {
+      "epoch": 67.40384615384616,
+      "grad_norm": 1.318103313446045,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 21030
+    },
+    {
+      "epoch": 67.43589743589743,
+      "grad_norm": 0.9804000854492188,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 21040
+    },
+    {
+      "epoch": 67.46794871794872,
+      "grad_norm": 1.2302172183990479,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 21050
+    },
+    {
+      "epoch": 67.5,
+      "grad_norm": 1.3402796983718872,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 21060
+    },
+    {
+      "epoch": 67.53205128205128,
+      "grad_norm": 0.9791256189346313,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 21070
+    },
+    {
+      "epoch": 67.56410256410257,
+      "grad_norm": 1.1994988918304443,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 21080
+    },
+    {
+      "epoch": 67.59615384615384,
+      "grad_norm": 1.1227425336837769,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 21090
+    },
+    {
+      "epoch": 67.62820512820512,
+      "grad_norm": 0.8565211892127991,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 21100
+    },
+    {
+      "epoch": 67.66025641025641,
+      "grad_norm": 1.1558128595352173,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 21110
+    },
+    {
+      "epoch": 67.6923076923077,
+      "grad_norm": 1.0336121320724487,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 21120
+    },
+    {
+      "epoch": 67.72435897435898,
+      "grad_norm": 1.2457127571105957,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 21130
+    },
+    {
+      "epoch": 67.75641025641026,
+      "grad_norm": 1.183409571647644,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 21140
+    },
+    {
+      "epoch": 67.78846153846153,
+      "grad_norm": 1.1405125856399536,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 21150
+    },
+    {
+      "epoch": 67.82051282051282,
+      "grad_norm": 1.0883015394210815,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 21160
+    },
+    {
+      "epoch": 67.8525641025641,
+      "grad_norm": 1.0231276750564575,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 21170
+    },
+    {
+      "epoch": 67.88461538461539,
+      "grad_norm": 1.0373027324676514,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 21180
+    },
+    {
+      "epoch": 67.91666666666667,
+      "grad_norm": 0.7807128429412842,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 21190
+    },
+    {
+      "epoch": 67.94871794871794,
+      "grad_norm": 1.07514226436615,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 21200
+    },
+    {
+      "epoch": 67.98076923076923,
+      "grad_norm": 0.77271968126297,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 21210
+    },
+    {
+      "epoch": 68.01282051282051,
+      "grad_norm": 1.0570074319839478,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 21220
+    },
+    {
+      "epoch": 68.0448717948718,
+      "grad_norm": 1.1132692098617554,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 21230
+    },
+    {
+      "epoch": 68.07692307692308,
+      "grad_norm": 0.9430263638496399,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 21240
+    },
+    {
+      "epoch": 68.10897435897436,
+      "grad_norm": 1.0121809244155884,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 21250
+    },
+    {
+      "epoch": 68.14102564102564,
+      "grad_norm": 1.0029675960540771,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 21260
+    },
+    {
+      "epoch": 68.17307692307692,
+      "grad_norm": 1.2481883764266968,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 21270
+    },
+    {
+      "epoch": 68.2051282051282,
+      "grad_norm": 1.024654746055603,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 21280
+    },
+    {
+      "epoch": 68.23717948717949,
+      "grad_norm": 0.9101797938346863,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 21290
+    },
+    {
+      "epoch": 68.26923076923077,
+      "grad_norm": 1.1160744428634644,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 21300
+    },
+    {
+      "epoch": 68.30128205128206,
+      "grad_norm": 0.8344446420669556,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 21310
+    },
+    {
+      "epoch": 68.33333333333333,
+      "grad_norm": 1.2929381132125854,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 21320
+    },
+    {
+      "epoch": 68.36538461538461,
+      "grad_norm": 1.1849400997161865,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 21330
+    },
+    {
+      "epoch": 68.3974358974359,
+      "grad_norm": 0.9838926792144775,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 21340
+    },
+    {
+      "epoch": 68.42948717948718,
+      "grad_norm": 1.2018362283706665,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 21350
+    },
+    {
+      "epoch": 68.46153846153847,
+      "grad_norm": 1.0491492748260498,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 21360
+    },
+    {
+      "epoch": 68.49358974358974,
+      "grad_norm": 1.079832673072815,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 21370
+    },
+    {
+      "epoch": 68.52564102564102,
+      "grad_norm": 0.9626755118370056,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 21380
+    },
+    {
+      "epoch": 68.5576923076923,
+      "grad_norm": 0.8196954131126404,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 21390
+    },
+    {
+      "epoch": 68.58974358974359,
+      "grad_norm": 1.1458532810211182,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 21400
+    },
+    {
+      "epoch": 68.62179487179488,
+      "grad_norm": 0.9018722176551819,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 21410
+    },
+    {
+      "epoch": 68.65384615384616,
+      "grad_norm": 0.9204859137535095,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 21420
+    },
+    {
+      "epoch": 68.68589743589743,
+      "grad_norm": 1.224970817565918,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 21430
+    },
+    {
+      "epoch": 68.71794871794872,
+      "grad_norm": 1.474047064781189,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 21440
+    },
+    {
+      "epoch": 68.75,
+      "grad_norm": 1.3234896659851074,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 21450
+    },
+    {
+      "epoch": 68.78205128205128,
+      "grad_norm": 1.1033358573913574,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 21460
+    },
+    {
+      "epoch": 68.81410256410257,
+      "grad_norm": 0.9653211832046509,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 21470
+    },
+    {
+      "epoch": 68.84615384615384,
+      "grad_norm": 1.142232894897461,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 21480
+    },
+    {
+      "epoch": 68.87820512820512,
+      "grad_norm": 0.837661862373352,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 21490
+    },
+    {
+      "epoch": 68.91025641025641,
+      "grad_norm": 0.6845487952232361,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 21500
+    },
+    {
+      "epoch": 68.9423076923077,
+      "grad_norm": 0.8773005604743958,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 21510
+    },
+    {
+      "epoch": 68.97435897435898,
+      "grad_norm": 0.8137258291244507,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 21520
+    },
+    {
+      "epoch": 69.00641025641026,
+      "grad_norm": 1.4425920248031616,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 21530
+    },
+    {
+      "epoch": 69.03846153846153,
+      "grad_norm": 1.2688331604003906,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 21540
+    },
+    {
+      "epoch": 69.07051282051282,
+      "grad_norm": 1.1128665208816528,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 21550
+    },
+    {
+      "epoch": 69.1025641025641,
+      "grad_norm": 1.0490168333053589,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 21560
+    },
+    {
+      "epoch": 69.13461538461539,
+      "grad_norm": 1.0743523836135864,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 21570
+    },
+    {
+      "epoch": 69.16666666666667,
+      "grad_norm": 0.8655350804328918,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 21580
+    },
+    {
+      "epoch": 69.19871794871794,
+      "grad_norm": 1.0281404256820679,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 21590
+    },
+    {
+      "epoch": 69.23076923076923,
+      "grad_norm": 1.1021145582199097,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 21600
+    },
+    {
+      "epoch": 69.26282051282051,
+      "grad_norm": 0.9166769981384277,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 21610
+    },
+    {
+      "epoch": 69.2948717948718,
+      "grad_norm": 1.030643343925476,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 21620
+    },
+    {
+      "epoch": 69.32692307692308,
+      "grad_norm": 0.9919961094856262,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 21630
+    },
+    {
+      "epoch": 69.35897435897436,
+      "grad_norm": 1.1842389106750488,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 21640
+    },
+    {
+      "epoch": 69.39102564102564,
+      "grad_norm": 1.3451770544052124,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 21650
+    },
+    {
+      "epoch": 69.42307692307692,
+      "grad_norm": 1.5017776489257812,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 21660
+    },
+    {
+      "epoch": 69.4551282051282,
+      "grad_norm": 1.6675996780395508,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 21670
+    },
+    {
+      "epoch": 69.48717948717949,
+      "grad_norm": 1.3192131519317627,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 21680
+    },
+    {
+      "epoch": 69.51923076923077,
+      "grad_norm": 1.0303090810775757,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 21690
+    },
+    {
+      "epoch": 69.55128205128206,
+      "grad_norm": 1.1584347486495972,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 21700
+    },
+    {
+      "epoch": 69.58333333333333,
+      "grad_norm": 1.2209255695343018,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 21710
+    },
+    {
+      "epoch": 69.61538461538461,
+      "grad_norm": 1.2050788402557373,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 21720
+    },
+    {
+      "epoch": 69.6474358974359,
+      "grad_norm": 0.9080017805099487,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 21730
+    },
+    {
+      "epoch": 69.67948717948718,
+      "grad_norm": 1.0606938600540161,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 21740
+    },
+    {
+      "epoch": 69.71153846153847,
+      "grad_norm": 1.168997049331665,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 21750
+    },
+    {
+      "epoch": 69.74358974358974,
+      "grad_norm": 1.17426598072052,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 21760
+    },
+    {
+      "epoch": 69.77564102564102,
+      "grad_norm": 1.0944291353225708,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 21770
+    },
+    {
+      "epoch": 69.8076923076923,
+      "grad_norm": 1.8605629205703735,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 21780
+    },
+    {
+      "epoch": 69.83974358974359,
+      "grad_norm": 1.4447283744812012,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 21790
+    },
+    {
+      "epoch": 69.87179487179488,
+      "grad_norm": 1.3026185035705566,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 21800
+    },
+    {
+      "epoch": 69.90384615384616,
+      "grad_norm": 1.6348493099212646,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 21810
+    },
+    {
+      "epoch": 69.93589743589743,
+      "grad_norm": 0.943996012210846,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 21820
+    },
+    {
+      "epoch": 69.96794871794872,
+      "grad_norm": 1.0233864784240723,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 21830
+    },
+    {
+      "epoch": 70.0,
+      "grad_norm": 1.1841388940811157,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 21840
+    },
+    {
+      "epoch": 70.03205128205128,
+      "grad_norm": 1.411307454109192,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 21850
+    },
+    {
+      "epoch": 70.06410256410257,
+      "grad_norm": 0.7894370555877686,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 21860
+    },
+    {
+      "epoch": 70.09615384615384,
+      "grad_norm": 1.17959463596344,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 21870
+    },
+    {
+      "epoch": 70.12820512820512,
+      "grad_norm": 0.8727943301200867,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 21880
+    },
+    {
+      "epoch": 70.16025641025641,
+      "grad_norm": 0.8408998250961304,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 21890
+    },
+    {
+      "epoch": 70.1923076923077,
+      "grad_norm": 0.7969034314155579,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 21900
+    },
+    {
+      "epoch": 70.22435897435898,
+      "grad_norm": 0.808008074760437,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 21910
+    },
+    {
+      "epoch": 70.25641025641026,
+      "grad_norm": 1.201475739479065,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 21920
+    },
+    {
+      "epoch": 70.28846153846153,
+      "grad_norm": 0.815958559513092,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 21930
+    },
+    {
+      "epoch": 70.32051282051282,
+      "grad_norm": 1.3786721229553223,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 21940
+    },
+    {
+      "epoch": 70.3525641025641,
+      "grad_norm": 0.8593316078186035,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 21950
+    },
+    {
+      "epoch": 70.38461538461539,
+      "grad_norm": 1.5219522714614868,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 21960
+    },
+    {
+      "epoch": 70.41666666666667,
+      "grad_norm": 1.569323182106018,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 21970
+    },
+    {
+      "epoch": 70.44871794871794,
+      "grad_norm": 1.441701054573059,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 21980
+    },
+    {
+      "epoch": 70.48076923076923,
+      "grad_norm": 1.3184441328048706,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 21990
+    },
+    {
+      "epoch": 70.51282051282051,
+      "grad_norm": 1.444514513015747,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 22000
+    },
+    {
+      "epoch": 70.5448717948718,
+      "grad_norm": 1.2359987497329712,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 22010
+    },
+    {
+      "epoch": 70.57692307692308,
+      "grad_norm": 1.022252082824707,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 22020
+    },
+    {
+      "epoch": 70.60897435897436,
+      "grad_norm": 1.239318609237671,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 22030
+    },
+    {
+      "epoch": 70.64102564102564,
+      "grad_norm": 1.0253301858901978,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 22040
+    },
+    {
+      "epoch": 70.67307692307692,
+      "grad_norm": 1.0989872217178345,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 22050
+    },
+    {
+      "epoch": 70.7051282051282,
+      "grad_norm": 0.8639981746673584,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 22060
+    },
+    {
+      "epoch": 70.73717948717949,
+      "grad_norm": 1.1387650966644287,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 22070
+    },
+    {
+      "epoch": 70.76923076923077,
+      "grad_norm": 1.1771845817565918,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 22080
+    },
+    {
+      "epoch": 70.80128205128206,
+      "grad_norm": 1.110611915588379,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 22090
+    },
+    {
+      "epoch": 70.83333333333333,
+      "grad_norm": 0.9870943427085876,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 22100
+    },
+    {
+      "epoch": 70.86538461538461,
+      "grad_norm": 1.487204909324646,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 22110
+    },
+    {
+      "epoch": 70.8974358974359,
+      "grad_norm": 1.2206605672836304,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 22120
+    },
+    {
+      "epoch": 70.92948717948718,
+      "grad_norm": 0.7777026891708374,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 22130
+    },
+    {
+      "epoch": 70.96153846153847,
+      "grad_norm": 1.0929845571517944,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 22140
+    },
+    {
+      "epoch": 70.99358974358974,
+      "grad_norm": 1.1708590984344482,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 22150
+    },
+    {
+      "epoch": 71.02564102564102,
+      "grad_norm": 0.9005795121192932,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 22160
+    },
+    {
+      "epoch": 71.0576923076923,
+      "grad_norm": 0.951298177242279,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 22170
+    },
+    {
+      "epoch": 71.08974358974359,
+      "grad_norm": 1.233211636543274,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 22180
+    },
+    {
+      "epoch": 71.12179487179488,
+      "grad_norm": 1.4104111194610596,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 22190
+    },
+    {
+      "epoch": 71.15384615384616,
+      "grad_norm": 1.0660350322723389,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 22200
+    },
+    {
+      "epoch": 71.18589743589743,
+      "grad_norm": 1.2588825225830078,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 22210
+    },
+    {
+      "epoch": 71.21794871794872,
+      "grad_norm": 1.5880379676818848,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 22220
+    },
+    {
+      "epoch": 71.25,
+      "grad_norm": 1.0948808193206787,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 22230
+    },
+    {
+      "epoch": 71.28205128205128,
+      "grad_norm": 0.978401243686676,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 22240
+    },
+    {
+      "epoch": 71.31410256410257,
+      "grad_norm": 1.0528440475463867,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 22250
+    },
+    {
+      "epoch": 71.34615384615384,
+      "grad_norm": 1.6622352600097656,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 22260
+    },
+    {
+      "epoch": 71.37820512820512,
+      "grad_norm": 1.3432953357696533,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 22270
+    },
+    {
+      "epoch": 71.41025641025641,
+      "grad_norm": 1.5021804571151733,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 22280
+    },
+    {
+      "epoch": 71.4423076923077,
+      "grad_norm": 1.2206896543502808,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 22290
+    },
+    {
+      "epoch": 71.47435897435898,
+      "grad_norm": 1.6573207378387451,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 22300
+    },
+    {
+      "epoch": 71.50641025641026,
+      "grad_norm": 2.1175267696380615,
+      "learning_rate": 0.0001,
+      "loss": 0.0131,
+      "step": 22310
+    },
+    {
+      "epoch": 71.53846153846153,
+      "grad_norm": 1.6601604223251343,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 22320
+    },
+    {
+      "epoch": 71.57051282051282,
+      "grad_norm": 1.545833945274353,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 22330
+    },
+    {
+      "epoch": 71.6025641025641,
+      "grad_norm": 1.608451247215271,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 22340
+    },
+    {
+      "epoch": 71.63461538461539,
+      "grad_norm": 1.4358985424041748,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 22350
+    },
+    {
+      "epoch": 71.66666666666667,
+      "grad_norm": 1.4596225023269653,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 22360
+    },
+    {
+      "epoch": 71.69871794871794,
+      "grad_norm": 1.2297017574310303,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 22370
+    },
+    {
+      "epoch": 71.73076923076923,
+      "grad_norm": 1.45010244846344,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 22380
+    },
+    {
+      "epoch": 71.76282051282051,
+      "grad_norm": 1.198768973350525,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 22390
+    },
+    {
+      "epoch": 71.7948717948718,
+      "grad_norm": 1.3941881656646729,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 22400
+    },
+    {
+      "epoch": 71.82692307692308,
+      "grad_norm": 3.170471668243408,
+      "learning_rate": 0.0001,
+      "loss": 0.0162,
+      "step": 22410
+    },
+    {
+      "epoch": 71.85897435897436,
+      "grad_norm": 2.29394793510437,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 22420
+    },
+    {
+      "epoch": 71.89102564102564,
+      "grad_norm": 1.8145262002944946,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 22430
+    },
+    {
+      "epoch": 71.92307692307692,
+      "grad_norm": 1.560196042060852,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 22440
+    },
+    {
+      "epoch": 71.9551282051282,
+      "grad_norm": 1.4545729160308838,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 22450
+    },
+    {
+      "epoch": 71.98717948717949,
+      "grad_norm": 1.619586706161499,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 22460
+    },
+    {
+      "epoch": 72.01923076923077,
+      "grad_norm": 1.3181650638580322,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 22470
+    },
+    {
+      "epoch": 72.05128205128206,
+      "grad_norm": 1.2641513347625732,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 22480
+    },
+    {
+      "epoch": 72.08333333333333,
+      "grad_norm": 1.223113775253296,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 22490
+    },
+    {
+      "epoch": 72.11538461538461,
+      "grad_norm": 1.4815893173217773,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 22500
+    },
+    {
+      "epoch": 72.1474358974359,
+      "grad_norm": 1.38503098487854,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 22510
+    },
+    {
+      "epoch": 72.17948717948718,
+      "grad_norm": 1.2423961162567139,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 22520
+    },
+    {
+      "epoch": 72.21153846153847,
+      "grad_norm": 1.551204800605774,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 22530
+    },
+    {
+      "epoch": 72.24358974358974,
+      "grad_norm": 1.2142387628555298,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 22540
+    },
+    {
+      "epoch": 72.27564102564102,
+      "grad_norm": 1.162362813949585,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 22550
+    },
+    {
+      "epoch": 72.3076923076923,
+      "grad_norm": 1.0226199626922607,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 22560
+    },
+    {
+      "epoch": 72.33974358974359,
+      "grad_norm": 1.2889257669448853,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 22570
+    },
+    {
+      "epoch": 72.37179487179488,
+      "grad_norm": 1.1243654489517212,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 22580
+    },
+    {
+      "epoch": 72.40384615384616,
+      "grad_norm": 1.091307282447815,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 22590
+    },
+    {
+      "epoch": 72.43589743589743,
+      "grad_norm": 0.8900591135025024,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 22600
+    },
+    {
+      "epoch": 72.46794871794872,
+      "grad_norm": 1.103355884552002,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 22610
+    },
+    {
+      "epoch": 72.5,
+      "grad_norm": 1.113856315612793,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 22620
+    },
+    {
+      "epoch": 72.53205128205128,
+      "grad_norm": 1.153470516204834,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 22630
+    },
+    {
+      "epoch": 72.56410256410257,
+      "grad_norm": 0.9256693720817566,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 22640
+    },
+    {
+      "epoch": 72.59615384615384,
+      "grad_norm": 1.1751114130020142,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 22650
+    },
+    {
+      "epoch": 72.62820512820512,
+      "grad_norm": 1.045146107673645,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 22660
+    },
+    {
+      "epoch": 72.66025641025641,
+      "grad_norm": 1.317740797996521,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 22670
+    },
+    {
+      "epoch": 72.6923076923077,
+      "grad_norm": 0.989246129989624,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 22680
+    },
+    {
+      "epoch": 72.72435897435898,
+      "grad_norm": 1.1197302341461182,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 22690
+    },
+    {
+      "epoch": 72.75641025641026,
+      "grad_norm": 1.0313619375228882,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 22700
+    },
+    {
+      "epoch": 72.78846153846153,
+      "grad_norm": 0.8264133334159851,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 22710
+    },
+    {
+      "epoch": 72.82051282051282,
+      "grad_norm": 0.7796728014945984,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 22720
+    },
+    {
+      "epoch": 72.8525641025641,
+      "grad_norm": 1.0669898986816406,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 22730
+    },
+    {
+      "epoch": 72.88461538461539,
+      "grad_norm": 0.9533177614212036,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 22740
+    },
+    {
+      "epoch": 72.91666666666667,
+      "grad_norm": 1.2003836631774902,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 22750
+    },
+    {
+      "epoch": 72.94871794871794,
+      "grad_norm": 1.4166383743286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 22760
+    },
+    {
+      "epoch": 72.98076923076923,
+      "grad_norm": 1.219267725944519,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 22770
+    },
+    {
+      "epoch": 73.01282051282051,
+      "grad_norm": 1.3406226634979248,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 22780
+    },
+    {
+      "epoch": 73.0448717948718,
+      "grad_norm": 1.0167282819747925,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 22790
+    },
+    {
+      "epoch": 73.07692307692308,
+      "grad_norm": 0.7454362511634827,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 22800
+    },
+    {
+      "epoch": 73.10897435897436,
+      "grad_norm": 1.1204825639724731,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 22810
+    },
+    {
+      "epoch": 73.14102564102564,
+      "grad_norm": 1.3630123138427734,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 22820
+    },
+    {
+      "epoch": 73.17307692307692,
+      "grad_norm": 1.788458228111267,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 22830
+    },
+    {
+      "epoch": 73.2051282051282,
+      "grad_norm": 1.2656046152114868,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 22840
+    },
+    {
+      "epoch": 73.23717948717949,
+      "grad_norm": 1.2695754766464233,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 22850
+    },
+    {
+      "epoch": 73.26923076923077,
+      "grad_norm": 0.9540714025497437,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 22860
+    },
+    {
+      "epoch": 73.30128205128206,
+      "grad_norm": 1.1862077713012695,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 22870
+    },
+    {
+      "epoch": 73.33333333333333,
+      "grad_norm": 1.233452320098877,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 22880
+    },
+    {
+      "epoch": 73.36538461538461,
+      "grad_norm": 1.1547080278396606,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 22890
+    },
+    {
+      "epoch": 73.3974358974359,
+      "grad_norm": 1.1446338891983032,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 22900
+    },
+    {
+      "epoch": 73.42948717948718,
+      "grad_norm": 1.1892647743225098,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 22910
+    },
+    {
+      "epoch": 73.46153846153847,
+      "grad_norm": 1.555613398551941,
+      "learning_rate": 0.0001,
+      "loss": 0.0126,
+      "step": 22920
+    },
+    {
+      "epoch": 73.49358974358974,
+      "grad_norm": 1.5941691398620605,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 22930
+    },
+    {
+      "epoch": 73.52564102564102,
+      "grad_norm": 1.2884693145751953,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 22940
+    },
+    {
+      "epoch": 73.5576923076923,
+      "grad_norm": 1.4644513130187988,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 22950
+    },
+    {
+      "epoch": 73.58974358974359,
+      "grad_norm": 0.9223923683166504,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 22960
+    },
+    {
+      "epoch": 73.62179487179488,
+      "grad_norm": 1.1907747983932495,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 22970
+    },
+    {
+      "epoch": 73.65384615384616,
+      "grad_norm": 1.0113152265548706,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 22980
+    },
+    {
+      "epoch": 73.68589743589743,
+      "grad_norm": 1.130860447883606,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 22990
+    },
+    {
+      "epoch": 73.71794871794872,
+      "grad_norm": 0.9200186133384705,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 23000
+    },
+    {
+      "epoch": 73.75,
+      "grad_norm": 1.0416611433029175,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 23010
+    },
+    {
+      "epoch": 73.78205128205128,
+      "grad_norm": 1.0671043395996094,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 23020
+    },
+    {
+      "epoch": 73.81410256410257,
+      "grad_norm": 0.9657081365585327,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 23030
+    },
+    {
+      "epoch": 73.84615384615384,
+      "grad_norm": 0.8526078462600708,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 23040
+    },
+    {
+      "epoch": 73.87820512820512,
+      "grad_norm": 1.5561065673828125,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 23050
+    },
+    {
+      "epoch": 73.91025641025641,
+      "grad_norm": 1.6794977188110352,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 23060
+    },
+    {
+      "epoch": 73.9423076923077,
+      "grad_norm": 1.2285000085830688,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 23070
+    },
+    {
+      "epoch": 73.97435897435898,
+      "grad_norm": 1.009986400604248,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 23080
+    },
+    {
+      "epoch": 74.00641025641026,
+      "grad_norm": 1.0035324096679688,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 23090
+    },
+    {
+      "epoch": 74.03846153846153,
+      "grad_norm": 1.008438229560852,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 23100
+    },
+    {
+      "epoch": 74.07051282051282,
+      "grad_norm": 0.9231150150299072,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 23110
+    },
+    {
+      "epoch": 74.1025641025641,
+      "grad_norm": 0.9562298655509949,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 23120
+    },
+    {
+      "epoch": 74.13461538461539,
+      "grad_norm": 0.8329399824142456,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 23130
+    },
+    {
+      "epoch": 74.16666666666667,
+      "grad_norm": 1.016670823097229,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 23140
+    },
+    {
+      "epoch": 74.19871794871794,
+      "grad_norm": 1.2349071502685547,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 23150
+    },
+    {
+      "epoch": 74.23076923076923,
+      "grad_norm": 0.8963037133216858,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 23160
+    },
+    {
+      "epoch": 74.26282051282051,
+      "grad_norm": 1.2074215412139893,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 23170
+    },
+    {
+      "epoch": 74.2948717948718,
+      "grad_norm": 1.1156834363937378,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 23180
+    },
+    {
+      "epoch": 74.32692307692308,
+      "grad_norm": 1.0876637697219849,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 23190
+    },
+    {
+      "epoch": 74.35897435897436,
+      "grad_norm": 1.3397687673568726,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 23200
+    },
+    {
+      "epoch": 74.39102564102564,
+      "grad_norm": 1.2300732135772705,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 23210
+    },
+    {
+      "epoch": 74.42307692307692,
+      "grad_norm": 0.9896567463874817,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 23220
+    },
+    {
+      "epoch": 74.4551282051282,
+      "grad_norm": 1.3409839868545532,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 23230
+    },
+    {
+      "epoch": 74.48717948717949,
+      "grad_norm": 1.5044469833374023,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 23240
+    },
+    {
+      "epoch": 74.51923076923077,
+      "grad_norm": 1.497040033340454,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 23250
+    },
+    {
+      "epoch": 74.55128205128206,
+      "grad_norm": 1.385583758354187,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 23260
+    },
+    {
+      "epoch": 74.58333333333333,
+      "grad_norm": 1.4853061437606812,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 23270
+    },
+    {
+      "epoch": 74.61538461538461,
+      "grad_norm": 1.3121049404144287,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 23280
+    },
+    {
+      "epoch": 74.6474358974359,
+      "grad_norm": 1.4631023406982422,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 23290
+    },
+    {
+      "epoch": 74.67948717948718,
+      "grad_norm": 1.022379755973816,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 23300
+    },
+    {
+      "epoch": 74.71153846153847,
+      "grad_norm": 1.0721116065979004,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 23310
+    },
+    {
+      "epoch": 74.74358974358974,
+      "grad_norm": 0.8957459926605225,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 23320
+    },
+    {
+      "epoch": 74.77564102564102,
+      "grad_norm": 0.958529531955719,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 23330
+    },
+    {
+      "epoch": 74.8076923076923,
+      "grad_norm": 0.9326498508453369,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 23340
+    },
+    {
+      "epoch": 74.83974358974359,
+      "grad_norm": 1.2265844345092773,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 23350
+    },
+    {
+      "epoch": 74.87179487179488,
+      "grad_norm": 0.9387669563293457,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 23360
+    },
+    {
+      "epoch": 74.90384615384616,
+      "grad_norm": 0.7937279939651489,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 23370
+    },
+    {
+      "epoch": 74.93589743589743,
+      "grad_norm": 0.8505454063415527,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 23380
+    },
+    {
+      "epoch": 74.96794871794872,
+      "grad_norm": 1.0549218654632568,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 23390
+    },
+    {
+      "epoch": 75.0,
+      "grad_norm": 1.0170587301254272,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 23400
+    },
+    {
+      "epoch": 75.03205128205128,
+      "grad_norm": 0.9728469252586365,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 23410
+    },
+    {
+      "epoch": 75.06410256410257,
+      "grad_norm": 1.044040560722351,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 23420
+    },
+    {
+      "epoch": 75.09615384615384,
+      "grad_norm": 1.6548206806182861,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 23430
+    },
+    {
+      "epoch": 75.12820512820512,
+      "grad_norm": 1.4760040044784546,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 23440
+    },
+    {
+      "epoch": 75.16025641025641,
+      "grad_norm": 1.4212307929992676,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 23450
+    },
+    {
+      "epoch": 75.1923076923077,
+      "grad_norm": 1.7683526277542114,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 23460
+    },
+    {
+      "epoch": 75.22435897435898,
+      "grad_norm": 2.44692325592041,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 23470
+    },
+    {
+      "epoch": 75.25641025641026,
+      "grad_norm": 1.82981538772583,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 23480
+    },
+    {
+      "epoch": 75.28846153846153,
+      "grad_norm": 1.2977721691131592,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 23490
+    },
+    {
+      "epoch": 75.32051282051282,
+      "grad_norm": 1.2476203441619873,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 23500
+    },
+    {
+      "epoch": 75.3525641025641,
+      "grad_norm": 1.2362083196640015,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 23510
+    },
+    {
+      "epoch": 75.38461538461539,
+      "grad_norm": 1.2586133480072021,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 23520
+    },
+    {
+      "epoch": 75.41666666666667,
+      "grad_norm": 0.9889112114906311,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 23530
+    },
+    {
+      "epoch": 75.44871794871794,
+      "grad_norm": 1.4871841669082642,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 23540
+    },
+    {
+      "epoch": 75.48076923076923,
+      "grad_norm": 1.16875159740448,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 23550
+    },
+    {
+      "epoch": 75.51282051282051,
+      "grad_norm": 1.0264406204223633,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 23560
+    },
+    {
+      "epoch": 75.5448717948718,
+      "grad_norm": 1.054238200187683,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 23570
+    },
+    {
+      "epoch": 75.57692307692308,
+      "grad_norm": 1.1949560642242432,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 23580
+    },
+    {
+      "epoch": 75.60897435897436,
+      "grad_norm": 1.0544307231903076,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 23590
+    },
+    {
+      "epoch": 75.64102564102564,
+      "grad_norm": 1.3121360540390015,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 23600
+    },
+    {
+      "epoch": 75.67307692307692,
+      "grad_norm": 1.0366966724395752,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 23610
+    },
+    {
+      "epoch": 75.7051282051282,
+      "grad_norm": 0.8694853186607361,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 23620
+    },
+    {
+      "epoch": 75.73717948717949,
+      "grad_norm": 1.2168978452682495,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 23630
+    },
+    {
+      "epoch": 75.76923076923077,
+      "grad_norm": 0.9907625913619995,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 23640
+    },
+    {
+      "epoch": 75.80128205128206,
+      "grad_norm": 0.9182089567184448,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 23650
+    },
+    {
+      "epoch": 75.83333333333333,
+      "grad_norm": 1.0413330793380737,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 23660
+    },
+    {
+      "epoch": 75.86538461538461,
+      "grad_norm": 1.243119478225708,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 23670
+    },
+    {
+      "epoch": 75.8974358974359,
+      "grad_norm": 1.0993205308914185,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 23680
+    },
+    {
+      "epoch": 75.92948717948718,
+      "grad_norm": 1.1615561246871948,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 23690
+    },
+    {
+      "epoch": 75.96153846153847,
+      "grad_norm": 1.0669715404510498,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 23700
+    },
+    {
+      "epoch": 75.99358974358974,
+      "grad_norm": 1.4474767446517944,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 23710
+    },
+    {
+      "epoch": 76.02564102564102,
+      "grad_norm": 0.9980424046516418,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 23720
+    },
+    {
+      "epoch": 76.0576923076923,
+      "grad_norm": 1.1048709154129028,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 23730
+    },
+    {
+      "epoch": 76.08974358974359,
+      "grad_norm": 1.1250922679901123,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 23740
+    },
+    {
+      "epoch": 76.12179487179488,
+      "grad_norm": 1.4004101753234863,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 23750
+    },
+    {
+      "epoch": 76.15384615384616,
+      "grad_norm": 0.835440993309021,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 23760
+    },
+    {
+      "epoch": 76.18589743589743,
+      "grad_norm": 0.8955557346343994,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 23770
+    },
+    {
+      "epoch": 76.21794871794872,
+      "grad_norm": 1.0637493133544922,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 23780
+    },
+    {
+      "epoch": 76.25,
+      "grad_norm": 1.156909465789795,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 23790
+    },
+    {
+      "epoch": 76.28205128205128,
+      "grad_norm": 1.9046804904937744,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 23800
+    },
+    {
+      "epoch": 76.31410256410257,
+      "grad_norm": 1.727977991104126,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 23810
+    },
+    {
+      "epoch": 76.34615384615384,
+      "grad_norm": 2.2183775901794434,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 23820
+    },
+    {
+      "epoch": 76.37820512820512,
+      "grad_norm": 1.9325116872787476,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 23830
+    },
+    {
+      "epoch": 76.41025641025641,
+      "grad_norm": 1.3281320333480835,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 23840
+    },
+    {
+      "epoch": 76.4423076923077,
+      "grad_norm": 1.2898705005645752,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 23850
+    },
+    {
+      "epoch": 76.47435897435898,
+      "grad_norm": 1.5387107133865356,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 23860
+    },
+    {
+      "epoch": 76.50641025641026,
+      "grad_norm": 1.636444091796875,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 23870
+    },
+    {
+      "epoch": 76.53846153846153,
+      "grad_norm": 1.4112106561660767,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 23880
+    },
+    {
+      "epoch": 76.57051282051282,
+      "grad_norm": 1.1933516263961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 23890
+    },
+    {
+      "epoch": 76.6025641025641,
+      "grad_norm": 1.4015284776687622,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 23900
+    },
+    {
+      "epoch": 76.63461538461539,
+      "grad_norm": 1.1778011322021484,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 23910
+    },
+    {
+      "epoch": 76.66666666666667,
+      "grad_norm": 1.2730836868286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 23920
+    },
+    {
+      "epoch": 76.69871794871794,
+      "grad_norm": 1.4393808841705322,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 23930
+    },
+    {
+      "epoch": 76.73076923076923,
+      "grad_norm": 2.308753728866577,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 23940
+    },
+    {
+      "epoch": 76.76282051282051,
+      "grad_norm": 1.7957841157913208,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 23950
+    },
+    {
+      "epoch": 76.7948717948718,
+      "grad_norm": 1.4418909549713135,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 23960
+    },
+    {
+      "epoch": 76.82692307692308,
+      "grad_norm": 0.927993655204773,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 23970
+    },
+    {
+      "epoch": 76.85897435897436,
+      "grad_norm": 1.209771752357483,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 23980
+    },
+    {
+      "epoch": 76.89102564102564,
+      "grad_norm": 1.4708088636398315,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 23990
+    },
+    {
+      "epoch": 76.92307692307692,
+      "grad_norm": 1.6128956079483032,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 24000
+    },
+    {
+      "epoch": 76.9551282051282,
+      "grad_norm": 1.173920750617981,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 24010
+    },
+    {
+      "epoch": 76.98717948717949,
+      "grad_norm": 1.2910981178283691,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 24020
+    },
+    {
+      "epoch": 77.01923076923077,
+      "grad_norm": 1.3123128414154053,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 24030
+    },
+    {
+      "epoch": 77.05128205128206,
+      "grad_norm": 1.5560908317565918,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 24040
+    },
+    {
+      "epoch": 77.08333333333333,
+      "grad_norm": 1.1801789999008179,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24050
+    },
+    {
+      "epoch": 77.11538461538461,
+      "grad_norm": 1.3076590299606323,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 24060
+    },
+    {
+      "epoch": 77.1474358974359,
+      "grad_norm": 1.2771937847137451,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 24070
+    },
+    {
+      "epoch": 77.17948717948718,
+      "grad_norm": 0.8669016361236572,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 24080
+    },
+    {
+      "epoch": 77.21153846153847,
+      "grad_norm": 1.62409508228302,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 24090
+    },
+    {
+      "epoch": 77.24358974358974,
+      "grad_norm": 1.3016700744628906,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24100
+    },
+    {
+      "epoch": 77.27564102564102,
+      "grad_norm": 1.4643843173980713,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24110
+    },
+    {
+      "epoch": 77.3076923076923,
+      "grad_norm": 1.3523285388946533,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 24120
+    },
+    {
+      "epoch": 77.33974358974359,
+      "grad_norm": 1.3723353147506714,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 24130
+    },
+    {
+      "epoch": 77.37179487179488,
+      "grad_norm": 1.0601520538330078,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 24140
+    },
+    {
+      "epoch": 77.40384615384616,
+      "grad_norm": 1.6559317111968994,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 24150
+    },
+    {
+      "epoch": 77.43589743589743,
+      "grad_norm": 0.9548357725143433,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 24160
+    },
+    {
+      "epoch": 77.46794871794872,
+      "grad_norm": 0.7328200340270996,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 24170
+    },
+    {
+      "epoch": 77.5,
+      "grad_norm": 0.8272359371185303,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 24180
+    },
+    {
+      "epoch": 77.53205128205128,
+      "grad_norm": 1.0998408794403076,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24190
+    },
+    {
+      "epoch": 77.56410256410257,
+      "grad_norm": 1.202176809310913,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 24200
+    },
+    {
+      "epoch": 77.59615384615384,
+      "grad_norm": 1.2061103582382202,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 24210
+    },
+    {
+      "epoch": 77.62820512820512,
+      "grad_norm": 1.0799371004104614,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 24220
+    },
+    {
+      "epoch": 77.66025641025641,
+      "grad_norm": 1.1370104551315308,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24230
+    },
+    {
+      "epoch": 77.6923076923077,
+      "grad_norm": 1.1132006645202637,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 24240
+    },
+    {
+      "epoch": 77.72435897435898,
+      "grad_norm": 1.3807865381240845,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 24250
+    },
+    {
+      "epoch": 77.75641025641026,
+      "grad_norm": 1.3525806665420532,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 24260
+    },
+    {
+      "epoch": 77.78846153846153,
+      "grad_norm": 1.3122093677520752,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 24270
+    },
+    {
+      "epoch": 77.82051282051282,
+      "grad_norm": 0.9093636870384216,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 24280
+    },
+    {
+      "epoch": 77.8525641025641,
+      "grad_norm": 1.22516930103302,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 24290
+    },
+    {
+      "epoch": 77.88461538461539,
+      "grad_norm": 0.8261937499046326,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 24300
+    },
+    {
+      "epoch": 77.91666666666667,
+      "grad_norm": 1.023329734802246,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 24310
+    },
+    {
+      "epoch": 77.94871794871794,
+      "grad_norm": 1.058618187904358,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 24320
+    },
+    {
+      "epoch": 77.98076923076923,
+      "grad_norm": 1.0821921825408936,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 24330
+    },
+    {
+      "epoch": 78.01282051282051,
+      "grad_norm": 1.2909674644470215,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 24340
+    },
+    {
+      "epoch": 78.0448717948718,
+      "grad_norm": 1.6586790084838867,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 24350
+    },
+    {
+      "epoch": 78.07692307692308,
+      "grad_norm": 1.3420673608779907,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24360
+    },
+    {
+      "epoch": 78.10897435897436,
+      "grad_norm": 1.4243910312652588,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 24370
+    },
+    {
+      "epoch": 78.14102564102564,
+      "grad_norm": 1.310510516166687,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 24380
+    },
+    {
+      "epoch": 78.17307692307692,
+      "grad_norm": 1.426168441772461,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 24390
+    },
+    {
+      "epoch": 78.2051282051282,
+      "grad_norm": 1.139032006263733,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 24400
+    },
+    {
+      "epoch": 78.23717948717949,
+      "grad_norm": 1.5096755027770996,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 24410
+    },
+    {
+      "epoch": 78.26923076923077,
+      "grad_norm": 0.999944269657135,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 24420
+    },
+    {
+      "epoch": 78.30128205128206,
+      "grad_norm": 0.9265159964561462,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 24430
+    },
+    {
+      "epoch": 78.33333333333333,
+      "grad_norm": 1.6290737390518188,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 24440
+    },
+    {
+      "epoch": 78.36538461538461,
+      "grad_norm": 0.9225745797157288,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 24450
+    },
+    {
+      "epoch": 78.3974358974359,
+      "grad_norm": 0.9096898436546326,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 24460
+    },
+    {
+      "epoch": 78.42948717948718,
+      "grad_norm": 1.0090771913528442,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 24470
+    },
+    {
+      "epoch": 78.46153846153847,
+      "grad_norm": 1.0671703815460205,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 24480
+    },
+    {
+      "epoch": 78.49358974358974,
+      "grad_norm": 1.1279855966567993,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24490
+    },
+    {
+      "epoch": 78.52564102564102,
+      "grad_norm": 1.32305908203125,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 24500
+    },
+    {
+      "epoch": 78.5576923076923,
+      "grad_norm": 1.3059961795806885,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 24510
+    },
+    {
+      "epoch": 78.58974358974359,
+      "grad_norm": 1.083932638168335,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 24520
+    },
+    {
+      "epoch": 78.62179487179488,
+      "grad_norm": 1.2665563821792603,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 24530
+    },
+    {
+      "epoch": 78.65384615384616,
+      "grad_norm": 1.3050695657730103,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24540
+    },
+    {
+      "epoch": 78.68589743589743,
+      "grad_norm": 0.9649658799171448,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 24550
+    },
+    {
+      "epoch": 78.71794871794872,
+      "grad_norm": 1.2202805280685425,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 24560
+    },
+    {
+      "epoch": 78.75,
+      "grad_norm": 1.2065457105636597,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 24570
+    },
+    {
+      "epoch": 78.78205128205128,
+      "grad_norm": 1.4675824642181396,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 24580
+    },
+    {
+      "epoch": 78.81410256410257,
+      "grad_norm": 0.7933533787727356,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 24590
+    },
+    {
+      "epoch": 78.84615384615384,
+      "grad_norm": 1.6299159526824951,
+      "learning_rate": 0.0001,
+      "loss": 0.0127,
+      "step": 24600
+    },
+    {
+      "epoch": 78.87820512820512,
+      "grad_norm": 1.6347192525863647,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 24610
+    },
+    {
+      "epoch": 78.91025641025641,
+      "grad_norm": 1.5468289852142334,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 24620
+    },
+    {
+      "epoch": 78.9423076923077,
+      "grad_norm": 0.9780817031860352,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 24630
+    },
+    {
+      "epoch": 78.97435897435898,
+      "grad_norm": 1.10318922996521,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 24640
+    },
+    {
+      "epoch": 79.00641025641026,
+      "grad_norm": 0.9460640549659729,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 24650
+    },
+    {
+      "epoch": 79.03846153846153,
+      "grad_norm": 0.9054620265960693,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 24660
+    },
+    {
