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": 142.04545454545453,
+  "eval_steps": 500,
+  "global_step": 50000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.028409090909090908,
+      "grad_norm": 145.83482360839844,
+      "learning_rate": 0.0001,
+      "loss": 19.629,
+      "step": 10
+    },
+    {
+      "epoch": 0.056818181818181816,
+      "grad_norm": 76.42560577392578,
+      "learning_rate": 0.0001,
+      "loss": 4.9944,
+      "step": 20
+    },
+    {
+      "epoch": 0.08522727272727272,
+      "grad_norm": 75.80607604980469,
+      "learning_rate": 0.0001,
+      "loss": 4.0315,
+      "step": 30
+    },
+    {
+      "epoch": 0.11363636363636363,
+      "grad_norm": 73.45606231689453,
+      "learning_rate": 0.0001,
+      "loss": 3.6022,
+      "step": 40
+    },
+    {
+      "epoch": 0.14204545454545456,
+      "grad_norm": 63.843379974365234,
+      "learning_rate": 0.0001,
+      "loss": 3.4001,
+      "step": 50
+    },
+    {
+      "epoch": 0.17045454545454544,
+      "grad_norm": 62.33827209472656,
+      "learning_rate": 0.0001,
+      "loss": 3.1379,
+      "step": 60
+    },
+    {
+      "epoch": 0.19886363636363635,
+      "grad_norm": 48.174617767333984,
+      "learning_rate": 0.0001,
+      "loss": 2.9382,
+      "step": 70
+    },
+    {
+      "epoch": 0.22727272727272727,
+      "grad_norm": 51.949546813964844,
+      "learning_rate": 0.0001,
+      "loss": 2.7898,
+      "step": 80
+    },
+    {
+      "epoch": 0.2556818181818182,
+      "grad_norm": 55.129215240478516,
+      "learning_rate": 0.0001,
+      "loss": 2.6821,
+      "step": 90
+    },
+    {
+      "epoch": 0.2840909090909091,
+      "grad_norm": 57.89436721801758,
+      "learning_rate": 0.0001,
+      "loss": 2.6053,
+      "step": 100
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 60.28406524658203,
+      "learning_rate": 0.0001,
+      "loss": 2.4816,
+      "step": 110
+    },
+    {
+      "epoch": 0.3409090909090909,
+      "grad_norm": 56.817726135253906,
+      "learning_rate": 0.0001,
+      "loss": 2.3979,
+      "step": 120
+    },
+    {
+      "epoch": 0.3693181818181818,
+      "grad_norm": 50.40308380126953,
+      "learning_rate": 0.0001,
+      "loss": 2.2238,
+      "step": 130
+    },
+    {
+      "epoch": 0.3977272727272727,
+      "grad_norm": 50.979698181152344,
+      "learning_rate": 0.0001,
+      "loss": 2.1042,
+      "step": 140
+    },
+    {
+      "epoch": 0.42613636363636365,
+      "grad_norm": 49.087684631347656,
+      "learning_rate": 0.0001,
+      "loss": 1.9842,
+      "step": 150
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 42.376319885253906,
+      "learning_rate": 0.0001,
+      "loss": 1.9271,
+      "step": 160
+    },
+    {
+      "epoch": 0.48295454545454547,
+      "grad_norm": 52.510475158691406,
+      "learning_rate": 0.0001,
+      "loss": 1.8932,
+      "step": 170
+    },
+    {
+      "epoch": 0.5113636363636364,
+      "grad_norm": 47.899627685546875,
+      "learning_rate": 0.0001,
+      "loss": 1.835,
+      "step": 180
+    },
+    {
+      "epoch": 0.5397727272727273,
+      "grad_norm": 49.3372688293457,
+      "learning_rate": 0.0001,
+      "loss": 1.723,
+      "step": 190
+    },
+    {
+      "epoch": 0.5681818181818182,
+      "grad_norm": 53.45163345336914,
+      "learning_rate": 0.0001,
+      "loss": 1.6617,
+      "step": 200
+    },
+    {
+      "epoch": 0.5965909090909091,
+      "grad_norm": 40.74541091918945,
+      "learning_rate": 0.0001,
+      "loss": 1.6427,
+      "step": 210
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 45.38197708129883,
+      "learning_rate": 0.0001,
+      "loss": 1.5676,
+      "step": 220
+    },
+    {
+      "epoch": 0.6534090909090909,
+      "grad_norm": 44.96558380126953,
+      "learning_rate": 0.0001,
+      "loss": 1.4756,
+      "step": 230
+    },
+    {
+      "epoch": 0.6818181818181818,
+      "grad_norm": 39.52370071411133,
+      "learning_rate": 0.0001,
+      "loss": 1.4001,