+      "epoch": 79.07051282051282,
+      "grad_norm": 1.1870341300964355,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 24670
+    },
+    {
+      "epoch": 79.1025641025641,
+      "grad_norm": 1.1864299774169922,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 24680
+    },
+    {
+      "epoch": 79.13461538461539,
+      "grad_norm": 1.1578251123428345,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 24690
+    },
+    {
+      "epoch": 79.16666666666667,
+      "grad_norm": 1.1330559253692627,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 24700
+    },
+    {
+      "epoch": 79.19871794871794,
+      "grad_norm": 1.1073421239852905,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 24710
+    },
+    {
+      "epoch": 79.23076923076923,
+      "grad_norm": 1.2814985513687134,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 24720
+    },
+    {
+      "epoch": 79.26282051282051,
+      "grad_norm": 1.2319680452346802,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 24730
+    },
+    {
+      "epoch": 79.2948717948718,
+      "grad_norm": 0.9407679438591003,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 24740
+    },
+    {
+      "epoch": 79.32692307692308,
+      "grad_norm": 0.9832383394241333,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 24750
+    },
+    {
+      "epoch": 79.35897435897436,
+      "grad_norm": 1.0220801830291748,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 24760
+    },
+    {
+      "epoch": 79.39102564102564,
+      "grad_norm": 0.7334434390068054,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 24770
+    },
+    {
+      "epoch": 79.42307692307692,
+      "grad_norm": 0.9029101729393005,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 24780
+    },
+    {
+      "epoch": 79.4551282051282,
+      "grad_norm": 1.0615936517715454,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 24790
+    },
+    {
+      "epoch": 79.48717948717949,
+      "grad_norm": 0.9625355005264282,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24800
+    },
+    {
+      "epoch": 79.51923076923077,
+      "grad_norm": 1.0103977918624878,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 24810
+    },
+    {
+      "epoch": 79.55128205128206,
+      "grad_norm": 1.1967467069625854,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 24820
+    },
+    {
+      "epoch": 79.58333333333333,
+      "grad_norm": 1.3825644254684448,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 24830
+    },
+    {
+      "epoch": 79.61538461538461,
+      "grad_norm": 1.101934790611267,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 24840
+    },
+    {
+      "epoch": 79.6474358974359,
+      "grad_norm": 1.163187861442566,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 24850
+    },
+    {
+      "epoch": 79.67948717948718,
+      "grad_norm": 1.0998090505599976,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 24860
+    },
+    {
+      "epoch": 79.71153846153847,
+      "grad_norm": 0.9553349018096924,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 24870
+    },
+    {
+      "epoch": 79.74358974358974,
+      "grad_norm": 1.104256272315979,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 24880
+    },
+    {
+      "epoch": 79.77564102564102,
+      "grad_norm": 1.326561450958252,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 24890
+    },
+    {
+      "epoch": 79.8076923076923,
+      "grad_norm": 1.0895440578460693,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 24900
+    },
+    {
+      "epoch": 79.83974358974359,
+      "grad_norm": 1.0732141733169556,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 24910
+    },
+    {
+      "epoch": 79.87179487179488,
+      "grad_norm": 0.9499208331108093,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 24920
+    },
+    {
+      "epoch": 79.90384615384616,
+      "grad_norm": 1.0702316761016846,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 24930
+    },
+    {
+      "epoch": 79.93589743589743,
+      "grad_norm": 1.0388367176055908,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 24940
+    },
+    {
+      "epoch": 79.96794871794872,
+      "grad_norm": 0.9515267610549927,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 24950
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 1.541578769683838,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 24960
+    },
+    {
+      "epoch": 80.03205128205128,
+      "grad_norm": 1.0198297500610352,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 24970
+    },
+    {
+      "epoch": 80.06410256410257,
+      "grad_norm": 1.2727006673812866,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 24980
+    },
+    {
+      "epoch": 80.09615384615384,
+      "grad_norm": 1.2669686079025269,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 24990
+    },
+    {
+      "epoch": 80.12820512820512,
+      "grad_norm": 1.3126643896102905,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 25000
+    },
+    {
+      "epoch": 80.16025641025641,
+      "grad_norm": 1.1459689140319824,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 25010
+    },
+    {
+      "epoch": 80.1923076923077,
+      "grad_norm": 1.502988338470459,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 25020
+    },
+    {
+      "epoch": 80.22435897435898,
+      "grad_norm": 1.1102573871612549,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 25030
+    },
+    {
+      "epoch": 80.25641025641026,
+      "grad_norm": 1.064913272857666,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 25040
+    },
+    {
+      "epoch": 80.28846153846153,
+      "grad_norm": 1.1329489946365356,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 25050
+    },
+    {
+      "epoch": 80.32051282051282,
+      "grad_norm": 1.0553516149520874,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 25060
+    },
+    {
+      "epoch": 80.3525641025641,
+      "grad_norm": 1.0856764316558838,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 25070
+    },
+    {
+      "epoch": 80.38461538461539,
+      "grad_norm": 1.1710989475250244,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 25080
+    },
+    {
+      "epoch": 80.41666666666667,
+      "grad_norm": 1.378645420074463,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 25090
+    },
+    {
+      "epoch": 80.44871794871794,
+      "grad_norm": 1.445517659187317,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 25100
+    },
+    {
+      "epoch": 80.48076923076923,
+      "grad_norm": 1.5673753023147583,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 25110
+    },
+    {
+      "epoch": 80.51282051282051,
+      "grad_norm": 1.1098631620407104,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 25120
+    },
+    {
+      "epoch": 80.5448717948718,
+      "grad_norm": 1.079301118850708,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 25130
+    },
+    {
+      "epoch": 80.57692307692308,
+      "grad_norm": 0.8931571245193481,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 25140
+    },
+    {
+      "epoch": 80.60897435897436,
+      "grad_norm": 0.918247640132904,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 25150
+    },
+    {
+      "epoch": 80.64102564102564,
+      "grad_norm": 1.0056253671646118,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 25160
+    },
+    {
+      "epoch": 80.67307692307692,
+      "grad_norm": 1.0326783657073975,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 25170
+    },
+    {
+      "epoch": 80.7051282051282,
+      "grad_norm": 1.2231738567352295,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 25180
+    },
+    {
+      "epoch": 80.73717948717949,
+      "grad_norm": 1.2501399517059326,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 25190
+    },
+    {
+      "epoch": 80.76923076923077,
+      "grad_norm": 1.528644323348999,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 25200
+    },
+    {
+      "epoch": 80.80128205128206,
+      "grad_norm": 1.3074228763580322,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 25210
+    },
+    {
+      "epoch": 80.83333333333333,
+      "grad_norm": 1.0288280248641968,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 25220
+    },
+    {
+      "epoch": 80.86538461538461,
+      "grad_norm": 0.897842288017273,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 25230
+    },
+    {
+      "epoch": 80.8974358974359,
+      "grad_norm": 0.8206071853637695,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 25240
+    },
+    {
+      "epoch": 80.92948717948718,
+      "grad_norm": 1.0519222021102905,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 25250
+    },
+    {
+      "epoch": 80.96153846153847,
+      "grad_norm": 1.0776216983795166,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 25260
+    },
+    {
+      "epoch": 80.99358974358974,
+      "grad_norm": 0.9972928762435913,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 25270
+    },
+    {
+      "epoch": 81.02564102564102,
+      "grad_norm": 1.243283987045288,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 25280
+    },
+    {
+      "epoch": 81.0576923076923,
+      "grad_norm": 1.020598292350769,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 25290
+    },
+    {
+      "epoch": 81.08974358974359,
+      "grad_norm": 1.0588959455490112,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 25300
+    },
+    {
+      "epoch": 81.12179487179488,
+      "grad_norm": 1.0566685199737549,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 25310
+    },
+    {
+      "epoch": 81.15384615384616,
+      "grad_norm": 0.8876959085464478,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 25320
+    },
+    {
+      "epoch": 81.18589743589743,
+      "grad_norm": 1.2576346397399902,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 25330
+    },
+    {
+      "epoch": 81.21794871794872,
+      "grad_norm": 1.2684239149093628,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 25340
+    },
+    {
+      "epoch": 81.25,
+      "grad_norm": 0.8207217454910278,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 25350
+    },
+    {
+      "epoch": 81.28205128205128,
+      "grad_norm": 1.3853724002838135,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 25360
+    },
+    {
+      "epoch": 81.31410256410257,
+      "grad_norm": 1.286219596862793,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 25370
+    },
+    {
+      "epoch": 81.34615384615384,
+      "grad_norm": 0.9253241419792175,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 25380
+    },
+    {
+      "epoch": 81.37820512820512,
+      "grad_norm": 1.062477946281433,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 25390
+    },
+    {
+      "epoch": 81.41025641025641,
+      "grad_norm": 0.9040235877037048,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 25400
+    },
+    {
+      "epoch": 81.4423076923077,
+      "grad_norm": 1.0645352602005005,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 25410
+    },
+    {
+      "epoch": 81.47435897435898,
+      "grad_norm": 1.048466682434082,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 25420
+    },
+    {
+      "epoch": 81.50641025641026,
+      "grad_norm": 1.3247127532958984,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 25430
+    },
+    {
+      "epoch": 81.53846153846153,
+      "grad_norm": 1.061428427696228,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 25440
+    },
+    {
+      "epoch": 81.57051282051282,
+      "grad_norm": 0.9499216675758362,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 25450
+    },
+    {
+      "epoch": 81.6025641025641,
+      "grad_norm": 0.9186298847198486,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 25460
+    },
+    {
+      "epoch": 81.63461538461539,
+      "grad_norm": 0.7672370076179504,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 25470
+    },
+    {
+      "epoch": 81.66666666666667,
+      "grad_norm": 1.090721607208252,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 25480
+    },
+    {
+      "epoch": 81.69871794871794,
+      "grad_norm": 0.7625296711921692,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 25490
+    },
+    {
+      "epoch": 81.73076923076923,
+      "grad_norm": 0.998805820941925,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 25500
+    },
+    {
+      "epoch": 81.76282051282051,
+      "grad_norm": 1.1688019037246704,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 25510
+    },
+    {
+      "epoch": 81.7948717948718,
+      "grad_norm": 1.039423942565918,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 25520
+    },
+    {
+      "epoch": 81.82692307692308,
+      "grad_norm": 1.0023764371871948,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 25530
+    },
+    {
+      "epoch": 81.85897435897436,
+      "grad_norm": 1.4034488201141357,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 25540
+    },
+    {
+      "epoch": 81.89102564102564,
+      "grad_norm": 1.1571396589279175,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 25550
+    },
+    {
+      "epoch": 81.92307692307692,
+      "grad_norm": 1.2380181550979614,
+      "learning_rate": 0.0001,
+      "loss": 0.0134,
+      "step": 25560
+    },
+    {
+      "epoch": 81.9551282051282,
+      "grad_norm": 1.0662986040115356,
+      "learning_rate": 0.0001,
+      "loss": 0.0124,
+      "step": 25570
+    },
+    {
+      "epoch": 81.98717948717949,
+      "grad_norm": 1.3564658164978027,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 25580
+    },
+    {
+      "epoch": 82.01923076923077,
+      "grad_norm": 1.107182264328003,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 25590
+    },
+    {
+      "epoch": 82.05128205128206,
+      "grad_norm": 0.8545488119125366,
+      "learning_rate": 0.0001,
+      "loss": 0.0121,
+      "step": 25600
+    },
+    {
+      "epoch": 82.08333333333333,
+      "grad_norm": 1.174623727798462,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 25610
+    },
+    {
+      "epoch": 82.11538461538461,
+      "grad_norm": 1.5771738290786743,
+      "learning_rate": 0.0001,
+      "loss": 0.0118,
+      "step": 25620
+    },
+    {
+      "epoch": 82.1474358974359,
+      "grad_norm": 1.2624553442001343,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 25630
+    },
+    {
+      "epoch": 82.17948717948718,
+      "grad_norm": 1.1085845232009888,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 25640
+    },
+    {
+      "epoch": 82.21153846153847,
+      "grad_norm": 1.1148704290390015,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 25650
+    },
+    {
+      "epoch": 82.24358974358974,
+      "grad_norm": 0.9677826762199402,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 25660
+    },
+    {
+      "epoch": 82.27564102564102,
+      "grad_norm": 1.2091593742370605,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 25670
+    },
+    {
+      "epoch": 82.3076923076923,
+      "grad_norm": 1.2651182413101196,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 25680
+    },
+    {
+      "epoch": 82.33974358974359,
+      "grad_norm": 1.019974708557129,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 25690
+    },
+    {
+      "epoch": 82.37179487179488,
+      "grad_norm": 1.2069085836410522,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 25700
+    },
+    {
+      "epoch": 82.40384615384616,
+      "grad_norm": 1.5133763551712036,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 25710
+    },
+    {
+      "epoch": 82.43589743589743,
+      "grad_norm": 1.1848459243774414,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 25720
+    },
+    {
+      "epoch": 82.46794871794872,
+      "grad_norm": 1.1966145038604736,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 25730
+    },
+    {
+      "epoch": 82.5,
+      "grad_norm": 1.0954393148422241,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 25740
+    },
+    {
+      "epoch": 82.53205128205128,
+      "grad_norm": 1.0871526002883911,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 25750
+    },
+    {
+      "epoch": 82.56410256410257,
+      "grad_norm": 0.8073726296424866,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 25760
+    },
+    {
+      "epoch": 82.59615384615384,
+      "grad_norm": 1.305765986442566,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 25770
+    },
+    {
+      "epoch": 82.62820512820512,
+      "grad_norm": 1.1869663000106812,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 25780
+    },
+    {
+      "epoch": 82.66025641025641,
+      "grad_norm": 1.2988473176956177,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 25790
+    },
+    {
+      "epoch": 82.6923076923077,
+      "grad_norm": 1.1118910312652588,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 25800
+    },
+    {
+      "epoch": 82.72435897435898,
+      "grad_norm": 1.0298088788986206,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 25810
+    },
+    {
+      "epoch": 82.75641025641026,
+      "grad_norm": 0.7746372222900391,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 25820
+    },
+    {
+      "epoch": 82.78846153846153,
+      "grad_norm": 0.6971079111099243,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 25830
+    },
+    {
+      "epoch": 82.82051282051282,
+      "grad_norm": 0.8046109080314636,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 25840
+    },
+    {
+      "epoch": 82.8525641025641,
+      "grad_norm": 1.0052965879440308,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 25850
+    },
+    {
+      "epoch": 82.88461538461539,
+      "grad_norm": 1.3909823894500732,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 25860
+    },
+    {
+      "epoch": 82.91666666666667,
+      "grad_norm": 1.2948739528656006,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 25870
+    },
+    {
+      "epoch": 82.94871794871794,
+      "grad_norm": 0.9948426485061646,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 25880
+    },
+    {
+      "epoch": 82.98076923076923,
+      "grad_norm": 1.021682858467102,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 25890
+    },
+    {
+      "epoch": 83.01282051282051,
+      "grad_norm": 1.1611346006393433,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 25900
+    },
+    {
+      "epoch": 83.0448717948718,
+      "grad_norm": 0.9742225408554077,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 25910
+    },
+    {
+      "epoch": 83.07692307692308,
+      "grad_norm": 0.9796162843704224,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 25920
+    },
+    {
+      "epoch": 83.10897435897436,
+      "grad_norm": 0.9410825371742249,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 25930
+    },
+    {
+      "epoch": 83.14102564102564,
+      "grad_norm": 0.9119295477867126,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 25940
+    },
+    {
+      "epoch": 83.17307692307692,
+      "grad_norm": 0.9407249093055725,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 25950
+    },
+    {
+      "epoch": 83.2051282051282,
+      "grad_norm": 0.8094700574874878,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 25960
+    },
+    {
+      "epoch": 83.23717948717949,
+      "grad_norm": 1.2110366821289062,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 25970
+    },
+    {
+      "epoch": 83.26923076923077,
+      "grad_norm": 0.9162928462028503,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 25980
+    },
+    {
+      "epoch": 83.30128205128206,
+      "grad_norm": 0.9125942587852478,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 25990
+    },
+    {
+      "epoch": 83.33333333333333,
+      "grad_norm": 1.3107157945632935,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 26000
+    },
+    {
+      "epoch": 83.36538461538461,
+      "grad_norm": 0.9358218312263489,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 26010
+    },
+    {
+      "epoch": 83.3974358974359,
+      "grad_norm": 1.0799893140792847,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 26020
+    },
+    {
+      "epoch": 83.42948717948718,
+      "grad_norm": 1.276165246963501,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 26030
+    },
+    {
+      "epoch": 83.46153846153847,
+      "grad_norm": 0.8341509103775024,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 26040
+    },
+    {
+      "epoch": 83.49358974358974,
+      "grad_norm": 1.1043777465820312,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 26050
+    },
+    {
+      "epoch": 83.52564102564102,
+      "grad_norm": 1.0655503273010254,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 26060
+    },
+    {
+      "epoch": 83.5576923076923,
+      "grad_norm": 0.918998122215271,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 26070
+    },
+    {
+      "epoch": 83.58974358974359,
+      "grad_norm": 1.0990623235702515,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 26080
+    },
+    {
+      "epoch": 83.62179487179488,
+      "grad_norm": 1.0108534097671509,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 26090
+    },
+    {
+      "epoch": 83.65384615384616,
+      "grad_norm": 1.0416992902755737,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 26100
+    },
+    {
+      "epoch": 83.68589743589743,
+      "grad_norm": 1.0372314453125,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 26110
+    },
+    {
+      "epoch": 83.71794871794872,
+      "grad_norm": 0.8896980285644531,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 26120
+    },
+    {
+      "epoch": 83.75,
+      "grad_norm": 1.152010202407837,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 26130
+    },
+    {
+      "epoch": 83.78205128205128,
+      "grad_norm": 1.2861690521240234,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 26140
+    },
+    {
+      "epoch": 83.81410256410257,
+      "grad_norm": 0.9783819913864136,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 26150
+    },
+    {
+      "epoch": 83.84615384615384,
+      "grad_norm": 1.180214524269104,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 26160
+    },
+    {
+      "epoch": 83.87820512820512,
+      "grad_norm": 0.9600754380226135,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 26170
+    },
+    {
+      "epoch": 83.91025641025641,
+      "grad_norm": 1.07351815700531,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 26180
+    },
+    {
+      "epoch": 83.9423076923077,
+      "grad_norm": 1.0318653583526611,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 26190
+    },
+    {
+      "epoch": 83.97435897435898,
+      "grad_norm": 1.0435547828674316,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 26200
+    },
+    {
+      "epoch": 84.00641025641026,
+      "grad_norm": 1.0957964658737183,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 26210
+    },
+    {
+      "epoch": 84.03846153846153,
+      "grad_norm": 0.9489453434944153,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 26220
+    },
+    {
+      "epoch": 84.07051282051282,
+      "grad_norm": 0.9501461982727051,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 26230
+    },
+    {
+      "epoch": 84.1025641025641,
+      "grad_norm": 0.8168841600418091,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 26240
+    },
+    {
+      "epoch": 84.13461538461539,
+      "grad_norm": 1.4365408420562744,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 26250
+    },
+    {
+      "epoch": 84.16666666666667,
+      "grad_norm": 1.520309567451477,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 26260
+    },
+    {
+      "epoch": 84.19871794871794,
+      "grad_norm": 1.1911076307296753,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 26270
+    },
+    {
+      "epoch": 84.23076923076923,
+      "grad_norm": 1.3347254991531372,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 26280
+    },
+    {
+      "epoch": 84.26282051282051,
+      "grad_norm": 1.0263869762420654,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 26290
+    },
+    {
+      "epoch": 84.2948717948718,
+      "grad_norm": 1.0699623823165894,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26300
+    },
+    {
+      "epoch": 84.32692307692308,
+      "grad_norm": 1.148584246635437,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 26310
+    },
+    {
+      "epoch": 84.35897435897436,
+      "grad_norm": 1.1310503482818604,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 26320
+    },
+    {
+      "epoch": 84.39102564102564,
+      "grad_norm": 0.852129340171814,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 26330
+    },
+    {
+      "epoch": 84.42307692307692,
+      "grad_norm": 1.007787823677063,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 26340
+    },
+    {
+      "epoch": 84.4551282051282,
+      "grad_norm": 0.8387292623519897,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 26350
+    },
+    {
+      "epoch": 84.48717948717949,
+      "grad_norm": 0.916240930557251,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26360
+    },
+    {
+      "epoch": 84.51923076923077,
+      "grad_norm": 1.162636160850525,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 26370
+    },
+    {
+      "epoch": 84.55128205128206,
+      "grad_norm": 1.0432791709899902,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 26380
+    },
+    {
+      "epoch": 84.58333333333333,
+      "grad_norm": 0.9204535484313965,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 26390
+    },
+    {
+      "epoch": 84.61538461538461,
+      "grad_norm": 1.1308108568191528,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 26400
+    },
+    {
+      "epoch": 84.6474358974359,
+      "grad_norm": 0.9431716799736023,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 26410
+    },
+    {
+      "epoch": 84.67948717948718,
+      "grad_norm": 0.8949158787727356,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 26420
+    },
+    {
+      "epoch": 84.71153846153847,
+      "grad_norm": 1.270753026008606,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 26430
+    },
+    {
+      "epoch": 84.74358974358974,
+      "grad_norm": 1.2355232238769531,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 26440
+    },
+    {
+      "epoch": 84.77564102564102,
+      "grad_norm": 0.9979893565177917,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26450
+    },
+    {
+      "epoch": 84.8076923076923,
+      "grad_norm": 1.0741347074508667,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 26460
+    },
+    {
+      "epoch": 84.83974358974359,
+      "grad_norm": 0.7996665239334106,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 26470
+    },
+    {
+      "epoch": 84.87179487179488,
+      "grad_norm": 1.0574846267700195,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 26480
+    },
+    {
+      "epoch": 84.90384615384616,
+      "grad_norm": 0.8661272525787354,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 26490
+    },
+    {
+      "epoch": 84.93589743589743,
+      "grad_norm": 1.0418760776519775,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 26500
+    },
+    {
+      "epoch": 84.96794871794872,
+      "grad_norm": 0.9873527884483337,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 26510
+    },
+    {
+      "epoch": 85.0,
+      "grad_norm": 0.7175167202949524,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 26520
+    },
+    {
+      "epoch": 85.03205128205128,
+      "grad_norm": 1.259170651435852,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 26530
+    },
+    {
+      "epoch": 85.06410256410257,
+      "grad_norm": 1.126401424407959,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26540
+    },
+    {
+      "epoch": 85.09615384615384,
+      "grad_norm": 0.6966883540153503,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 26550
+    },
+    {
+      "epoch": 85.12820512820512,
+      "grad_norm": 0.7834030389785767,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 26560
+    },
+    {
+      "epoch": 85.16025641025641,
+      "grad_norm": 0.7649627327919006,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 26570
+    },
+    {
+      "epoch": 85.1923076923077,
+      "grad_norm": 1.4918720722198486,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 26580
+    },
+    {
+      "epoch": 85.22435897435898,
+      "grad_norm": 1.1720247268676758,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26590
+    },
+    {
+      "epoch": 85.25641025641026,
+      "grad_norm": 1.0403311252593994,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 26600
+    },
+    {
+      "epoch": 85.28846153846153,
+      "grad_norm": 1.3912709951400757,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 26610
+    },
+    {
+      "epoch": 85.32051282051282,
+      "grad_norm": 1.0267012119293213,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 26620
+    },
+    {
+      "epoch": 85.3525641025641,
+      "grad_norm": 1.2007029056549072,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 26630
+    },
+    {
+      "epoch": 85.38461538461539,
+      "grad_norm": 1.3645838499069214,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 26640
+    },
+    {
+      "epoch": 85.41666666666667,
+      "grad_norm": 1.4266469478607178,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 26650
+    },
+    {
+      "epoch": 85.44871794871794,
+      "grad_norm": 1.0990915298461914,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 26660
+    },
+    {
+      "epoch": 85.48076923076923,
+      "grad_norm": 0.8452649116516113,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 26670
+    },
+    {
+      "epoch": 85.51282051282051,
+      "grad_norm": 0.8579877614974976,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 26680
+    },
+    {
+      "epoch": 85.5448717948718,
+      "grad_norm": 0.7246649861335754,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 26690
+    },
+    {
+      "epoch": 85.57692307692308,
+      "grad_norm": 0.8458319306373596,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26700
+    },
+    {
+      "epoch": 85.60897435897436,
+      "grad_norm": 0.88551265001297,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26710
+    },
+    {
+      "epoch": 85.64102564102564,
+      "grad_norm": 1.1923192739486694,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 26720
+    },
+    {
+      "epoch": 85.67307692307692,
+      "grad_norm": 1.1061265468597412,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26730
+    },
+    {
+      "epoch": 85.7051282051282,
+      "grad_norm": 1.284785509109497,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 26740
+    },
+    {
+      "epoch": 85.73717948717949,
+      "grad_norm": 1.8727518320083618,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 26750
+    },
+    {
+      "epoch": 85.76923076923077,
+      "grad_norm": 1.11116361618042,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 26760
+    },
+    {
+      "epoch": 85.80128205128206,
+      "grad_norm": 1.3268496990203857,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 26770
+    },
+    {
+      "epoch": 85.83333333333333,
+      "grad_norm": 1.163639783859253,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 26780
+    },
+    {
+      "epoch": 85.86538461538461,
+      "grad_norm": 1.039463996887207,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 26790
+    },
+    {
+      "epoch": 85.8974358974359,
+      "grad_norm": 0.840183436870575,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 26800
+    },
+    {
+      "epoch": 85.92948717948718,
+      "grad_norm": 1.0350215435028076,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 26810
+    },
+    {
+      "epoch": 85.96153846153847,
+      "grad_norm": 1.0550730228424072,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 26820
+    },
+    {
+      "epoch": 85.99358974358974,
+      "grad_norm": 0.7962012887001038,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 26830
+    },
+    {
+      "epoch": 86.02564102564102,
+      "grad_norm": 1.0931552648544312,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 26840
+    },
+    {
+      "epoch": 86.0576923076923,
+      "grad_norm": 1.4182804822921753,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 26850
+    },
+    {
+      "epoch": 86.08974358974359,
+      "grad_norm": 0.9369016289710999,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26860
+    },
+    {
+      "epoch": 86.12179487179488,
+      "grad_norm": 0.9805184006690979,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 26870
+    },
+    {
+      "epoch": 86.15384615384616,
+      "grad_norm": 0.9323002696037292,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 26880
+    },
+    {
+      "epoch": 86.18589743589743,
+      "grad_norm": 1.3885878324508667,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 26890
+    },
+    {
+      "epoch": 86.21794871794872,
+      "grad_norm": 1.1175038814544678,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 26900
+    },
+    {
+      "epoch": 86.25,
+      "grad_norm": 1.3702661991119385,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 26910
+    },
+    {
+      "epoch": 86.28205128205128,
+      "grad_norm": 1.420781135559082,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 26920
+    },
+    {
+      "epoch": 86.31410256410257,
+      "grad_norm": 0.9861726760864258,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 26930
+    },
+    {
+      "epoch": 86.34615384615384,
+      "grad_norm": 0.9139097929000854,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 26940
+    },
+    {
+      "epoch": 86.37820512820512,
+      "grad_norm": 0.8702027201652527,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 26950
+    },
+    {
+      "epoch": 86.41025641025641,
+      "grad_norm": 1.1227682828903198,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 26960
+    },
+    {
+      "epoch": 86.4423076923077,
+      "grad_norm": 1.4654505252838135,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 26970
+    },
+    {
+      "epoch": 86.47435897435898,
+      "grad_norm": 1.0370961427688599,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 26980
+    },
+    {
+      "epoch": 86.50641025641026,
+      "grad_norm": 1.0972424745559692,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 26990
+    },
+    {
+      "epoch": 86.53846153846153,
+      "grad_norm": 0.9726055264472961,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 27000
+    },
+    {
+      "epoch": 86.57051282051282,
+      "grad_norm": 1.0907995700836182,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 27010
+    },
+    {
+      "epoch": 86.6025641025641,
+      "grad_norm": 1.4719023704528809,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 27020
+    },
+    {
+      "epoch": 86.63461538461539,
+      "grad_norm": 1.1843310594558716,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 27030
+    },
+    {
+      "epoch": 86.66666666666667,
+      "grad_norm": 1.4570854902267456,
+      "learning_rate": 0.0001,
+      "loss": 0.0114,
+      "step": 27040
+    },
+    {
+      "epoch": 86.69871794871794,
+      "grad_norm": 1.165031909942627,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 27050
+    },
+    {
+      "epoch": 86.73076923076923,
+      "grad_norm": 1.2503888607025146,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 27060
+    },
+    {
+      "epoch": 86.76282051282051,
+      "grad_norm": 1.1604160070419312,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 27070
+    },
+    {
+      "epoch": 86.7948717948718,
+      "grad_norm": 0.9850698113441467,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 27080
+    },
+    {
+      "epoch": 86.82692307692308,
+      "grad_norm": 1.0099742412567139,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 27090
+    },
+    {
+      "epoch": 86.85897435897436,
+      "grad_norm": 1.1363497972488403,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 27100
+    },
+    {
+      "epoch": 86.89102564102564,
+      "grad_norm": 1.0219416618347168,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 27110
+    },
+    {
+      "epoch": 86.92307692307692,
+      "grad_norm": 0.9885398149490356,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 27120
+    },
+    {
+      "epoch": 86.9551282051282,
+      "grad_norm": 0.8215135931968689,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 27130
+    },
+    {
+      "epoch": 86.98717948717949,
+      "grad_norm": 0.8087760806083679,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 27140
+    },
+    {
+      "epoch": 87.01923076923077,
+      "grad_norm": 0.9331963062286377,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 27150
+    },
+    {
+      "epoch": 87.05128205128206,
+      "grad_norm": 1.285672664642334,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 27160
+    },
+    {
+      "epoch": 87.08333333333333,
+      "grad_norm": 0.907340943813324,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 27170
+    },
+    {
+      "epoch": 87.11538461538461,
+      "grad_norm": 1.231503963470459,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 27180
+    },
+    {
+      "epoch": 87.1474358974359,
+      "grad_norm": 1.4046545028686523,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 27190
+    },
+    {
+      "epoch": 87.17948717948718,
+      "grad_norm": 1.289543867111206,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 27200
+    },
+    {
+      "epoch": 87.21153846153847,
+      "grad_norm": 1.2016764879226685,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 27210
+    },
+    {
+      "epoch": 87.24358974358974,
+      "grad_norm": 1.4158663749694824,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 27220
+    },
+    {
+      "epoch": 87.27564102564102,
+      "grad_norm": 1.2144899368286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 27230
+    },
+    {
+      "epoch": 87.3076923076923,
+      "grad_norm": 1.219285011291504,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 27240
+    },
+    {
+      "epoch": 87.33974358974359,
+      "grad_norm": 1.4477324485778809,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 27250
+    },
+    {
+      "epoch": 87.37179487179488,
+      "grad_norm": 1.135096549987793,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 27260
+    },
+    {
+      "epoch": 87.40384615384616,
+      "grad_norm": 1.4753986597061157,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27270
+    },
+    {
+      "epoch": 87.43589743589743,
+      "grad_norm": 1.2284066677093506,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27280
+    },
+    {
+      "epoch": 87.46794871794872,
+      "grad_norm": 0.8683377504348755,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 27290
+    },
+    {
+      "epoch": 87.5,
+      "grad_norm": 0.7445430755615234,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 27300
+    },
+    {
+      "epoch": 87.53205128205128,
+      "grad_norm": 1.0849111080169678,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27310
+    },
+    {
+      "epoch": 87.56410256410257,
+      "grad_norm": 0.926053524017334,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 27320
+    },
+    {
+      "epoch": 87.59615384615384,
+      "grad_norm": 1.057822585105896,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 27330
+    },
+    {
+      "epoch": 87.62820512820512,
+      "grad_norm": 1.3814646005630493,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 27340
+    },
+    {
+      "epoch": 87.66025641025641,
+      "grad_norm": 0.9638282656669617,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 27350
+    },
+    {
+      "epoch": 87.6923076923077,
+      "grad_norm": 1.0407716035842896,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 27360
+    },
+    {
+      "epoch": 87.72435897435898,
+      "grad_norm": 1.0335736274719238,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27370
+    },
+    {
+      "epoch": 87.75641025641026,
+      "grad_norm": 0.9242628216743469,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27380
+    },
+    {
+      "epoch": 87.78846153846153,
+      "grad_norm": 1.026410460472107,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 27390
+    },
+    {
+      "epoch": 87.82051282051282,
+      "grad_norm": 0.9200119376182556,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 27400
+    },
+    {
+      "epoch": 87.8525641025641,
+      "grad_norm": 1.1037755012512207,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 27410
+    },
+    {
+      "epoch": 87.88461538461539,
+      "grad_norm": 1.0269988775253296,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 27420
+    },
+    {
+      "epoch": 87.91666666666667,
+      "grad_norm": 1.208242416381836,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 27430
+    },
+    {
+      "epoch": 87.94871794871794,
+      "grad_norm": 1.5667381286621094,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 27440
+    },
+    {
+      "epoch": 87.98076923076923,
+      "grad_norm": 1.2918819189071655,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 27450
+    },
+    {
+      "epoch": 88.01282051282051,
+      "grad_norm": 1.2391234636306763,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 27460
+    },
+    {
+      "epoch": 88.0448717948718,
+      "grad_norm": 1.2932060956954956,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27470
+    },
+    {
+      "epoch": 88.07692307692308,
+      "grad_norm": 1.085540771484375,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 27480
+    },
+    {
+      "epoch": 88.10897435897436,
+      "grad_norm": 0.8330854773521423,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 27490
+    },
+    {
+      "epoch": 88.14102564102564,
+      "grad_norm": 0.8389071822166443,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 27500
+    },
+    {
+      "epoch": 88.17307692307692,
+      "grad_norm": 0.971132755279541,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 27510
+    },
+    {
+      "epoch": 88.2051282051282,
+      "grad_norm": 1.2678768634796143,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 27520
+    },
+    {
+      "epoch": 88.23717948717949,
+      "grad_norm": 0.7640508413314819,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 27530
+    },
+    {
+      "epoch": 88.26923076923077,
+      "grad_norm": 1.512778401374817,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 27540
+    },
+    {
+      "epoch": 88.30128205128206,
+      "grad_norm": 1.1802122592926025,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 27550
+    },
+    {
+      "epoch": 88.33333333333333,
+      "grad_norm": 1.1152549982070923,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 27560
+    },
+    {
+      "epoch": 88.36538461538461,
+      "grad_norm": 0.8941317796707153,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 27570
+    },
+    {
+      "epoch": 88.3974358974359,
+      "grad_norm": 1.0477169752120972,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 27580
+    },
+    {
+      "epoch": 88.42948717948718,
+      "grad_norm": 1.0268667936325073,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 27590
+    },
+    {
+      "epoch": 88.46153846153847,
+      "grad_norm": 1.423592448234558,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 27600
+    },
+    {
+      "epoch": 88.49358974358974,
+      "grad_norm": 1.2157814502716064,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 27610
+    },
+    {
+      "epoch": 88.52564102564102,
+      "grad_norm": 1.0159616470336914,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 27620
+    },
+    {
+      "epoch": 88.5576923076923,
+      "grad_norm": 1.1314655542373657,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 27630
+    },
+    {
+      "epoch": 88.58974358974359,
+      "grad_norm": 1.0533576011657715,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 27640
+    },
+    {
+      "epoch": 88.62179487179488,
+      "grad_norm": 0.9346412420272827,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 27650
+    },
+    {
+      "epoch": 88.65384615384616,
+      "grad_norm": 1.4933240413665771,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 27660
+    },
+    {
+      "epoch": 88.68589743589743,
+      "grad_norm": 5.125031471252441,
+      "learning_rate": 0.0001,
+      "loss": 0.0231,
+      "step": 27670
+    },
+    {
+      "epoch": 88.71794871794872,
+      "grad_norm": 2.6261403560638428,
+      "learning_rate": 0.0001,
+      "loss": 0.0128,
+      "step": 27680
+    },
+    {
+      "epoch": 88.75,
+      "grad_norm": 2.0673489570617676,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 27690
+    },
+    {
+      "epoch": 88.78205128205128,
+      "grad_norm": 1.8317197561264038,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 27700
+    },
+    {
+      "epoch": 88.81410256410257,
+      "grad_norm": 1.5864702463150024,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 27710
+    },
+    {
+      "epoch": 88.84615384615384,
+      "grad_norm": 1.2285490036010742,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 27720
+    },
+    {
+      "epoch": 88.87820512820512,
+      "grad_norm": 1.4478930234909058,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 27730
+    },
+    {
+      "epoch": 88.91025641025641,
+      "grad_norm": 1.3760044574737549,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 27740
+    },
+    {
+      "epoch": 88.9423076923077,
+      "grad_norm": 1.43417489528656,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27750
+    },
+    {
+      "epoch": 88.97435897435898,
+      "grad_norm": 1.4247156381607056,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 27760
+    },
+    {
+      "epoch": 89.00641025641026,
+      "grad_norm": 1.183577299118042,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 27770
+    },
+    {
+      "epoch": 89.03846153846153,
+      "grad_norm": 1.3394482135772705,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 27780
+    },
+    {
+      "epoch": 89.07051282051282,
+      "grad_norm": 1.4393645524978638,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 27790
+    },
+    {
+      "epoch": 89.1025641025641,
+      "grad_norm": 1.155265212059021,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 27800
+    },
+    {
+      "epoch": 89.13461538461539,
+      "grad_norm": 0.8964120745658875,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 27810
+    },
+    {
+      "epoch": 89.16666666666667,
+      "grad_norm": 0.8323544263839722,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 27820
+    },
+    {
+      "epoch": 89.19871794871794,
+      "grad_norm": 1.5648187398910522,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 27830
+    },
+    {
+      "epoch": 89.23076923076923,
+      "grad_norm": 1.2339624166488647,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27840
+    },
+    {
+      "epoch": 89.26282051282051,
+      "grad_norm": 1.1935561895370483,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 27850
+    },
+    {
+      "epoch": 89.2948717948718,
+      "grad_norm": 1.1193430423736572,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 27860
+    },
+    {
+      "epoch": 89.32692307692308,
+      "grad_norm": 1.059342384338379,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27870
+    },
+    {
+      "epoch": 89.35897435897436,
+      "grad_norm": 1.6699055433273315,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 27880
+    },
+    {
+      "epoch": 89.39102564102564,
+      "grad_norm": 1.0735670328140259,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 27890
+    },
+    {
+      "epoch": 89.42307692307692,
+      "grad_norm": 1.1222681999206543,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 27900
+    },
+    {
+      "epoch": 89.4551282051282,
+      "grad_norm": 1.0809561014175415,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 27910
+    },
+    {
+      "epoch": 89.48717948717949,
+      "grad_norm": 0.9475740790367126,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 27920
+    },
+    {
+      "epoch": 89.51923076923077,
+      "grad_norm": 1.4341444969177246,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 27930
+    },
+    {
+      "epoch": 89.55128205128206,
+      "grad_norm": 1.426371455192566,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 27940
+    },
+    {
+      "epoch": 89.58333333333333,
+      "grad_norm": 0.8423776626586914,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 27950
+    },
+    {
+      "epoch": 89.61538461538461,
+      "grad_norm": 0.961849570274353,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 27960
+    },
+    {
+      "epoch": 89.6474358974359,
+      "grad_norm": 0.9001820087432861,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 27970
+    },
+    {
+      "epoch": 89.67948717948718,
+      "grad_norm": 0.829327404499054,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 27980
+    },
+    {
+      "epoch": 89.71153846153847,
+      "grad_norm": 1.1387144327163696,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 27990
+    },
+    {
+      "epoch": 89.74358974358974,
+      "grad_norm": 0.8194755911827087,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 28000
+    },
+    {
+      "epoch": 89.77564102564102,
+      "grad_norm": 0.9404492378234863,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 28010
+    },
+    {
+      "epoch": 89.8076923076923,
+      "grad_norm": 1.2933814525604248,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 28020
+    },
+    {
+      "epoch": 89.83974358974359,
+      "grad_norm": 1.087354063987732,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 28030
+    },
+    {
+      "epoch": 89.87179487179488,
+      "grad_norm": 0.8047097325325012,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 28040
+    },
+    {
+      "epoch": 89.90384615384616,
+      "grad_norm": 1.3227614164352417,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 28050
+    },
+    {
+      "epoch": 89.93589743589743,
+      "grad_norm": 1.0239931344985962,
+      "learning_rate": 0.0001,
+      "loss": 0.0113,
+      "step": 28060
+    },
+    {
+      "epoch": 89.96794871794872,
+      "grad_norm": 1.201944351196289,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 28070
+    },
+    {
+      "epoch": 90.0,
+      "grad_norm": 0.8760789632797241,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 28080
+    },
+    {
+      "epoch": 90.03205128205128,
+      "grad_norm": 1.0655488967895508,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 28090
+    },
+    {
+      "epoch": 90.06410256410257,
+      "grad_norm": 1.2284702062606812,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 28100
+    },
+    {
+      "epoch": 90.09615384615384,
+      "grad_norm": 1.040578842163086,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 28110
+    },
+    {
+      "epoch": 90.12820512820512,
+      "grad_norm": 1.0520074367523193,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 28120
+    },
+    {
+      "epoch": 90.16025641025641,
+      "grad_norm": 1.0738314390182495,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 28130
+    },
+    {
+      "epoch": 90.1923076923077,
+      "grad_norm": 1.0717884302139282,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 28140
+    },
+    {
+      "epoch": 90.22435897435898,
+      "grad_norm": 0.8118210434913635,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 28150
+    },
+    {
+      "epoch": 90.25641025641026,
+      "grad_norm": 0.9052904844284058,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 28160
+    },
+    {
+      "epoch": 90.28846153846153,
+      "grad_norm": 0.8692615032196045,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 28170
+    },
+    {
+      "epoch": 90.32051282051282,
+      "grad_norm": 0.768993616104126,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 28180
+    },
+    {
+      "epoch": 90.3525641025641,
+      "grad_norm": 1.1043215990066528,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 28190
+    },
+    {
+      "epoch": 90.38461538461539,
+      "grad_norm": 0.7817684412002563,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 28200
+    },
+    {
+      "epoch": 90.41666666666667,
+      "grad_norm": 1.072939395904541,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 28210
+    },
+    {
+      "epoch": 90.44871794871794,
+      "grad_norm": 1.4935352802276611,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 28220
+    },
+    {
+      "epoch": 90.48076923076923,
+      "grad_norm": 1.0987670421600342,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 28230
+    },
+    {
+      "epoch": 90.51282051282051,
+      "grad_norm": 1.1646724939346313,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 28240
+    },
+    {
+      "epoch": 90.5448717948718,
+      "grad_norm": 1.0551669597625732,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 28250
+    },
+    {
+      "epoch": 90.57692307692308,
+      "grad_norm": 0.9614415764808655,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 28260
+    },
+    {
+      "epoch": 90.60897435897436,
+      "grad_norm": 1.0417048931121826,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 28270
+    },
+    {
+      "epoch": 90.64102564102564,
+      "grad_norm": 0.8284194469451904,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 28280
+    },
+    {
+      "epoch": 90.67307692307692,
+      "grad_norm": 0.8223841190338135,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 28290
+    },
+    {
+      "epoch": 90.7051282051282,
+      "grad_norm": 0.7318913340568542,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 28300
+    },
+    {
+      "epoch": 90.73717948717949,
+      "grad_norm": 1.1203962564468384,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 28310
+    },
+    {
+      "epoch": 90.76923076923077,
+      "grad_norm": 1.1374503374099731,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 28320
+    },
+    {
+      "epoch": 90.80128205128206,
+      "grad_norm": 1.0151300430297852,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 28330
+    },
+    {
+      "epoch": 90.83333333333333,
+      "grad_norm": 1.7323276996612549,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 28340
+    },
+    {
+      "epoch": 90.86538461538461,
+      "grad_norm": 1.197790503501892,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 28350
+    },
+    {
+      "epoch": 90.8974358974359,
+      "grad_norm": 1.1220053434371948,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 28360
+    },
+    {
+      "epoch": 90.92948717948718,
+      "grad_norm": 1.1403708457946777,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 28370
+    },
+    {
+      "epoch": 90.96153846153847,
+      "grad_norm": 1.4526711702346802,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 28380
+    },
+    {
+      "epoch": 90.99358974358974,
+      "grad_norm": 1.1199020147323608,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 28390
+    },
+    {
+      "epoch": 91.02564102564102,
+      "grad_norm": 1.1805237531661987,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 28400
+    },
+    {
+      "epoch": 91.0576923076923,
+      "grad_norm": 1.1483365297317505,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 28410
+    },
+    {
+      "epoch": 91.08974358974359,
+      "grad_norm": 1.0971534252166748,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 28420
+    },
+    {
+      "epoch": 91.12179487179488,
+      "grad_norm": 1.1056830883026123,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 28430
+    },
+    {
+      "epoch": 91.15384615384616,
+      "grad_norm": 0.8532825112342834,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 28440
+    },
+    {
+      "epoch": 91.18589743589743,
+      "grad_norm": 0.8221752047538757,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 28450
+    },
+    {
+      "epoch": 91.21794871794872,
+      "grad_norm": 1.0717806816101074,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 28460
+    },
+    {
+      "epoch": 91.25,
+      "grad_norm": 1.31162428855896,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 28470
+    },
+    {
+      "epoch": 91.28205128205128,
+      "grad_norm": 0.9340906143188477,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 28480
+    },
+    {
+      "epoch": 91.31410256410257,
+      "grad_norm": 1.0304266214370728,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 28490
+    },
+    {
+      "epoch": 91.34615384615384,
+      "grad_norm": 1.2808486223220825,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 28500
+    },
+    {
+      "epoch": 91.37820512820512,
+      "grad_norm": 1.0184332132339478,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 28510
+    },
+    {
+      "epoch": 91.41025641025641,
+      "grad_norm": 1.419906735420227,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 28520
+    },
+    {
+      "epoch": 91.4423076923077,
+      "grad_norm": 1.0944197177886963,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 28530
+    },
+    {
+      "epoch": 91.47435897435898,
+      "grad_norm": 1.0563242435455322,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 28540
+    },
+    {
+      "epoch": 91.50641025641026,
+      "grad_norm": 1.0987980365753174,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 28550
+    },
+    {
+      "epoch": 91.53846153846153,
+      "grad_norm": 1.1243349313735962,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 28560
+    },
+    {
+      "epoch": 91.57051282051282,
+      "grad_norm": 1.3600138425827026,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 28570
+    },
+    {
+      "epoch": 91.6025641025641,
+      "grad_norm": 0.9718136191368103,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 28580
+    },
+    {
+      "epoch": 91.63461538461539,
+      "grad_norm": 1.0374574661254883,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 28590
+    },
+    {
+      "epoch": 91.66666666666667,
+      "grad_norm": 0.9795124530792236,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 28600
+    },
+    {
+      "epoch": 91.69871794871794,
+      "grad_norm": 0.8945016860961914,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 28610
+    },
+    {
+      "epoch": 91.73076923076923,
+      "grad_norm": 0.8639125823974609,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 28620
+    },
+    {
+      "epoch": 91.76282051282051,
+      "grad_norm": 1.0644510984420776,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 28630
+    },
+    {
+      "epoch": 91.7948717948718,
+      "grad_norm": 1.1102157831192017,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 28640
+    },
+    {
+      "epoch": 91.82692307692308,
+      "grad_norm": 1.4365415573120117,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 28650
+    },
+    {
+      "epoch": 91.85897435897436,
+      "grad_norm": 1.1636052131652832,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 28660
+    },
+    {
+      "epoch": 91.89102564102564,
+      "grad_norm": 1.063966155052185,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 28670
+    },
+    {
+      "epoch": 91.92307692307692,
+      "grad_norm": 1.0715863704681396,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 28680
+    },
+    {
+      "epoch": 91.9551282051282,
+      "grad_norm": 0.9494783282279968,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 28690
+    },
+    {
+      "epoch": 91.98717948717949,
+      "grad_norm": 1.1644641160964966,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 28700
+    },
+    {
+      "epoch": 92.01923076923077,
+      "grad_norm": 0.9460110068321228,
+      "learning_rate": 0.0001,
+      "loss": 0.0116,
+      "step": 28710
+    },
+    {
+      "epoch": 92.05128205128206,
+      "grad_norm": 1.4648184776306152,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 28720
+    },
+    {
+      "epoch": 92.08333333333333,
+      "grad_norm": 0.9927682876586914,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 28730
+    },
+    {
+      "epoch": 92.11538461538461,
+      "grad_norm": 1.176065444946289,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 28740
+    },
+    {
+      "epoch": 92.1474358974359,
+      "grad_norm": 1.077075719833374,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 28750
+    },
+    {
+      "epoch": 92.17948717948718,
+      "grad_norm": 0.6825456023216248,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 28760
+    },
+    {
+      "epoch": 92.21153846153847,
+      "grad_norm": 1.1457738876342773,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 28770
+    },
+    {
+      "epoch": 92.24358974358974,
+      "grad_norm": 0.8125004172325134,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 28780
+    },
+    {
+      "epoch": 92.27564102564102,
+      "grad_norm": 1.0079777240753174,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 28790
+    },
+    {
+      "epoch": 92.3076923076923,
+      "grad_norm": 1.0483089685440063,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 28800
+    },
+    {
+      "epoch": 92.33974358974359,
+      "grad_norm": 1.4312201738357544,
+      "learning_rate": 0.0001,
+      "loss": 0.0111,
+      "step": 28810
+    },
+    {
+      "epoch": 92.37179487179488,
+      "grad_norm": 1.0436593294143677,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 28820
+    },
+    {
+      "epoch": 92.40384615384616,
+      "grad_norm": 1.1867754459381104,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 28830
+    },
+    {
+      "epoch": 92.43589743589743,
+      "grad_norm": 1.2549397945404053,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 28840
+    },
+    {
+      "epoch": 92.46794871794872,
+      "grad_norm": 1.3372987508773804,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 28850
+    },
+    {
+      "epoch": 92.5,
+      "grad_norm": 1.2272616624832153,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 28860
+    },
+    {
+      "epoch": 92.53205128205128,
+      "grad_norm": 0.988530695438385,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 28870
+    },
+    {
+      "epoch": 92.56410256410257,
+      "grad_norm": 0.9237765073776245,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 28880
+    },
+    {
+      "epoch": 92.59615384615384,
+      "grad_norm": 1.1870150566101074,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 28890
+    },
+    {
+      "epoch": 92.62820512820512,
+      "grad_norm": 1.1641806364059448,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 28900
+    },
+    {
+      "epoch": 92.66025641025641,
+      "grad_norm": 0.9005507826805115,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 28910
+    },
+    {
+      "epoch": 92.6923076923077,
+      "grad_norm": 0.9542353749275208,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 28920
+    },
+    {
+      "epoch": 92.72435897435898,
+      "grad_norm": 0.845251739025116,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 28930
+    },
+    {
+      "epoch": 92.75641025641026,
+      "grad_norm": 0.7599291801452637,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 28940
+    },
+    {
+      "epoch": 92.78846153846153,
+      "grad_norm": 0.8272253274917603,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 28950
+    },
+    {
+      "epoch": 92.82051282051282,
+      "grad_norm": 0.7508731484413147,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 28960
+    },
+    {
+      "epoch": 92.8525641025641,
+      "grad_norm": 0.7820129990577698,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 28970
+    },
+    {
+      "epoch": 92.88461538461539,
+      "grad_norm": 1.077419400215149,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 28980
+    },
+    {
+      "epoch": 92.91666666666667,
+      "grad_norm": 0.9351970553398132,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 28990
+    },
+    {
+      "epoch": 92.94871794871794,
+      "grad_norm": 0.9406611919403076,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 29000
+    },
+    {
+      "epoch": 92.98076923076923,
+      "grad_norm": 1.3225135803222656,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 29010
+    },
+    {
+      "epoch": 93.01282051282051,
+      "grad_norm": 1.362971544265747,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 29020
+    },
+    {
+      "epoch": 93.0448717948718,
+      "grad_norm": 1.2444289922714233,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 29030
+    },
+    {
+      "epoch": 93.07692307692308,
+      "grad_norm": 1.1266038417816162,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 29040
+    },
+    {
+      "epoch": 93.10897435897436,
+      "grad_norm": 0.8148980140686035,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29050
+    },
+    {
+      "epoch": 93.14102564102564,
+      "grad_norm": 1.2963305711746216,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 29060
+    },
+    {
+      "epoch": 93.17307692307692,
+      "grad_norm": 0.9989068508148193,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29070
+    },
+    {
+      "epoch": 93.2051282051282,
+      "grad_norm": 0.8131892681121826,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 29080
+    },
+    {
+      "epoch": 93.23717948717949,
+      "grad_norm": 0.8659066557884216,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 29090
+    },
+    {
+      "epoch": 93.26923076923077,
+      "grad_norm": 0.750195324420929,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29100
+    },
+    {
+      "epoch": 93.30128205128206,
+      "grad_norm": 0.894476056098938,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29110
+    },
+    {
+      "epoch": 93.33333333333333,
+      "grad_norm": 1.0347065925598145,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 29120
+    },
+    {
+      "epoch": 93.36538461538461,
+      "grad_norm": 0.9128508567810059,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 29130
+    },
+    {
+      "epoch": 93.3974358974359,
+      "grad_norm": 1.055043339729309,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 29140
+    },
+    {
+      "epoch": 93.42948717948718,
+      "grad_norm": 0.9750173687934875,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 29150
+    },
+    {
+      "epoch": 93.46153846153847,
+      "grad_norm": 1.0037808418273926,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 29160
+    },
+    {
+      "epoch": 93.49358974358974,
+      "grad_norm": 0.8471161723136902,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29170
+    },
+    {
+      "epoch": 93.52564102564102,
+      "grad_norm": 0.8943414092063904,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 29180
+    },
+    {
+      "epoch": 93.5576923076923,
+      "grad_norm": 0.6343587636947632,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 29190
+    },
+    {
+      "epoch": 93.58974358974359,
+      "grad_norm": 0.6261102557182312,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29200
+    },
+    {
+      "epoch": 93.62179487179488,
+      "grad_norm": 0.7162252068519592,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 29210
+    },
+    {
+      "epoch": 93.65384615384616,
+      "grad_norm": 0.8985969424247742,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29220
+    },
+    {
+      "epoch": 93.68589743589743,
+      "grad_norm": 0.9556020498275757,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 29230
+    },
+    {
+      "epoch": 93.71794871794872,
+      "grad_norm": 0.8684034943580627,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 29240
+    },
+    {
+      "epoch": 93.75,
+      "grad_norm": 1.3074047565460205,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 29250
+    },
+    {
+      "epoch": 93.78205128205128,
+      "grad_norm": 1.0871189832687378,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 29260
+    },
+    {
+      "epoch": 93.81410256410257,
+      "grad_norm": 1.183151125907898,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 29270
+    },
+    {
+      "epoch": 93.84615384615384,
+      "grad_norm": 1.4089962244033813,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 29280
+    },
+    {
+      "epoch": 93.87820512820512,
+      "grad_norm": 0.9743595123291016,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 29290
+    },
+    {
+      "epoch": 93.91025641025641,
+      "grad_norm": 1.1530818939208984,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 29300
+    },
+    {
+      "epoch": 93.9423076923077,
+      "grad_norm": 1.004894495010376,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 29310
+    },
+    {
+      "epoch": 93.97435897435898,
+      "grad_norm": 1.116240382194519,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 29320
+    },
+    {
+      "epoch": 94.00641025641026,
+      "grad_norm": 0.8370277285575867,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 29330
+    },
+    {
+      "epoch": 94.03846153846153,
+      "grad_norm": 0.804837167263031,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 29340
+    },
+    {
+      "epoch": 94.07051282051282,
+      "grad_norm": 1.1252539157867432,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 29350
+    },
+    {
+      "epoch": 94.1025641025641,
+      "grad_norm": 1.1337395906448364,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 29360
+    },
+    {
+      "epoch": 94.13461538461539,
+      "grad_norm": 0.9703201651573181,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 29370
+    },
+    {
+      "epoch": 94.16666666666667,
+      "grad_norm": 0.7305422425270081,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 29380
+    },
+    {
+      "epoch": 94.19871794871794,
+      "grad_norm": 0.8227375745773315,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 29390
+    },
+    {
+      "epoch": 94.23076923076923,
+      "grad_norm": 0.8815519213676453,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 29400
+    },
+    {
+      "epoch": 94.26282051282051,
+      "grad_norm": 0.8792902231216431,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 29410
+    },
+    {
+      "epoch": 94.2948717948718,
+      "grad_norm": 1.2412927150726318,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 29420
+    },
+    {
+      "epoch": 94.32692307692308,
+      "grad_norm": 1.3938851356506348,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 29430
+    },
+    {
+      "epoch": 94.35897435897436,
+      "grad_norm": 1.648319959640503,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 29440
+    },
+    {
+      "epoch": 94.39102564102564,
+      "grad_norm": 1.790331482887268,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 29450
+    },
+    {
+      "epoch": 94.42307692307692,
+      "grad_norm": 1.417850375175476,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 29460
+    },
+    {
+      "epoch": 94.4551282051282,
+      "grad_norm": 1.1215909719467163,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 29470
+    },
+    {
+      "epoch": 94.48717948717949,
+      "grad_norm": 1.0406571626663208,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 29480
+    },
+    {
+      "epoch": 94.51923076923077,
+      "grad_norm": 0.9620739221572876,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 29490
+    },
+    {
+      "epoch": 94.55128205128206,
+      "grad_norm": 0.8626880049705505,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 29500
+    },
+    {
+      "epoch": 94.58333333333333,
+      "grad_norm": 0.8712671995162964,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29510
+    },
+    {
+      "epoch": 94.61538461538461,
+      "grad_norm": 0.9128088355064392,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 29520
+    },
+    {
+      "epoch": 94.6474358974359,
+      "grad_norm": 1.0133453607559204,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 29530
+    },
+    {
+      "epoch": 94.67948717948718,
+      "grad_norm": 1.0725053548812866,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 29540
+    },
+    {
+      "epoch": 94.71153846153847,
+      "grad_norm": 1.0330123901367188,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 29550
+    },
+    {
+      "epoch": 94.74358974358974,
+      "grad_norm": 1.2678139209747314,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29560
+    },
+    {
+      "epoch": 94.77564102564102,
+      "grad_norm": 0.9273304343223572,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 29570
+    },
+    {
+      "epoch": 94.8076923076923,
+      "grad_norm": 0.9408043622970581,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 29580
+    },
+    {
+      "epoch": 94.83974358974359,
+      "grad_norm": 0.9495874643325806,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 29590
+    },
+    {
+      "epoch": 94.87179487179488,
+      "grad_norm": 1.3754467964172363,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29600
+    },
+    {
+      "epoch": 94.90384615384616,
+      "grad_norm": 1.1390533447265625,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 29610
+    },
+    {
+      "epoch": 94.93589743589743,
+      "grad_norm": 0.9797568321228027,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 29620
+    },
+    {
+      "epoch": 94.96794871794872,
+      "grad_norm": 1.055306315422058,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29630
+    },
+    {
+      "epoch": 95.0,
+      "grad_norm": 0.9698041677474976,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 29640
+    },
+    {
+      "epoch": 95.03205128205128,
+      "grad_norm": 1.152928352355957,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 29650
+    },
+    {
+      "epoch": 95.06410256410257,
+      "grad_norm": 1.1152628660202026,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 29660
+    },
+    {
+      "epoch": 95.09615384615384,
+      "grad_norm": 1.1973178386688232,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29670
+    },
+    {
+      "epoch": 95.12820512820512,
+      "grad_norm": 0.9419043660163879,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29680
+    },
+    {
+      "epoch": 95.16025641025641,
+      "grad_norm": 0.8008570075035095,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 29690
+    },
+    {
+      "epoch": 95.1923076923077,
+      "grad_norm": 0.8279953598976135,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 29700
+    },
+    {
+      "epoch": 95.22435897435898,
+      "grad_norm": 1.1259633302688599,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 29710
+    },
+    {
+      "epoch": 95.25641025641026,
+      "grad_norm": 1.1547572612762451,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 29720
+    },
+    {
+      "epoch": 95.28846153846153,
+      "grad_norm": 1.039164423942566,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 29730
+    },
+    {
+      "epoch": 95.32051282051282,
+      "grad_norm": 0.9760857820510864,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 29740
+    },
+    {
+      "epoch": 95.3525641025641,
+      "grad_norm": 0.8577026724815369,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 29750
+    },
+    {
+      "epoch": 95.38461538461539,
+      "grad_norm": 0.9697113633155823,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 29760
+    },
+    {
+      "epoch": 95.41666666666667,
+      "grad_norm": 0.8354844450950623,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 29770
+    },
+    {
+      "epoch": 95.44871794871794,
+      "grad_norm": 1.3231652975082397,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29780
+    },
+    {
+      "epoch": 95.48076923076923,
+      "grad_norm": 0.7189549207687378,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 29790
+    },
+    {
+      "epoch": 95.51282051282051,
+      "grad_norm": 0.8349175453186035,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 29800
+    },
+    {
+      "epoch": 95.5448717948718,
+      "grad_norm": 0.9250572919845581,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 29810
+    },
+    {
+      "epoch": 95.57692307692308,
+      "grad_norm": 1.13448166847229,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 29820
+    },
+    {
+      "epoch": 95.60897435897436,
+      "grad_norm": 0.9322713017463684,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 29830
+    },
+    {
+      "epoch": 95.64102564102564,
+      "grad_norm": 0.8138542771339417,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 29840
+    },
+    {
+      "epoch": 95.67307692307692,
+      "grad_norm": 1.1056312322616577,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 29850
+    },
+    {
+      "epoch": 95.7051282051282,
+      "grad_norm": 0.9851621389389038,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 29860
+    },
+    {
+      "epoch": 95.73717948717949,
+      "grad_norm": 0.9687674641609192,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 29870
+    },
+    {
+      "epoch": 95.76923076923077,
+      "grad_norm": 1.274790644645691,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 29880
+    },
+    {
+      "epoch": 95.80128205128206,
+      "grad_norm": 0.9387258291244507,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29890
+    },
+    {
+      "epoch": 95.83333333333333,
+      "grad_norm": 0.9499697685241699,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29900
+    },
+    {
+      "epoch": 95.86538461538461,
+      "grad_norm": 1.082606315612793,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29910
+    },
+    {
+      "epoch": 95.8974358974359,
+      "grad_norm": 1.3048943281173706,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 29920
+    },
+    {
+      "epoch": 95.92948717948718,
+      "grad_norm": 1.271277904510498,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29930
+    },
+    {
+      "epoch": 95.96153846153847,
+      "grad_norm": 1.2169809341430664,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 29940
+    },
+    {
+      "epoch": 95.99358974358974,
+      "grad_norm": 1.3270673751831055,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 29950
+    },
+    {
+      "epoch": 96.02564102564102,
+      "grad_norm": 1.072044849395752,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 29960
+    },
+    {
+      "epoch": 96.0576923076923,
+      "grad_norm": 0.9607635140419006,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 29970
+    },
+    {
+      "epoch": 96.08974358974359,
+      "grad_norm": 0.9160192012786865,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 29980
+    },
+    {
+      "epoch": 96.12179487179488,
+      "grad_norm": 0.9196527600288391,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 29990
+    },
+    {
+      "epoch": 96.15384615384616,
+      "grad_norm": 0.6967593431472778,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 30000
+    },
+    {
+      "epoch": 96.18589743589743,
+      "grad_norm": 0.9979833960533142,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 30010
+    },
+    {
+      "epoch": 96.21794871794872,
+      "grad_norm": 1.0538158416748047,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 30020
+    },
+    {
+      "epoch": 96.25,
+      "grad_norm": 1.1540195941925049,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 30030
+    },
+    {
+      "epoch": 96.28205128205128,
+      "grad_norm": 0.9779187440872192,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 30040
+    },
+    {
+      "epoch": 96.31410256410257,
+      "grad_norm": 1.065081000328064,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 30050
+    },
+    {
+      "epoch": 96.34615384615384,
+      "grad_norm": 0.9001664519309998,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 30060
+    },
+    {
+      "epoch": 96.37820512820512,
+      "grad_norm": 0.96477210521698,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 30070
+    },
+    {
+      "epoch": 96.41025641025641,
+      "grad_norm": 0.9198656678199768,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 30080
+    },
+    {
+      "epoch": 96.4423076923077,
+      "grad_norm": 1.0044944286346436,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 30090
+    },
+    {
+      "epoch": 96.47435897435898,
+      "grad_norm": 1.4076038599014282,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 30100
+    },
+    {
+      "epoch": 96.50641025641026,
+      "grad_norm": 0.787113606929779,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 30110
+    },
+    {
+      "epoch": 96.53846153846153,
+      "grad_norm": 0.8142930865287781,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 30120
+    },
+    {
+      "epoch": 96.57051282051282,
+      "grad_norm": 1.2032959461212158,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 30130
+    },
+    {
+      "epoch": 96.6025641025641,
+      "grad_norm": 0.9988319873809814,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 30140
+    },
+    {
+      "epoch": 96.63461538461539,
+      "grad_norm": 0.951921284198761,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 30150
+    },
+    {
+      "epoch": 96.66666666666667,
+      "grad_norm": 1.2056597471237183,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 30160
+    },
+    {
+      "epoch": 96.69871794871794,
+      "grad_norm": 1.490012288093567,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 30170
+    },
+    {
+      "epoch": 96.73076923076923,
+      "grad_norm": 1.387589931488037,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 30180
+    },
+    {
+      "epoch": 96.76282051282051,
+      "grad_norm": 1.3152523040771484,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 30190
+    },
+    {
+      "epoch": 96.7948717948718,
+      "grad_norm": 1.0985549688339233,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 30200
+    },
+    {
+      "epoch": 96.82692307692308,
+      "grad_norm": 0.9332179427146912,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 30210
+    },
+    {
+      "epoch": 96.85897435897436,
+      "grad_norm": 1.1849873065948486,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 30220
+    },
+    {
+      "epoch": 96.89102564102564,
+      "grad_norm": 1.346021056175232,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30230
+    },
+    {
+      "epoch": 96.92307692307692,
+      "grad_norm": 1.1336413621902466,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30240
+    },
+    {
+      "epoch": 96.9551282051282,
+      "grad_norm": 1.3273988962173462,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30250
+    },
+    {
+      "epoch": 96.98717948717949,
+      "grad_norm": 1.3515723943710327,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 30260
+    },
+    {
+      "epoch": 97.01923076923077,
+      "grad_norm": 1.5498100519180298,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 30270
+    },
+    {
+      "epoch": 97.05128205128206,
+      "grad_norm": 0.9283611178398132,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 30280
+    },
+    {
+      "epoch": 97.08333333333333,
+      "grad_norm": 1.120369553565979,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30290
+    },
+    {
+      "epoch": 97.11538461538461,
+      "grad_norm": 1.1480189561843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 30300
+    },
+    {
+      "epoch": 97.1474358974359,
+      "grad_norm": 1.3063420057296753,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 30310
+    },
+    {
+      "epoch": 97.17948717948718,
+      "grad_norm": 0.9092293977737427,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30320
+    },
+    {
+      "epoch": 97.21153846153847,
+      "grad_norm": 0.9000687003135681,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 30330
+    },
+    {
+      "epoch": 97.24358974358974,
+      "grad_norm": 0.9905684590339661,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30340
+    },
+    {
+      "epoch": 97.27564102564102,
+      "grad_norm": 1.095186710357666,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 30350
+    },
+    {
+      "epoch": 97.3076923076923,
+      "grad_norm": 1.1197947263717651,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30360
+    },
+    {
+      "epoch": 97.33974358974359,
+      "grad_norm": 0.8946802020072937,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 30370
+    },
+    {
+      "epoch": 97.37179487179488,
+      "grad_norm": 1.1427544355392456,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30380
+    },
+    {
+      "epoch": 97.40384615384616,
+      "grad_norm": 0.9852504134178162,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 30390
+    },
+    {
+      "epoch": 97.43589743589743,
+      "grad_norm": 0.815701425075531,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30400
+    },
+    {
+      "epoch": 97.46794871794872,
+      "grad_norm": 1.2773734331130981,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30410
+    },
+    {
+      "epoch": 97.5,
+      "grad_norm": 1.3894448280334473,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30420
+    },
+    {
+      "epoch": 97.53205128205128,
+      "grad_norm": 0.9824554920196533,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 30430
+    },
+    {
+      "epoch": 97.56410256410257,
+      "grad_norm": 0.9622264504432678,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 30440
+    },
+    {
+      "epoch": 97.59615384615384,
+      "grad_norm": 0.9713234901428223,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 30450
+    },
+    {
+      "epoch": 97.62820512820512,
+      "grad_norm": 1.0194907188415527,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30460
+    },
+    {
+      "epoch": 97.66025641025641,
+      "grad_norm": 0.8545103073120117,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30470
+    },
+    {
+      "epoch": 97.6923076923077,
+      "grad_norm": 0.8379747867584229,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 30480
+    },
+    {
+      "epoch": 97.72435897435898,
+      "grad_norm": 0.7695945501327515,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 30490
+    },
+    {
+      "epoch": 97.75641025641026,
+      "grad_norm": 0.8329702019691467,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 30500
+    },
+    {
+      "epoch": 97.78846153846153,
+      "grad_norm": 0.7467461824417114,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 30510
+    },
+    {
+      "epoch": 97.82051282051282,
+      "grad_norm": 0.653804361820221,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 30520
+    },
+    {
+      "epoch": 97.8525641025641,
+      "grad_norm": 0.7265058755874634,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 30530
+    },
+    {
+      "epoch": 97.88461538461539,
+      "grad_norm": 0.7803757786750793,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30540
+    },
+    {
+      "epoch": 97.91666666666667,
+      "grad_norm": 0.9205565452575684,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30550
+    },
+    {
+      "epoch": 97.94871794871794,
+      "grad_norm": 0.9536099433898926,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 30560
+    },
+    {
+      "epoch": 97.98076923076923,
+      "grad_norm": 0.9282727837562561,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 30570
+    },
+    {
+      "epoch": 98.01282051282051,
+      "grad_norm": 1.5871555805206299,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 30580
+    },
+    {
+      "epoch": 98.0448717948718,
+      "grad_norm": 1.1079682111740112,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 30590
+    },
+    {
+      "epoch": 98.07692307692308,
+      "grad_norm": 0.8366027474403381,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 30600
+    },
+    {
+      "epoch": 98.10897435897436,
+      "grad_norm": 0.8468282222747803,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 30610
+    },
+    {
+      "epoch": 98.14102564102564,
+      "grad_norm": 0.8458372354507446,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 30620
+    },
+    {
+      "epoch": 98.17307692307692,
+      "grad_norm": 0.852669358253479,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30630
+    },
+    {
+      "epoch": 98.2051282051282,
+      "grad_norm": 0.8848656415939331,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 30640
+    },
+    {
+      "epoch": 98.23717948717949,
+      "grad_norm": 0.9323807954788208,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 30650
+    },
+    {
+      "epoch": 98.26923076923077,
+      "grad_norm": 1.1894923448562622,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 30660
+    },
+    {
+      "epoch": 98.30128205128206,
+      "grad_norm": 0.8370342254638672,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 30670
+    },
+    {
+      "epoch": 98.33333333333333,
+      "grad_norm": 1.226685881614685,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 30680
+    },
+    {
+      "epoch": 98.36538461538461,
+      "grad_norm": 0.793327271938324,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 30690
+    },
+    {
+      "epoch": 98.3974358974359,
+      "grad_norm": 0.9912601113319397,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 30700
+    },
+    {
+      "epoch": 98.42948717948718,
+      "grad_norm": 0.9260236620903015,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 30710
+    },
+    {
+      "epoch": 98.46153846153847,
+      "grad_norm": 1.098226547241211,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 30720
+    },
+    {
+      "epoch": 98.49358974358974,
+      "grad_norm": 1.497388243675232,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 30730
+    },
+    {
+      "epoch": 98.52564102564102,
+      "grad_norm": 1.072384238243103,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 30740
+    },
+    {
+      "epoch": 98.5576923076923,
+      "grad_norm": 1.1668776273727417,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 30750
+    },
+    {
+      "epoch": 98.58974358974359,
+      "grad_norm": 1.041835069656372,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30760
+    },
+    {
+      "epoch": 98.62179487179488,
+      "grad_norm": 0.8520818948745728,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30770
+    },
+    {
+      "epoch": 98.65384615384616,
+      "grad_norm": 1.157824158668518,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 30780
+    },
+    {
+      "epoch": 98.68589743589743,
+      "grad_norm": 1.138601303100586,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 30790
+    },
+    {
+      "epoch": 98.71794871794872,
+      "grad_norm": 0.8706300258636475,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30800
+    },
+    {
+      "epoch": 98.75,
+      "grad_norm": 0.610787034034729,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 30810
+    },
+    {
+      "epoch": 98.78205128205128,
+      "grad_norm": 1.1509610414505005,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30820
+    },
+    {
+      "epoch": 98.81410256410257,
+      "grad_norm": 0.7177321314811707,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 30830
+    },
+    {
+      "epoch": 98.84615384615384,
+      "grad_norm": 0.961208164691925,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 30840
+    },
+    {
+      "epoch": 98.87820512820512,
+      "grad_norm": 0.64850914478302,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30850
+    },
+    {
+      "epoch": 98.91025641025641,
+      "grad_norm": 0.774963915348053,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30860
+    },
+    {
+      "epoch": 98.9423076923077,
+      "grad_norm": 0.8816413283348083,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 30870
+    },
+    {
+      "epoch": 98.97435897435898,
+      "grad_norm": 1.1564134359359741,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30880
+    },
+    {
+      "epoch": 99.00641025641026,
+      "grad_norm": 0.7105777859687805,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 30890
+    },
+    {
+      "epoch": 99.03846153846153,
+      "grad_norm": 0.9117348790168762,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 30900
+    },
+    {
+      "epoch": 99.07051282051282,
+      "grad_norm": 0.9415019154548645,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 30910
+    },
+    {
+      "epoch": 99.1025641025641,
+      "grad_norm": 0.9867045879364014,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 30920
+    },
+    {
+      "epoch": 99.13461538461539,
+      "grad_norm": 0.7480201125144958,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 30930
+    },
+    {
+      "epoch": 99.16666666666667,
+      "grad_norm": 0.6586176753044128,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 30940
+    },
+    {
+      "epoch": 99.19871794871794,
+      "grad_norm": 0.8653594851493835,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 30950
+    },
+    {
+      "epoch": 99.23076923076923,
+      "grad_norm": 2.9124183654785156,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 30960
+    },
+    {
+      "epoch": 99.26282051282051,
+      "grad_norm": 1.6115987300872803,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 30970
+    },
+    {
+      "epoch": 99.2948717948718,
+      "grad_norm": 1.9272806644439697,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 30980
+    },
+    {
+      "epoch": 99.32692307692308,
+      "grad_norm": 1.7457468509674072,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 30990
+    },
+    {
+      "epoch": 99.35897435897436,
+      "grad_norm": 1.7888545989990234,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 31000
+    },
+    {
+      "epoch": 99.39102564102564,
+      "grad_norm": 1.8599727153778076,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 31010
+    },
+    {
+      "epoch": 99.42307692307692,
+      "grad_norm": 1.0302637815475464,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 31020
+    },
+    {
+      "epoch": 99.4551282051282,
+      "grad_norm": 0.8894798159599304,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 31030
+    },
+    {
+      "epoch": 99.48717948717949,
+      "grad_norm": 0.9087692499160767,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 31040
+    },
+    {
+      "epoch": 99.51923076923077,
+      "grad_norm": 1.0197643041610718,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 31050
+    },
+    {
+      "epoch": 99.55128205128206,
+      "grad_norm": 1.373435378074646,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 31060
+    },
+    {
+      "epoch": 99.58333333333333,
+      "grad_norm": 1.2151329517364502,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 31070
+    },
+    {
+      "epoch": 99.61538461538461,
+      "grad_norm": 0.8429015874862671,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31080
+    },
+    {
+      "epoch": 99.6474358974359,
+      "grad_norm": 1.0141538381576538,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 31090
+    },
+    {
+      "epoch": 99.67948717948718,
+      "grad_norm": 1.1431126594543457,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 31100
+    },
+    {
+      "epoch": 99.71153846153847,
+      "grad_norm": 1.2662638425827026,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 31110
+    },
+    {
+      "epoch": 99.74358974358974,
+      "grad_norm": 0.9626235365867615,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 31120
+    },
+    {
+      "epoch": 99.77564102564102,
+      "grad_norm": 1.1287992000579834,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 31130
+    },
+    {
+      "epoch": 99.8076923076923,
+      "grad_norm": 1.0874022245407104,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 31140
+    },
+    {
+      "epoch": 99.83974358974359,
+      "grad_norm": 0.8721771240234375,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 31150
+    },
+    {
+      "epoch": 99.87179487179488,
+      "grad_norm": 0.95539391040802,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 31160
+    },
+    {
+      "epoch": 99.90384615384616,
+      "grad_norm": 0.7638134956359863,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 31170
+    },
+    {
+      "epoch": 99.93589743589743,
+      "grad_norm": 1.1967852115631104,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 31180
+    },
+    {
+      "epoch": 99.96794871794872,
+      "grad_norm": 1.0417975187301636,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 31190
+    },
+    {
+      "epoch": 100.0,
+      "grad_norm": 1.0326917171478271,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31200
+    },
+    {
+      "epoch": 100.03205128205128,
+      "grad_norm": 1.0002565383911133,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 31210
+    },
+    {
+      "epoch": 100.06410256410257,
+      "grad_norm": 0.9318867325782776,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 31220
+    },
+    {
+      "epoch": 100.09615384615384,
+      "grad_norm": 1.150520920753479,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 31230
+    },
+    {
+      "epoch": 100.12820512820512,
+      "grad_norm": 1.0451037883758545,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 31240
+    },
+    {
+      "epoch": 100.16025641025641,
+      "grad_norm": 1.0879089832305908,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 31250
+    },
+    {
+      "epoch": 100.1923076923077,
+      "grad_norm": 1.195737600326538,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 31260
+    },
+    {
+      "epoch": 100.22435897435898,
+      "grad_norm": 1.0355607271194458,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 31270
+    },
+    {
+      "epoch": 100.25641025641026,
+      "grad_norm": 0.9092247486114502,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 31280
+    },
+    {
+      "epoch": 100.28846153846153,
+      "grad_norm": 0.8770983815193176,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 31290
+    },
+    {
+      "epoch": 100.32051282051282,
+      "grad_norm": 1.200512409210205,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 31300
+    },
+    {
+      "epoch": 100.3525641025641,
+      "grad_norm": 1.2089014053344727,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 31310
+    },
+    {
+      "epoch": 100.38461538461539,
+      "grad_norm": 1.1373724937438965,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 31320
+    },
+    {
+      "epoch": 100.41666666666667,
+      "grad_norm": 0.6583201885223389,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 31330
+    },
+    {
+      "epoch": 100.44871794871794,
+      "grad_norm": 0.8903700113296509,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31340
+    },
+    {
+      "epoch": 100.48076923076923,
+      "grad_norm": 1.026665449142456,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31350
+    },
+    {
+      "epoch": 100.51282051282051,
+      "grad_norm": 1.1131681203842163,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31360
+    },
+    {
+      "epoch": 100.5448717948718,
+      "grad_norm": 1.3122880458831787,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 31370
+    },
+    {
+      "epoch": 100.57692307692308,
+      "grad_norm": 1.115187644958496,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 31380
+    },
+    {
+      "epoch": 100.60897435897436,
+      "grad_norm": 0.7986509799957275,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 31390
+    },
+    {
+      "epoch": 100.64102564102564,
+      "grad_norm": 0.997673511505127,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31400
+    },
+    {
+      "epoch": 100.67307692307692,
+      "grad_norm": 1.1208261251449585,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 31410
+    },
+    {
+      "epoch": 100.7051282051282,
+      "grad_norm": 0.8164052963256836,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 31420
+    },
+    {
+      "epoch": 100.73717948717949,
+      "grad_norm": 0.9595080018043518,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31430
+    },
+    {
+      "epoch": 100.76923076923077,
+      "grad_norm": 0.8583412766456604,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 31440
+    },
+    {
+      "epoch": 100.80128205128206,
+      "grad_norm": 0.9395642280578613,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 31450
+    },
+    {
+      "epoch": 100.83333333333333,
+      "grad_norm": 0.8325936794281006,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 31460
+    },
+    {
+      "epoch": 100.86538461538461,
+      "grad_norm": 0.8576821088790894,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 31470
+    },
+    {
+      "epoch": 100.8974358974359,
+      "grad_norm": 0.9073320031166077,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31480
+    },
+    {
+      "epoch": 100.92948717948718,
+      "grad_norm": 1.019034743309021,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 31490
+    },
+    {
+      "epoch": 100.96153846153847,
+      "grad_norm": 1.1553974151611328,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 31500
+    },
+    {
+      "epoch": 100.99358974358974,
+      "grad_norm": 1.2547258138656616,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31510
+    },
+    {
+      "epoch": 101.02564102564102,
+      "grad_norm": 1.0984326601028442,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 31520
+    },
+    {
+      "epoch": 101.0576923076923,
+      "grad_norm": 0.9818124175071716,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 31530
+    },
+    {
+      "epoch": 101.08974358974359,
+      "grad_norm": 0.9469954967498779,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 31540
+    },
+    {
+      "epoch": 101.12179487179488,
+      "grad_norm": 0.817945122718811,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 31550
+    },
+    {
+      "epoch": 101.15384615384616,
+      "grad_norm": 0.8023650050163269,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 31560
+    },
+    {
+      "epoch": 101.18589743589743,
+      "grad_norm": 0.9219418168067932,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31570
+    },
+    {
+      "epoch": 101.21794871794872,
+      "grad_norm": 0.7077937126159668,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 31580
+    },
+    {
+      "epoch": 101.25,
+      "grad_norm": 1.0817831754684448,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 31590
+    },
+    {
+      "epoch": 101.28205128205128,
+      "grad_norm": 1.0397425889968872,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 31600
+    },
+    {
+      "epoch": 101.31410256410257,
+      "grad_norm": 1.120374083518982,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 31610
+    },
+    {
+      "epoch": 101.34615384615384,
+      "grad_norm": 0.7550649046897888,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 31620
+    },
+    {
+      "epoch": 101.37820512820512,
+      "grad_norm": 0.761455237865448,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31630
+    },
+    {
+      "epoch": 101.41025641025641,
+      "grad_norm": 0.9833846688270569,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31640
+    },
+    {
+      "epoch": 101.4423076923077,
+      "grad_norm": 1.2119204998016357,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 31650
+    },
+    {
+      "epoch": 101.47435897435898,
+      "grad_norm": 1.1763670444488525,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 31660
+    },
+    {
+      "epoch": 101.50641025641026,
+      "grad_norm": 1.1194134950637817,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31670
+    },
+    {
+      "epoch": 101.53846153846153,
+      "grad_norm": 1.072326421737671,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 31680
+    },
+    {
+      "epoch": 101.57051282051282,
+      "grad_norm": 1.100206971168518,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 31690
+    },
+    {
+      "epoch": 101.6025641025641,
+      "grad_norm": 1.09629225730896,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 31700
+    },
+    {
+      "epoch": 101.63461538461539,
+      "grad_norm": 0.8254122138023376,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 31710
+    },
+    {
+      "epoch": 101.66666666666667,
+      "grad_norm": 0.8064016103744507,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 31720
+    },
+    {
+      "epoch": 101.69871794871794,
+      "grad_norm": 0.9167472720146179,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31730
+    },
+    {
+      "epoch": 101.73076923076923,
+      "grad_norm": 0.8509678840637207,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 31740
+    },
+    {
+      "epoch": 101.76282051282051,
+      "grad_norm": 0.8907046318054199,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 31750
+    },
+    {
+      "epoch": 101.7948717948718,
+      "grad_norm": 1.1079720258712769,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31760
+    },
+    {
+      "epoch": 101.82692307692308,
+      "grad_norm": 1.053181529045105,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31770
+    },
+    {
+      "epoch": 101.85897435897436,
+      "grad_norm": 1.0599677562713623,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31780
+    },
+    {
+      "epoch": 101.89102564102564,
+      "grad_norm": 0.9726989269256592,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 31790
+    },
+    {
+      "epoch": 101.92307692307692,
+      "grad_norm": 1.097075343132019,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 31800
+    },
+    {
+      "epoch": 101.9551282051282,
+      "grad_norm": 1.1299819946289062,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 31810
+    },
+    {
+      "epoch": 101.98717948717949,
+      "grad_norm": 1.054342269897461,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31820
+    },
+    {
+      "epoch": 102.01923076923077,
+      "grad_norm": 0.8541196584701538,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 31830
+    },
+    {
+      "epoch": 102.05128205128206,
+      "grad_norm": 1.3153631687164307,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 31840
+    },
+    {
+      "epoch": 102.08333333333333,
+      "grad_norm": 0.9673150181770325,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31850
+    },
+    {
+      "epoch": 102.11538461538461,
+      "grad_norm": 1.0887483358383179,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 31860
+    },
+    {
+      "epoch": 102.1474358974359,
+      "grad_norm": 0.9809402227401733,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 31870
+    },
+    {
+      "epoch": 102.17948717948718,
+      "grad_norm": 0.8767188191413879,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 31880
+    },
+    {
+      "epoch": 102.21153846153847,
+      "grad_norm": 0.9457308053970337,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 31890
+    },
+    {
+      "epoch": 102.24358974358974,
+      "grad_norm": 0.831484854221344,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 31900
+    },
+    {
+      "epoch": 102.27564102564102,
+      "grad_norm": 0.8854587078094482,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 31910
+    },
+    {
+      "epoch": 102.3076923076923,
+      "grad_norm": 0.9425541758537292,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 31920
+    },
+    {
+      "epoch": 102.33974358974359,
+      "grad_norm": 0.9164042472839355,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 31930
+    },
+    {
+      "epoch": 102.37179487179488,
+      "grad_norm": 1.103002667427063,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 31940
+    },
+    {
+      "epoch": 102.40384615384616,
+      "grad_norm": 0.8039612174034119,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 31950
+    },
+    {
+      "epoch": 102.43589743589743,
+      "grad_norm": 0.8169320225715637,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 31960
+    },
+    {
+      "epoch": 102.46794871794872,
+      "grad_norm": 0.8716517686843872,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 31970
+    },
+    {
+      "epoch": 102.5,
+      "grad_norm": 1.230852723121643,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 31980
+    },
+    {
+      "epoch": 102.53205128205128,
+      "grad_norm": 1.1143795251846313,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 31990
+    },
+    {
+      "epoch": 102.56410256410257,
+      "grad_norm": 1.0625293254852295,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32000
+    },
+    {
+      "epoch": 102.59615384615384,
+      "grad_norm": 0.9873017072677612,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32010
+    },
+    {
+      "epoch": 102.62820512820512,
+      "grad_norm": 0.711355447769165,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32020
+    },
+    {
+      "epoch": 102.66025641025641,
+      "grad_norm": 0.7498918175697327,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32030
+    },
+    {
+      "epoch": 102.6923076923077,
+      "grad_norm": 1.4504958391189575,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 32040
+    },
+    {
+      "epoch": 102.72435897435898,
+      "grad_norm": 1.2904976606369019,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32050
+    },
+    {
+      "epoch": 102.75641025641026,
+      "grad_norm": 1.3049110174179077,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 32060
+    },
+    {
+      "epoch": 102.78846153846153,
+      "grad_norm": 1.3837952613830566,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 32070
+    },
+    {
+      "epoch": 102.82051282051282,
+      "grad_norm": 1.1941587924957275,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 32080
+    },
+    {
+      "epoch": 102.8525641025641,
+      "grad_norm": 0.8934855461120605,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 32090
+    },
+    {
+      "epoch": 102.88461538461539,
+      "grad_norm": 0.8319123983383179,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 32100
+    },
+    {
+      "epoch": 102.91666666666667,
+      "grad_norm": 1.0874130725860596,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 32110
+    },
+    {
+      "epoch": 102.94871794871794,
+      "grad_norm": 1.1251428127288818,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 32120
+    },
+    {
+      "epoch": 102.98076923076923,
+      "grad_norm": 1.2262020111083984,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32130
+    },
+    {
+      "epoch": 103.01282051282051,
+      "grad_norm": 1.2287790775299072,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32140
+    },
+    {
+      "epoch": 103.0448717948718,
+      "grad_norm": 0.9319519996643066,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 32150
+    },
+    {
+      "epoch": 103.07692307692308,
+      "grad_norm": 1.228039026260376,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 32160
+    },
+    {
+      "epoch": 103.10897435897436,
+      "grad_norm": 1.075895071029663,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 32170
+    },
+    {
+      "epoch": 103.14102564102564,
+      "grad_norm": 1.2476003170013428,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 32180
+    },
+    {
+      "epoch": 103.17307692307692,
+      "grad_norm": 1.6356513500213623,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32190
+    },
+    {
+      "epoch": 103.2051282051282,
+      "grad_norm": 1.1813724040985107,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32200
+    },
+    {
+      "epoch": 103.23717948717949,
+      "grad_norm": 1.4343928098678589,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 32210
+    },
+    {
+      "epoch": 103.26923076923077,
+      "grad_norm": 1.438004493713379,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32220
+    },
+    {
+      "epoch": 103.30128205128206,
+      "grad_norm": 1.2498557567596436,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32230
+    },
+    {
+      "epoch": 103.33333333333333,
+      "grad_norm": 1.2029050588607788,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 32240
+    },
+    {
+      "epoch": 103.36538461538461,
+      "grad_norm": 1.1751854419708252,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32250
+    },
+    {
+      "epoch": 103.3974358974359,
+      "grad_norm": 0.8899516463279724,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32260
+    },
+    {
+      "epoch": 103.42948717948718,
+      "grad_norm": 0.9517676830291748,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 32270
+    },
+    {
+      "epoch": 103.46153846153847,
+      "grad_norm": 0.8730244040489197,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 32280
+    },
+    {
+      "epoch": 103.49358974358974,
+      "grad_norm": 1.4241180419921875,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 32290
+    },
+    {
+      "epoch": 103.52564102564102,
+      "grad_norm": 0.9220625758171082,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32300
+    },
+    {
+      "epoch": 103.5576923076923,
+      "grad_norm": 1.2176034450531006,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 32310
+    },
+    {
+      "epoch": 103.58974358974359,
+      "grad_norm": 0.7408306002616882,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 32320
+    },
+    {
+      "epoch": 103.62179487179488,
+      "grad_norm": 1.0203922986984253,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 32330
+    },
+    {
+      "epoch": 103.65384615384616,
+      "grad_norm": 1.136041283607483,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32340
+    },
+    {
+      "epoch": 103.68589743589743,
+      "grad_norm": 1.2420742511749268,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 32350
+    },
+    {
+      "epoch": 103.71794871794872,
+      "grad_norm": 0.6043910980224609,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 32360
+    },
+    {
+      "epoch": 103.75,
+      "grad_norm": 0.6990523338317871,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 32370
+    },
+    {
+      "epoch": 103.78205128205128,
+      "grad_norm": 0.9473005533218384,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32380
+    },
+    {
+      "epoch": 103.81410256410257,
+      "grad_norm": 0.9155024290084839,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32390
+    },
+    {
+      "epoch": 103.84615384615384,
+      "grad_norm": 0.8790653347969055,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 32400
+    },
+    {
+      "epoch": 103.87820512820512,
+      "grad_norm": 1.0599877834320068,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 32410
+    },
+    {
+      "epoch": 103.91025641025641,
+      "grad_norm": 1.090511441230774,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 32420
+    },
+    {
+      "epoch": 103.9423076923077,
+      "grad_norm": 0.7103379964828491,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 32430
+    },
+    {
+      "epoch": 103.97435897435898,
+      "grad_norm": 1.2665454149246216,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32440
+    },
+    {
+      "epoch": 104.00641025641026,
+      "grad_norm": 1.275608777999878,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32450
+    },
+    {
+      "epoch": 104.03846153846153,
+      "grad_norm": 1.050180196762085,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32460
+    },
+    {
+      "epoch": 104.07051282051282,
+      "grad_norm": 1.2392091751098633,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 32470
+    },
+    {
+      "epoch": 104.1025641025641,
+      "grad_norm": 1.25461745262146,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 32480
+    },
+    {
+      "epoch": 104.13461538461539,
+      "grad_norm": 1.1211342811584473,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32490
+    },
+    {
+      "epoch": 104.16666666666667,
+      "grad_norm": 1.074284553527832,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 32500
+    },
+    {
+      "epoch": 104.19871794871794,
+      "grad_norm": 1.4421110153198242,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 32510
+    },
+    {
+      "epoch": 104.23076923076923,
+      "grad_norm": 1.1271684169769287,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32520
+    },
+    {
+      "epoch": 104.26282051282051,
+      "grad_norm": 1.5261832475662231,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 32530
+    },
+    {
+      "epoch": 104.2948717948718,
+      "grad_norm": 0.8837817907333374,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 32540
+    },
+    {
+      "epoch": 104.32692307692308,
+      "grad_norm": 0.9068406820297241,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 32550
+    },
+    {
+      "epoch": 104.35897435897436,
+      "grad_norm": 0.9489185214042664,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32560
+    },
+    {
+      "epoch": 104.39102564102564,
+      "grad_norm": 0.9959732294082642,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32570
+    },
+    {
+      "epoch": 104.42307692307692,
+      "grad_norm": 0.9962406158447266,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32580
+    },
+    {
+      "epoch": 104.4551282051282,
+      "grad_norm": 1.0926258563995361,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32590
+    },
+    {
+      "epoch": 104.48717948717949,
+      "grad_norm": 1.2101036310195923,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32600
+    },
+    {
+      "epoch": 104.51923076923077,
+      "grad_norm": 1.102863073348999,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 32610
+    },
+    {
+      "epoch": 104.55128205128206,
+      "grad_norm": 1.3056912422180176,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 32620
+    },
+    {
+      "epoch": 104.58333333333333,
+      "grad_norm": 0.9311241507530212,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32630
+    },
+    {
+      "epoch": 104.61538461538461,
+      "grad_norm": 1.0740422010421753,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 32640
+    },
+    {
+      "epoch": 104.6474358974359,
+      "grad_norm": 1.1015931367874146,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 32650
+    },
+    {
+      "epoch": 104.67948717948718,
+      "grad_norm": 1.1736297607421875,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32660
+    },
+    {
+      "epoch": 104.71153846153847,
+      "grad_norm": 0.927139163017273,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 32670
+    },
+    {
+      "epoch": 104.74358974358974,
+      "grad_norm": 0.8249550461769104,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 32680
+    },
+    {
+      "epoch": 104.77564102564102,
+      "grad_norm": 1.1602442264556885,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 32690
+    },
+    {
+      "epoch": 104.8076923076923,
+      "grad_norm": 0.9334272742271423,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 32700
+    },
+    {
+      "epoch": 104.83974358974359,
+      "grad_norm": 0.7936388850212097,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 32710
+    },
+    {
+      "epoch": 104.87179487179488,
+      "grad_norm": 0.8747595548629761,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 32720
+    },
+    {
+      "epoch": 104.90384615384616,
+      "grad_norm": 1.2525811195373535,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 32730
+    },
+    {
+      "epoch": 104.93589743589743,
+      "grad_norm": 0.9851282835006714,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 32740
+    },
+    {
+      "epoch": 104.96794871794872,
+      "grad_norm": 0.9389690160751343,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32750
+    },
+    {
+      "epoch": 105.0,
+      "grad_norm": 0.8586122989654541,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32760
+    },
+    {
+      "epoch": 105.03205128205128,
+      "grad_norm": 0.9320475459098816,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32770
+    },
+    {
+      "epoch": 105.06410256410257,
+      "grad_norm": 1.0669373273849487,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32780
+    },
+    {
+      "epoch": 105.09615384615384,
+      "grad_norm": 0.9107142686843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 32790
+    },
+    {
+      "epoch": 105.12820512820512,
+      "grad_norm": 1.446686029434204,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 32800
+    },
+    {
+      "epoch": 105.16025641025641,
+      "grad_norm": 0.8168612122535706,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32810
+    },
+    {
+      "epoch": 105.1923076923077,
+      "grad_norm": 0.929972767829895,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32820
+    },
+    {
+      "epoch": 105.22435897435898,
+      "grad_norm": 0.7650881409645081,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 32830
+    },
+    {
+      "epoch": 105.25641025641026,
+      "grad_norm": 0.6487380862236023,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 32840
+    },
+    {
+      "epoch": 105.28846153846153,
+      "grad_norm": 0.7337898015975952,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32850
+    },
+    {
+      "epoch": 105.32051282051282,
+      "grad_norm": 1.0633516311645508,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32860
+    },
+    {
+      "epoch": 105.3525641025641,
+      "grad_norm": 0.8016203045845032,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 32870
+    },
+    {
+      "epoch": 105.38461538461539,
+      "grad_norm": 0.9693610072135925,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 32880
+    },
+    {
+      "epoch": 105.41666666666667,
+      "grad_norm": 0.7678739428520203,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32890
+    },
+    {
+      "epoch": 105.44871794871794,
+      "grad_norm": 1.0410531759262085,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 32900
+    },
+    {
+      "epoch": 105.48076923076923,
+      "grad_norm": 0.9421747922897339,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 32910
+    },
+    {
+      "epoch": 105.51282051282051,
+      "grad_norm": 0.8698601722717285,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 32920
+    },
+    {
+      "epoch": 105.5448717948718,
+      "grad_norm": 0.8644634485244751,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 32930
+    },
+    {
+      "epoch": 105.57692307692308,
+      "grad_norm": 1.0383771657943726,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32940
+    },
+    {
+      "epoch": 105.60897435897436,
+      "grad_norm": 1.1860803365707397,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 32950
+    },
+    {
+      "epoch": 105.64102564102564,
+      "grad_norm": 1.1308879852294922,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 32960
+    },
+    {
+      "epoch": 105.67307692307692,
+      "grad_norm": 0.9055660367012024,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 32970
+    },
+    {
+      "epoch": 105.7051282051282,
+      "grad_norm": 0.9347453117370605,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 32980
+    },
+    {
+      "epoch": 105.73717948717949,
+      "grad_norm": 1.3765008449554443,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 32990
+    },
+    {
+      "epoch": 105.76923076923077,
+      "grad_norm": 1.4630372524261475,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 33000
+    },
+    {
+      "epoch": 105.80128205128206,
+      "grad_norm": 1.3660123348236084,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 33010
+    },
+    {
+      "epoch": 105.83333333333333,
+      "grad_norm": 1.225045919418335,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 33020
+    },
+    {
+      "epoch": 105.86538461538461,
+      "grad_norm": 1.0398187637329102,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 33030
+    },
+    {
+      "epoch": 105.8974358974359,
+      "grad_norm": 0.9922576546669006,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33040
+    },
+    {
+      "epoch": 105.92948717948718,
+      "grad_norm": 0.7886695861816406,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 33050
+    },
+    {
+      "epoch": 105.96153846153847,
+      "grad_norm": 0.8077449202537537,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 33060
+    },
+    {
+      "epoch": 105.99358974358974,
+      "grad_norm": 0.7622653245925903,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 33070
+    },
+    {
+      "epoch": 106.02564102564102,
+      "grad_norm": 0.6558428406715393,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33080
+    },
+    {
+      "epoch": 106.0576923076923,
+      "grad_norm": 0.7902133464813232,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 33090
+    },
+    {
+      "epoch": 106.08974358974359,
+      "grad_norm": 0.8891046643257141,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 33100
+    },
+    {
+      "epoch": 106.12179487179488,
+      "grad_norm": 0.8277244567871094,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 33110
+    },
+    {
+      "epoch": 106.15384615384616,
+      "grad_norm": 0.9163572192192078,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 33120
+    },
+    {
+      "epoch": 106.18589743589743,
+      "grad_norm": 0.8726097345352173,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 33130
+    },
+    {
+      "epoch": 106.21794871794872,
+      "grad_norm": 0.6690422892570496,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 33140
+    },
+    {
+      "epoch": 106.25,
+      "grad_norm": 0.823326587677002,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 33150
+    },
+    {
+      "epoch": 106.28205128205128,
+      "grad_norm": 0.9198511838912964,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 33160
+    },
+    {
+      "epoch": 106.31410256410257,
+      "grad_norm": 0.9117867350578308,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 33170
+    },
+    {
+      "epoch": 106.34615384615384,
+      "grad_norm": 0.6912775635719299,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33180
+    },
+    {
+      "epoch": 106.37820512820512,
+      "grad_norm": 0.7725376486778259,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 33190
+    },
+    {
+      "epoch": 106.41025641025641,
+      "grad_norm": 1.0630896091461182,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 33200
+    },
+    {
+      "epoch": 106.4423076923077,
+      "grad_norm": 1.1875184774398804,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 33210
+    },
+    {
+      "epoch": 106.47435897435898,
+      "grad_norm": 0.9866331219673157,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 33220
+    },
+    {
+      "epoch": 106.50641025641026,
+      "grad_norm": 0.9602553844451904,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 33230
+    },
+    {
+      "epoch": 106.53846153846153,
+      "grad_norm": 1.24833083152771,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 33240
+    },
+    {
+      "epoch": 106.57051282051282,
+      "grad_norm": 1.1557798385620117,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 33250
+    },
+    {
+      "epoch": 106.6025641025641,
+      "grad_norm": 0.9565214514732361,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 33260
+    },
+    {
+      "epoch": 106.63461538461539,
+      "grad_norm": 1.1369109153747559,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 33270
+    },
+    {
+      "epoch": 106.66666666666667,
+      "grad_norm": 1.0724023580551147,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 33280
+    },
+    {
+      "epoch": 106.69871794871794,
+      "grad_norm": 1.2608166933059692,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 33290
+    },
+    {
+      "epoch": 106.73076923076923,
+      "grad_norm": 1.2945586442947388,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 33300
+    },
+    {
+      "epoch": 106.76282051282051,
+      "grad_norm": 1.3012363910675049,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 33310
+    },
+    {
+      "epoch": 106.7948717948718,
+      "grad_norm": 1.2148568630218506,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 33320
+    },
+    {
+      "epoch": 106.82692307692308,
+      "grad_norm": 1.3564491271972656,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 33330
+    },
+    {
+      "epoch": 106.85897435897436,
+      "grad_norm": 1.0824799537658691,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 33340
+    },
+    {
+      "epoch": 106.89102564102564,
+      "grad_norm": 1.2466249465942383,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 33350
+    },
+    {
+      "epoch": 106.92307692307692,
+      "grad_norm": 1.3274850845336914,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 33360
+    },
+    {
+      "epoch": 106.9551282051282,
+      "grad_norm": 0.9621427059173584,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 33370
+    },
+    {
+      "epoch": 106.98717948717949,
+      "grad_norm": 1.3851635456085205,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 33380
+    },
+    {
+      "epoch": 107.01923076923077,
+      "grad_norm": 1.4412254095077515,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 33390
+    },
+    {
+      "epoch": 107.05128205128206,
+      "grad_norm": 1.5531322956085205,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 33400
+    },
+    {
+      "epoch": 107.08333333333333,
+      "grad_norm": 1.1500027179718018,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 33410
+    },
+    {
+      "epoch": 107.11538461538461,
+      "grad_norm": 1.0065501928329468,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 33420
+    },
+    {
+      "epoch": 107.1474358974359,
+      "grad_norm": 1.0397504568099976,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 33430
+    },
+    {
+      "epoch": 107.17948717948718,
+      "grad_norm": 1.186741828918457,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 33440
+    },
+    {
+      "epoch": 107.21153846153847,
+      "grad_norm": 0.7999820113182068,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 33450
+    },
+    {
+      "epoch": 107.24358974358974,
+      "grad_norm": 0.9187130331993103,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33460
+    },
+    {
+      "epoch": 107.27564102564102,
+      "grad_norm": 1.1057325601577759,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 33470
+    },
+    {
+      "epoch": 107.3076923076923,
+      "grad_norm": 1.1684679985046387,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 33480
+    },
+    {
+      "epoch": 107.33974358974359,
+      "grad_norm": 1.2443907260894775,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 33490
+    },
+    {
+      "epoch": 107.37179487179488,
+      "grad_norm": 1.5025193691253662,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 33500
+    },
+    {
+      "epoch": 107.40384615384616,
+      "grad_norm": 1.581925868988037,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 33510
+    },
+    {
+      "epoch": 107.43589743589743,
+      "grad_norm": 1.6550366878509521,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 33520
+    },
+    {
+      "epoch": 107.46794871794872,
+      "grad_norm": 1.3542249202728271,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 33530
+    },
+    {
+      "epoch": 107.5,
+      "grad_norm": 1.2455040216445923,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 33540
+    },
+    {
+      "epoch": 107.53205128205128,
+      "grad_norm": 1.2502199411392212,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 33550
+    },
+    {
+      "epoch": 107.56410256410257,
+      "grad_norm": 1.201119303703308,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 33560
+    },
+    {
+      "epoch": 107.59615384615384,
+      "grad_norm": 1.1203515529632568,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 33570
+    },
+    {
+      "epoch": 107.62820512820512,
+      "grad_norm": 1.2272642850875854,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33580
+    },
+    {
+      "epoch": 107.66025641025641,
+      "grad_norm": 1.1720044612884521,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 33590
+    },
+    {
+      "epoch": 107.6923076923077,
+      "grad_norm": 0.945874035358429,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 33600
+    },
+    {
+      "epoch": 107.72435897435898,
+      "grad_norm": 0.8840534687042236,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 33610
+    },
+    {
+      "epoch": 107.75641025641026,
+      "grad_norm": 1.220062494277954,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 33620
+    },
+    {
+      "epoch": 107.78846153846153,
+      "grad_norm": 1.5745981931686401,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 33630
+    },
+    {
+      "epoch": 107.82051282051282,
+      "grad_norm": 1.5506103038787842,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 33640
+    },
+    {
+      "epoch": 107.8525641025641,
+      "grad_norm": 1.196818232536316,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 33650
+    },
+    {
+      "epoch": 107.88461538461539,
+      "grad_norm": 1.2970088720321655,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 33660
+    },
+    {
+      "epoch": 107.91666666666667,
+      "grad_norm": 1.6514456272125244,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33670
+    },
+    {
+      "epoch": 107.94871794871794,
+      "grad_norm": 1.0735890865325928,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 33680
+    },
+    {
+      "epoch": 107.98076923076923,
+      "grad_norm": 1.2541842460632324,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 33690
+    },
+    {
+      "epoch": 108.01282051282051,
+      "grad_norm": 0.9915783405303955,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 33700
+    },
+    {
+      "epoch": 108.0448717948718,
+      "grad_norm": 1.146491289138794,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33710
+    },
+    {
+      "epoch": 108.07692307692308,
+      "grad_norm": 0.9971730709075928,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 33720
+    },
+    {
+      "epoch": 108.10897435897436,
+      "grad_norm": 1.1796011924743652,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 33730
+    },
+    {
+      "epoch": 108.14102564102564,
+      "grad_norm": 1.2482353448867798,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33740
+    },
+    {
+      "epoch": 108.17307692307692,
+      "grad_norm": 0.9247346520423889,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 33750
+    },
+    {
+      "epoch": 108.2051282051282,
+      "grad_norm": 0.784359335899353,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 33760
+    },
+    {
+      "epoch": 108.23717948717949,
+      "grad_norm": 0.8802067637443542,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 33770
+    },
+    {
+      "epoch": 108.26923076923077,
+      "grad_norm": 1.087322473526001,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 33780
+    },
+    {
+      "epoch": 108.30128205128206,
+      "grad_norm": 0.833400309085846,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 33790
+    },
+    {
+      "epoch": 108.33333333333333,
+      "grad_norm": 1.1809109449386597,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 33800
+    },
+    {
+      "epoch": 108.36538461538461,
+      "grad_norm": 1.2994091510772705,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 33810
+    },
+    {
+      "epoch": 108.3974358974359,
+      "grad_norm": 1.4942547082901,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 33820
+    },
+    {
+      "epoch": 108.42948717948718,
+      "grad_norm": 1.2374008893966675,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33830
+    },
+    {
+      "epoch": 108.46153846153847,
+      "grad_norm": 1.6098361015319824,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 33840
+    },
+    {
+      "epoch": 108.49358974358974,
+      "grad_norm": 1.1142244338989258,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 33850
+    },
+    {
+      "epoch": 108.52564102564102,
+      "grad_norm": 1.5012084245681763,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 33860
+    },
+    {
+      "epoch": 108.5576923076923,
+      "grad_norm": 1.0210154056549072,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 33870
+    },
+    {
+      "epoch": 108.58974358974359,
+      "grad_norm": 1.4841159582138062,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 33880
+    },
+    {
+      "epoch": 108.62179487179488,
+      "grad_norm": 0.7834526300430298,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 33890
+    },
+    {
+      "epoch": 108.65384615384616,
+      "grad_norm": 0.7023696303367615,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 33900
+    },
+    {
+      "epoch": 108.68589743589743,
+      "grad_norm": 0.8844192624092102,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 33910
+    },
+    {
+      "epoch": 108.71794871794872,
+      "grad_norm": 0.6393696665763855,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 33920
+    },
+    {
+      "epoch": 108.75,
+      "grad_norm": 0.7790671586990356,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 33930
+    },
+    {
+      "epoch": 108.78205128205128,
+      "grad_norm": 0.8863510489463806,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 33940
+    },
+    {
+      "epoch": 108.81410256410257,
+      "grad_norm": 0.8157210946083069,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 33950
+    },
+    {
+      "epoch": 108.84615384615384,
+      "grad_norm": 0.7282866835594177,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33960
+    },
+    {
+      "epoch": 108.87820512820512,
+      "grad_norm": 0.9400081634521484,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 33970
+    },
+    {
+      "epoch": 108.91025641025641,
+      "grad_norm": 0.5986185073852539,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 33980
+    },
+    {
+      "epoch": 108.9423076923077,
+      "grad_norm": 0.650469958782196,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 33990
+    },
+    {
+      "epoch": 108.97435897435898,
+      "grad_norm": 0.7117701172828674,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 34000
+    },
+    {
+      "epoch": 109.00641025641026,
+      "grad_norm": 0.6674757599830627,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 34010
+    },
+    {
+      "epoch": 109.03846153846153,
+      "grad_norm": 1.0260192155838013,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 34020
+    },
+    {
+      "epoch": 109.07051282051282,
+      "grad_norm": 0.6784473657608032,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 34030
+    },
+    {
+      "epoch": 109.1025641025641,
+      "grad_norm": 0.9133219718933105,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 34040
+    },
+    {
+      "epoch": 109.13461538461539,
+      "grad_norm": 0.7874703407287598,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 34050
+    },
+    {
+      "epoch": 109.16666666666667,
+      "grad_norm": 0.8986002206802368,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 34060
+    },
+    {
+      "epoch": 109.19871794871794,
+      "grad_norm": 0.5833228826522827,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 34070
+    },
+    {
+      "epoch": 109.23076923076923,
+      "grad_norm": 0.6761743426322937,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 34080
+    },
+    {
+      "epoch": 109.26282051282051,
+      "grad_norm": 0.7825962901115417,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34090
+    },
+    {
+      "epoch": 109.2948717948718,
+      "grad_norm": 1.5648020505905151,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 34100
+    },
+    {
+      "epoch": 109.32692307692308,
+      "grad_norm": 1.2049318552017212,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 34110
+    },
+    {
+      "epoch": 109.35897435897436,
+      "grad_norm": 1.1632270812988281,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 34120
+    },
+    {
+      "epoch": 109.39102564102564,
+      "grad_norm": 1.1322294473648071,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 34130
+    },
+    {
+      "epoch": 109.42307692307692,
+      "grad_norm": 1.2899034023284912,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 34140
+    },
+    {
+      "epoch": 109.4551282051282,
+      "grad_norm": 1.2655348777770996,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 34150
+    },
+    {
+      "epoch": 109.48717948717949,
+      "grad_norm": 1.0696076154708862,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 34160
+    },
+    {
+      "epoch": 109.51923076923077,
+      "grad_norm": 0.7835086584091187,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 34170
+    },
+    {
+      "epoch": 109.55128205128206,
+      "grad_norm": 1.0576872825622559,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 34180
+    },
+    {
+      "epoch": 109.58333333333333,
+      "grad_norm": 1.1246038675308228,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 34190
+    },
+    {
+      "epoch": 109.61538461538461,
+      "grad_norm": 1.328602910041809,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 34200
+    },
+    {
+      "epoch": 109.6474358974359,
+      "grad_norm": 1.1119916439056396,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 34210
+    },
+    {
+      "epoch": 109.67948717948718,
+      "grad_norm": 1.102603554725647,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 34220
+    },
+    {
+      "epoch": 109.71153846153847,
+      "grad_norm": 1.0553879737854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 34230
+    },
+    {
+      "epoch": 109.74358974358974,
+      "grad_norm": 1.1431869268417358,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34240
+    },
+    {
+      "epoch": 109.77564102564102,
+      "grad_norm": 1.057543396949768,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 34250
+    },
+    {
+      "epoch": 109.8076923076923,
+      "grad_norm": 1.0541483163833618,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 34260
+    },
+    {
+      "epoch": 109.83974358974359,
+      "grad_norm": 1.0451209545135498,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 34270
+    },
+    {
+      "epoch": 109.87179487179488,
+      "grad_norm": 0.8176800608634949,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 34280
+    },
+    {
+      "epoch": 109.90384615384616,
+      "grad_norm": 0.8832671046257019,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 34290
+    },
+    {
+      "epoch": 109.93589743589743,
+      "grad_norm": 0.9940178990364075,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34300
+    },
+    {
+      "epoch": 109.96794871794872,
+      "grad_norm": 1.111122727394104,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 34310
+    },
+    {
+      "epoch": 110.0,
+      "grad_norm": 1.322989583015442,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 34320
+    },
+    {
+      "epoch": 110.03205128205128,
+      "grad_norm": 1.0758558511734009,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34330
+    },
+    {
+      "epoch": 110.06410256410257,
+      "grad_norm": 1.1175016164779663,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34340
+    },
+    {
+      "epoch": 110.09615384615384,
+      "grad_norm": 1.2070716619491577,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 34350
+    },
+    {
+      "epoch": 110.12820512820512,
+      "grad_norm": 1.0097620487213135,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 34360
+    },
+    {
+      "epoch": 110.16025641025641,
+      "grad_norm": 1.4673477411270142,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 34370
+    },
+    {
+      "epoch": 110.1923076923077,
+      "grad_norm": 1.5588078498840332,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34380
+    },
+    {
+      "epoch": 110.22435897435898,
+      "grad_norm": 1.389692783355713,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 34390
+    },
+    {
+      "epoch": 110.25641025641026,
+      "grad_norm": 1.639846682548523,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 34400
+    },
+    {
+      "epoch": 110.28846153846153,
+      "grad_norm": 1.325696587562561,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 34410
+    },
+    {
+      "epoch": 110.32051282051282,
+      "grad_norm": 1.6608549356460571,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 34420
+    },
+    {
+      "epoch": 110.3525641025641,
+      "grad_norm": 1.4510254859924316,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 34430
+    },
+    {
+      "epoch": 110.38461538461539,
+      "grad_norm": 1.408737063407898,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 34440
+    },
+    {
+      "epoch": 110.41666666666667,
+      "grad_norm": 1.2830785512924194,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 34450
+    },
+    {
+      "epoch": 110.44871794871794,
+      "grad_norm": 1.2500698566436768,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 34460
+    },
+    {
+      "epoch": 110.48076923076923,
+      "grad_norm": 1.1225138902664185,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 34470
+    },
+    {
+      "epoch": 110.51282051282051,
+      "grad_norm": 0.6997758746147156,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 34480
+    },
+    {
+      "epoch": 110.5448717948718,
+      "grad_norm": 0.8601786494255066,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 34490
+    },
+    {
+      "epoch": 110.57692307692308,
+      "grad_norm": 0.7737405300140381,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 34500
+    },
+    {
+      "epoch": 110.60897435897436,
+      "grad_norm": 0.6076096296310425,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 34510
+    },
+    {
+      "epoch": 110.64102564102564,
+      "grad_norm": 0.9305208921432495,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 34520
+    },
+    {
+      "epoch": 110.67307692307692,
+      "grad_norm": 0.7088750600814819,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 34530
+    },
+    {
+      "epoch": 110.7051282051282,
+      "grad_norm": 0.5952097773551941,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 34540
+    },
+    {
+      "epoch": 110.73717948717949,
+      "grad_norm": 1.1074635982513428,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 34550
+    },
+    {
+      "epoch": 110.76923076923077,
+      "grad_norm": 0.7817279696464539,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34560
+    },
+    {
+      "epoch": 110.80128205128206,
+      "grad_norm": 0.7279987931251526,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34570
+    },
+    {
+      "epoch": 110.83333333333333,
+      "grad_norm": 0.8127354383468628,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 34580
+    },
+    {
+      "epoch": 110.86538461538461,
+      "grad_norm": 0.9565805196762085,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 34590
+    },
+    {
+      "epoch": 110.8974358974359,
+      "grad_norm": 0.9565560817718506,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 34600
+    },
+    {
+      "epoch": 110.92948717948718,
+      "grad_norm": 1.0345351696014404,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 34610
+    },
+    {
+      "epoch": 110.96153846153847,
+      "grad_norm": 0.8294216394424438,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 34620
+    },
+    {
+      "epoch": 110.99358974358974,
+      "grad_norm": 0.9895102977752686,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 34630
+    },
+    {
+      "epoch": 111.02564102564102,
+      "grad_norm": 0.9142376780509949,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 34640
+    },
+    {
+      "epoch": 111.0576923076923,
+      "grad_norm": 0.9160284399986267,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 34650
+    },
+    {
+      "epoch": 111.08974358974359,
+      "grad_norm": 0.8997448682785034,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 34660
+    },
+    {
+      "epoch": 111.12179487179488,
+      "grad_norm": 1.1252777576446533,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 34670
+    },
+    {
+      "epoch": 111.15384615384616,
+      "grad_norm": 1.0730657577514648,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 34680
+    },
+    {
+      "epoch": 111.18589743589743,
+      "grad_norm": 1.0317295789718628,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 34690
+    },
+    {
+      "epoch": 111.21794871794872,
+      "grad_norm": 0.8978215456008911,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 34700
+    },
+    {
+      "epoch": 111.25,
+      "grad_norm": 1.2034265995025635,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 34710
+    },
+    {
+      "epoch": 111.28205128205128,
+      "grad_norm": 1.0515291690826416,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 34720
+    },
+    {
+      "epoch": 111.31410256410257,
+      "grad_norm": 0.9301179647445679,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 34730
+    },
+    {
+      "epoch": 111.34615384615384,
+      "grad_norm": 0.7197339534759521,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 34740
+    },
+    {
+      "epoch": 111.37820512820512,
+      "grad_norm": 0.7008622884750366,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34750
+    },
+    {
+      "epoch": 111.41025641025641,
+      "grad_norm": 0.6930691003799438,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34760
+    },
+    {
+      "epoch": 111.4423076923077,
+      "grad_norm": 0.897777259349823,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 34770
+    },
+    {
+      "epoch": 111.47435897435898,
+      "grad_norm": 0.8869208693504333,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 34780
+    },
+    {
+      "epoch": 111.50641025641026,
+      "grad_norm": 0.7022380232810974,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 34790
+    },
+    {
+      "epoch": 111.53846153846153,
+      "grad_norm": 0.9040806293487549,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 34800
+    },
+    {
+      "epoch": 111.57051282051282,
+      "grad_norm": 0.891967236995697,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 34810
+    },
+    {
+      "epoch": 111.6025641025641,
+      "grad_norm": 0.902060329914093,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 34820
+    },
+    {
+      "epoch": 111.63461538461539,
+      "grad_norm": 0.6735020279884338,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 34830
+    },
+    {
+      "epoch": 111.66666666666667,
+      "grad_norm": 0.7780515551567078,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 34840
+    },
+    {
+      "epoch": 111.69871794871794,
+      "grad_norm": 1.7445306777954102,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 34850
+    },
+    {
+      "epoch": 111.73076923076923,
+      "grad_norm": 0.9174376130104065,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 34860
+    },
+    {
+      "epoch": 111.76282051282051,
+      "grad_norm": 1.075122594833374,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 34870
+    },
+    {
+      "epoch": 111.7948717948718,
+      "grad_norm": 1.333669900894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 34880
+    },
+    {
+      "epoch": 111.82692307692308,
+      "grad_norm": 1.0999832153320312,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 34890
+    },
+    {
+      "epoch": 111.85897435897436,
+      "grad_norm": 1.216084599494934,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 34900
+    },
+    {
+      "epoch": 111.89102564102564,
+      "grad_norm": 1.474242091178894,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 34910
+    },
+    {
+      "epoch": 111.92307692307692,
+      "grad_norm": 1.3323888778686523,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 34920
+    },
+    {
+      "epoch": 111.9551282051282,
+      "grad_norm": 1.0383822917938232,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 34930
+    },
+    {
+      "epoch": 111.98717948717949,
+      "grad_norm": 1.2216930389404297,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 34940
+    },
+    {
+      "epoch": 112.01923076923077,
+      "grad_norm": 1.2176706790924072,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 34950
+    },
+    {
+      "epoch": 112.05128205128206,
+      "grad_norm": 1.3152118921279907,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 34960
+    },
+    {
+      "epoch": 112.08333333333333,
+      "grad_norm": 0.8718500137329102,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 34970
+    },
+    {
+      "epoch": 112.11538461538461,
+      "grad_norm": 0.9505050182342529,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 34980
+    },
+    {
+      "epoch": 112.1474358974359,
+      "grad_norm": 0.9456276893615723,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 34990
+    },
+    {
+      "epoch": 112.17948717948718,
+      "grad_norm": 0.7304359078407288,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 35000
+    },
+    {
+      "epoch": 112.21153846153847,
+      "grad_norm": 0.8227810859680176,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35010
+    },
+    {
+      "epoch": 112.24358974358974,
+      "grad_norm": 0.8111754059791565,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 35020
+    },
+    {
+      "epoch": 112.27564102564102,
+      "grad_norm": 0.9024219512939453,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 35030
+    },
+    {
+      "epoch": 112.3076923076923,
+      "grad_norm": 0.5909203290939331,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35040
+    },
+    {
+      "epoch": 112.33974358974359,
+      "grad_norm": 0.9463586211204529,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 35050
+    },
+    {
+      "epoch": 112.37179487179488,
+      "grad_norm": 0.9586068987846375,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35060
+    },
+    {
+      "epoch": 112.40384615384616,
+      "grad_norm": 1.1366328001022339,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35070
+    },
+    {
+      "epoch": 112.43589743589743,
+      "grad_norm": 1.0722743272781372,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 35080
+    },
+    {
+      "epoch": 112.46794871794872,
+      "grad_norm": 1.311224102973938,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 35090
+    },
+    {
+      "epoch": 112.5,
+      "grad_norm": 1.0864028930664062,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35100
+    },
+    {
+      "epoch": 112.53205128205128,
+      "grad_norm": 1.1019556522369385,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 35110
+    },
+    {
+      "epoch": 112.56410256410257,
+      "grad_norm": 1.067219853401184,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35120
+    },
+    {
+      "epoch": 112.59615384615384,
+      "grad_norm": 1.1555657386779785,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 35130
+    },
+    {
+      "epoch": 112.62820512820512,
+      "grad_norm": 0.8361935615539551,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35140
+    },
+    {
+      "epoch": 112.66025641025641,
+      "grad_norm": 0.9822496175765991,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35150
+    },
+    {
+      "epoch": 112.6923076923077,
+      "grad_norm": 1.0597524642944336,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35160
+    },
+    {
+      "epoch": 112.72435897435898,
+      "grad_norm": 0.9246448278427124,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 35170
+    },
+    {
+      "epoch": 112.75641025641026,
+      "grad_norm": 1.0878424644470215,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 35180
+    },
+    {
+      "epoch": 112.78846153846153,
+      "grad_norm": 0.9272701740264893,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35190
+    },
+    {
+      "epoch": 112.82051282051282,
+      "grad_norm": 1.0747030973434448,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35200
+    },
+    {
+      "epoch": 112.8525641025641,
+      "grad_norm": 1.0395036935806274,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35210
+    },
+    {
+      "epoch": 112.88461538461539,
+      "grad_norm": 0.874455451965332,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 35220
+    },
+    {
+      "epoch": 112.91666666666667,
+      "grad_norm": 0.9112585186958313,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35230
+    },
+    {
+      "epoch": 112.94871794871794,
+      "grad_norm": 0.6932641267776489,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 35240
+    },
+    {
+      "epoch": 112.98076923076923,
+      "grad_norm": 0.8880487084388733,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35250
+    },
+    {
+      "epoch": 113.01282051282051,
+      "grad_norm": 1.1079466342926025,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 35260
+    },
+    {
+      "epoch": 113.0448717948718,
+      "grad_norm": 1.0269263982772827,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 35270
+    },
+    {
+      "epoch": 113.07692307692308,
+      "grad_norm": 0.8676454424858093,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35280
+    },
+    {
+      "epoch": 113.10897435897436,
+      "grad_norm": 1.1284599304199219,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35290
+    },
+    {
+      "epoch": 113.14102564102564,
+      "grad_norm": 0.8673512935638428,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35300
+    },
+    {
+      "epoch": 113.17307692307692,
+      "grad_norm": 0.5973019599914551,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 35310
+    },
+    {
+      "epoch": 113.2051282051282,
+      "grad_norm": 0.6606478691101074,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35320
+    },
+    {
+      "epoch": 113.23717948717949,
+      "grad_norm": 0.6272337436676025,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 35330
+    },
+    {
+      "epoch": 113.26923076923077,
+      "grad_norm": 0.7502418756484985,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 35340
+    },
+    {
+      "epoch": 113.30128205128206,
+      "grad_norm": 0.8329353332519531,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 35350
+    },
+    {
+      "epoch": 113.33333333333333,
+      "grad_norm": 0.980131983757019,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35360
+    },
+    {
+      "epoch": 113.36538461538461,
+      "grad_norm": 1.258017897605896,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35370
+    },
+    {
+      "epoch": 113.3974358974359,
+      "grad_norm": 0.9383469223976135,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 35380
+    },
+    {
+      "epoch": 113.42948717948718,
+      "grad_norm": 0.8187419772148132,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 35390
+    },
+    {
+      "epoch": 113.46153846153847,
+      "grad_norm": 1.148239254951477,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35400
+    },
+    {
+      "epoch": 113.49358974358974,
+      "grad_norm": 0.8913676142692566,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35410
+    },
+    {
+      "epoch": 113.52564102564102,
+      "grad_norm": 0.8188424110412598,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35420
+    },
+    {
+      "epoch": 113.5576923076923,
+      "grad_norm": 1.0009900331497192,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 35430
+    },
+    {
+      "epoch": 113.58974358974359,
+      "grad_norm": 1.1893550157546997,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 35440
+    },
+    {
+      "epoch": 113.62179487179488,
+      "grad_norm": 1.179569125175476,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 35450
+    },
+    {
+      "epoch": 113.65384615384616,
+      "grad_norm": 0.7887077331542969,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 35460
+    },
+    {
+      "epoch": 113.68589743589743,
+      "grad_norm": 0.983808696269989,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 35470
+    },
+    {
+      "epoch": 113.71794871794872,
+      "grad_norm": 1.3310878276824951,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35480
+    },
+    {
+      "epoch": 113.75,
+      "grad_norm": 0.818787693977356,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 35490
+    },
+    {
+      "epoch": 113.78205128205128,
+      "grad_norm": 0.9680303931236267,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 35500
+    },
+    {
+      "epoch": 113.81410256410257,
+      "grad_norm": 1.1821353435516357,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 35510
+    },
+    {
+      "epoch": 113.84615384615384,
+      "grad_norm": 1.0101124048233032,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 35520
+    },
+    {
+      "epoch": 113.87820512820512,
+      "grad_norm": 0.8890703916549683,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35530
+    },
+    {
+      "epoch": 113.91025641025641,
+      "grad_norm": 0.7848128080368042,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35540
+    },
+    {
+      "epoch": 113.9423076923077,
+      "grad_norm": 0.6089345216751099,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35550
+    },
+    {
+      "epoch": 113.97435897435898,
+      "grad_norm": 0.9361271858215332,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35560
+    },
+    {
+      "epoch": 114.00641025641026,
+      "grad_norm": 1.4255293607711792,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 35570
+    },
+    {
+      "epoch": 114.03846153846153,
+      "grad_norm": 0.9973898530006409,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35580
+    },
+    {
+      "epoch": 114.07051282051282,
+      "grad_norm": 1.0127886533737183,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35590
+    },
+    {
+      "epoch": 114.1025641025641,
+      "grad_norm": 1.0149319171905518,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 35600
+    },
+    {
+      "epoch": 114.13461538461539,
+      "grad_norm": 0.6760701537132263,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 35610
+    },
+    {
+      "epoch": 114.16666666666667,
+      "grad_norm": 1.1918506622314453,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 35620
+    },
+    {
+      "epoch": 114.19871794871794,
+      "grad_norm": 1.774768352508545,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35630
+    },
+    {
+      "epoch": 114.23076923076923,
+      "grad_norm": 1.3247172832489014,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35640
+    },
+    {
+      "epoch": 114.26282051282051,
+      "grad_norm": 1.3465096950531006,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 35650
+    },
+    {
+      "epoch": 114.2948717948718,
+      "grad_norm": 1.093914270401001,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35660
+    },
+    {
+      "epoch": 114.32692307692308,
+      "grad_norm": 1.2012385129928589,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 35670
+    },
+    {
+      "epoch": 114.35897435897436,
+      "grad_norm": 0.8929421901702881,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 35680
+    },
+    {
+      "epoch": 114.39102564102564,
+      "grad_norm": 1.0518995523452759,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35690
+    },
+    {
+      "epoch": 114.42307692307692,
+      "grad_norm": 1.2981057167053223,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 35700
+    },
+    {
+      "epoch": 114.4551282051282,
+      "grad_norm": 1.1891868114471436,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35710
+    },
+    {
+      "epoch": 114.48717948717949,
+      "grad_norm": 1.0315462350845337,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35720
+    },
+    {
+      "epoch": 114.51923076923077,
+      "grad_norm": 0.8428895473480225,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 35730
+    },
+    {
+      "epoch": 114.55128205128206,
+      "grad_norm": 0.8957397937774658,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35740
+    },
+    {
+      "epoch": 114.58333333333333,
+      "grad_norm": 0.9292715191841125,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35750
+    },
+    {
+      "epoch": 114.61538461538461,
+      "grad_norm": 0.8292759656906128,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 35760
+    },
+    {
+      "epoch": 114.6474358974359,
+      "grad_norm": 0.9409943222999573,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 35770
+    },
+    {
+      "epoch": 114.67948717948718,
+      "grad_norm": 1.3534847497940063,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 35780
+    },
+    {
+      "epoch": 114.71153846153847,
+      "grad_norm": 0.8104700446128845,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 35790
+    },
+    {
+      "epoch": 114.74358974358974,
+      "grad_norm": 1.0025343894958496,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35800
+    },
+    {
+      "epoch": 114.77564102564102,
+      "grad_norm": 0.7922455072402954,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35810
+    },
+    {
+      "epoch": 114.8076923076923,
+      "grad_norm": 1.0667495727539062,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 35820
+    },
+    {
+      "epoch": 114.83974358974359,
+      "grad_norm": 1.9966543912887573,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 35830
+    },
+    {
+      "epoch": 114.87179487179488,
+      "grad_norm": 1.6114463806152344,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 35840
+    },
+    {
+      "epoch": 114.90384615384616,
+      "grad_norm": 1.4336788654327393,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35850
+    },
+    {
+      "epoch": 114.93589743589743,
+      "grad_norm": 0.9872815012931824,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35860
+    },
+    {
+      "epoch": 114.96794871794872,
+      "grad_norm": 1.4436378479003906,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 35870
+    },
+    {
+      "epoch": 115.0,
+      "grad_norm": 1.1280333995819092,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 35880
+    },
+    {
+      "epoch": 115.03205128205128,
+      "grad_norm": 1.2551544904708862,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 35890
+    },
+    {
+      "epoch": 115.06410256410257,
+      "grad_norm": 1.0902167558670044,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 35900
+    },
+    {
+      "epoch": 115.09615384615384,
+      "grad_norm": 1.025434970855713,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 35910
+    },
+    {
+      "epoch": 115.12820512820512,
+      "grad_norm": 1.0646862983703613,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35920
+    },
+    {
+      "epoch": 115.16025641025641,
+      "grad_norm": 1.2802343368530273,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 35930
+    },
+    {
+      "epoch": 115.1923076923077,
+      "grad_norm": 1.0355526208877563,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 35940
+    },
+    {
+      "epoch": 115.22435897435898,
+      "grad_norm": 0.8883004784584045,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 35950
+    },
+    {
+      "epoch": 115.25641025641026,
+      "grad_norm": 0.7793373465538025,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 35960
+    },
+    {
+      "epoch": 115.28846153846153,
+      "grad_norm": 0.9096797108650208,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 35970
+    },
+    {
+      "epoch": 115.32051282051282,
+      "grad_norm": 1.1496793031692505,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 35980
+    },
+    {
+      "epoch": 115.3525641025641,
+      "grad_norm": 0.9760134816169739,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 35990
+    },
+    {
+      "epoch": 115.38461538461539,
+      "grad_norm": 1.6267887353897095,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 36000
+    },
+    {
+      "epoch": 115.41666666666667,
+      "grad_norm": 1.241400122642517,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 36010
+    },
+    {
+      "epoch": 115.44871794871794,
+      "grad_norm": 1.3385789394378662,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36020
+    },
+    {
+      "epoch": 115.48076923076923,
+      "grad_norm": 1.2090448141098022,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 36030
+    },
+    {
+      "epoch": 115.51282051282051,
+      "grad_norm": 1.048966646194458,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 36040
+    },
+    {
+      "epoch": 115.5448717948718,
+      "grad_norm": 1.1476104259490967,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 36050
+    },
+    {
+      "epoch": 115.57692307692308,
+      "grad_norm": 1.4643290042877197,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36060
+    },
+    {
+      "epoch": 115.60897435897436,
+      "grad_norm": 1.2463600635528564,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36070
+    },
+    {
+      "epoch": 115.64102564102564,
+      "grad_norm": 1.1056742668151855,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 36080
+    },
+    {
+      "epoch": 115.67307692307692,
+      "grad_norm": 1.0344951152801514,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 36090
+    },
+    {
+      "epoch": 115.7051282051282,
+      "grad_norm": 0.7796273231506348,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 36100
+    },
+    {
+      "epoch": 115.73717948717949,
+      "grad_norm": 0.9786306023597717,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 36110
+    },
+    {
+      "epoch": 115.76923076923077,
+      "grad_norm": 0.8525427579879761,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 36120
+    },
+    {
+      "epoch": 115.80128205128206,
+      "grad_norm": 0.7534030675888062,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 36130
+    },
+    {
+      "epoch": 115.83333333333333,
+      "grad_norm": 0.8590915203094482,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 36140
+    },
+    {
+      "epoch": 115.86538461538461,
+      "grad_norm": 0.9556967616081238,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 36150
+    },
+    {
+      "epoch": 115.8974358974359,
+      "grad_norm": 0.889697253704071,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 36160
+    },
+    {
+      "epoch": 115.92948717948718,
+      "grad_norm": 0.9504598379135132,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 36170
+    },
+    {
+      "epoch": 115.96153846153847,
+      "grad_norm": 0.9794744253158569,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36180
+    },
+    {
+      "epoch": 115.99358974358974,
+      "grad_norm": 1.2558276653289795,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 36190
+    },
+    {
+      "epoch": 116.02564102564102,
+      "grad_norm": 0.7131901979446411,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 36200
+    },
+    {
+      "epoch": 116.0576923076923,
+      "grad_norm": 0.6972107291221619,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 36210
+    },
+    {
+      "epoch": 116.08974358974359,
+      "grad_norm": 0.7796922326087952,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 36220
+    },
+    {
+      "epoch": 116.12179487179488,
+      "grad_norm": 0.7518036365509033,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 36230
+    },
+    {
+      "epoch": 116.15384615384616,
+      "grad_norm": 0.9847820401191711,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 36240
+    },
+    {
+      "epoch": 116.18589743589743,
+      "grad_norm": 0.8609389066696167,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 36250
+    },
+    {
+      "epoch": 116.21794871794872,
+      "grad_norm": 0.9751351475715637,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36260
+    },
+    {
+      "epoch": 116.25,
+      "grad_norm": 1.3638896942138672,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 36270
+    },
+    {
+      "epoch": 116.28205128205128,
+      "grad_norm": 1.99062979221344,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36280
+    },
+    {
+      "epoch": 116.31410256410257,
+      "grad_norm": 1.5893847942352295,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 36290
+    },
+    {
+      "epoch": 116.34615384615384,
+      "grad_norm": 1.3926559686660767,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 36300
+    },
+    {
+      "epoch": 116.37820512820512,
+      "grad_norm": 1.1700364351272583,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 36310
+    },
+    {
+      "epoch": 116.41025641025641,
+      "grad_norm": 1.3581068515777588,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 36320
+    },
+    {
+      "epoch": 116.4423076923077,
+      "grad_norm": 1.3297077417373657,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 36330
+    },
+    {
+      "epoch": 116.47435897435898,
+      "grad_norm": 1.0370789766311646,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 36340
+    },
+    {
+      "epoch": 116.50641025641026,
+      "grad_norm": 1.004262089729309,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 36350
+    },
+    {
+      "epoch": 116.53846153846153,
+      "grad_norm": 1.2942055463790894,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36360
+    },
+    {
+      "epoch": 116.57051282051282,
+      "grad_norm": 1.5884029865264893,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 36370
+    },
+    {
+      "epoch": 116.6025641025641,
+      "grad_norm": 1.6395597457885742,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 36380
+    },
+    {
+      "epoch": 116.63461538461539,
+      "grad_norm": 1.7193987369537354,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 36390
+    },
+    {
+      "epoch": 116.66666666666667,
+      "grad_norm": 1.1398732662200928,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 36400
+    },
+    {
+      "epoch": 116.69871794871794,
+      "grad_norm": 1.0670888423919678,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 36410
+    },
+    {
+      "epoch": 116.73076923076923,
+      "grad_norm": 1.2067244052886963,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 36420
+    },
+    {
+      "epoch": 116.76282051282051,
+      "grad_norm": 1.1968103647232056,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 36430
+    },
+    {
+      "epoch": 116.7948717948718,
+      "grad_norm": 1.133968472480774,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 36440
+    },
+    {
+      "epoch": 116.82692307692308,
+      "grad_norm": 1.3593436479568481,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 36450
+    },
+    {
+      "epoch": 116.85897435897436,
+      "grad_norm": 1.0189651250839233,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 36460
+    },
+    {
+      "epoch": 116.89102564102564,
+      "grad_norm": 1.3122587203979492,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36470
+    },
+    {
+      "epoch": 116.92307692307692,
+      "grad_norm": 1.2137296199798584,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36480
+    },
+    {
+      "epoch": 116.9551282051282,
+      "grad_norm": 1.3130732774734497,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 36490
+    },
+    {
+      "epoch": 116.98717948717949,
+      "grad_norm": 1.1677205562591553,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 36500
+    },
+    {
+      "epoch": 117.01923076923077,
+      "grad_norm": 0.9951295256614685,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36510
+    },
+    {
+      "epoch": 117.05128205128206,
+      "grad_norm": 1.002666711807251,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 36520
+    },
+    {
+      "epoch": 117.08333333333333,
+      "grad_norm": 0.9221408367156982,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 36530
+    },
+    {
+      "epoch": 117.11538461538461,
+      "grad_norm": 1.0738126039505005,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 36540
+    },
+    {
+      "epoch": 117.1474358974359,
+      "grad_norm": 0.9025449752807617,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 36550
+    },
+    {
+      "epoch": 117.17948717948718,
+      "grad_norm": 0.8644457459449768,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 36560
+    },
+    {
+      "epoch": 117.21153846153847,
+      "grad_norm": 1.0142635107040405,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 36570
+    },
+    {
+      "epoch": 117.24358974358974,
+      "grad_norm": 0.8254585862159729,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 36580
+    },
+    {
+      "epoch": 117.27564102564102,
+      "grad_norm": 0.9724637866020203,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 36590
+    },
+    {
+      "epoch": 117.3076923076923,
+      "grad_norm": 0.9432185888290405,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 36600
+    },
+    {
+      "epoch": 117.33974358974359,
+      "grad_norm": 0.7894449830055237,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36610
+    },
+    {
+      "epoch": 117.37179487179488,
+      "grad_norm": 1.3503042459487915,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 36620
+    },
+    {
+      "epoch": 117.40384615384616,
+      "grad_norm": 1.2821942567825317,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 36630
+    },
+    {
+      "epoch": 117.43589743589743,
+      "grad_norm": 1.197762370109558,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 36640
+    },
+    {
+      "epoch": 117.46794871794872,
+      "grad_norm": 1.1641517877578735,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 36650
+    },
+    {
+      "epoch": 117.5,
+      "grad_norm": 0.768038809299469,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 36660
+    },
+    {
+      "epoch": 117.53205128205128,
+      "grad_norm": 1.2366490364074707,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 36670
+    },
+    {
+      "epoch": 117.56410256410257,
+      "grad_norm": 0.8339612483978271,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 36680
+    },
+    {
+      "epoch": 117.59615384615384,
+      "grad_norm": 0.8909167647361755,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 36690
+    },
+    {
+      "epoch": 117.62820512820512,
+      "grad_norm": 1.0436495542526245,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 36700
+    },
+    {
+      "epoch": 117.66025641025641,
+      "grad_norm": 1.5050634145736694,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 36710
+    },
+    {
+      "epoch": 117.6923076923077,
+      "grad_norm": 1.2290029525756836,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 36720
+    },
+    {
+      "epoch": 117.72435897435898,
+      "grad_norm": 0.7505238056182861,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36730
+    },
+    {
+      "epoch": 117.75641025641026,
+      "grad_norm": 1.1592241525650024,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36740
+    },
+    {
+      "epoch": 117.78846153846153,
+      "grad_norm": 1.4114667177200317,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36750
+    },
+    {
+      "epoch": 117.82051282051282,
+      "grad_norm": 1.2711656093597412,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 36760
+    },
+    {
+      "epoch": 117.8525641025641,
+      "grad_norm": 1.3661960363388062,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 36770
+    },
+    {
+      "epoch": 117.88461538461539,
+      "grad_norm": 1.335207223892212,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 36780
+    },
+    {
+      "epoch": 117.91666666666667,
+      "grad_norm": 1.0093327760696411,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 36790
+    },
+    {
+      "epoch": 117.94871794871794,
+      "grad_norm": 0.7778162360191345,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 36800
+    },
+    {
+      "epoch": 117.98076923076923,
+      "grad_norm": 0.8359770178794861,
+      "learning_rate": 0.0001,
+      "loss": 0.0099,
+      "step": 36810
+    },
+    {
+      "epoch": 118.01282051282051,
+      "grad_norm": 0.8301272988319397,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 36820
+    },
+    {
+      "epoch": 118.0448717948718,
+      "grad_norm": 1.1685984134674072,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36830
+    },
+    {
+      "epoch": 118.07692307692308,
+      "grad_norm": 0.8358983397483826,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36840
+    },
+    {
+      "epoch": 118.10897435897436,
+      "grad_norm": 1.2407375574111938,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36850
+    },
+    {
+      "epoch": 118.14102564102564,
+      "grad_norm": 1.1837193965911865,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 36860
+    },
+    {
+      "epoch": 118.17307692307692,
+      "grad_norm": 1.365870714187622,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 36870
+    },
+    {
+      "epoch": 118.2051282051282,
+      "grad_norm": 0.7776281237602234,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 36880
+    },
+    {
+      "epoch": 118.23717948717949,
+      "grad_norm": 1.3914923667907715,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36890
+    },
+    {
+      "epoch": 118.26923076923077,
+      "grad_norm": 1.0448061227798462,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 36900
+    },
+    {
+      "epoch": 118.30128205128206,
+      "grad_norm": 0.9456466436386108,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 36910
+    },
+    {
+      "epoch": 118.33333333333333,
+      "grad_norm": 1.1002576351165771,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36920
+    },
+    {
+      "epoch": 118.36538461538461,
+      "grad_norm": 1.0046789646148682,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 36930
+    },
+    {
+      "epoch": 118.3974358974359,
+      "grad_norm": 0.9848045706748962,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 36940
+    },
+    {
+      "epoch": 118.42948717948718,
+      "grad_norm": 0.7974382638931274,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36950
+    },
+    {
+      "epoch": 118.46153846153847,
+      "grad_norm": 1.179329514503479,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 36960
+    },
+    {
+      "epoch": 118.49358974358974,
+      "grad_norm": 1.1153396368026733,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 36970
+    },
+    {
+      "epoch": 118.52564102564102,
+      "grad_norm": 1.0547434091567993,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 36980
+    },
+    {
+      "epoch": 118.5576923076923,
+      "grad_norm": 0.8892029523849487,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 36990
+    },
+    {
+      "epoch": 118.58974358974359,
+      "grad_norm": 0.9579285383224487,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 37000
+    },
+    {
+      "epoch": 118.62179487179488,
+      "grad_norm": 0.9576907157897949,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 37010
+    },
+    {
+      "epoch": 118.65384615384616,
+      "grad_norm": 1.0754719972610474,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 37020
+    },
+    {
+      "epoch": 118.68589743589743,
+      "grad_norm": 0.953140914440155,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 37030
+    },
+    {
+      "epoch": 118.71794871794872,
+      "grad_norm": 0.9028884172439575,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 37040
+    },
+    {
+      "epoch": 118.75,
+      "grad_norm": 0.8152345418930054,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 37050
+    },
+    {
+      "epoch": 118.78205128205128,
+      "grad_norm": 1.053157925605774,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 37060
+    },
+    {
+      "epoch": 118.81410256410257,
+      "grad_norm": 1.0114665031433105,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 37070
+    },
+    {
+      "epoch": 118.84615384615384,
+      "grad_norm": 1.2854266166687012,
+      "learning_rate": 0.0001,
+      "loss": 0.01,
+      "step": 37080
+    },
+    {
+      "epoch": 118.87820512820512,
+      "grad_norm": 1.1853556632995605,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 37090
+    },
+    {
+      "epoch": 118.91025641025641,
+      "grad_norm": 1.2492280006408691,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 37100
+    },
+    {
+      "epoch": 118.9423076923077,
+      "grad_norm": 0.9992641806602478,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 37110
+    },
+    {
+      "epoch": 118.97435897435898,
+      "grad_norm": 1.2440236806869507,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 37120
+    },
+    {
+      "epoch": 119.00641025641026,
+      "grad_norm": 1.5629132986068726,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 37130
+    },
+    {
+      "epoch": 119.03846153846153,
+      "grad_norm": 1.038429856300354,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37140
+    },
+    {
+      "epoch": 119.07051282051282,
+      "grad_norm": 1.3030531406402588,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37150
+    },
+    {
+      "epoch": 119.1025641025641,
+      "grad_norm": 0.90859055519104,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 37160
+    },
+    {
+      "epoch": 119.13461538461539,
+      "grad_norm": 0.8204979300498962,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 37170
+    },
+    {
+      "epoch": 119.16666666666667,
+      "grad_norm": 1.364283800125122,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37180
+    },
+    {
+      "epoch": 119.19871794871794,
+      "grad_norm": 0.972510039806366,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 37190
+    },
+    {
+      "epoch": 119.23076923076923,
+      "grad_norm": 0.943085253238678,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 37200
+    },
+    {
+      "epoch": 119.26282051282051,
+      "grad_norm": 1.0313957929611206,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 37210
+    },
+    {
+      "epoch": 119.2948717948718,
+      "grad_norm": 1.0821967124938965,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 37220
+    },
+    {
+      "epoch": 119.32692307692308,
+      "grad_norm": 0.8521621823310852,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 37230
+    },
+    {
+      "epoch": 119.35897435897436,
+      "grad_norm": 1.0269619226455688,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 37240
+    },
+    {
+      "epoch": 119.39102564102564,
+      "grad_norm": 1.2158182859420776,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 37250
+    },
+    {
+      "epoch": 119.42307692307692,
+      "grad_norm": 1.2556642293930054,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 37260
+    },
+    {
+      "epoch": 119.4551282051282,
+      "grad_norm": 1.1150671243667603,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 37270
+    },
+    {
+      "epoch": 119.48717948717949,
+      "grad_norm": 0.9753519892692566,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 37280
+    },
+    {
+      "epoch": 119.51923076923077,
+      "grad_norm": 1.5012890100479126,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 37290
+    },
+    {
+      "epoch": 119.55128205128206,
+      "grad_norm": 1.0090817213058472,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 37300
+    },
+    {
+      "epoch": 119.58333333333333,
+      "grad_norm": 0.8751769065856934,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 37310
+    },
+    {
+      "epoch": 119.61538461538461,
+      "grad_norm": 0.892868161201477,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 37320
+    },
+    {
+      "epoch": 119.6474358974359,
+      "grad_norm": 0.9372827410697937,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 37330
+    },
+    {
+      "epoch": 119.67948717948718,
+      "grad_norm": 0.9513546824455261,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 37340
+    },
+    {
+      "epoch": 119.71153846153847,
+      "grad_norm": 1.392795205116272,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 37350
+    },
+    {
+      "epoch": 119.74358974358974,
+      "grad_norm": 0.7589762806892395,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 37360
+    },
+    {
+      "epoch": 119.77564102564102,
+      "grad_norm": 0.9090687036514282,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 37370
+    },
+    {
+      "epoch": 119.8076923076923,
+      "grad_norm": 1.4228520393371582,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 37380
+    },
+    {
+      "epoch": 119.83974358974359,
+      "grad_norm": 1.2681548595428467,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 37390
+    },
+    {
+      "epoch": 119.87179487179488,
+      "grad_norm": 0.9315091967582703,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 37400
+    },
+    {
+      "epoch": 119.90384615384616,
+      "grad_norm": 0.8245327472686768,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 37410
+    },
+    {
+      "epoch": 119.93589743589743,
+      "grad_norm": 0.7165843844413757,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 37420
+    },
+    {
+      "epoch": 119.96794871794872,
+      "grad_norm": 1.2519831657409668,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37430
+    },
+    {
+      "epoch": 120.0,
+      "grad_norm": 1.140870213508606,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 37440
+    },
+    {
+      "epoch": 120.03205128205128,
+      "grad_norm": 0.873918890953064,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37450
+    },
+    {
+      "epoch": 120.06410256410257,
+      "grad_norm": 0.8864120841026306,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 37460
+    },
+    {
+      "epoch": 120.09615384615384,
+      "grad_norm": 1.0961915254592896,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 37470
+    },
+    {
+      "epoch": 120.12820512820512,
+      "grad_norm": 0.7221135497093201,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 37480
+    },
+    {
+      "epoch": 120.16025641025641,
+      "grad_norm": 1.292258381843567,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 37490
+    },
+    {
+      "epoch": 120.1923076923077,
+      "grad_norm": 0.9487662315368652,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 37500
+    },
+    {
+      "epoch": 120.22435897435898,
+      "grad_norm": 0.7183855772018433,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 37510
+    },
+    {
+      "epoch": 120.25641025641026,
+      "grad_norm": 1.0286792516708374,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 37520
+    },
+    {
+      "epoch": 120.28846153846153,
+      "grad_norm": 0.7679458260536194,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 37530
+    },
+    {
+      "epoch": 120.32051282051282,
+      "grad_norm": 0.7706106305122375,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 37540
+    },
+    {
+      "epoch": 120.3525641025641,
+      "grad_norm": 0.9112149477005005,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37550
+    },
+    {
+      "epoch": 120.38461538461539,
+      "grad_norm": 0.9655150771141052,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 37560
+    },
+    {
+      "epoch": 120.41666666666667,
+      "grad_norm": 1.1305304765701294,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 37570
+    },
+    {
+      "epoch": 120.44871794871794,
+      "grad_norm": 1.1805723905563354,
+      "learning_rate": 0.0001,
+      "loss": 0.0112,
+      "step": 37580
+    },
+    {
+      "epoch": 120.48076923076923,
+      "grad_norm": 1.5254597663879395,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 37590
+    },
+    {
+      "epoch": 120.51282051282051,
+      "grad_norm": 1.05934476852417,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 37600
+    },
+    {
+      "epoch": 120.5448717948718,
+      "grad_norm": 1.0628738403320312,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 37610
+    },
+    {
+      "epoch": 120.57692307692308,
+      "grad_norm": 0.9894538521766663,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 37620
+    },
+    {
+      "epoch": 120.60897435897436,
+      "grad_norm": 1.2212821245193481,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 37630
+    },
+    {
+      "epoch": 120.64102564102564,
+      "grad_norm": 1.0113123655319214,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 37640
+    },
+    {
+      "epoch": 120.67307692307692,
+      "grad_norm": 1.0287967920303345,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 37650
+    },
+    {
+      "epoch": 120.7051282051282,
+      "grad_norm": 1.305213212966919,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 37660
+    },
+    {
+      "epoch": 120.73717948717949,
+      "grad_norm": 1.270689845085144,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37670
+    },
+    {
+      "epoch": 120.76923076923077,
+      "grad_norm": 1.0723679065704346,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 37680
+    },
+    {
+      "epoch": 120.80128205128206,
+      "grad_norm": 1.5003395080566406,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 37690
+    },
+    {
+      "epoch": 120.83333333333333,
+      "grad_norm": 1.3104630708694458,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37700
+    },
+    {
+      "epoch": 120.86538461538461,
+      "grad_norm": 0.9401475787162781,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 37710
+    },
+    {
+      "epoch": 120.8974358974359,
+      "grad_norm": 0.9448487162590027,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 37720
+    },
+    {
+      "epoch": 120.92948717948718,
+      "grad_norm": 0.8658837676048279,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 37730
+    },
+    {
+      "epoch": 120.96153846153847,
+      "grad_norm": 1.0394607782363892,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 37740
+    },
+    {
+      "epoch": 120.99358974358974,
+      "grad_norm": 1.0935457944869995,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 37750
+    },
+    {
+      "epoch": 121.02564102564102,
+      "grad_norm": 0.8717172741889954,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37760
+    },
+    {
+      "epoch": 121.0576923076923,
+      "grad_norm": 0.8224765658378601,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 37770
+    },
+    {
+      "epoch": 121.08974358974359,
+      "grad_norm": 0.8842837810516357,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 37780
+    },
+    {
+      "epoch": 121.12179487179488,
+      "grad_norm": 0.977579653263092,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 37790
+    },
+    {
+      "epoch": 121.15384615384616,
+      "grad_norm": 1.1447718143463135,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 37800
+    },
+    {
+      "epoch": 121.18589743589743,
+      "grad_norm": 0.944974958896637,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 37810
+    },
+    {
+      "epoch": 121.21794871794872,
+      "grad_norm": 0.9219497442245483,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 37820
+    },
+    {
+      "epoch": 121.25,
+      "grad_norm": 0.8822305798530579,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 37830
+    },
+    {
+      "epoch": 121.28205128205128,
+      "grad_norm": 0.9163874387741089,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 37840
+    },
+    {
+      "epoch": 121.31410256410257,
+      "grad_norm": 1.2067246437072754,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 37850
+    },
+    {
+      "epoch": 121.34615384615384,
+      "grad_norm": 1.0459383726119995,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 37860
+    },
+    {
+      "epoch": 121.37820512820512,
+      "grad_norm": 1.4731372594833374,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 37870
+    },
+    {
+      "epoch": 121.41025641025641,
+      "grad_norm": 1.787795901298523,
+      "learning_rate": 0.0001,
+      "loss": 0.0101,
+      "step": 37880
+    },
+    {
+      "epoch": 121.4423076923077,
+      "grad_norm": 1.080737590789795,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 37890
+    },
+    {
+      "epoch": 121.47435897435898,
+      "grad_norm": 1.6581826210021973,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 37900
+    },
+    {
+      "epoch": 121.50641025641026,
+      "grad_norm": 0.8877742886543274,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 37910
+    },
+    {
+      "epoch": 121.53846153846153,
+      "grad_norm": 1.073733925819397,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 37920
+    },
+    {
+      "epoch": 121.57051282051282,
+      "grad_norm": 0.9844526052474976,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 37930
+    },
+    {
+      "epoch": 121.6025641025641,
+      "grad_norm": 0.9768216609954834,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 37940
+    },
+    {
+      "epoch": 121.63461538461539,
+      "grad_norm": 1.0610164403915405,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 37950
+    },
+    {
+      "epoch": 121.66666666666667,
+      "grad_norm": 0.9092679619789124,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 37960
+    },
+    {
+      "epoch": 121.69871794871794,
+      "grad_norm": 1.3586238622665405,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 37970
+    },
+    {
+      "epoch": 121.73076923076923,
+      "grad_norm": 0.8189979791641235,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 37980
+    },
+    {
+      "epoch": 121.76282051282051,
+      "grad_norm": 2.027216672897339,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 37990
+    },
+    {
+      "epoch": 121.7948717948718,
+      "grad_norm": 1.3056910037994385,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 38000
+    },
+    {
+      "epoch": 121.82692307692308,
+      "grad_norm": 1.086320161819458,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 38010
+    },
+    {
+      "epoch": 121.85897435897436,
+      "grad_norm": 1.2434576749801636,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38020
+    },
+    {
+      "epoch": 121.89102564102564,
+      "grad_norm": 1.1754564046859741,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 38030
+    },
+    {
+      "epoch": 121.92307692307692,
+      "grad_norm": 1.2429568767547607,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 38040
+    },
+    {
+      "epoch": 121.9551282051282,
+      "grad_norm": 1.08347749710083,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 38050
+    },
+    {
+      "epoch": 121.98717948717949,
+      "grad_norm": 1.231427550315857,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 38060
+    },
+    {
+      "epoch": 122.01923076923077,
+      "grad_norm": 1.3169761896133423,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 38070
+    },
+    {
+      "epoch": 122.05128205128206,
+      "grad_norm": 1.7613811492919922,
+      "learning_rate": 0.0001,
+      "loss": 0.0104,
+      "step": 38080
+    },
+    {
+      "epoch": 122.08333333333333,
+      "grad_norm": 1.539383888244629,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 38090
+    },
+    {
+      "epoch": 122.11538461538461,
+      "grad_norm": 1.1597728729248047,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 38100
+    },
+    {
+      "epoch": 122.1474358974359,
+      "grad_norm": 1.306006908416748,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 38110
+    },
+    {
+      "epoch": 122.17948717948718,
+      "grad_norm": 1.496312141418457,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 38120
+    },
+    {
+      "epoch": 122.21153846153847,
+      "grad_norm": 1.1867165565490723,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 38130
+    },
+    {
+      "epoch": 122.24358974358974,
+      "grad_norm": 1.316125512123108,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38140
+    },
+    {
+      "epoch": 122.27564102564102,
+      "grad_norm": 1.2291179895401,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 38150
+    },
+    {
+      "epoch": 122.3076923076923,
+      "grad_norm": 1.1859768629074097,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38160
+    },
+    {
+      "epoch": 122.33974358974359,
+      "grad_norm": 0.8237673044204712,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 38170
+    },
+    {
+      "epoch": 122.37179487179488,
+      "grad_norm": 0.8602513670921326,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 38180
+    },
+    {
+      "epoch": 122.40384615384616,
+      "grad_norm": 0.8516953587532043,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 38190
+    },
+    {
+      "epoch": 122.43589743589743,
+      "grad_norm": 0.9156997799873352,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38200
+    },
+    {
+      "epoch": 122.46794871794872,
+      "grad_norm": 1.1865869760513306,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 38210
+    },
+    {
+      "epoch": 122.5,
+      "grad_norm": 0.7694472670555115,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 38220
+    },
+    {
+      "epoch": 122.53205128205128,
+      "grad_norm": 1.1037179231643677,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 38230
+    },
+    {
+      "epoch": 122.56410256410257,
+      "grad_norm": 1.1107829809188843,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 38240
+    },
+    {
+      "epoch": 122.59615384615384,
+      "grad_norm": 0.9626089334487915,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 38250
+    },
+    {
+      "epoch": 122.62820512820512,
+      "grad_norm": 1.108228087425232,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 38260
+    },
+    {
+      "epoch": 122.66025641025641,
+      "grad_norm": 1.1997582912445068,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 38270
+    },
+    {
+      "epoch": 122.6923076923077,
+      "grad_norm": 1.0256119966506958,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 38280
+    },
+    {
+      "epoch": 122.72435897435898,
+      "grad_norm": 1.007524847984314,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 38290
+    },
+    {
+      "epoch": 122.75641025641026,
+      "grad_norm": 0.8311702013015747,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 38300
+    },
+    {
+      "epoch": 122.78846153846153,
+      "grad_norm": 0.6781300902366638,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 38310
+    },
+    {
+      "epoch": 122.82051282051282,
+      "grad_norm": 0.9936074018478394,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 38320
+    },
+    {
+      "epoch": 122.8525641025641,
+      "grad_norm": 1.0264308452606201,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 38330
+    },
+    {
+      "epoch": 122.88461538461539,
+      "grad_norm": 0.9022648334503174,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 38340
+    },
+    {
+      "epoch": 122.91666666666667,
+      "grad_norm": 0.9871827363967896,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38350
+    },
+    {
+      "epoch": 122.94871794871794,
+      "grad_norm": 1.009092926979065,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 38360
+    },
+    {
+      "epoch": 122.98076923076923,
+      "grad_norm": 0.959388792514801,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 38370
+    },
+    {
+      "epoch": 123.01282051282051,
+      "grad_norm": 0.9542678594589233,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38380
+    },
+    {
+      "epoch": 123.0448717948718,
+      "grad_norm": 0.8461974859237671,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38390
+    },
+    {
+      "epoch": 123.07692307692308,
+      "grad_norm": 0.9967362284660339,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 38400
+    },
+    {
+      "epoch": 123.10897435897436,
+      "grad_norm": 1.1398873329162598,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38410
+    },
+    {
+      "epoch": 123.14102564102564,
+      "grad_norm": 1.4437137842178345,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 38420
+    },
+    {
+      "epoch": 123.17307692307692,
+      "grad_norm": 0.9274420142173767,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 38430
+    },
+    {
+      "epoch": 123.2051282051282,
+      "grad_norm": 0.81486576795578,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 38440
+    },
+    {
+      "epoch": 123.23717948717949,
+      "grad_norm": 0.897555947303772,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 38450
+    },
+    {
+      "epoch": 123.26923076923077,
+      "grad_norm": 1.0148229598999023,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 38460
+    },
+    {
+      "epoch": 123.30128205128206,
+      "grad_norm": 0.780492901802063,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38470
+    },
+    {
+      "epoch": 123.33333333333333,
+      "grad_norm": 0.8472212553024292,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 38480
+    },
+    {
+      "epoch": 123.36538461538461,
+      "grad_norm": 1.1869184970855713,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38490
+    },
+    {
+      "epoch": 123.3974358974359,
+      "grad_norm": 1.268882393836975,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 38500
+    },
+    {
+      "epoch": 123.42948717948718,
+      "grad_norm": 1.0924943685531616,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 38510
+    },
+    {
+      "epoch": 123.46153846153847,
+      "grad_norm": 1.1451411247253418,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 38520
+    },
+    {
+      "epoch": 123.49358974358974,
+      "grad_norm": 0.9232407212257385,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 38530
+    },
+    {
+      "epoch": 123.52564102564102,
+      "grad_norm": 1.2505323886871338,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38540
+    },
+    {
+      "epoch": 123.5576923076923,
+      "grad_norm": 1.0468837022781372,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38550
+    },
+    {
+      "epoch": 123.58974358974359,
+      "grad_norm": 0.9577003717422485,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38560
+    },
+    {
+      "epoch": 123.62179487179488,
+      "grad_norm": 0.6712190508842468,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38570
+    },
+    {
+      "epoch": 123.65384615384616,
+      "grad_norm": 1.112716794013977,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 38580
+    },
+    {
+      "epoch": 123.68589743589743,
+      "grad_norm": 0.9134427309036255,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 38590
+    },
+    {
+      "epoch": 123.71794871794872,
+      "grad_norm": 0.8722566962242126,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 38600
+    },
+    {
+      "epoch": 123.75,
+      "grad_norm": 0.9633263349533081,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 38610
+    },
+    {
+      "epoch": 123.78205128205128,
+      "grad_norm": 0.8448625802993774,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 38620
+    },
+    {
+      "epoch": 123.81410256410257,
+      "grad_norm": 1.0955870151519775,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38630
+    },
+    {
+      "epoch": 123.84615384615384,
+      "grad_norm": 0.8688932061195374,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38640
+    },
+    {
+      "epoch": 123.87820512820512,
+      "grad_norm": 1.1955755949020386,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 38650
+    },
+    {
+      "epoch": 123.91025641025641,
+      "grad_norm": 1.136197566986084,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38660
+    },
+    {
+      "epoch": 123.9423076923077,
+      "grad_norm": 0.8654513955116272,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 38670
+    },
+    {
+      "epoch": 123.97435897435898,
+      "grad_norm": 1.056085467338562,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38680
+    },
+    {
+      "epoch": 124.00641025641026,
+      "grad_norm": 0.9820038080215454,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38690
+    },
+    {
+      "epoch": 124.03846153846153,
+      "grad_norm": 1.0208890438079834,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38700
+    },
+    {
+      "epoch": 124.07051282051282,
+      "grad_norm": 1.3993744850158691,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 38710
+    },
+    {
+      "epoch": 124.1025641025641,
+      "grad_norm": 1.309738039970398,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 38720
+    },
+    {
+      "epoch": 124.13461538461539,
+      "grad_norm": 1.7182036638259888,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 38730
+    },
+    {
+      "epoch": 124.16666666666667,
+      "grad_norm": 1.1063226461410522,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 38740
+    },
+    {
+      "epoch": 124.19871794871794,
+      "grad_norm": 1.3547861576080322,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 38750
+    },
+    {
+      "epoch": 124.23076923076923,
+      "grad_norm": 1.4306471347808838,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38760
+    },
+    {
+      "epoch": 124.26282051282051,
+      "grad_norm": 1.188254952430725,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 38770
+    },
+    {
+      "epoch": 124.2948717948718,
+      "grad_norm": 0.9612206220626831,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 38780
+    },
+    {
+      "epoch": 124.32692307692308,
+      "grad_norm": 0.8345605731010437,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 38790
+    },
+    {
+      "epoch": 124.35897435897436,
+      "grad_norm": 0.7034676671028137,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38800
+    },
+    {
+      "epoch": 124.39102564102564,
+      "grad_norm": 1.0068942308425903,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38810
+    },
+    {
+      "epoch": 124.42307692307692,
+      "grad_norm": 0.8288251757621765,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38820
+    },
+    {
+      "epoch": 124.4551282051282,
+      "grad_norm": 0.7677650451660156,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38830
+    },
+    {
+      "epoch": 124.48717948717949,
+      "grad_norm": 1.165030837059021,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 38840
+    },
+    {
+      "epoch": 124.51923076923077,
+      "grad_norm": 0.7278369665145874,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38850
+    },
+    {
+      "epoch": 124.55128205128206,
+      "grad_norm": 0.6589283347129822,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 38860
+    },
+    {
+      "epoch": 124.58333333333333,
+      "grad_norm": 0.8573265075683594,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 38870
+    },
+    {
+      "epoch": 124.61538461538461,
+      "grad_norm": 1.0131523609161377,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 38880
+    },
+    {
+      "epoch": 124.6474358974359,
+      "grad_norm": 1.0030937194824219,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 38890
+    },
+    {
+      "epoch": 124.67948717948718,
+      "grad_norm": 0.9330103993415833,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 38900
+    },
+    {
+      "epoch": 124.71153846153847,
+      "grad_norm": 0.6826152205467224,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38910
+    },
+    {
+      "epoch": 124.74358974358974,
+      "grad_norm": 0.9866458773612976,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 38920
+    },
+    {
+      "epoch": 124.77564102564102,
+      "grad_norm": 0.8811094760894775,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 38930
+    },
+    {
+      "epoch": 124.8076923076923,
+      "grad_norm": 0.888728141784668,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 38940
+    },
+    {
+      "epoch": 124.83974358974359,
+      "grad_norm": 1.0838638544082642,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 38950
+    },
+    {
+      "epoch": 124.87179487179488,
+      "grad_norm": 1.015923261642456,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 38960
+    },
+    {
+      "epoch": 124.90384615384616,
+      "grad_norm": 0.999843180179596,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 38970
+    },
+    {
+      "epoch": 124.93589743589743,
+      "grad_norm": 1.2710967063903809,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 38980
+    },
+    {
+      "epoch": 124.96794871794872,
+      "grad_norm": 0.9060564041137695,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 38990
+    },
+    {
+      "epoch": 125.0,
+      "grad_norm": 0.758036732673645,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 39000
+    },
+    {
+      "epoch": 125.03205128205128,
+      "grad_norm": 0.8229281306266785,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 39010
+    },
+    {
+      "epoch": 125.06410256410257,
+      "grad_norm": 1.0202244520187378,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 39020
+    },
+    {
+      "epoch": 125.09615384615384,
+      "grad_norm": 0.8539890646934509,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39030
+    },
+    {
+      "epoch": 125.12820512820512,
+      "grad_norm": 0.5889796614646912,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39040
+    },
+    {
+      "epoch": 125.16025641025641,
+      "grad_norm": 0.7456795573234558,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39050
+    },
+    {
+      "epoch": 125.1923076923077,
+      "grad_norm": 0.9887012243270874,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 39060
+    },
+    {
+      "epoch": 125.22435897435898,
+      "grad_norm": 0.8845901489257812,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 39070
+    },
+    {
+      "epoch": 125.25641025641026,
+      "grad_norm": 0.8112621307373047,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39080
+    },
+    {
+      "epoch": 125.28846153846153,
+      "grad_norm": 0.8526967167854309,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 39090
+    },
+    {
+      "epoch": 125.32051282051282,
+      "grad_norm": 0.5408542156219482,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39100
+    },
+    {
+      "epoch": 125.3525641025641,
+      "grad_norm": 0.7082494497299194,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 39110
+    },
+    {
+      "epoch": 125.38461538461539,
+      "grad_norm": 1.0076625347137451,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 39120
+    },
+    {
+      "epoch": 125.41666666666667,
+      "grad_norm": 0.6562144756317139,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39130
+    },
+    {
+      "epoch": 125.44871794871794,
+      "grad_norm": 0.8369951248168945,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39140
+    },
+    {
+      "epoch": 125.48076923076923,
+      "grad_norm": 1.273681402206421,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 39150
+    },
+    {
+      "epoch": 125.51282051282051,
+      "grad_norm": 0.9491506814956665,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39160
+    },
+    {
+      "epoch": 125.5448717948718,
+      "grad_norm": 0.949055016040802,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 39170
+    },
+    {
+      "epoch": 125.57692307692308,
+      "grad_norm": 0.8828176856040955,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 39180
+    },
+    {
+      "epoch": 125.60897435897436,
+      "grad_norm": 0.9834796786308289,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 39190
+    },
+    {
+      "epoch": 125.64102564102564,
+      "grad_norm": 0.7818616032600403,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39200
+    },
+    {
+      "epoch": 125.67307692307692,
+      "grad_norm": 0.884906530380249,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39210
+    },
+    {
+      "epoch": 125.7051282051282,
+      "grad_norm": 0.9519088864326477,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39220
+    },
+    {
+      "epoch": 125.73717948717949,
+      "grad_norm": 1.1145578622817993,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 39230
+    },
+    {
+      "epoch": 125.76923076923077,
+      "grad_norm": 1.0290894508361816,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39240
+    },
+    {
+      "epoch": 125.80128205128206,
+      "grad_norm": 1.0575909614562988,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 39250
+    },
+    {
+      "epoch": 125.83333333333333,
+      "grad_norm": 0.9596019983291626,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 39260
+    },
+    {
+      "epoch": 125.86538461538461,
+      "grad_norm": 1.364061951637268,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39270
+    },
+    {
+      "epoch": 125.8974358974359,
+      "grad_norm": 0.9600426554679871,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39280
+    },
+    {
+      "epoch": 125.92948717948718,
+      "grad_norm": 0.8981009721755981,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 39290
+    },
+    {
+      "epoch": 125.96153846153847,
+      "grad_norm": 1.086188554763794,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 39300
+    },
+    {
+      "epoch": 125.99358974358974,
+      "grad_norm": 1.0241005420684814,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 39310
+    },
+    {
+      "epoch": 126.02564102564102,
+      "grad_norm": 1.0659921169281006,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 39320
+    },
+    {
+      "epoch": 126.0576923076923,
+      "grad_norm": 1.077957034111023,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39330
+    },
+    {
+      "epoch": 126.08974358974359,
+      "grad_norm": 1.0862088203430176,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 39340
+    },
+    {
+      "epoch": 126.12179487179488,
+      "grad_norm": 1.149412989616394,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 39350
+    },
+    {
+      "epoch": 126.15384615384616,
+      "grad_norm": 0.7921169400215149,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39360
+    },
+    {
+      "epoch": 126.18589743589743,
+      "grad_norm": 1.0487414598464966,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 39370
+    },
+    {
+      "epoch": 126.21794871794872,
+      "grad_norm": 1.1538461446762085,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39380
+    },
+    {
+      "epoch": 126.25,
+      "grad_norm": 1.026017665863037,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39390
+    },
+    {
+      "epoch": 126.28205128205128,
+      "grad_norm": 1.2047306299209595,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39400
+    },
+    {
+      "epoch": 126.31410256410257,
+      "grad_norm": 0.9121538400650024,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39410
+    },
+    {
+      "epoch": 126.34615384615384,
+      "grad_norm": 0.542550802230835,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39420
+    },
+    {
+      "epoch": 126.37820512820512,
+      "grad_norm": 1.2510408163070679,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39430
+    },
+    {
+      "epoch": 126.41025641025641,
+      "grad_norm": 1.0414520502090454,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 39440
+    },
+    {
+      "epoch": 126.4423076923077,
+      "grad_norm": 0.6576949954032898,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 39450
+    },
+    {
+      "epoch": 126.47435897435898,
+      "grad_norm": 0.7586349844932556,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 39460
+    },
+    {
+      "epoch": 126.50641025641026,
+      "grad_norm": 1.1157252788543701,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39470
+    },
+    {
+      "epoch": 126.53846153846153,
+      "grad_norm": 0.8899763822555542,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 39480
+    },
+    {
+      "epoch": 126.57051282051282,
+      "grad_norm": 0.9209287762641907,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 39490
+    },
+    {
+      "epoch": 126.6025641025641,
+      "grad_norm": 1.1588634252548218,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39500
+    },
+    {
+      "epoch": 126.63461538461539,
+      "grad_norm": 0.8047305345535278,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 39510
+    },
+    {
+      "epoch": 126.66666666666667,
+      "grad_norm": 1.0244567394256592,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 39520
+    },
+    {
+      "epoch": 126.69871794871794,
+      "grad_norm": 0.9543430209159851,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39530
+    },
+    {
+      "epoch": 126.73076923076923,
+      "grad_norm": 1.361498475074768,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 39540
+    },
+    {
+      "epoch": 126.76282051282051,
+      "grad_norm": 1.7954063415527344,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39550
+    },
+    {
+      "epoch": 126.7948717948718,
+      "grad_norm": 0.9814450144767761,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 39560
+    },
+    {
+      "epoch": 126.82692307692308,
+      "grad_norm": 1.1139119863510132,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 39570
+    },
+    {
+      "epoch": 126.85897435897436,
+      "grad_norm": 0.768962025642395,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 39580
+    },
+    {
+      "epoch": 126.89102564102564,
+      "grad_norm": 0.7492997050285339,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 39590
+    },
+    {
+      "epoch": 126.92307692307692,
+      "grad_norm": 0.9768083691596985,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 39600
+    },
+    {
+      "epoch": 126.9551282051282,
+      "grad_norm": 0.6380500793457031,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 39610
+    },
+    {
+      "epoch": 126.98717948717949,
+      "grad_norm": 0.8277572393417358,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 39620
+    },
+    {
+      "epoch": 127.01923076923077,
+      "grad_norm": 1.0916587114334106,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 39630
+    },
+    {
+      "epoch": 127.05128205128206,
+      "grad_norm": 1.0720793008804321,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 39640
+    },
+    {
+      "epoch": 127.08333333333333,
+      "grad_norm": 0.9529857635498047,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 39650
+    },
+    {
+      "epoch": 127.11538461538461,
+      "grad_norm": 1.0344326496124268,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 39660
+    },
+    {
+      "epoch": 127.1474358974359,
+      "grad_norm": 1.5250375270843506,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 39670
+    },
+    {
+      "epoch": 127.17948717948718,
+      "grad_norm": 1.1471507549285889,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 39680
+    },
+    {
+      "epoch": 127.21153846153847,
+      "grad_norm": 0.8738937377929688,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 39690
+    },
+    {
+      "epoch": 127.24358974358974,
+      "grad_norm": 1.087622880935669,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39700
+    },
+    {
+      "epoch": 127.27564102564102,
+      "grad_norm": 1.0543031692504883,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39710
+    },
+    {
+      "epoch": 127.3076923076923,
+      "grad_norm": 1.161056637763977,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 39720
+    },
+    {
+      "epoch": 127.33974358974359,
+      "grad_norm": 1.3229137659072876,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39730
+    },
+    {
+      "epoch": 127.37179487179488,
+      "grad_norm": 1.0491135120391846,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39740
+    },
+    {
+      "epoch": 127.40384615384616,
+      "grad_norm": 1.8914945125579834,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39750
+    },
+    {
+      "epoch": 127.43589743589743,
+      "grad_norm": 1.1148899793624878,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 39760
+    },
+    {
+      "epoch": 127.46794871794872,
+      "grad_norm": 1.193085789680481,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39770
+    },
+    {
+      "epoch": 127.5,
+      "grad_norm": 0.7942069172859192,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 39780
+    },
+    {
+      "epoch": 127.53205128205128,
+      "grad_norm": 0.8707871437072754,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 39790
+    },
+    {
+      "epoch": 127.56410256410257,
+      "grad_norm": 0.8801212906837463,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 39800
+    },
+    {
+      "epoch": 127.59615384615384,
+      "grad_norm": 0.8320527672767639,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 39810
+    },
+    {
+      "epoch": 127.62820512820512,
+      "grad_norm": 0.9485070109367371,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 39820
+    },
+    {
+      "epoch": 127.66025641025641,
+      "grad_norm": 0.7693749070167542,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 39830
+    },
+    {
+      "epoch": 127.6923076923077,
+      "grad_norm": 0.913823664188385,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 39840
+    },
+    {
+      "epoch": 127.72435897435898,
+      "grad_norm": 1.0221679210662842,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 39850
+    },
+    {
+      "epoch": 127.75641025641026,
+      "grad_norm": 1.1704120635986328,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 39860
+    },
+    {
+      "epoch": 127.78846153846153,
+      "grad_norm": 1.0989817380905151,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 39870
+    },
+    {
+      "epoch": 127.82051282051282,
+      "grad_norm": 1.260799527168274,
+      "learning_rate": 0.0001,
+      "loss": 0.0098,
+      "step": 39880
+    },
+    {
+      "epoch": 127.8525641025641,
+      "grad_norm": 1.1412057876586914,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 39890
+    },
+    {
+      "epoch": 127.88461538461539,
+      "grad_norm": 1.4795066118240356,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 39900
+    },
+    {
+      "epoch": 127.91666666666667,
+      "grad_norm": 1.478244423866272,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 39910
+    },
+    {
+      "epoch": 127.94871794871794,
+      "grad_norm": 1.3007595539093018,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 39920
+    },
+    {
+      "epoch": 127.98076923076923,
+      "grad_norm": 0.9378211498260498,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 39930
+    },
+    {
+      "epoch": 128.01282051282053,
+      "grad_norm": 0.8112432956695557,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 39940
+    },
+    {
+      "epoch": 128.0448717948718,
+      "grad_norm": 0.9108434319496155,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 39950
+    },
+    {
+      "epoch": 128.07692307692307,
+      "grad_norm": 0.954410195350647,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 39960
+    },
+    {
+      "epoch": 128.10897435897436,
+      "grad_norm": 0.8351195454597473,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 39970
+    },
+    {
+      "epoch": 128.14102564102564,
+      "grad_norm": 0.7267666459083557,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 39980
+    },
+    {
+      "epoch": 128.17307692307693,
+      "grad_norm": 0.8122282028198242,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 39990
+    },
+    {
+      "epoch": 128.2051282051282,
+      "grad_norm": 0.8008731007575989,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 40000
+    },
+    {
+      "epoch": 128.23717948717947,
+      "grad_norm": 0.821221649646759,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 40010
+    },
+    {
+      "epoch": 128.26923076923077,
+      "grad_norm": 0.8462360501289368,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 40020
+    },
+    {
+      "epoch": 128.30128205128204,
+      "grad_norm": 1.028199553489685,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40030
+    },
+    {
+      "epoch": 128.33333333333334,
+      "grad_norm": 0.9447816014289856,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 40040
+    },
+    {
+      "epoch": 128.3653846153846,
+      "grad_norm": 1.0166338682174683,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40050
+    },
+    {
+      "epoch": 128.39743589743588,
+      "grad_norm": 0.7016328573226929,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 40060
+    },
+    {
+      "epoch": 128.42948717948718,
+      "grad_norm": 0.5777688026428223,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 40070
+    },
+    {
+      "epoch": 128.46153846153845,
+      "grad_norm": 0.5507752299308777,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 40080
+    },
+    {
+      "epoch": 128.49358974358975,
+      "grad_norm": 0.809963583946228,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 40090
+    },
+    {
+      "epoch": 128.52564102564102,
+      "grad_norm": 1.1626081466674805,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40100
+    },
+    {
+      "epoch": 128.55769230769232,
+      "grad_norm": 1.082632064819336,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 40110
+    },
+    {
+      "epoch": 128.5897435897436,
+      "grad_norm": 0.9105743169784546,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40120
+    },
+    {
+      "epoch": 128.62179487179486,
+      "grad_norm": 1.017638087272644,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40130
+    },
+    {
+      "epoch": 128.65384615384616,
+      "grad_norm": 1.162660002708435,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40140
+    },
+    {
+      "epoch": 128.68589743589743,
+      "grad_norm": 1.1187268495559692,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40150
+    },
+    {
+      "epoch": 128.71794871794873,
+      "grad_norm": 1.0334343910217285,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 40160
+    },
+    {
+      "epoch": 128.75,
+      "grad_norm": 0.6769480109214783,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40170
+    },
+    {
+      "epoch": 128.78205128205127,
+      "grad_norm": 0.763494610786438,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40180
+    },
+    {
+      "epoch": 128.81410256410257,
+      "grad_norm": 0.7498688697814941,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 40190
+    },
+    {
+      "epoch": 128.84615384615384,
+      "grad_norm": 0.824165403842926,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 40200
+    },
+    {
+      "epoch": 128.87820512820514,
+      "grad_norm": 0.683300256729126,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40210
+    },
+    {
+      "epoch": 128.9102564102564,
+      "grad_norm": 0.7581266760826111,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40220
+    },
+    {
+      "epoch": 128.94230769230768,
+      "grad_norm": 1.3345515727996826,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40230
+    },
+    {
+      "epoch": 128.97435897435898,
+      "grad_norm": 1.009885311126709,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 40240
+    },
+    {
+      "epoch": 129.00641025641025,
+      "grad_norm": 0.7895881533622742,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 40250
+    },
+    {
+      "epoch": 129.03846153846155,
+      "grad_norm": 0.6133207082748413,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 40260
+    },
+    {
+      "epoch": 129.07051282051282,
+      "grad_norm": 0.8255252838134766,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 40270
+    },
+    {
+      "epoch": 129.10256410256412,
+      "grad_norm": 1.0223073959350586,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40280
+    },
+    {
+      "epoch": 129.1346153846154,
+      "grad_norm": 1.2490885257720947,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 40290
+    },
+    {
+      "epoch": 129.16666666666666,
+      "grad_norm": 1.3888219594955444,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40300
+    },
+    {
+      "epoch": 129.19871794871796,
+      "grad_norm": 0.9710935354232788,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40310
+    },
+    {
+      "epoch": 129.23076923076923,
+      "grad_norm": 0.9784860014915466,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 40320
+    },
+    {
+      "epoch": 129.26282051282053,
+      "grad_norm": 0.8571407198905945,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40330
+    },
+    {
+      "epoch": 129.2948717948718,
+      "grad_norm": 0.6442854404449463,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40340
+    },
+    {
+      "epoch": 129.32692307692307,
+      "grad_norm": 0.8334605693817139,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40350
+    },
+    {
+      "epoch": 129.35897435897436,
+      "grad_norm": 0.8638678789138794,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 40360
+    },
+    {
+      "epoch": 129.39102564102564,
+      "grad_norm": 0.8971635103225708,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40370
+    },
+    {
+      "epoch": 129.42307692307693,
+      "grad_norm": 0.8557437062263489,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 40380
+    },
+    {
+      "epoch": 129.4551282051282,
+      "grad_norm": 1.1281142234802246,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 40390
+    },
+    {
+      "epoch": 129.48717948717947,
+      "grad_norm": 1.196190595626831,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40400
+    },
+    {
+      "epoch": 129.51923076923077,
+      "grad_norm": 1.2224334478378296,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 40410
+    },
+    {
+      "epoch": 129.55128205128204,
+      "grad_norm": 0.9111282229423523,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 40420
+    },
+    {
+      "epoch": 129.58333333333334,
+      "grad_norm": 1.4658267498016357,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 40430
+    },
+    {
+      "epoch": 129.6153846153846,
+      "grad_norm": 1.2281274795532227,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 40440
+    },
+    {
+      "epoch": 129.64743589743588,
+      "grad_norm": 0.9478588104248047,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 40450
+    },
+    {
+      "epoch": 129.67948717948718,
+      "grad_norm": 0.748001754283905,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 40460
+    },
+    {
+      "epoch": 129.71153846153845,
+      "grad_norm": 0.7860796451568604,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 40470
+    },
+    {
+      "epoch": 129.74358974358975,
+      "grad_norm": 0.7841623425483704,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 40480
+    },
+    {
+      "epoch": 129.77564102564102,
+      "grad_norm": 1.135008692741394,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40490
+    },
+    {
+      "epoch": 129.80769230769232,
+      "grad_norm": 1.1300421953201294,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40500
+    },
+    {
+      "epoch": 129.8397435897436,
+      "grad_norm": 0.9486061334609985,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40510
+    },
+    {
+      "epoch": 129.87179487179486,
+      "grad_norm": 0.9468240737915039,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 40520
+    },
+    {
+      "epoch": 129.90384615384616,
+      "grad_norm": 1.0004769563674927,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 40530
+    },
+    {
+      "epoch": 129.93589743589743,
+      "grad_norm": 0.8989123106002808,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 40540
+    },
+    {
+      "epoch": 129.96794871794873,
+      "grad_norm": 0.9721859097480774,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 40550
+    },
+    {
+      "epoch": 130.0,
+      "grad_norm": 0.856829822063446,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 40560
+    },
+    {
+      "epoch": 130.03205128205127,
+      "grad_norm": 1.072182297706604,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 40570
+    },
+    {
+      "epoch": 130.06410256410257,
+      "grad_norm": 0.8518576622009277,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40580
+    },
+    {
+      "epoch": 130.09615384615384,
+      "grad_norm": 0.6961578130722046,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 40590
+    },
+    {
+      "epoch": 130.12820512820514,
+      "grad_norm": 0.920546293258667,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40600
+    },
+    {
+      "epoch": 130.1602564102564,
+      "grad_norm": 1.072540044784546,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 40610
+    },
+    {
+      "epoch": 130.19230769230768,
+      "grad_norm": 0.5170711874961853,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 40620
+    },
+    {
+      "epoch": 130.22435897435898,
+      "grad_norm": 0.8336690664291382,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 40630
+    },
+    {
+      "epoch": 130.25641025641025,
+      "grad_norm": 1.067299723625183,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40640
+    },
+    {
+      "epoch": 130.28846153846155,
+      "grad_norm": 1.1630576848983765,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 40650
+    },
+    {
+      "epoch": 130.32051282051282,
+      "grad_norm": 0.8949191570281982,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 40660
+    },
+    {
+      "epoch": 130.35256410256412,
+      "grad_norm": 0.7457284927368164,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 40670
+    },
+    {
+      "epoch": 130.3846153846154,
+      "grad_norm": 1.1882988214492798,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 40680
+    },
+    {
+      "epoch": 130.41666666666666,
+      "grad_norm": 1.624743103981018,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 40690
+    },
+    {
+      "epoch": 130.44871794871796,
+      "grad_norm": 1.4876890182495117,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 40700
+    },
+    {
+      "epoch": 130.48076923076923,
+      "grad_norm": 1.3916891813278198,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 40710
+    },
+    {
+      "epoch": 130.51282051282053,
+      "grad_norm": 1.5563504695892334,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 40720
+    },
+    {
+      "epoch": 130.5448717948718,
+      "grad_norm": 1.1379281282424927,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 40730
+    },
+    {
+      "epoch": 130.57692307692307,
+      "grad_norm": 1.143289566040039,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40740
+    },
+    {
+      "epoch": 130.60897435897436,
+      "grad_norm": 1.3009567260742188,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40750
+    },
+    {
+      "epoch": 130.64102564102564,
+      "grad_norm": 1.1561301946640015,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 40760
+    },
+    {
+      "epoch": 130.67307692307693,
+      "grad_norm": 1.053537130355835,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 40770
+    },
+    {
+      "epoch": 130.7051282051282,
+      "grad_norm": 1.0330591201782227,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 40780
+    },
+    {
+      "epoch": 130.73717948717947,
+      "grad_norm": 1.4868332147598267,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40790
+    },
+    {
+      "epoch": 130.76923076923077,
+      "grad_norm": 1.0093119144439697,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 40800
+    },
+    {
+      "epoch": 130.80128205128204,
+      "grad_norm": 0.9155437350273132,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40810
+    },
+    {
+      "epoch": 130.83333333333334,
+      "grad_norm": 1.3271325826644897,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 40820
+    },
+    {
+      "epoch": 130.8653846153846,
+      "grad_norm": 0.9319993853569031,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 40830
+    },
+    {
+      "epoch": 130.89743589743588,
+      "grad_norm": 1.2464168071746826,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 40840
+    },
+    {
+      "epoch": 130.92948717948718,
+      "grad_norm": 0.9242552518844604,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 40850
+    },
+    {
+      "epoch": 130.96153846153845,
+      "grad_norm": 0.9945878386497498,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 40860
+    },
+    {
+      "epoch": 130.99358974358975,
+      "grad_norm": 0.9030364155769348,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 40870
+    },
+    {
+      "epoch": 131.02564102564102,
+      "grad_norm": 0.7210484743118286,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 40880
+    },
+    {
+      "epoch": 131.05769230769232,
+      "grad_norm": 1.1570329666137695,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 40890
+    },
+    {
+      "epoch": 131.0897435897436,
+      "grad_norm": 1.1119487285614014,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 40900
+    },
+    {
+      "epoch": 131.12179487179486,
+      "grad_norm": 0.9954478144645691,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 40910
+    },
+    {
+      "epoch": 131.15384615384616,
+      "grad_norm": 0.9773839116096497,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 40920
+    },
+    {
+      "epoch": 131.18589743589743,
+      "grad_norm": 0.9760628938674927,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 40930
+    },
+    {
+      "epoch": 131.21794871794873,
+      "grad_norm": 0.8868107795715332,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 40940
+    },
+    {
+      "epoch": 131.25,
+      "grad_norm": 1.0387372970581055,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 40950
+    },
+    {
+      "epoch": 131.28205128205127,
+      "grad_norm": 0.9633218050003052,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 40960
+    },
+    {
+      "epoch": 131.31410256410257,
+      "grad_norm": 0.7792038321495056,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 40970
+    },
+    {
+      "epoch": 131.34615384615384,
+      "grad_norm": 1.0529659986495972,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 40980
+    },
+    {
+      "epoch": 131.37820512820514,
+      "grad_norm": 0.8471903204917908,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 40990
+    },
+    {
+      "epoch": 131.4102564102564,
+      "grad_norm": 0.9164130687713623,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 41000
+    },
+    {
+      "epoch": 131.44230769230768,
+      "grad_norm": 0.8490406274795532,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 41010
+    },
+    {
+      "epoch": 131.47435897435898,
+      "grad_norm": 0.8357127904891968,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 41020
+    },
+    {
+      "epoch": 131.50641025641025,
+      "grad_norm": 0.791365385055542,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 41030
+    },
+    {
+      "epoch": 131.53846153846155,
+      "grad_norm": 0.7629724740982056,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 41040
+    },
+    {
+      "epoch": 131.57051282051282,
+      "grad_norm": 0.8455222845077515,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 41050
+    },
+    {
+      "epoch": 131.60256410256412,
+      "grad_norm": 0.8448671698570251,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 41060
+    },
+    {
+      "epoch": 131.6346153846154,
+      "grad_norm": 0.9633859992027283,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 41070
+    },
+    {
+      "epoch": 131.66666666666666,
+      "grad_norm": 0.9244256615638733,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 41080
+    },
+    {
+      "epoch": 131.69871794871796,
+      "grad_norm": 0.9395145773887634,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 41090
+    },
+    {
+      "epoch": 131.73076923076923,
+      "grad_norm": 0.9403217434883118,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 41100
+    },
+    {
+      "epoch": 131.76282051282053,
+      "grad_norm": 0.9047693014144897,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41110
+    },
+    {
+      "epoch": 131.7948717948718,
+      "grad_norm": 0.7504073977470398,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 41120
+    },
+    {
+      "epoch": 131.82692307692307,
+      "grad_norm": 0.9209052920341492,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 41130
+    },
+    {
+      "epoch": 131.85897435897436,
+      "grad_norm": 1.1460788249969482,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 41140
+    },
+    {
+      "epoch": 131.89102564102564,
+      "grad_norm": 0.805779218673706,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 41150
+    },
+    {
+      "epoch": 131.92307692307693,
+      "grad_norm": 1.2456859350204468,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 41160
+    },
+    {
+      "epoch": 131.9551282051282,
+      "grad_norm": 1.2310653924942017,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 41170
+    },
+    {
+      "epoch": 131.98717948717947,
+      "grad_norm": 0.7187948226928711,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 41180
+    },
+    {
+      "epoch": 132.01923076923077,
+      "grad_norm": 0.9879278540611267,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 41190
+    },
+    {
+      "epoch": 132.05128205128204,
+      "grad_norm": 0.9449487924575806,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 41200
+    },
+    {
+      "epoch": 132.08333333333334,
+      "grad_norm": 0.5736438035964966,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 41210
+    },
+    {
+      "epoch": 132.1153846153846,
+      "grad_norm": 0.7247572541236877,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 41220
+    },
+    {
+      "epoch": 132.14743589743588,
+      "grad_norm": 0.9201165437698364,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 41230
+    },
+    {
+      "epoch": 132.17948717948718,
+      "grad_norm": 1.362163782119751,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 41240
+    },
+    {
+      "epoch": 132.21153846153845,
+      "grad_norm": 1.6294606924057007,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 41250
+    },
+    {
+      "epoch": 132.24358974358975,
+      "grad_norm": 1.5649573802947998,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 41260
+    },
+    {
+      "epoch": 132.27564102564102,
+      "grad_norm": 1.1449552774429321,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 41270
+    },
+    {
+      "epoch": 132.30769230769232,
+      "grad_norm": 0.8285214900970459,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 41280
+    },
+    {
+      "epoch": 132.3397435897436,
+      "grad_norm": 1.0165528059005737,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41290
+    },
+    {
+      "epoch": 132.37179487179486,
+      "grad_norm": 0.8818365931510925,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 41300
+    },
+    {
+      "epoch": 132.40384615384616,
+      "grad_norm": 0.8480932116508484,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 41310
+    },
+    {
+      "epoch": 132.43589743589743,
+      "grad_norm": 0.8920019865036011,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 41320
+    },
+    {
+      "epoch": 132.46794871794873,
+      "grad_norm": 0.6783831715583801,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 41330
+    },
+    {
+      "epoch": 132.5,
+      "grad_norm": 1.1031618118286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41340
+    },
+    {
+      "epoch": 132.53205128205127,
+      "grad_norm": 0.7046658396720886,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 41350
+    },
+    {
+      "epoch": 132.56410256410257,
+      "grad_norm": 0.9559024572372437,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41360
+    },
+    {
+      "epoch": 132.59615384615384,
+      "grad_norm": 0.9884719848632812,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 41370
+    },
+    {
+      "epoch": 132.62820512820514,
+      "grad_norm": 1.0395368337631226,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 41380
+    },
+    {
+      "epoch": 132.6602564102564,
+      "grad_norm": 1.049433708190918,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 41390
+    },
+    {
+      "epoch": 132.69230769230768,
+      "grad_norm": 1.2448092699050903,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 41400
+    },
+    {
+      "epoch": 132.72435897435898,
+      "grad_norm": 1.256861686706543,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 41410
+    },
+    {
+      "epoch": 132.75641025641025,
+      "grad_norm": 1.380922555923462,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 41420
+    },
+    {
+      "epoch": 132.78846153846155,
+      "grad_norm": 1.2527562379837036,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 41430
+    },
+    {
+      "epoch": 132.82051282051282,
+      "grad_norm": 1.1867939233779907,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 41440
+    },
+    {
+      "epoch": 132.85256410256412,
+      "grad_norm": 1.271929144859314,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 41450
+    },
+    {
+      "epoch": 132.8846153846154,
+      "grad_norm": 1.1202125549316406,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 41460
+    },
+    {
+      "epoch": 132.91666666666666,
+      "grad_norm": 1.0556081533432007,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 41470
+    },
+    {
+      "epoch": 132.94871794871796,
+      "grad_norm": 0.9942350387573242,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 41480
+    },
+    {
+      "epoch": 132.98076923076923,
+      "grad_norm": 1.3279690742492676,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 41490
+    },
+    {
+      "epoch": 133.01282051282053,
+      "grad_norm": 1.2225079536437988,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 41500
+    },
+    {
+      "epoch": 133.0448717948718,
+      "grad_norm": 1.1289117336273193,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 41510
+    },
+    {
+      "epoch": 133.07692307692307,
+      "grad_norm": 1.1678565740585327,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41520
+    },
+    {
+      "epoch": 133.10897435897436,
+      "grad_norm": 1.1180061101913452,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 41530
+    },
+    {
+      "epoch": 133.14102564102564,
+      "grad_norm": 1.4301514625549316,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 41540
+    },
+    {
+      "epoch": 133.17307692307693,
+      "grad_norm": 1.1077994108200073,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 41550
+    },
+    {
+      "epoch": 133.2051282051282,
+      "grad_norm": 1.0626121759414673,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 41560
+    },
+    {
+      "epoch": 133.23717948717947,
+      "grad_norm": 1.4238107204437256,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 41570
+    },
+    {
+      "epoch": 133.26923076923077,
+      "grad_norm": 2.0956029891967773,
+      "learning_rate": 0.0001,
+      "loss": 0.0188,
+      "step": 41580
+    },
+    {
+      "epoch": 133.30128205128204,
+      "grad_norm": 2.3587613105773926,
+      "learning_rate": 0.0001,
+      "loss": 0.0106,
+      "step": 41590
+    },
+    {
+      "epoch": 133.33333333333334,
+      "grad_norm": 1.6055891513824463,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 41600
+    },
+    {
+      "epoch": 133.3653846153846,
+      "grad_norm": 1.7625079154968262,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41610
+    },
+    {
+      "epoch": 133.39743589743588,
+      "grad_norm": 1.1708691120147705,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 41620
+    },
+    {
+      "epoch": 133.42948717948718,
+      "grad_norm": 1.4879144430160522,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 41630
+    },
+    {
+      "epoch": 133.46153846153845,
+      "grad_norm": 1.3468555212020874,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 41640
+    },
+    {
+      "epoch": 133.49358974358975,
+      "grad_norm": 1.3325177431106567,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 41650
+    },
+    {
+      "epoch": 133.52564102564102,
+      "grad_norm": 0.9995689392089844,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 41660
+    },
+    {
+      "epoch": 133.55769230769232,
+      "grad_norm": 1.038002371788025,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 41670
+    },
+    {
+      "epoch": 133.5897435897436,
+      "grad_norm": 1.1690970659255981,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 41680
+    },
+    {
+      "epoch": 133.62179487179486,
+      "grad_norm": 0.6326062083244324,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41690
+    },
+    {
+      "epoch": 133.65384615384616,
+      "grad_norm": 0.8575053215026855,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 41700
+    },
+    {
+      "epoch": 133.68589743589743,
+      "grad_norm": 0.9287829995155334,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 41710
+    },
+    {
+      "epoch": 133.71794871794873,
+      "grad_norm": 1.015012502670288,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41720
+    },
+    {
+      "epoch": 133.75,
+      "grad_norm": 0.8345351815223694,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 41730
+    },
+    {
+      "epoch": 133.78205128205127,
+      "grad_norm": 0.9694735407829285,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 41740
+    },
+    {
+      "epoch": 133.81410256410257,
+      "grad_norm": 0.9517322182655334,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41750
+    },
+    {
+      "epoch": 133.84615384615384,
+      "grad_norm": 0.941601037979126,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41760
+    },
+    {
+      "epoch": 133.87820512820514,
+      "grad_norm": 0.9894121289253235,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 41770
+    },
+    {
+      "epoch": 133.9102564102564,
+      "grad_norm": 1.350033164024353,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41780
+    },
+    {
+      "epoch": 133.94230769230768,
+      "grad_norm": 1.380984902381897,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 41790
+    },
+    {
+      "epoch": 133.97435897435898,
+      "grad_norm": 1.1026928424835205,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 41800
+    },
+    {
+      "epoch": 134.00641025641025,
+      "grad_norm": 1.2516635656356812,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 41810
+    },
+    {
+      "epoch": 134.03846153846155,
+      "grad_norm": 0.9879319667816162,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 41820
+    },
+    {
+      "epoch": 134.07051282051282,
+      "grad_norm": 1.0220760107040405,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 41830
+    },
+    {
+      "epoch": 134.10256410256412,
+      "grad_norm": 1.0225168466567993,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 41840
+    },
+    {
+      "epoch": 134.1346153846154,
+      "grad_norm": 0.6372552514076233,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 41850
+    },
+    {
+      "epoch": 134.16666666666666,
+      "grad_norm": 0.759159505367279,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 41860
+    },
+    {
+      "epoch": 134.19871794871796,
+      "grad_norm": 0.88011634349823,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 41870
+    },
+    {
+      "epoch": 134.23076923076923,
+      "grad_norm": 1.275669813156128,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 41880
+    },
+    {
+      "epoch": 134.26282051282053,
+      "grad_norm": 1.4843533039093018,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 41890
+    },
+    {
+      "epoch": 134.2948717948718,
+      "grad_norm": 1.1293914318084717,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 41900
+    },
+    {
+      "epoch": 134.32692307692307,
+      "grad_norm": 0.9685555696487427,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 41910
+    },
+    {
+      "epoch": 134.35897435897436,
+      "grad_norm": 0.9069461822509766,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 41920
+    },
+    {
+      "epoch": 134.39102564102564,
+      "grad_norm": 1.0790650844573975,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 41930
+    },
+    {
+      "epoch": 134.42307692307693,
+      "grad_norm": 1.120959758758545,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 41940
+    },
+    {
+      "epoch": 134.4551282051282,
+      "grad_norm": 0.8841703534126282,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 41950
+    },
+    {
+      "epoch": 134.48717948717947,
+      "grad_norm": 0.8294675350189209,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 41960
+    },
+    {
+      "epoch": 134.51923076923077,
+      "grad_norm": 0.7982137799263,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 41970
+    },
+    {
+      "epoch": 134.55128205128204,
+      "grad_norm": 0.7113240361213684,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 41980
+    },
+    {
+      "epoch": 134.58333333333334,
+      "grad_norm": 1.0555437803268433,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 41990
+    },
+    {
+      "epoch": 134.6153846153846,
+      "grad_norm": 0.8066710233688354,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42000
+    },
+    {
+      "epoch": 134.64743589743588,
+      "grad_norm": 0.9635194540023804,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 42010
+    },
+    {
+      "epoch": 134.67948717948718,
+      "grad_norm": 1.0641534328460693,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42020
+    },
+    {
+      "epoch": 134.71153846153845,
+      "grad_norm": 0.8662307262420654,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 42030
+    },
+    {
+      "epoch": 134.74358974358975,
+      "grad_norm": 0.7776861786842346,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42040
+    },
+    {
+      "epoch": 134.77564102564102,
+      "grad_norm": 0.9182864427566528,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42050
+    },
+    {
+      "epoch": 134.80769230769232,
+      "grad_norm": 0.8380518555641174,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 42060
+    },
+    {
+      "epoch": 134.8397435897436,
+      "grad_norm": 0.6761800646781921,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 42070
+    },
+    {
+      "epoch": 134.87179487179486,
+      "grad_norm": 0.7350406646728516,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 42080
+    },
+    {
+      "epoch": 134.90384615384616,
+      "grad_norm": 0.7375661730766296,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42090
+    },
+    {
+      "epoch": 134.93589743589743,
+      "grad_norm": 0.7939698696136475,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 42100
+    },
+    {
+      "epoch": 134.96794871794873,
+      "grad_norm": 0.8130255937576294,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42110
+    },
+    {
+      "epoch": 135.0,
+      "grad_norm": 0.7787204384803772,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42120
+    },
+    {
+      "epoch": 135.03205128205127,
+      "grad_norm": 0.8014622926712036,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42130
+    },
+    {
+      "epoch": 135.06410256410257,
+      "grad_norm": 0.8781328201293945,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 42140
+    },
+    {
+      "epoch": 135.09615384615384,
+      "grad_norm": 0.9737539291381836,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42150
+    },
+    {
+      "epoch": 135.12820512820514,
+      "grad_norm": 0.9640907049179077,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 42160
+    },
+    {
+      "epoch": 135.1602564102564,
+      "grad_norm": 1.2288602590560913,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 42170
+    },
+    {
+      "epoch": 135.19230769230768,
+      "grad_norm": 0.7526292204856873,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 42180
+    },
+    {
+      "epoch": 135.22435897435898,
+      "grad_norm": 1.1471903324127197,
+      "learning_rate": 0.0001,
+      "loss": 0.0094,
+      "step": 42190
+    },
+    {
+      "epoch": 135.25641025641025,
+      "grad_norm": 1.0224624872207642,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 42200
+    },
+    {
+      "epoch": 135.28846153846155,
+      "grad_norm": 1.1950265169143677,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 42210
+    },
+    {
+      "epoch": 135.32051282051282,
+      "grad_norm": 0.9715636968612671,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42220
+    },
+    {
+      "epoch": 135.35256410256412,
+      "grad_norm": 0.7395468354225159,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 42230
+    },
+    {
+      "epoch": 135.3846153846154,
+      "grad_norm": 0.8946119546890259,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 42240
+    },
+    {
+      "epoch": 135.41666666666666,
+      "grad_norm": 0.8142320513725281,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 42250
+    },
+    {
+      "epoch": 135.44871794871796,
+      "grad_norm": 0.8859590888023376,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 42260
+    },
+    {
+      "epoch": 135.48076923076923,
+      "grad_norm": 0.8457803130149841,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42270
+    },
+    {
+      "epoch": 135.51282051282053,
+      "grad_norm": 1.00390625,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42280
+    },
+    {
+      "epoch": 135.5448717948718,
+      "grad_norm": 0.5190619826316833,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42290
+    },
+    {
+      "epoch": 135.57692307692307,
+      "grad_norm": 0.7304899096488953,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42300
+    },
+    {
+      "epoch": 135.60897435897436,
+      "grad_norm": 0.5950448513031006,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42310
+    },
+    {
+      "epoch": 135.64102564102564,
+      "grad_norm": 0.9532377123832703,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 42320
+    },
+    {
+      "epoch": 135.67307692307693,
+      "grad_norm": 1.3117504119873047,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42330
+    },
+    {
+      "epoch": 135.7051282051282,
+      "grad_norm": 0.9900763630867004,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42340
+    },
+    {
+      "epoch": 135.73717948717947,
+      "grad_norm": 1.185456395149231,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42350
+    },
+    {
+      "epoch": 135.76923076923077,
+      "grad_norm": 0.8857272863388062,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42360
+    },
+    {
+      "epoch": 135.80128205128204,
+      "grad_norm": 1.0635133981704712,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42370
+    },
+    {
+      "epoch": 135.83333333333334,
+      "grad_norm": 0.913626492023468,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 42380
+    },
+    {
+      "epoch": 135.8653846153846,
+      "grad_norm": 1.0728349685668945,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 42390
+    },
+    {
+      "epoch": 135.89743589743588,
+      "grad_norm": 0.6236194372177124,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42400
+    },
+    {
+      "epoch": 135.92948717948718,
+      "grad_norm": 0.7910710573196411,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 42410
+    },
+    {
+      "epoch": 135.96153846153845,
+      "grad_norm": 1.085669994354248,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42420
+    },
+    {
+      "epoch": 135.99358974358975,
+      "grad_norm": 0.8769057989120483,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 42430
+    },
+    {
+      "epoch": 136.02564102564102,
+      "grad_norm": 0.9891975522041321,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 42440
+    },
+    {
+      "epoch": 136.05769230769232,
+      "grad_norm": 0.9652897119522095,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 42450
+    },
+    {
+      "epoch": 136.0897435897436,
+      "grad_norm": 1.0903688669204712,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42460
+    },
+    {
+      "epoch": 136.12179487179486,
+      "grad_norm": 0.8364027142524719,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 42470
+    },
+    {
+      "epoch": 136.15384615384616,
+      "grad_norm": 1.369253158569336,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 42480
+    },
+    {
+      "epoch": 136.18589743589743,
+      "grad_norm": 0.9096805453300476,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42490
+    },
+    {
+      "epoch": 136.21794871794873,
+      "grad_norm": 0.8152179718017578,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 42500
+    },
+    {
+      "epoch": 136.25,
+      "grad_norm": 0.7228220701217651,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42510
+    },
+    {
+      "epoch": 136.28205128205127,
+      "grad_norm": 0.6018578410148621,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42520
+    },
+    {
+      "epoch": 136.31410256410257,
+      "grad_norm": 0.7247589826583862,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42530
+    },
+    {
+      "epoch": 136.34615384615384,
+      "grad_norm": 0.8871558904647827,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 42540
+    },
+    {
+      "epoch": 136.37820512820514,
+      "grad_norm": 0.9222078919410706,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 42550
+    },
+    {
+      "epoch": 136.4102564102564,
+      "grad_norm": 0.8293497562408447,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42560
+    },
+    {
+      "epoch": 136.44230769230768,
+      "grad_norm": 0.8355336785316467,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42570
+    },
+    {
+      "epoch": 136.47435897435898,
+      "grad_norm": 1.0259640216827393,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42580
+    },
+    {
+      "epoch": 136.50641025641025,
+      "grad_norm": 0.8521203994750977,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 42590
+    },
+    {
+      "epoch": 136.53846153846155,
+      "grad_norm": 0.8805991411209106,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 42600
+    },
+    {
+      "epoch": 136.57051282051282,
+      "grad_norm": 1.0667550563812256,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42610
+    },
+    {
+      "epoch": 136.60256410256412,
+      "grad_norm": 0.7554386258125305,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42620
+    },
+    {
+      "epoch": 136.6346153846154,
+      "grad_norm": 0.9585873484611511,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 42630
+    },
+    {
+      "epoch": 136.66666666666666,
+      "grad_norm": 0.8050658702850342,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42640
+    },
+    {
+      "epoch": 136.69871794871796,
+      "grad_norm": 1.2035479545593262,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 42650
+    },
+    {
+      "epoch": 136.73076923076923,
+      "grad_norm": 1.2822036743164062,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 42660
+    },
+    {
+      "epoch": 136.76282051282053,
+      "grad_norm": 0.8207159042358398,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 42670
+    },
+    {
+      "epoch": 136.7948717948718,
+      "grad_norm": 0.8836487531661987,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42680
+    },
+    {
+      "epoch": 136.82692307692307,
+      "grad_norm": 1.09138023853302,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42690
+    },
+    {
+      "epoch": 136.85897435897436,
+      "grad_norm": 1.1367130279541016,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42700
+    },
+    {
+      "epoch": 136.89102564102564,
+      "grad_norm": 0.8651005029678345,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 42710
+    },
+    {
+      "epoch": 136.92307692307693,
+      "grad_norm": 1.0003151893615723,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42720
+    },
+    {
+      "epoch": 136.9551282051282,
+      "grad_norm": 1.1446912288665771,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 42730
+    },
+    {
+      "epoch": 136.98717948717947,
+      "grad_norm": 0.9998494982719421,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42740
+    },
+    {
+      "epoch": 137.01923076923077,
+      "grad_norm": 1.1719685792922974,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42750
+    },
+    {
+      "epoch": 137.05128205128204,
+      "grad_norm": 0.7847386598587036,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42760
+    },
+    {
+      "epoch": 137.08333333333334,
+      "grad_norm": 0.7589125037193298,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 42770
+    },
+    {
+      "epoch": 137.1153846153846,
+      "grad_norm": 0.7059719562530518,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 42780
+    },
+    {
+      "epoch": 137.14743589743588,
+      "grad_norm": 0.8783110976219177,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 42790
+    },
+    {
+      "epoch": 137.17948717948718,
+      "grad_norm": 1.0578157901763916,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42800
+    },
+    {
+      "epoch": 137.21153846153845,
+      "grad_norm": 1.0426218509674072,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42810
+    },
+    {
+      "epoch": 137.24358974358975,
+      "grad_norm": 1.1713966131210327,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42820
+    },
+    {
+      "epoch": 137.27564102564102,
+      "grad_norm": 0.9017342329025269,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 42830
+    },
+    {
+      "epoch": 137.30769230769232,
+      "grad_norm": 1.0597835779190063,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 42840
+    },
+    {
+      "epoch": 137.3397435897436,
+      "grad_norm": 1.2390542030334473,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 42850
+    },
+    {
+      "epoch": 137.37179487179486,
+      "grad_norm": 0.8812397718429565,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 42860
+    },
+    {
+      "epoch": 137.40384615384616,
+      "grad_norm": 0.9313947558403015,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42870
+    },
+    {
+      "epoch": 137.43589743589743,
+      "grad_norm": 0.8761280179023743,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 42880
+    },
+    {
+      "epoch": 137.46794871794873,
+      "grad_norm": 0.6877323985099792,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42890
+    },
+    {
+      "epoch": 137.5,
+      "grad_norm": 0.7714307904243469,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42900
+    },
+    {
+      "epoch": 137.53205128205127,
+      "grad_norm": 0.890799343585968,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42910
+    },
+    {
+      "epoch": 137.56410256410257,
+      "grad_norm": 0.8749904036521912,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 42920
+    },
+    {
+      "epoch": 137.59615384615384,
+      "grad_norm": 0.7845653295516968,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42930
+    },
+    {
+      "epoch": 137.62820512820514,
+      "grad_norm": 0.8477080464363098,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 42940
+    },
+    {
+      "epoch": 137.6602564102564,
+      "grad_norm": 0.7920386791229248,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42950
+    },
+    {
+      "epoch": 137.69230769230768,
+      "grad_norm": 1.154887080192566,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 42960
+    },
+    {
+      "epoch": 137.72435897435898,
+      "grad_norm": 1.0686864852905273,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 42970
+    },
+    {
+      "epoch": 137.75641025641025,
+      "grad_norm": 1.1323206424713135,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42980
+    },
+    {
+      "epoch": 137.78846153846155,
+      "grad_norm": 1.0869622230529785,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 42990
+    },
+    {
+      "epoch": 137.82051282051282,
+      "grad_norm": 1.1352031230926514,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 43000
+    },
+    {
+      "epoch": 137.85256410256412,
+      "grad_norm": 0.9843942523002625,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 43010
+    },
+    {
+      "epoch": 137.8846153846154,
+      "grad_norm": 0.874221920967102,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 43020
+    },
+    {
+      "epoch": 137.91666666666666,
+      "grad_norm": 1.068870186805725,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 43030
+    },
+    {
+      "epoch": 137.94871794871796,
+      "grad_norm": 1.1108791828155518,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43040
+    },
+    {
+      "epoch": 137.98076923076923,
+      "grad_norm": 1.151734709739685,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 43050
+    },
+    {
+      "epoch": 138.01282051282053,
+      "grad_norm": 1.053589105606079,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 43060
+    },
+    {
+      "epoch": 138.0448717948718,
+      "grad_norm": 0.892074704170227,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 43070
+    },
+    {
+      "epoch": 138.07692307692307,
+      "grad_norm": 1.2547091245651245,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 43080
+    },
+    {
+      "epoch": 138.10897435897436,
+      "grad_norm": 1.1482923030853271,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 43090
+    },
+    {
+      "epoch": 138.14102564102564,
+      "grad_norm": 1.3627434968948364,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 43100
+    },
+    {
+      "epoch": 138.17307692307693,
+      "grad_norm": 1.3675791025161743,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 43110
+    },
+    {
+      "epoch": 138.2051282051282,
+      "grad_norm": 1.1068394184112549,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43120
+    },
+    {
+      "epoch": 138.23717948717947,
+      "grad_norm": 1.1762914657592773,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 43130
+    },
+    {
+      "epoch": 138.26923076923077,
+      "grad_norm": 1.082706332206726,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 43140
+    },
+    {
+      "epoch": 138.30128205128204,
+      "grad_norm": 1.5856122970581055,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 43150
+    },
+    {
+      "epoch": 138.33333333333334,
+      "grad_norm": 1.2991719245910645,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 43160
+    },
+    {
+      "epoch": 138.3653846153846,
+      "grad_norm": 1.527813196182251,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 43170
+    },
+    {
+      "epoch": 138.39743589743588,
+      "grad_norm": 1.8533713817596436,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43180
+    },
+    {
+      "epoch": 138.42948717948718,
+      "grad_norm": 1.4100902080535889,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 43190
+    },
+    {
+      "epoch": 138.46153846153845,
+      "grad_norm": 1.391253113746643,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 43200
+    },
+    {
+      "epoch": 138.49358974358975,
+      "grad_norm": 1.4635891914367676,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 43210
+    },
+    {
+      "epoch": 138.52564102564102,
+      "grad_norm": 1.1834546327590942,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 43220
+    },
+    {
+      "epoch": 138.55769230769232,
+      "grad_norm": 1.1128984689712524,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 43230
+    },
+    {
+      "epoch": 138.5897435897436,
+      "grad_norm": 1.1534010171890259,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 43240
+    },
+    {
+      "epoch": 138.62179487179486,
+      "grad_norm": 1.1708475351333618,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 43250
+    },
+    {
+      "epoch": 138.65384615384616,
+      "grad_norm": 1.1696356534957886,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43260
+    },
+    {
+      "epoch": 138.68589743589743,
+      "grad_norm": 1.102606177330017,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43270
+    },
+    {
+      "epoch": 138.71794871794873,
+      "grad_norm": 1.1363025903701782,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 43280
+    },
+    {
+      "epoch": 138.75,
+      "grad_norm": 1.155537486076355,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 43290
+    },
+    {
+      "epoch": 138.78205128205127,
+      "grad_norm": 0.9189514517784119,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 43300
+    },
+    {
+      "epoch": 138.81410256410257,
+      "grad_norm": 1.113067388534546,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43310
+    },
+    {
+      "epoch": 138.84615384615384,
+      "grad_norm": 0.9292428493499756,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 43320
+    },
+    {
+      "epoch": 138.87820512820514,
+      "grad_norm": 1.0205073356628418,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 43330
+    },
+    {
+      "epoch": 138.9102564102564,
+      "grad_norm": 0.8594955801963806,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 43340
+    },
+    {
+      "epoch": 138.94230769230768,
+      "grad_norm": 0.867431104183197,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 43350
+    },
+    {
+      "epoch": 138.97435897435898,
+      "grad_norm": 1.1726086139678955,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 43360
+    },
+    {
+      "epoch": 139.00641025641025,
+      "grad_norm": 1.485252022743225,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 43370
+    },
+    {
+      "epoch": 139.03846153846155,
+      "grad_norm": 1.3244273662567139,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 43380
+    },
+    {
+      "epoch": 139.07051282051282,
+      "grad_norm": 1.079722285270691,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 43390
+    },
+    {
+      "epoch": 139.10256410256412,
+      "grad_norm": 1.0714560747146606,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 43400
+    },
+    {
+      "epoch": 139.1346153846154,
+      "grad_norm": 1.087631106376648,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43410
+    },
+    {
+      "epoch": 139.16666666666666,
+      "grad_norm": 1.2032290697097778,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 43420
+    },
+    {
+      "epoch": 139.19871794871796,
+      "grad_norm": 0.9391984343528748,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 43430
+    },
+    {
+      "epoch": 139.23076923076923,
+      "grad_norm": 0.903366208076477,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 43440
+    },
+    {
+      "epoch": 139.26282051282053,
+      "grad_norm": 1.2079994678497314,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43450
+    },
+    {
+      "epoch": 139.2948717948718,
+      "grad_norm": 0.8530546426773071,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 43460
+    },
+    {
+      "epoch": 139.32692307692307,
+      "grad_norm": 1.0882861614227295,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 43470
+    },
+    {
+      "epoch": 139.35897435897436,
+      "grad_norm": 1.3810807466506958,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 43480
+    },
+    {
+      "epoch": 139.39102564102564,
+      "grad_norm": 1.0789598226547241,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 43490
+    },
+    {
+      "epoch": 139.42307692307693,
+      "grad_norm": 1.0022602081298828,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 43500
+    },
+    {
+      "epoch": 139.4551282051282,
+      "grad_norm": 0.6634025573730469,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43510
+    },
+    {
+      "epoch": 139.48717948717947,
+      "grad_norm": 1.2340095043182373,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43520
+    },
+    {
+      "epoch": 139.51923076923077,
+      "grad_norm": 1.3220624923706055,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 43530
+    },
+    {
+      "epoch": 139.55128205128204,
+      "grad_norm": 1.0687506198883057,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 43540
+    },
+    {
+      "epoch": 139.58333333333334,
+      "grad_norm": 1.0636520385742188,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43550
+    },
+    {
+      "epoch": 139.6153846153846,
+      "grad_norm": 1.1297968626022339,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 43560
+    },
+    {
+      "epoch": 139.64743589743588,
+      "grad_norm": 1.0599294900894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43570
+    },
+    {
+      "epoch": 139.67948717948718,
+      "grad_norm": 1.1052438020706177,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 43580
+    },
+    {
+      "epoch": 139.71153846153845,
+      "grad_norm": 1.1554138660430908,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43590
+    },
+    {
+      "epoch": 139.74358974358975,
+      "grad_norm": 1.2115942239761353,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43600
+    },
+    {
+      "epoch": 139.77564102564102,
+      "grad_norm": 0.9647664427757263,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43610
+    },
+    {
+      "epoch": 139.80769230769232,
+      "grad_norm": 0.753227949142456,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 43620
+    },
+    {
+      "epoch": 139.8397435897436,
+      "grad_norm": 0.6989542841911316,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43630
+    },
+    {
+      "epoch": 139.87179487179486,
+      "grad_norm": 1.113453984260559,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43640
+    },
+    {
+      "epoch": 139.90384615384616,
+      "grad_norm": 1.056209683418274,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 43650
+    },
+    {
+      "epoch": 139.93589743589743,
+      "grad_norm": 1.0730665922164917,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43660
+    },
+    {
+      "epoch": 139.96794871794873,
+      "grad_norm": 1.017750859260559,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43670
+    },
+    {
+      "epoch": 140.0,
+      "grad_norm": 1.0378674268722534,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 43680
+    },
+    {
+      "epoch": 140.03205128205127,
+      "grad_norm": 0.7793622016906738,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43690
+    },
+    {
+      "epoch": 140.06410256410257,
+      "grad_norm": 1.0494550466537476,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43700
+    },
+    {
+      "epoch": 140.09615384615384,
+      "grad_norm": 0.9376341700553894,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 43710
+    },
+    {
+      "epoch": 140.12820512820514,
+      "grad_norm": 0.9389193058013916,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43720
+    },
+    {
+      "epoch": 140.1602564102564,
+      "grad_norm": 0.8216260671615601,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43730
+    },
+    {
+      "epoch": 140.19230769230768,
+      "grad_norm": 0.912428081035614,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43740
+    },
+    {
+      "epoch": 140.22435897435898,
+      "grad_norm": 1.2668297290802002,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 43750
+    },
+    {
+      "epoch": 140.25641025641025,
+      "grad_norm": 1.1128742694854736,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 43760
+    },
+    {
+      "epoch": 140.28846153846155,
+      "grad_norm": 0.9027514457702637,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43770
+    },
+    {
+      "epoch": 140.32051282051282,
+      "grad_norm": 0.7449844479560852,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43780
+    },
+    {
+      "epoch": 140.35256410256412,
+      "grad_norm": 0.7598353028297424,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 43790
+    },
+    {
+      "epoch": 140.3846153846154,
+      "grad_norm": 1.0876195430755615,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 43800
+    },
+    {
+      "epoch": 140.41666666666666,
+      "grad_norm": 0.9069765210151672,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 43810
+    },
+    {
+      "epoch": 140.44871794871796,
+      "grad_norm": 1.0712124109268188,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43820
+    },
+    {
+      "epoch": 140.48076923076923,
+      "grad_norm": 1.258585810661316,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43830
+    },
+    {
+      "epoch": 140.51282051282053,
+      "grad_norm": 0.7951087355613708,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 43840
+    },
+    {
+      "epoch": 140.5448717948718,
+      "grad_norm": 0.8801838755607605,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43850
+    },
+    {
+      "epoch": 140.57692307692307,
+      "grad_norm": 0.9070515036582947,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 43860
+    },
+    {
+      "epoch": 140.60897435897436,
+      "grad_norm": 0.8086984753608704,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 43870
+    },
+    {
+      "epoch": 140.64102564102564,
+      "grad_norm": 0.7418122291564941,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43880
+    },
+    {
+      "epoch": 140.67307692307693,
+      "grad_norm": 0.7873529195785522,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43890
+    },
+    {
+      "epoch": 140.7051282051282,
+      "grad_norm": 0.6638549566268921,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 43900
+    },
+    {
+      "epoch": 140.73717948717947,
+      "grad_norm": 0.6931739449501038,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 43910
+    },
+    {
+      "epoch": 140.76923076923077,
+      "grad_norm": 0.9474515914916992,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 43920
+    },
+    {
+      "epoch": 140.80128205128204,
+      "grad_norm": 0.7679346799850464,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43930
+    },
+    {
+      "epoch": 140.83333333333334,
+      "grad_norm": 1.0330655574798584,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 43940
+    },
+    {
+      "epoch": 140.8653846153846,
+      "grad_norm": 0.9159137606620789,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 43950
+    },
+    {
+      "epoch": 140.89743589743588,
+      "grad_norm": 0.6723729372024536,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 43960
+    },
+    {
+      "epoch": 140.92948717948718,
+      "grad_norm": 0.702799379825592,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43970
+    },
+    {
+      "epoch": 140.96153846153845,
+      "grad_norm": 0.8168069124221802,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 43980
+    },
+    {
+      "epoch": 140.99358974358975,
+      "grad_norm": 0.7648693919181824,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 43990
+    },
+    {
+      "epoch": 141.02564102564102,
+      "grad_norm": 1.0240288972854614,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 44000
+    },
+    {
+      "epoch": 141.05769230769232,
+      "grad_norm": 1.261512279510498,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 44010
+    },
+    {
+      "epoch": 141.0897435897436,
+      "grad_norm": 1.022743582725525,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44020
+    },
+    {
+      "epoch": 141.12179487179486,
+      "grad_norm": 0.6150705218315125,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 44030
+    },
+    {
+      "epoch": 141.15384615384616,
+      "grad_norm": 0.9560233950614929,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 44040
+    },
+    {
+      "epoch": 141.18589743589743,
+      "grad_norm": 0.9558258056640625,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 44050
+    },
+    {
+      "epoch": 141.21794871794873,
+      "grad_norm": 0.862571656703949,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 44060
+    },
+    {
+      "epoch": 141.25,
+      "grad_norm": 0.9906817674636841,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 44070
+    },
+    {
+      "epoch": 141.28205128205127,
+      "grad_norm": 0.7192110419273376,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 44080
+    },
+    {
+      "epoch": 141.31410256410257,
+      "grad_norm": 0.9688276648521423,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44090
+    },
+    {
+      "epoch": 141.34615384615384,
+      "grad_norm": 1.0947387218475342,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 44100
+    },
+    {
+      "epoch": 141.37820512820514,
+      "grad_norm": 0.6638801097869873,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 44110
+    },
+    {
+      "epoch": 141.4102564102564,
+      "grad_norm": 1.1665483713150024,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 44120
+    },
+    {
+      "epoch": 141.44230769230768,
+      "grad_norm": 1.10096275806427,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 44130
+    },
+    {
+      "epoch": 141.47435897435898,
+      "grad_norm": 0.7263680696487427,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 44140
+    },
+    {
+      "epoch": 141.50641025641025,
+      "grad_norm": 0.8854547142982483,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 44150
+    },
+    {
+      "epoch": 141.53846153846155,
+      "grad_norm": 0.8457266092300415,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 44160
+    },
+    {
+      "epoch": 141.57051282051282,
+      "grad_norm": 0.9345922470092773,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 44170
+    },
+    {
+      "epoch": 141.60256410256412,
+      "grad_norm": 0.9163205027580261,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44180
+    },
+    {
+      "epoch": 141.6346153846154,
+      "grad_norm": 1.0762957334518433,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44190
+    },
+    {
+      "epoch": 141.66666666666666,
+      "grad_norm": 0.9425115585327148,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 44200
+    },
+    {
+      "epoch": 141.69871794871796,
+      "grad_norm": 0.6525254249572754,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 44210
+    },
+    {
+      "epoch": 141.73076923076923,
+      "grad_norm": 1.1534725427627563,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 44220
+    },
+    {
+      "epoch": 141.76282051282053,
+      "grad_norm": 1.189234733581543,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 44230
+    },
+    {
+      "epoch": 141.7948717948718,
+      "grad_norm": 0.6665402054786682,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 44240
+    },
+    {
+      "epoch": 141.82692307692307,
+      "grad_norm": 0.9670297503471375,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 44250
+    },
+    {
+      "epoch": 141.85897435897436,
+      "grad_norm": 0.6199517846107483,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 44260
+    },
+    {
+      "epoch": 141.89102564102564,
+      "grad_norm": 0.802922248840332,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 44270
+    },
+    {
+      "epoch": 141.92307692307693,
+      "grad_norm": 0.8995755910873413,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 44280
+    },
+    {
+      "epoch": 141.9551282051282,
+      "grad_norm": 0.950206995010376,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 44290
+    },
+    {
+      "epoch": 141.98717948717947,
+      "grad_norm": 0.950037956237793,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 44300
+    },
+    {
+      "epoch": 142.01923076923077,
+      "grad_norm": 1.1843194961547852,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 44310
+    },
+    {
+      "epoch": 142.05128205128204,
+      "grad_norm": 1.0125387907028198,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 44320
+    },
+    {
+      "epoch": 142.08333333333334,
+      "grad_norm": 0.8039528131484985,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 44330
+    },
+    {
+      "epoch": 142.1153846153846,
+      "grad_norm": 0.9118345379829407,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 44340
+    },
+    {
+      "epoch": 142.14743589743588,
+      "grad_norm": 0.876322865486145,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 44350
+    },
+    {
+      "epoch": 142.17948717948718,
+      "grad_norm": 1.31216299533844,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 44360
+    },
+    {
+      "epoch": 142.21153846153845,
+      "grad_norm": 0.9139251708984375,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 44370
+    },
+    {
+      "epoch": 142.24358974358975,
+      "grad_norm": 0.6392374634742737,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 44380
+    },
+    {
+      "epoch": 142.27564102564102,
+      "grad_norm": 0.746773362159729,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 44390
+    },
+    {
+      "epoch": 142.30769230769232,
+      "grad_norm": 0.6974177360534668,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 44400
+    },
+    {
+      "epoch": 142.3397435897436,
+      "grad_norm": 0.856221079826355,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 44410
+    },
+    {
+      "epoch": 142.37179487179486,
+      "grad_norm": 0.6639910936355591,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 44420
+    },
+    {
+      "epoch": 142.40384615384616,
+      "grad_norm": 0.9185786843299866,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 44430
+    },
+    {
+      "epoch": 142.43589743589743,
+      "grad_norm": 0.8004684448242188,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 44440
+    },
+    {
+      "epoch": 142.46794871794873,
+      "grad_norm": 0.9241353273391724,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44450
+    },
+    {
+      "epoch": 142.5,
+      "grad_norm": 1.0590401887893677,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 44460
+    },
+    {
+      "epoch": 142.53205128205127,
+      "grad_norm": 1.0719493627548218,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44470
+    },
+    {
+      "epoch": 142.56410256410257,
+      "grad_norm": 1.012537956237793,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 44480
+    },
+    {
+      "epoch": 142.59615384615384,
+      "grad_norm": 0.660118043422699,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 44490
+    },
+    {
+      "epoch": 142.62820512820514,
+      "grad_norm": 0.7194827795028687,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44500
+    },
+    {
+      "epoch": 142.6602564102564,
+      "grad_norm": 0.6530608534812927,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 44510
+    },
+    {
+      "epoch": 142.69230769230768,
+      "grad_norm": 0.7802638411521912,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 44520
+    },
+    {
+      "epoch": 142.72435897435898,
+      "grad_norm": 0.9352722764015198,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 44530
+    },
+    {
+      "epoch": 142.75641025641025,
+      "grad_norm": 1.086318850517273,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 44540
+    },
+    {
+      "epoch": 142.78846153846155,
+      "grad_norm": 0.812022864818573,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 44550
+    },
+    {
+      "epoch": 142.82051282051282,
+      "grad_norm": 1.1601678133010864,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 44560
+    },
+    {
+      "epoch": 142.85256410256412,
+      "grad_norm": 0.7120773196220398,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 44570
+    },
+    {
+      "epoch": 142.8846153846154,
+      "grad_norm": 0.6015300154685974,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44580
+    },
+    {
+      "epoch": 142.91666666666666,
+      "grad_norm": 0.7413464188575745,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 44590
+    },
+    {
+      "epoch": 142.94871794871796,
+      "grad_norm": 0.6816237568855286,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 44600
+    },
+    {
+      "epoch": 142.98076923076923,
+      "grad_norm": 1.0420820713043213,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44610
+    },
+    {
+      "epoch": 143.01282051282053,
+      "grad_norm": 1.020647406578064,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44620
+    },
+    {
+      "epoch": 143.0448717948718,
+      "grad_norm": 0.7657058238983154,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 44630
+    },
+    {
+      "epoch": 143.07692307692307,
+      "grad_norm": 0.6070588827133179,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 44640
+    },
+    {
+      "epoch": 143.10897435897436,
+      "grad_norm": 0.7369689345359802,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44650
+    },
+    {
+      "epoch": 143.14102564102564,
+      "grad_norm": 0.7953053116798401,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44660
+    },
+    {
+      "epoch": 143.17307692307693,
+      "grad_norm": 0.7730791568756104,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 44670
+    },
+    {
+      "epoch": 143.2051282051282,
+      "grad_norm": 0.8095161318778992,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 44680
+    },
+    {
+      "epoch": 143.23717948717947,
+      "grad_norm": 0.8016958236694336,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 44690
+    },
+    {
+      "epoch": 143.26923076923077,
+      "grad_norm": 0.8540694713592529,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 44700
+    },
+    {
+      "epoch": 143.30128205128204,
+      "grad_norm": 0.7871220707893372,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 44710
+    },
+    {
+      "epoch": 143.33333333333334,
+      "grad_norm": 0.9141536951065063,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 44720
+    },
+    {
+      "epoch": 143.3653846153846,
+      "grad_norm": 0.7675531506538391,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 44730
+    },
+    {
+      "epoch": 143.39743589743588,
+      "grad_norm": 0.8091365694999695,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44740
+    },
+    {
+      "epoch": 143.42948717948718,
+      "grad_norm": 0.7266249060630798,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 44750
+    },
+    {
+      "epoch": 143.46153846153845,
+      "grad_norm": 0.6762871146202087,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 44760
+    },
+    {
+      "epoch": 143.49358974358975,
+      "grad_norm": 0.839911937713623,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 44770
+    },
+    {
+      "epoch": 143.52564102564102,
+      "grad_norm": 0.607191801071167,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 44780
+    },
+    {
+      "epoch": 143.55769230769232,
+      "grad_norm": 0.6751165986061096,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44790
+    },
+    {
+      "epoch": 143.5897435897436,
+      "grad_norm": 0.6884068250656128,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 44800
+    },
+    {
+      "epoch": 143.62179487179486,
+      "grad_norm": 0.8884255886077881,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 44810
+    },
+    {
+      "epoch": 143.65384615384616,
+      "grad_norm": 0.9773091673851013,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44820
+    },
+    {
+      "epoch": 143.68589743589743,
+      "grad_norm": 0.81741863489151,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44830
+    },
+    {
+      "epoch": 143.71794871794873,
+      "grad_norm": 0.7952264547348022,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 44840
+    },
+    {
+      "epoch": 143.75,
+      "grad_norm": 0.9762158393859863,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44850
+    },
+    {
+      "epoch": 143.78205128205127,
+      "grad_norm": 1.1221883296966553,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44860
+    },
+    {
+      "epoch": 143.81410256410257,
+      "grad_norm": 0.9141285419464111,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44870
+    },
+    {
+      "epoch": 143.84615384615384,
+      "grad_norm": 0.926311194896698,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 44880
+    },
+    {
+      "epoch": 143.87820512820514,
+      "grad_norm": 1.2420856952667236,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 44890
+    },
+    {
+      "epoch": 143.9102564102564,
+      "grad_norm": 1.4038039445877075,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 44900
+    },
+    {
+      "epoch": 143.94230769230768,
+      "grad_norm": 1.2024857997894287,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 44910
+    },
+    {
+      "epoch": 143.97435897435898,
+      "grad_norm": 1.1092438697814941,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 44920
+    },
+    {
+      "epoch": 144.00641025641025,
+      "grad_norm": 0.978106677532196,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 44930
+    },
+    {
+      "epoch": 144.03846153846155,
+      "grad_norm": 0.6517308354377747,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 44940
+    },
+    {
+      "epoch": 144.07051282051282,
+      "grad_norm": 0.5723773837089539,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 44950
+    },
+    {
+      "epoch": 144.10256410256412,
+      "grad_norm": 0.7317081689834595,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 44960
+    },
+    {
+      "epoch": 144.1346153846154,
+      "grad_norm": 1.205437183380127,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 44970
+    },
+    {
+      "epoch": 144.16666666666666,
+      "grad_norm": 1.2358423471450806,
+      "learning_rate": 0.0001,
+      "loss": 0.0095,
+      "step": 44980
+    },
+    {
+      "epoch": 144.19871794871796,
+      "grad_norm": 1.6438366174697876,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 44990
+    },
+    {
+      "epoch": 144.23076923076923,
+      "grad_norm": 1.5016038417816162,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45000
+    },
+    {
+      "epoch": 144.26282051282053,
+      "grad_norm": 1.5467901229858398,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45010
+    },
+    {
+      "epoch": 144.2948717948718,
+      "grad_norm": 1.2579247951507568,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45020
+    },
+    {
+      "epoch": 144.32692307692307,
+      "grad_norm": 1.477423071861267,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 45030
+    },
+    {
+      "epoch": 144.35897435897436,
+      "grad_norm": 1.2982001304626465,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 45040
+    },
+    {
+      "epoch": 144.39102564102564,
+      "grad_norm": 1.1610627174377441,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 45050
+    },
+    {
+      "epoch": 144.42307692307693,
+      "grad_norm": 1.1585185527801514,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45060
+    },
+    {
+      "epoch": 144.4551282051282,
+      "grad_norm": 1.0189965963363647,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 45070
+    },
+    {
+      "epoch": 144.48717948717947,
+      "grad_norm": 0.9954744577407837,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45080
+    },
+    {
+      "epoch": 144.51923076923077,
+      "grad_norm": 1.1030213832855225,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 45090
+    },
+    {
+      "epoch": 144.55128205128204,
+      "grad_norm": 1.0491254329681396,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 45100
+    },
+    {
+      "epoch": 144.58333333333334,
+      "grad_norm": 1.0965327024459839,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 45110
+    },
+    {
+      "epoch": 144.6153846153846,
+      "grad_norm": 1.2375671863555908,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 45120
+    },
+    {
+      "epoch": 144.64743589743588,
+      "grad_norm": 1.1305793523788452,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 45130
+    },
+    {
+      "epoch": 144.67948717948718,
+      "grad_norm": 1.1302865743637085,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 45140
+    },
+    {
+      "epoch": 144.71153846153845,
+      "grad_norm": 1.1252975463867188,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 45150
+    },
+    {
+      "epoch": 144.74358974358975,
+      "grad_norm": 0.7494484782218933,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 45160
+    },
+    {
+      "epoch": 144.77564102564102,
+      "grad_norm": 0.8330286145210266,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 45170
+    },
+    {
+      "epoch": 144.80769230769232,
+      "grad_norm": 0.9373483061790466,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 45180
+    },
+    {
+      "epoch": 144.8397435897436,
+      "grad_norm": 0.7865563035011292,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 45190
+    },
+    {
+      "epoch": 144.87179487179486,
+      "grad_norm": 0.8090651631355286,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 45200
+    },
+    {
+      "epoch": 144.90384615384616,
+      "grad_norm": 0.7767390608787537,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 45210
+    },
+    {
+      "epoch": 144.93589743589743,
+      "grad_norm": 0.6229214072227478,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45220
+    },
+    {
+      "epoch": 144.96794871794873,
+      "grad_norm": 1.0506739616394043,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45230
+    },
+    {
+      "epoch": 145.0,
+      "grad_norm": 0.854518711566925,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 45240
+    },
+    {
+      "epoch": 145.03205128205127,
+      "grad_norm": 0.7110360860824585,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45250
+    },
+    {
+      "epoch": 145.06410256410257,
+      "grad_norm": 0.6398422122001648,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 45260
+    },
+    {
+      "epoch": 145.09615384615384,
+      "grad_norm": 1.4473176002502441,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 45270
+    },
+    {
+      "epoch": 145.12820512820514,
+      "grad_norm": 1.5904476642608643,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 45280
+    },
+    {
+      "epoch": 145.1602564102564,
+      "grad_norm": 0.9127330780029297,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45290
+    },
+    {
+      "epoch": 145.19230769230768,
+      "grad_norm": 1.3553242683410645,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45300
+    },
+    {
+      "epoch": 145.22435897435898,
+      "grad_norm": 1.5794918537139893,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45310
+    },
+    {
+      "epoch": 145.25641025641025,
+      "grad_norm": 1.3089487552642822,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 45320
+    },
+    {
+      "epoch": 145.28846153846155,
+      "grad_norm": 1.3219020366668701,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 45330
+    },
+    {
+      "epoch": 145.32051282051282,
+      "grad_norm": 1.1172137260437012,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 45340
+    },
+    {
+      "epoch": 145.35256410256412,
+      "grad_norm": 1.0678706169128418,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 45350
+    },
+    {
+      "epoch": 145.3846153846154,
+      "grad_norm": 0.9602237343788147,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 45360
+    },
+    {
+      "epoch": 145.41666666666666,
+      "grad_norm": 0.9311193823814392,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45370
+    },
+    {
+      "epoch": 145.44871794871796,
+      "grad_norm": 0.8220816850662231,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45380
+    },
+    {
+      "epoch": 145.48076923076923,
+      "grad_norm": 0.9445400834083557,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 45390
+    },
+    {
+      "epoch": 145.51282051282053,
+      "grad_norm": 0.9373456239700317,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 45400
+    },
+    {
+      "epoch": 145.5448717948718,
+      "grad_norm": 1.1345022916793823,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 45410
+    },
+    {
+      "epoch": 145.57692307692307,
+      "grad_norm": 0.9854171872138977,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 45420
+    },
+    {
+      "epoch": 145.60897435897436,
+      "grad_norm": 1.0584802627563477,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 45430
+    },
+    {
+      "epoch": 145.64102564102564,
+      "grad_norm": 0.8641334176063538,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 45440
+    },
+    {
+      "epoch": 145.67307692307693,
+      "grad_norm": 1.133070468902588,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 45450
+    },
+    {
+      "epoch": 145.7051282051282,
+      "grad_norm": 0.9145009517669678,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45460
+    },
+    {
+      "epoch": 145.73717948717947,
+      "grad_norm": 1.0075539350509644,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45470
+    },
+    {
+      "epoch": 145.76923076923077,
+      "grad_norm": 0.7573880553245544,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45480
+    },
+    {
+      "epoch": 145.80128205128204,
+      "grad_norm": 0.8627190589904785,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 45490
+    },
+    {
+      "epoch": 145.83333333333334,
+      "grad_norm": 0.7084478735923767,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45500
+    },
+    {
+      "epoch": 145.8653846153846,
+      "grad_norm": 0.6968368291854858,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 45510
+    },
+    {
+      "epoch": 145.89743589743588,
+      "grad_norm": 0.7420106530189514,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 45520
+    },
+    {
+      "epoch": 145.92948717948718,
+      "grad_norm": 0.8722419738769531,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 45530
+    },
+    {
+      "epoch": 145.96153846153845,
+      "grad_norm": 1.0667341947555542,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 45540
+    },
+    {
+      "epoch": 145.99358974358975,
+      "grad_norm": 1.0108622312545776,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 45550
+    },
+    {
+      "epoch": 146.02564102564102,
+      "grad_norm": 0.9840543270111084,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45560
+    },
+    {
+      "epoch": 146.05769230769232,
+      "grad_norm": 0.8659612536430359,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45570
+    },
+    {
+      "epoch": 146.0897435897436,
+      "grad_norm": 0.8842259645462036,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45580
+    },
+    {
+      "epoch": 146.12179487179486,
+      "grad_norm": 1.0284441709518433,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45590
+    },
+    {
+      "epoch": 146.15384615384616,
+      "grad_norm": 0.8382779359817505,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45600
+    },
+    {
+      "epoch": 146.18589743589743,
+      "grad_norm": 0.9199274182319641,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 45610
+    },
+    {
+      "epoch": 146.21794871794873,
+      "grad_norm": 0.7010545134544373,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 45620
+    },
+    {
+      "epoch": 146.25,
+      "grad_norm": 0.8449309468269348,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 45630
+    },
+    {
+      "epoch": 146.28205128205127,
+      "grad_norm": 0.760442316532135,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 45640
+    },
+    {
+      "epoch": 146.31410256410257,
+      "grad_norm": 0.6987426280975342,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 45650
+    },
+    {
+      "epoch": 146.34615384615384,
+      "grad_norm": 0.5824467539787292,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 45660
+    },
+    {
+      "epoch": 146.37820512820514,
+      "grad_norm": 0.825019896030426,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 45670
+    },
+    {
+      "epoch": 146.4102564102564,
+      "grad_norm": 0.8736175298690796,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45680
+    },
+    {
+      "epoch": 146.44230769230768,
+      "grad_norm": 0.9683285355567932,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45690
+    },
+    {
+      "epoch": 146.47435897435898,
+      "grad_norm": 1.2891815900802612,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 45700
+    },
+    {
+      "epoch": 146.50641025641025,
+      "grad_norm": 1.0590016841888428,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 45710
+    },
+    {
+      "epoch": 146.53846153846155,
+      "grad_norm": 1.1108191013336182,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 45720
+    },
+    {
+      "epoch": 146.57051282051282,
+      "grad_norm": 0.9952238202095032,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 45730
+    },
+    {
+      "epoch": 146.60256410256412,
+      "grad_norm": 0.9382213950157166,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 45740
+    },
+    {
+      "epoch": 146.6346153846154,
+      "grad_norm": 0.8028233647346497,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 45750
+    },
+    {
+      "epoch": 146.66666666666666,
+      "grad_norm": 0.9772260785102844,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 45760
+    },
+    {
+      "epoch": 146.69871794871796,
+      "grad_norm": 0.9609761834144592,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45770
+    },
+    {
+      "epoch": 146.73076923076923,
+      "grad_norm": 0.95916748046875,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 45780
+    },
+    {
+      "epoch": 146.76282051282053,
+      "grad_norm": 1.2672356367111206,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 45790
+    },
+    {
+      "epoch": 146.7948717948718,
+      "grad_norm": 0.8898100852966309,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 45800
+    },
+    {
+      "epoch": 146.82692307692307,
+      "grad_norm": 0.9769667983055115,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 45810
+    },
+    {
+      "epoch": 146.85897435897436,
+      "grad_norm": 0.8366612195968628,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 45820
+    },
+    {
+      "epoch": 146.89102564102564,
+      "grad_norm": 1.2350268363952637,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 45830
+    },
+    {
+      "epoch": 146.92307692307693,
+      "grad_norm": 1.016572117805481,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 45840
+    },
+    {
+      "epoch": 146.9551282051282,
+      "grad_norm": 1.03544020652771,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 45850
+    },
+    {
+      "epoch": 146.98717948717947,
+      "grad_norm": 1.0459411144256592,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 45860
+    },
+    {
+      "epoch": 147.01923076923077,
+      "grad_norm": 0.943755030632019,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45870
+    },
+    {
+      "epoch": 147.05128205128204,
+      "grad_norm": 0.7563897967338562,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45880
+    },
+    {
+      "epoch": 147.08333333333334,
+      "grad_norm": 0.8732236623764038,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 45890
+    },
+    {
+      "epoch": 147.1153846153846,
+      "grad_norm": 0.8686173558235168,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 45900
+    },
+    {
+      "epoch": 147.14743589743588,
+      "grad_norm": 1.0728427171707153,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 45910
+    },
+    {
+      "epoch": 147.17948717948718,
+      "grad_norm": 0.9658101797103882,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 45920
+    },
+    {
+      "epoch": 147.21153846153845,
+      "grad_norm": 0.8797629475593567,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 45930
+    },
+    {
+      "epoch": 147.24358974358975,
+      "grad_norm": 1.089215636253357,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 45940
+    },
+    {
+      "epoch": 147.27564102564102,
+      "grad_norm": 0.6864970922470093,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 45950
+    },
+    {
+      "epoch": 147.30769230769232,
+      "grad_norm": 0.978845477104187,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 45960
+    },
+    {
+      "epoch": 147.3397435897436,
+      "grad_norm": 1.2689096927642822,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 45970
+    },
+    {
+      "epoch": 147.37179487179486,
+      "grad_norm": 0.8068153858184814,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 45980
+    },
+    {
+      "epoch": 147.40384615384616,
+      "grad_norm": 0.6760982871055603,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 45990
+    },
+    {
+      "epoch": 147.43589743589743,
+      "grad_norm": 1.0522366762161255,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46000
+    },
+    {
+      "epoch": 147.46794871794873,
+      "grad_norm": 0.8133437633514404,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 46010
+    },
+    {
+      "epoch": 147.5,
+      "grad_norm": 0.8401362299919128,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 46020
+    },
+    {
+      "epoch": 147.53205128205127,
+      "grad_norm": 0.7101181745529175,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 46030
+    },
+    {
+      "epoch": 147.56410256410257,
+      "grad_norm": 0.9902019500732422,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 46040
+    },
+    {
+      "epoch": 147.59615384615384,
+      "grad_norm": 0.7551693320274353,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 46050
+    },
+    {
+      "epoch": 147.62820512820514,
+      "grad_norm": 0.9657802581787109,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 46060
+    },
+    {
+      "epoch": 147.6602564102564,
+      "grad_norm": 0.781644880771637,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 46070
+    },
+    {
+      "epoch": 147.69230769230768,
+      "grad_norm": 0.640335202217102,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46080
+    },
+    {
+      "epoch": 147.72435897435898,
+      "grad_norm": 0.7604745626449585,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46090
+    },
+    {
+      "epoch": 147.75641025641025,
+      "grad_norm": 0.7049381732940674,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46100
+    },
+    {
+      "epoch": 147.78846153846155,
+      "grad_norm": 0.8013370633125305,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46110
+    },
+    {
+      "epoch": 147.82051282051282,
+      "grad_norm": 0.7244305610656738,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 46120
+    },
+    {
+      "epoch": 147.85256410256412,
+      "grad_norm": 0.8501131534576416,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 46130
+    },
+    {
+      "epoch": 147.8846153846154,
+      "grad_norm": 0.9117534756660461,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 46140
+    },
+    {
+      "epoch": 147.91666666666666,
+      "grad_norm": 0.9732885360717773,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 46150
+    },
+    {
+      "epoch": 147.94871794871796,
+      "grad_norm": 0.9408954977989197,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 46160
+    },
+    {
+      "epoch": 147.98076923076923,
+      "grad_norm": 0.913078784942627,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 46170
+    },
+    {
+      "epoch": 148.01282051282053,
+      "grad_norm": 0.8693295121192932,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46180
+    },
+    {
+      "epoch": 148.0448717948718,
+      "grad_norm": 0.9157660603523254,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 46190
+    },
+    {
+      "epoch": 148.07692307692307,
+      "grad_norm": 0.7011609077453613,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46200
+    },
+    {
+      "epoch": 148.10897435897436,
+      "grad_norm": 0.8579110503196716,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 46210
+    },
+    {
+      "epoch": 148.14102564102564,
+      "grad_norm": 0.730010986328125,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 46220
+    },
+    {
+      "epoch": 148.17307692307693,
+      "grad_norm": 0.8527998924255371,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 46230
+    },
+    {
+      "epoch": 148.2051282051282,
+      "grad_norm": 0.8844266533851624,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 46240
+    },
+    {
+      "epoch": 148.23717948717947,
+      "grad_norm": 1.30916428565979,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 46250
+    },
+    {
+      "epoch": 148.26923076923077,
+      "grad_norm": 0.7742210030555725,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 46260
+    },
+    {
+      "epoch": 148.30128205128204,
+      "grad_norm": 0.9443020820617676,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 46270
+    },
+    {
+      "epoch": 148.33333333333334,
+      "grad_norm": 0.8221555948257446,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 46280
+    },
+    {
+      "epoch": 148.3653846153846,
+      "grad_norm": 0.7556942701339722,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46290
+    },
+    {
+      "epoch": 148.39743589743588,
+      "grad_norm": 0.8888771533966064,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46300
+    },
+    {
+      "epoch": 148.42948717948718,
+      "grad_norm": 1.0420702695846558,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 46310
+    },
+    {
+      "epoch": 148.46153846153845,
+      "grad_norm": 0.8979423642158508,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 46320
+    },
+    {
+      "epoch": 148.49358974358975,
+      "grad_norm": 0.725291907787323,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 46330
+    },
+    {
+      "epoch": 148.52564102564102,
+      "grad_norm": 0.9726351499557495,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 46340
+    },
+    {
+      "epoch": 148.55769230769232,
+      "grad_norm": 1.00815749168396,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46350
+    },
+    {
+      "epoch": 148.5897435897436,
+      "grad_norm": 0.6373475790023804,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 46360
+    },
+    {
+      "epoch": 148.62179487179486,
+      "grad_norm": 0.7770876884460449,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46370
+    },
+    {
+      "epoch": 148.65384615384616,
+      "grad_norm": 0.885504961013794,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 46380
+    },
+    {
+      "epoch": 148.68589743589743,
+      "grad_norm": 0.6662806868553162,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46390
+    },
+    {
+      "epoch": 148.71794871794873,
+      "grad_norm": 0.7769169807434082,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 46400
+    },
+    {
+      "epoch": 148.75,
+      "grad_norm": 0.8605049252510071,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 46410
+    },
+    {
+      "epoch": 148.78205128205127,
+      "grad_norm": 0.8930333256721497,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 46420
+    },
+    {
+      "epoch": 148.81410256410257,
+      "grad_norm": 1.0542104244232178,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46430
+    },
+    {
+      "epoch": 148.84615384615384,
+      "grad_norm": 1.5255388021469116,
+      "learning_rate": 0.0001,
+      "loss": 0.009,
+      "step": 46440
+    },
+    {
+      "epoch": 148.87820512820514,
+      "grad_norm": 0.9063251614570618,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 46450
+    },
+    {
+      "epoch": 148.9102564102564,
+      "grad_norm": 0.8934366106987,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 46460
+    },
+    {
+      "epoch": 148.94230769230768,
+      "grad_norm": 1.4431812763214111,
+      "learning_rate": 0.0001,
+      "loss": 0.0091,
+      "step": 46470
+    },
+    {
+      "epoch": 148.97435897435898,
+      "grad_norm": 0.7444995641708374,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 46480
+    },
+    {
+      "epoch": 149.00641025641025,
+      "grad_norm": 0.8338679075241089,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 46490
+    },
+    {
+      "epoch": 149.03846153846155,
+      "grad_norm": 0.706529438495636,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 46500
+    },
+    {
+      "epoch": 149.07051282051282,
+      "grad_norm": 0.8024114966392517,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 46510
+    },
+    {
+      "epoch": 149.10256410256412,
+      "grad_norm": 1.032045841217041,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 46520
+    },
+    {
+      "epoch": 149.1346153846154,
+      "grad_norm": 0.7354134917259216,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 46530
+    },
+    {
+      "epoch": 149.16666666666666,
+      "grad_norm": 1.8741520643234253,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 46540
+    },
+    {
+      "epoch": 149.19871794871796,
+      "grad_norm": 1.8733103275299072,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 46550
+    },
+    {
+      "epoch": 149.23076923076923,
+      "grad_norm": 1.3654581308364868,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 46560
+    },
+    {
+      "epoch": 149.26282051282053,
+      "grad_norm": 1.6897554397583008,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 46570
+    },
+    {
+      "epoch": 149.2948717948718,
+      "grad_norm": 1.889331579208374,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 46580
+    },
+    {
+      "epoch": 149.32692307692307,
+      "grad_norm": 1.639663815498352,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 46590
+    },
+    {
+      "epoch": 149.35897435897436,
+      "grad_norm": 1.2622140645980835,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46600
+    },
+    {
+      "epoch": 149.39102564102564,
+      "grad_norm": 1.232759952545166,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "step": 46610
+    },
+    {
+      "epoch": 149.42307692307693,
+      "grad_norm": 1.1964030265808105,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 46620
+    },
+    {
+      "epoch": 149.4551282051282,
+      "grad_norm": 0.9396184682846069,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 46630
+    },
+    {
+      "epoch": 149.48717948717947,
+      "grad_norm": 0.990642249584198,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 46640
+    },
+    {
+      "epoch": 149.51923076923077,
+      "grad_norm": 1.0254772901535034,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 46650
+    },
+    {
+      "epoch": 149.55128205128204,
+      "grad_norm": 1.2773929834365845,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 46660
+    },
+    {
+      "epoch": 149.58333333333334,
+      "grad_norm": 0.9351746439933777,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 46670
+    },
+    {
+      "epoch": 149.6153846153846,
+      "grad_norm": 1.105401873588562,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 46680
+    },
+    {
+      "epoch": 149.64743589743588,
+      "grad_norm": 1.3094066381454468,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 46690
+    },
+    {
+      "epoch": 149.67948717948718,
+      "grad_norm": 0.997434139251709,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 46700
+    },
+    {
+      "epoch": 149.71153846153845,
+      "grad_norm": 0.8869137763977051,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 46710
+    },
+    {
+      "epoch": 149.74358974358975,
+      "grad_norm": 1.0171715021133423,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 46720
+    },
+    {
+      "epoch": 149.77564102564102,
+      "grad_norm": 0.9343420267105103,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 46730
+    },
+    {
+      "epoch": 149.80769230769232,
+      "grad_norm": 0.7901569604873657,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 46740
+    },
+    {
+      "epoch": 149.8397435897436,
+      "grad_norm": 0.9107661247253418,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 46750
+    },
+    {
+      "epoch": 149.87179487179486,
+      "grad_norm": 0.681766927242279,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 46760
+    },
+    {
+      "epoch": 149.90384615384616,
+      "grad_norm": 1.0007917881011963,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 46770
+    },
+    {
+      "epoch": 149.93589743589743,
+      "grad_norm": 0.9534011483192444,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 46780
+    },
+    {
+      "epoch": 149.96794871794873,
+      "grad_norm": 0.9521602392196655,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 46790
+    },
+    {
+      "epoch": 150.0,
+      "grad_norm": 1.0354636907577515,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 46800
+    },
+    {
+      "epoch": 150.03205128205127,
+      "grad_norm": 0.9464621543884277,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 46810
+    },
+    {
+      "epoch": 150.06410256410257,
+      "grad_norm": 0.9633666276931763,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 46820
+    },
+    {
+      "epoch": 150.09615384615384,
+      "grad_norm": 0.8897047638893127,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 46830
+    },
+    {
+      "epoch": 150.12820512820514,
+      "grad_norm": 0.7847090363502502,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 46840
+    },
+    {
+      "epoch": 150.1602564102564,
+      "grad_norm": 1.0076582431793213,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 46850
+    },
+    {
+      "epoch": 150.19230769230768,
+      "grad_norm": 0.9405493140220642,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46860
+    },
+    {
+      "epoch": 150.22435897435898,
+      "grad_norm": 0.8695074319839478,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46870
+    },
+    {
+      "epoch": 150.25641025641025,
+      "grad_norm": 1.0922536849975586,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46880
+    },
+    {
+      "epoch": 150.28846153846155,
+      "grad_norm": 0.8828272223472595,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 46890
+    },
+    {
+      "epoch": 150.32051282051282,
+      "grad_norm": 0.6613286733627319,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 46900
+    },
+    {
+      "epoch": 150.35256410256412,
+      "grad_norm": 0.6270273923873901,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 46910
+    },
+    {
+      "epoch": 150.3846153846154,
+      "grad_norm": 1.1519887447357178,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 46920
+    },
+    {
+      "epoch": 150.41666666666666,
+      "grad_norm": 1.0025699138641357,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 46930
+    },
+    {
+      "epoch": 150.44871794871796,
+      "grad_norm": 0.8382835388183594,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 46940
+    },
+    {
+      "epoch": 150.48076923076923,
+      "grad_norm": 1.4224618673324585,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 46950
+    },
+    {
+      "epoch": 150.51282051282053,
+      "grad_norm": 0.9295186996459961,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 46960
+    },
+    {
+      "epoch": 150.5448717948718,
+      "grad_norm": 1.0782382488250732,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46970
+    },
+    {
+      "epoch": 150.57692307692307,
+      "grad_norm": 0.8972647190093994,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 46980
+    },
+    {
+      "epoch": 150.60897435897436,
+      "grad_norm": 0.7982335090637207,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 46990
+    },
+    {
+      "epoch": 150.64102564102564,
+      "grad_norm": 1.2209959030151367,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 47000
+    },
+    {
+      "epoch": 150.67307692307693,
+      "grad_norm": 1.3147231340408325,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 47010
+    },
+    {
+      "epoch": 150.7051282051282,
+      "grad_norm": 1.0695289373397827,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47020
+    },
+    {
+      "epoch": 150.73717948717947,
+      "grad_norm": 1.2934048175811768,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47030
+    },
+    {
+      "epoch": 150.76923076923077,
+      "grad_norm": 1.5192657709121704,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 47040
+    },
+    {
+      "epoch": 150.80128205128204,
+      "grad_norm": 1.1308826208114624,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 47050
+    },
+    {
+      "epoch": 150.83333333333334,
+      "grad_norm": 1.5678565502166748,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47060
+    },
+    {
+      "epoch": 150.8653846153846,
+      "grad_norm": 1.319311261177063,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47070
+    },
+    {
+      "epoch": 150.89743589743588,
+      "grad_norm": 1.2850375175476074,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47080
+    },
+    {
+      "epoch": 150.92948717948718,
+      "grad_norm": 1.2641966342926025,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 47090
+    },
+    {
+      "epoch": 150.96153846153845,
+      "grad_norm": 1.092020869255066,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47100
+    },
+    {
+      "epoch": 150.99358974358975,
+      "grad_norm": 1.4318528175354004,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 47110
+    },
+    {
+      "epoch": 151.02564102564102,
+      "grad_norm": 1.159729242324829,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 47120
+    },
+    {
+      "epoch": 151.05769230769232,
+      "grad_norm": 1.0711417198181152,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47130
+    },
+    {
+      "epoch": 151.0897435897436,
+      "grad_norm": 1.0353615283966064,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 47140
+    },
+    {
+      "epoch": 151.12179487179486,
+      "grad_norm": 1.1433595418930054,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47150
+    },
+    {
+      "epoch": 151.15384615384616,
+      "grad_norm": 0.8709270358085632,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 47160
+    },
+    {
+      "epoch": 151.18589743589743,
+      "grad_norm": 0.9820587635040283,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47170
+    },
+    {
+      "epoch": 151.21794871794873,
+      "grad_norm": 0.9030230045318604,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 47180
+    },
+    {
+      "epoch": 151.25,
+      "grad_norm": 0.9259054660797119,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 47190
+    },
+    {
+      "epoch": 151.28205128205127,
+      "grad_norm": 0.9228596091270447,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 47200
+    },
+    {
+      "epoch": 151.31410256410257,
+      "grad_norm": 0.9815551042556763,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 47210
+    },
+    {
+      "epoch": 151.34615384615384,
+      "grad_norm": 1.1890475749969482,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 47220
+    },
+    {
+      "epoch": 151.37820512820514,
+      "grad_norm": 1.0556182861328125,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 47230
+    },
+    {
+      "epoch": 151.4102564102564,
+      "grad_norm": 0.899682879447937,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 47240
+    },
+    {
+      "epoch": 151.44230769230768,
+      "grad_norm": 0.9307608604431152,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 47250
+    },
+    {
+      "epoch": 151.47435897435898,
+      "grad_norm": 1.0605065822601318,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 47260
+    },
+    {
+      "epoch": 151.50641025641025,
+      "grad_norm": 0.6683368682861328,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47270
+    },
+    {
+      "epoch": 151.53846153846155,
+      "grad_norm": 0.7156051993370056,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 47280
+    },
+    {
+      "epoch": 151.57051282051282,
+      "grad_norm": 0.8192053437232971,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47290
+    },
+    {
+      "epoch": 151.60256410256412,
+      "grad_norm": 1.1849528551101685,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47300
+    },
+    {
+      "epoch": 151.6346153846154,
+      "grad_norm": 0.8787204623222351,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47310
+    },
+    {
+      "epoch": 151.66666666666666,
+      "grad_norm": 1.1756372451782227,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 47320
+    },
+    {
+      "epoch": 151.69871794871796,
+      "grad_norm": 0.7784785628318787,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 47330
+    },
+    {
+      "epoch": 151.73076923076923,
+      "grad_norm": 0.7999380230903625,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 47340
+    },
+    {
+      "epoch": 151.76282051282053,
+      "grad_norm": 1.0176254510879517,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 47350
+    },
+    {
+      "epoch": 151.7948717948718,
+      "grad_norm": 0.7134835720062256,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 47360
+    },
+    {
+      "epoch": 151.82692307692307,
+      "grad_norm": 0.7577230334281921,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 47370
+    },
+    {
+      "epoch": 151.85897435897436,
+      "grad_norm": 0.9539762735366821,
+      "learning_rate": 0.0001,
+      "loss": 0.007,
+      "step": 47380
+    },
+    {
+      "epoch": 151.89102564102564,
+      "grad_norm": 0.9607837200164795,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 47390
+    },
+    {
+      "epoch": 151.92307692307693,
+      "grad_norm": 0.606456995010376,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 47400
+    },
+    {
+      "epoch": 151.9551282051282,
+      "grad_norm": 1.4166818857192993,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47410
+    },
+    {
+      "epoch": 151.98717948717947,
+      "grad_norm": 0.9065709114074707,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 47420
+    },
+    {
+      "epoch": 152.01923076923077,
+      "grad_norm": 1.158111333847046,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 47430
+    },
+    {
+      "epoch": 152.05128205128204,
+      "grad_norm": 0.8633808493614197,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47440
+    },
+    {
+      "epoch": 152.08333333333334,
+      "grad_norm": 0.8967487812042236,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 47450
+    },
+    {
+      "epoch": 152.1153846153846,
+      "grad_norm": 1.1710387468338013,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47460
+    },
+    {
+      "epoch": 152.14743589743588,
+      "grad_norm": 1.1671667098999023,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47470
+    },
+    {
+      "epoch": 152.17948717948718,
+      "grad_norm": 0.909729540348053,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 47480
+    },
+    {
+      "epoch": 152.21153846153845,
+      "grad_norm": 1.2072278261184692,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 47490
+    },
+    {
+      "epoch": 152.24358974358975,
+      "grad_norm": 0.8716070652008057,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47500
+    },
+    {
+      "epoch": 152.27564102564102,
+      "grad_norm": 0.7001426219940186,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 47510
+    },
+    {
+      "epoch": 152.30769230769232,
+      "grad_norm": 0.7354575991630554,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47520
+    },
+    {
+      "epoch": 152.3397435897436,
+      "grad_norm": 1.0287494659423828,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47530
+    },
+    {
+      "epoch": 152.37179487179486,
+      "grad_norm": 1.019085168838501,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47540
+    },
+    {
+      "epoch": 152.40384615384616,
+      "grad_norm": 0.8369966745376587,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 47550
+    },
+    {
+      "epoch": 152.43589743589743,
+      "grad_norm": 0.9254984259605408,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 47560
+    },
+    {
+      "epoch": 152.46794871794873,
+      "grad_norm": 1.1485049724578857,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 47570
+    },
+    {
+      "epoch": 152.5,
+      "grad_norm": 1.246665120124817,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 47580
+    },
+    {
+      "epoch": 152.53205128205127,
+      "grad_norm": 1.1362632513046265,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 47590
+    },
+    {
+      "epoch": 152.56410256410257,
+      "grad_norm": 0.9765452742576599,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 47600
+    },
+    {
+      "epoch": 152.59615384615384,
+      "grad_norm": 1.0879384279251099,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 47610
+    },
+    {
+      "epoch": 152.62820512820514,
+      "grad_norm": 1.0315918922424316,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 47620
+    },
+    {
+      "epoch": 152.6602564102564,
+      "grad_norm": 0.9997173547744751,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47630
+    },
+    {
+      "epoch": 152.69230769230768,
+      "grad_norm": 1.2826389074325562,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47640
+    },
+    {
+      "epoch": 152.72435897435898,
+      "grad_norm": 0.9209719896316528,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47650
+    },
+    {
+      "epoch": 152.75641025641025,
+      "grad_norm": 0.8933044672012329,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 47660
+    },
+    {
+      "epoch": 152.78846153846155,
+      "grad_norm": 0.6572012305259705,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 47670
+    },
+    {
+      "epoch": 152.82051282051282,
+      "grad_norm": 0.7742297649383545,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47680
+    },
+    {
+      "epoch": 152.85256410256412,
+      "grad_norm": 0.7514411807060242,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 47690
+    },
+    {
+      "epoch": 152.8846153846154,
+      "grad_norm": 1.2676972150802612,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47700
+    },
+    {
+      "epoch": 152.91666666666666,
+      "grad_norm": 1.1973443031311035,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 47710
+    },
+    {
+      "epoch": 152.94871794871796,
+      "grad_norm": 1.285274863243103,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 47720
+    },
+    {
+      "epoch": 152.98076923076923,
+      "grad_norm": 0.8339930772781372,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47730
+    },
+    {
+      "epoch": 153.01282051282053,
+      "grad_norm": 1.2155197858810425,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 47740
+    },
+    {
+      "epoch": 153.0448717948718,
+      "grad_norm": 0.9439241886138916,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47750
+    },
+    {
+      "epoch": 153.07692307692307,
+      "grad_norm": 0.7250413298606873,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 47760
+    },
+    {
+      "epoch": 153.10897435897436,
+      "grad_norm": 0.6534104943275452,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47770
+    },
+    {
+      "epoch": 153.14102564102564,
+      "grad_norm": 0.6207537651062012,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 47780
+    },
+    {
+      "epoch": 153.17307692307693,
+      "grad_norm": 0.6967524290084839,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 47790
+    },
+    {
+      "epoch": 153.2051282051282,
+      "grad_norm": 0.5221480131149292,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47800
+    },
+    {
+      "epoch": 153.23717948717947,
+      "grad_norm": 0.761562168598175,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 47810
+    },
+    {
+      "epoch": 153.26923076923077,
+      "grad_norm": 0.6834405064582825,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 47820
+    },
+    {
+      "epoch": 153.30128205128204,
+      "grad_norm": 0.7461129426956177,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 47830
+    },
+    {
+      "epoch": 153.33333333333334,
+      "grad_norm": 0.598677933216095,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47840
+    },
+    {
+      "epoch": 153.3653846153846,
+      "grad_norm": 0.7064099907875061,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47850
+    },
+    {
+      "epoch": 153.39743589743588,
+      "grad_norm": 0.8404161930084229,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47860
+    },
+    {
+      "epoch": 153.42948717948718,
+      "grad_norm": 0.8819819092750549,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 47870
+    },
+    {
+      "epoch": 153.46153846153845,
+      "grad_norm": 0.8372803330421448,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47880
+    },
+    {
+      "epoch": 153.49358974358975,
+      "grad_norm": 0.8907155394554138,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47890
+    },
+    {
+      "epoch": 153.52564102564102,
+      "grad_norm": 1.188107967376709,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 47900
+    },
+    {
+      "epoch": 153.55769230769232,
+      "grad_norm": 1.4800262451171875,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 47910
+    },
+    {
+      "epoch": 153.5897435897436,
+      "grad_norm": 1.133283257484436,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47920
+    },
+    {
+      "epoch": 153.62179487179486,
+      "grad_norm": 1.0127747058868408,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 47930
+    },
+    {
+      "epoch": 153.65384615384616,
+      "grad_norm": 0.9422915577888489,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 47940
+    },
+    {
+      "epoch": 153.68589743589743,
+      "grad_norm": 1.6243470907211304,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47950
+    },
+    {
+      "epoch": 153.71794871794873,
+      "grad_norm": 1.177925705909729,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 47960
+    },
+    {
+      "epoch": 153.75,
+      "grad_norm": 1.064049482345581,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 47970
+    },
+    {
+      "epoch": 153.78205128205127,
+      "grad_norm": 1.2048500776290894,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 47980
+    },
+    {
+      "epoch": 153.81410256410257,
+      "grad_norm": 1.2040858268737793,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 47990
+    },
+    {
+      "epoch": 153.84615384615384,
+      "grad_norm": 1.3593003749847412,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 48000
+    },
+    {
+      "epoch": 153.87820512820514,
+      "grad_norm": 1.3997352123260498,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48010
+    },
+    {
+      "epoch": 153.9102564102564,
+      "grad_norm": 1.263550877571106,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 48020
+    },
+    {
+      "epoch": 153.94230769230768,
+      "grad_norm": 1.4323903322219849,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 48030
+    },
+    {
+      "epoch": 153.97435897435898,
+      "grad_norm": 1.0601181983947754,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 48040
+    },
+    {
+      "epoch": 154.00641025641025,
+      "grad_norm": 1.0240659713745117,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 48050
+    },
+    {
+      "epoch": 154.03846153846155,
+      "grad_norm": 1.243104338645935,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 48060
+    },
+    {
+      "epoch": 154.07051282051282,
+      "grad_norm": 1.107500672340393,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 48070
+    },
+    {
+      "epoch": 154.10256410256412,
+      "grad_norm": 0.89823317527771,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48080
+    },
+    {
+      "epoch": 154.1346153846154,
+      "grad_norm": 1.0978033542633057,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 48090
+    },
+    {
+      "epoch": 154.16666666666666,
+      "grad_norm": 1.1517369747161865,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 48100
+    },
+    {
+      "epoch": 154.19871794871796,
+      "grad_norm": 1.0661201477050781,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 48110
+    },
+    {
+      "epoch": 154.23076923076923,
+      "grad_norm": 0.8934845924377441,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 48120
+    },
+    {
+      "epoch": 154.26282051282053,
+      "grad_norm": 0.7364760637283325,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48130
+    },
+    {
+      "epoch": 154.2948717948718,
+      "grad_norm": 0.6615127921104431,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 48140
+    },
+    {
+      "epoch": 154.32692307692307,
+      "grad_norm": 0.7504852414131165,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 48150
+    },
+    {
+      "epoch": 154.35897435897436,
+      "grad_norm": 0.8223520517349243,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 48160
+    },
+    {
+      "epoch": 154.39102564102564,
+      "grad_norm": 1.2340352535247803,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 48170
+    },
+    {
+      "epoch": 154.42307692307693,
+      "grad_norm": 1.2998707294464111,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 48180
+    },
+    {
+      "epoch": 154.4551282051282,
+      "grad_norm": 0.947438657283783,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 48190
+    },
+    {
+      "epoch": 154.48717948717947,
+      "grad_norm": 0.7468838095664978,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 48200
+    },
+    {
+      "epoch": 154.51923076923077,
+      "grad_norm": 0.6706416010856628,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 48210
+    },
+    {
+      "epoch": 154.55128205128204,
+      "grad_norm": 1.091734528541565,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 48220
+    },
+    {
+      "epoch": 154.58333333333334,
+      "grad_norm": 1.283678650856018,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 48230
+    },
+    {
+      "epoch": 154.6153846153846,
+      "grad_norm": 1.1095494031906128,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 48240
+    },
+    {
+      "epoch": 154.64743589743588,
+      "grad_norm": 1.1335346698760986,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48250
+    },
+    {
+      "epoch": 154.67948717948718,
+      "grad_norm": 0.8227550387382507,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 48260
+    },
+    {
+      "epoch": 154.71153846153845,
+      "grad_norm": 0.9483147263526917,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 48270
+    },
+    {
+      "epoch": 154.74358974358975,
+      "grad_norm": 1.041998267173767,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48280
+    },
+    {
+      "epoch": 154.77564102564102,
+      "grad_norm": 1.2038744688034058,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 48290
+    },
+    {
+      "epoch": 154.80769230769232,
+      "grad_norm": 1.0476438999176025,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 48300
+    },
+    {
+      "epoch": 154.8397435897436,
+      "grad_norm": 0.9968962669372559,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 48310
+    },
+    {
+      "epoch": 154.87179487179486,
+      "grad_norm": 0.7208284735679626,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48320
+    },
+    {
+      "epoch": 154.90384615384616,
+      "grad_norm": 0.8328607082366943,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 48330
+    },
+    {
+      "epoch": 154.93589743589743,
+      "grad_norm": 1.1293331384658813,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 48340
+    },
+    {
+      "epoch": 154.96794871794873,
+      "grad_norm": 1.1062344312667847,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 48350
+    },
+    {
+      "epoch": 155.0,
+      "grad_norm": 0.8958812952041626,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48360
+    },
+    {
+      "epoch": 155.03205128205127,
+      "grad_norm": 0.6923726201057434,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 48370
+    },
+    {
+      "epoch": 155.06410256410257,
+      "grad_norm": 0.9175081253051758,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 48380
+    },
+    {
+      "epoch": 155.09615384615384,
+      "grad_norm": 0.9226924777030945,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 48390
+    },
+    {
+      "epoch": 155.12820512820514,
+      "grad_norm": 0.8707476854324341,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 48400
+    },
+    {
+      "epoch": 155.1602564102564,
+      "grad_norm": 0.8992027640342712,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 48410
+    },
+    {
+      "epoch": 155.19230769230768,
+      "grad_norm": 1.340314507484436,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 48420
+    },
+    {
+      "epoch": 155.22435897435898,
+      "grad_norm": 1.2011003494262695,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 48430
+    },
+    {
+      "epoch": 155.25641025641025,
+      "grad_norm": 0.9529276490211487,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 48440
+    },
+    {
+      "epoch": 155.28846153846155,
+      "grad_norm": 0.7297769784927368,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 48450
+    },
+    {
+      "epoch": 155.32051282051282,
+      "grad_norm": 0.8507109880447388,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 48460
+    },
+    {
+      "epoch": 155.35256410256412,
+      "grad_norm": 0.9919142723083496,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 48470
+    },
+    {
+      "epoch": 155.3846153846154,
+      "grad_norm": 0.8465596437454224,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 48480
+    },
+    {
+      "epoch": 155.41666666666666,
+      "grad_norm": 0.9474320411682129,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 48490
+    },
+    {
+      "epoch": 155.44871794871796,
+      "grad_norm": 0.7192931771278381,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 48500
+    },
+    {
+      "epoch": 155.48076923076923,
+      "grad_norm": 0.773494303226471,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 48510
+    },
+    {
+      "epoch": 155.51282051282053,
+      "grad_norm": 0.7796310186386108,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 48520
+    },
+    {
+      "epoch": 155.5448717948718,
+      "grad_norm": 0.8573026657104492,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 48530
+    },
+    {
+      "epoch": 155.57692307692307,
+      "grad_norm": 1.0261263847351074,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 48540
+    },
+    {
+      "epoch": 155.60897435897436,
+      "grad_norm": 0.8825958371162415,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48550
+    },
+    {
+      "epoch": 155.64102564102564,
+      "grad_norm": 1.082521915435791,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 48560
+    },
+    {
+      "epoch": 155.67307692307693,
+      "grad_norm": 1.1025274991989136,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 48570
+    },
+    {
+      "epoch": 155.7051282051282,
+      "grad_norm": 0.9585487246513367,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 48580
+    },
+    {
+      "epoch": 155.73717948717947,
+      "grad_norm": 0.9526180028915405,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 48590
+    },
+    {
+      "epoch": 155.76923076923077,
+      "grad_norm": 1.0881729125976562,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 48600
+    },
+    {
+      "epoch": 155.80128205128204,
+      "grad_norm": 0.8071385025978088,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 48610
+    },
+    {
+      "epoch": 155.83333333333334,
+      "grad_norm": 0.7068547606468201,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 48620
+    },
+    {
+      "epoch": 155.8653846153846,
+      "grad_norm": 0.9189907312393188,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 48630
+    },
+    {
+      "epoch": 155.89743589743588,
+      "grad_norm": 0.8681409358978271,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 48640
+    },
+    {
+      "epoch": 155.92948717948718,
+      "grad_norm": 1.3995455503463745,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 48650
+    },
+    {
+      "epoch": 155.96153846153845,
+      "grad_norm": 1.074598789215088,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 48660
+    },
+    {
+      "epoch": 155.99358974358975,
+      "grad_norm": 1.0807838439941406,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 48670
+    },
+    {
+      "epoch": 156.02564102564102,
+      "grad_norm": 1.0217137336730957,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 48680
+    },
+    {
+      "epoch": 156.05769230769232,
+      "grad_norm": 1.0150425434112549,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 48690
+    },
+    {
+      "epoch": 156.0897435897436,
+      "grad_norm": 1.0464472770690918,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 48700
+    },
+    {
+      "epoch": 156.12179487179486,
+      "grad_norm": 1.000022530555725,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 48710
+    },
+    {
+      "epoch": 156.15384615384616,
+      "grad_norm": 1.0737718343734741,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 48720
+    },
+    {
+      "epoch": 156.18589743589743,
+      "grad_norm": 0.9800658226013184,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 48730
+    },
+    {
+      "epoch": 156.21794871794873,
+      "grad_norm": 0.9285019040107727,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 48740
+    },
+    {
+      "epoch": 156.25,
+      "grad_norm": 0.9777775406837463,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 48750
+    },
+    {
+      "epoch": 156.28205128205127,
+      "grad_norm": 1.1482830047607422,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 48760
+    },
+    {
+      "epoch": 156.31410256410257,
+      "grad_norm": 1.2843098640441895,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 48770
+    },
+    {
+      "epoch": 156.34615384615384,
+      "grad_norm": 1.0358413457870483,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 48780
+    },
+    {
+      "epoch": 156.37820512820514,
+      "grad_norm": 1.3906159400939941,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 48790
+    },
+    {
+      "epoch": 156.4102564102564,
+      "grad_norm": 1.098139762878418,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 48800
+    },
+    {
+      "epoch": 156.44230769230768,
+      "grad_norm": 0.9058622717857361,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 48810
+    },
+    {
+      "epoch": 156.47435897435898,
+      "grad_norm": 1.505102515220642,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 48820
+    },
+    {
+      "epoch": 156.50641025641025,
+      "grad_norm": 1.0528188943862915,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 48830
+    },
+    {
+      "epoch": 156.53846153846155,
+      "grad_norm": 1.0600413084030151,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 48840
+    },
+    {
+      "epoch": 156.57051282051282,
+      "grad_norm": 0.9212043881416321,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 48850
+    },
+    {
+      "epoch": 156.60256410256412,
+      "grad_norm": 0.7134129405021667,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 48860
+    },
+    {
+      "epoch": 156.6346153846154,
+      "grad_norm": 0.8041413426399231,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 48870
+    },
+    {
+      "epoch": 156.66666666666666,
+      "grad_norm": 1.1428470611572266,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48880
+    },
+    {
+      "epoch": 156.69871794871796,
+      "grad_norm": 1.0702173709869385,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 48890
+    },
+    {
+      "epoch": 156.73076923076923,
+      "grad_norm": 0.9946589469909668,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 48900
+    },
+    {
+      "epoch": 156.76282051282053,
+      "grad_norm": 1.0584266185760498,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 48910
+    },
+    {
+      "epoch": 156.7948717948718,
+      "grad_norm": 0.9086068272590637,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 48920
+    },
+    {
+      "epoch": 156.82692307692307,
+      "grad_norm": 0.9068453907966614,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 48930
+    },
+    {
+      "epoch": 156.85897435897436,
+      "grad_norm": 1.4011365175247192,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 48940
+    },
+    {
+      "epoch": 156.89102564102564,
+      "grad_norm": 0.9249535202980042,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 48950
+    },
+    {
+      "epoch": 156.92307692307693,
+      "grad_norm": 1.0591628551483154,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 48960
+    },
+    {
+      "epoch": 156.9551282051282,
+      "grad_norm": 1.3762058019638062,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 48970
+    },
+    {
+      "epoch": 156.98717948717947,
+      "grad_norm": 1.192929744720459,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 48980
+    },
+    {
+      "epoch": 157.01923076923077,
+      "grad_norm": 1.4308019876480103,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 48990
+    },
+    {
+      "epoch": 157.05128205128204,
+      "grad_norm": 1.108561396598816,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 49000
+    },
+    {
+      "epoch": 157.08333333333334,
+      "grad_norm": 1.6896909475326538,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 49010
+    },
+    {
+      "epoch": 157.1153846153846,
+      "grad_norm": 1.559343934059143,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 49020
+    },
+    {
+      "epoch": 157.14743589743588,
+      "grad_norm": 1.046356439590454,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 49030
+    },
+    {
+      "epoch": 157.17948717948718,
+      "grad_norm": 1.0212172269821167,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49040
+    },
+    {
+      "epoch": 157.21153846153845,
+      "grad_norm": 1.0496689081192017,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49050
+    },
+    {
+      "epoch": 157.24358974358975,
+      "grad_norm": 1.0572686195373535,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 49060
+    },
+    {
+      "epoch": 157.27564102564102,
+      "grad_norm": 0.8760495781898499,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 49070
+    },
+    {
+      "epoch": 157.30769230769232,
+      "grad_norm": 0.8597815632820129,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 49080
+    },
+    {
+      "epoch": 157.3397435897436,
+      "grad_norm": 0.7977476716041565,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 49090
+    },
+    {
+      "epoch": 157.37179487179486,
+      "grad_norm": 0.8298639059066772,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 49100
+    },
+    {
+      "epoch": 157.40384615384616,
+      "grad_norm": 0.9703036546707153,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 49110
+    },
+    {
+      "epoch": 157.43589743589743,
+      "grad_norm": 0.8020877242088318,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49120
+    },
+    {
+      "epoch": 157.46794871794873,
+      "grad_norm": 1.0551992654800415,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 49130
+    },
+    {
+      "epoch": 157.5,
+      "grad_norm": 1.7279304265975952,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 49140
+    },
+    {
+      "epoch": 157.53205128205127,
+      "grad_norm": 1.3456729650497437,
+      "learning_rate": 0.0001,
+      "loss": 0.0092,
+      "step": 49150
+    },
+    {
+      "epoch": 157.56410256410257,
+      "grad_norm": 1.1825851202011108,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 49160
+    },
+    {
+      "epoch": 157.59615384615384,
+      "grad_norm": 1.4219955205917358,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 49170
+    },
+    {
+      "epoch": 157.62820512820514,
+      "grad_norm": 0.886629581451416,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 49180
+    },
+    {
+      "epoch": 157.6602564102564,
+      "grad_norm": 1.224158763885498,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 49190
+    },
+    {
+      "epoch": 157.69230769230768,
+      "grad_norm": 1.5118920803070068,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49200
+    },
+    {
+      "epoch": 157.72435897435898,
+      "grad_norm": 0.7693598866462708,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 49210
+    },
+    {
+      "epoch": 157.75641025641025,
+      "grad_norm": 0.852527379989624,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49220
+    },
+    {
+      "epoch": 157.78846153846155,
+      "grad_norm": 0.844401478767395,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49230
+    },
+    {
+      "epoch": 157.82051282051282,
+      "grad_norm": 0.9674721956253052,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49240
+    },
+    {
+      "epoch": 157.85256410256412,
+      "grad_norm": 1.107801079750061,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49250
+    },
+    {
+      "epoch": 157.8846153846154,
+      "grad_norm": 1.2925678491592407,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 49260
+    },
+    {
+      "epoch": 157.91666666666666,
+      "grad_norm": 0.9736142754554749,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49270
+    },
+    {
+      "epoch": 157.94871794871796,
+      "grad_norm": 1.0266528129577637,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49280
+    },
+    {
+      "epoch": 157.98076923076923,
+      "grad_norm": 0.8876248002052307,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 49290
+    },
+    {
+      "epoch": 158.01282051282053,
+      "grad_norm": 0.976776659488678,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 49300
+    },
+    {
+      "epoch": 158.0448717948718,
+      "grad_norm": 0.8730822801589966,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 49310
+    },
+    {
+      "epoch": 158.07692307692307,
+      "grad_norm": 0.847844660282135,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49320
+    },
+    {
+      "epoch": 158.10897435897436,
+      "grad_norm": 0.8894463777542114,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 49330
+    },
+    {
+      "epoch": 158.14102564102564,
+      "grad_norm": 0.9121226668357849,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 49340
+    },
+    {
+      "epoch": 158.17307692307693,
+      "grad_norm": 0.8763313889503479,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49350
+    },
+    {
+      "epoch": 158.2051282051282,
+      "grad_norm": 0.8956344127655029,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49360
+    },
+    {
+      "epoch": 158.23717948717947,
+      "grad_norm": 0.6479077339172363,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49370
+    },
+    {
+      "epoch": 158.26923076923077,
+      "grad_norm": 0.8698983192443848,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49380
+    },
+    {
+      "epoch": 158.30128205128204,
+      "grad_norm": 0.9349910616874695,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49390
+    },
+    {
+      "epoch": 158.33333333333334,
+      "grad_norm": 0.8956306576728821,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49400
+    },
+    {
+      "epoch": 158.3653846153846,
+      "grad_norm": 0.9412496089935303,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 49410
+    },
+    {
+      "epoch": 158.39743589743588,
+      "grad_norm": 0.9529260396957397,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 49420
+    },
+    {
+      "epoch": 158.42948717948718,
+      "grad_norm": 0.7706722021102905,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49430
+    },
+    {
+      "epoch": 158.46153846153845,
+      "grad_norm": 0.8043166399002075,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 49440
+    },
+    {
+      "epoch": 158.49358974358975,
+      "grad_norm": 1.0604608058929443,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 49450
+    },
+    {
+      "epoch": 158.52564102564102,
+      "grad_norm": 0.8926914930343628,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49460
+    },
+    {
+      "epoch": 158.55769230769232,
+      "grad_norm": 0.82020503282547,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 49470
+    },
+    {
+      "epoch": 158.5897435897436,
+      "grad_norm": 0.8535478115081787,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49480
+    },
+    {
+      "epoch": 158.62179487179486,
+      "grad_norm": 0.8290415406227112,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49490
+    },
+    {
+      "epoch": 158.65384615384616,
+      "grad_norm": 0.9744908809661865,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 49500
+    },
+    {
+      "epoch": 158.68589743589743,
+      "grad_norm": 1.4291648864746094,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49510
+    },
+    {
+      "epoch": 158.71794871794873,
+      "grad_norm": 1.094354271888733,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49520
+    },
+    {
+      "epoch": 158.75,
+      "grad_norm": 0.9798198342323303,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49530
+    },
+    {
+      "epoch": 158.78205128205127,
+      "grad_norm": 0.8428601622581482,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49540
+    },
+    {
+      "epoch": 158.81410256410257,
+      "grad_norm": 0.813828706741333,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49550
+    },
+    {
+      "epoch": 158.84615384615384,
+      "grad_norm": 0.700119137763977,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 49560
+    },
+    {
+      "epoch": 158.87820512820514,
+      "grad_norm": 0.7448760271072388,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 49570
+    },
+    {
+      "epoch": 158.9102564102564,
+      "grad_norm": 0.9478446841239929,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 49580
+    },
+    {
+      "epoch": 158.94230769230768,
+      "grad_norm": 0.9041128158569336,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49590
+    },
+    {
+      "epoch": 158.97435897435898,
+      "grad_norm": 1.279263973236084,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49600
+    },
+    {
+      "epoch": 159.00641025641025,
+      "grad_norm": 1.0376896858215332,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 49610
+    },
+    {
+      "epoch": 159.03846153846155,
+      "grad_norm": 0.9366222023963928,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 49620
+    },
+    {
+      "epoch": 159.07051282051282,
+      "grad_norm": 0.7542009949684143,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 49630
+    },
+    {
+      "epoch": 159.10256410256412,
+      "grad_norm": 0.9788053631782532,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 49640
+    },
+    {
+      "epoch": 159.1346153846154,
+      "grad_norm": 0.9132522344589233,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49650
+    },
+    {
+      "epoch": 159.16666666666666,
+      "grad_norm": 0.9936500191688538,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49660
+    },
+    {
+      "epoch": 159.19871794871796,
+      "grad_norm": 0.8494492173194885,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49670
+    },
+    {
+      "epoch": 159.23076923076923,
+      "grad_norm": 0.9253721237182617,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 49680
+    },
+    {
+      "epoch": 159.26282051282053,
+      "grad_norm": 0.7618947625160217,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 49690
+    },
+    {
+      "epoch": 159.2948717948718,
+      "grad_norm": 1.123360276222229,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49700
+    },
+    {
+      "epoch": 159.32692307692307,
+      "grad_norm": 1.0670242309570312,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49710
+    },
+    {
+      "epoch": 159.35897435897436,
+      "grad_norm": 0.667417049407959,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49720
+    },
+    {
+      "epoch": 159.39102564102564,
+      "grad_norm": 0.929340124130249,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49730
+    },
+    {
+      "epoch": 159.42307692307693,
+      "grad_norm": 1.0360277891159058,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 49740
+    },
+    {
+      "epoch": 159.4551282051282,
+      "grad_norm": 1.0367746353149414,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49750
+    },
+    {
+      "epoch": 159.48717948717947,
+      "grad_norm": 1.1009763479232788,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 49760
+    },
+    {
+      "epoch": 159.51923076923077,
+      "grad_norm": 0.9860029220581055,
+      "learning_rate": 0.0001,
+      "loss": 0.0075,
+      "step": 49770
+    },
+    {
+      "epoch": 159.55128205128204,
+      "grad_norm": 1.0814940929412842,
+      "learning_rate": 0.0001,
+      "loss": 0.0083,
+      "step": 49780
+    },
+    {
+      "epoch": 159.58333333333334,
+      "grad_norm": 1.137305736541748,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 49790
+    },
+    {
+      "epoch": 159.6153846153846,
+      "grad_norm": 1.3639808893203735,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 49800
+    },
+    {
+      "epoch": 159.64743589743588,
+      "grad_norm": 0.800050675868988,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49810
+    },
+    {
+      "epoch": 159.67948717948718,
+      "grad_norm": 1.202020525932312,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49820
+    },
+    {
+      "epoch": 159.71153846153845,
+      "grad_norm": 0.7799537181854248,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 49830
+    },
+    {
+      "epoch": 159.74358974358975,
+      "grad_norm": 1.0705054998397827,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49840
+    },
+    {
+      "epoch": 159.77564102564102,
+      "grad_norm": 1.0489952564239502,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 49850
+    },
+    {
+      "epoch": 159.80769230769232,
+      "grad_norm": 1.1312514543533325,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49860
+    },
+    {
+      "epoch": 159.8397435897436,
+      "grad_norm": 1.0176628828048706,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 49870
+    },
+    {
+      "epoch": 159.87179487179486,
+      "grad_norm": 0.8568329215049744,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49880
+    },
+    {
+      "epoch": 159.90384615384616,
+      "grad_norm": 1.1073591709136963,
+      "learning_rate": 0.0001,
+      "loss": 0.0082,
+      "step": 49890
+    },
+    {
+      "epoch": 159.93589743589743,
+      "grad_norm": 0.983454167842865,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 49900
+    },
+    {
+      "epoch": 159.96794871794873,
+      "grad_norm": 1.3169201612472534,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 49910
+    },
+    {
+      "epoch": 160.0,
+      "grad_norm": 1.1161298751831055,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 49920
+    },
+    {
+      "epoch": 160.03205128205127,
+      "grad_norm": 1.4464205503463745,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 49930
+    },
+    {
+      "epoch": 160.06410256410257,
+      "grad_norm": 1.0221689939498901,
+      "learning_rate": 0.0001,
+      "loss": 0.0077,
+      "step": 49940
+    },
+    {
+      "epoch": 160.09615384615384,
+      "grad_norm": 1.0924010276794434,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 49950
+    },
+    {
+      "epoch": 160.12820512820514,
+      "grad_norm": 1.043839693069458,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 49960
+    },
+    {
+      "epoch": 160.1602564102564,
+      "grad_norm": 0.9498154520988464,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49970
+    },
+    {
+      "epoch": 160.19230769230768,
+      "grad_norm": 1.0326838493347168,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 49980
+    },
+    {
+      "epoch": 160.22435897435898,
+      "grad_norm": 1.0223008394241333,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 49990
+    },
+    {
+      "epoch": 160.25641025641025,
+      "grad_norm": 1.1111880540847778,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 50000
+    },
+    {
+      "epoch": 160.25641025641025,
+      "step": 50000,
+      "total_flos": 0.0,
+      "train_loss": 0.03167831766262651,
+      "train_runtime": 22030.0504,
+      "train_samples_per_second": 145.256,
+      "train_steps_per_second": 2.27
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 50000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 161,
+  "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
+}