+      "step": 240
+    },
+    {
+      "epoch": 0.7102272727272727,
+      "grad_norm": 35.26641845703125,
+      "learning_rate": 0.0001,
+      "loss": 1.346,
+      "step": 250
+    },
+    {
+      "epoch": 0.7386363636363636,
+      "grad_norm": 43.04871368408203,
+      "learning_rate": 0.0001,
+      "loss": 1.2858,
+      "step": 260
+    },
+    {
+      "epoch": 0.7670454545454546,
+      "grad_norm": 41.421043395996094,
+      "learning_rate": 0.0001,
+      "loss": 1.2311,
+      "step": 270
+    },
+    {
+      "epoch": 0.7954545454545454,
+      "grad_norm": 34.36821365356445,
+      "learning_rate": 0.0001,
+      "loss": 1.1698,
+      "step": 280
+    },
+    {
+      "epoch": 0.8238636363636364,
+      "grad_norm": 28.939420700073242,
+      "learning_rate": 0.0001,
+      "loss": 1.1007,
+      "step": 290
+    },
+    {
+      "epoch": 0.8522727272727273,
+      "grad_norm": 34.10892868041992,
+      "learning_rate": 0.0001,
+      "loss": 1.0837,
+      "step": 300
+    },
+    {
+      "epoch": 0.8806818181818182,
+      "grad_norm": 32.707054138183594,
+      "learning_rate": 0.0001,
+      "loss": 1.037,
+      "step": 310
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 35.907508850097656,
+      "learning_rate": 0.0001,
+      "loss": 1.0093,
+      "step": 320
+    },
+    {
+      "epoch": 0.9375,
+      "grad_norm": 37.658451080322266,
+      "learning_rate": 0.0001,
+      "loss": 0.9501,
+      "step": 330
+    },
+    {
+      "epoch": 0.9659090909090909,
+      "grad_norm": 29.12462615966797,
+      "learning_rate": 0.0001,
+      "loss": 0.9148,
+      "step": 340
+    },
+    {
+      "epoch": 0.9943181818181818,
+      "grad_norm": 31.38657569885254,
+      "learning_rate": 0.0001,
+      "loss": 0.8863,
+      "step": 350
+    },
+    {
+      "epoch": 1.0227272727272727,
+      "grad_norm": 29.956890106201172,
+      "learning_rate": 0.0001,
+      "loss": 0.8712,
+      "step": 360
+    },
+    {
+      "epoch": 1.0511363636363635,
+      "grad_norm": 26.45981788635254,
+      "learning_rate": 0.0001,
+      "loss": 0.8092,
+      "step": 370
+    },
+    {
+      "epoch": 1.0795454545454546,
+      "grad_norm": 27.67877197265625,
+      "learning_rate": 0.0001,
+      "loss": 0.8083,
+      "step": 380
+    },
+    {
+      "epoch": 1.1079545454545454,
+      "grad_norm": 29.698911666870117,
+      "learning_rate": 0.0001,
+      "loss": 0.7937,
+      "step": 390
+    },
+    {
+      "epoch": 1.1363636363636362,
+      "grad_norm": 25.87833595275879,
+      "learning_rate": 0.0001,
+      "loss": 0.7842,
+      "step": 400
+    },
+    {
+      "epoch": 1.1647727272727273,
+      "grad_norm": 30.7982120513916,
+      "learning_rate": 0.0001,
+      "loss": 0.7507,
+      "step": 410
+    },
+    {
+      "epoch": 1.1931818181818181,
+      "grad_norm": 25.391246795654297,
+      "learning_rate": 0.0001,
+      "loss": 0.7208,
+      "step": 420
+    },
+    {
+      "epoch": 1.2215909090909092,
+      "grad_norm": 22.389162063598633,
+      "learning_rate": 0.0001,
+      "loss": 0.6947,
+      "step": 430
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 21.166950225830078,
+      "learning_rate": 0.0001,
+      "loss": 0.6735,
+      "step": 440
+    },
+    {
+      "epoch": 1.2784090909090908,
+      "grad_norm": 20.702722549438477,
+      "learning_rate": 0.0001,
+      "loss": 0.6466,
+      "step": 450
+    },
+    {
+      "epoch": 1.3068181818181819,
+      "grad_norm": 25.78806495666504,
+      "learning_rate": 0.0001,
+      "loss": 0.6449,
+      "step": 460
+    },
+    {
+      "epoch": 1.3352272727272727,
+      "grad_norm": 19.319217681884766,
+      "learning_rate": 0.0001,
+      "loss": 0.6416,
+      "step": 470
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 22.997730255126953,
+      "learning_rate": 0.0001,
+      "loss": 0.6296,
+      "step": 480
+    },
+    {
+      "epoch": 1.3920454545454546,
+      "grad_norm": 21.263296127319336,
+      "learning_rate": 0.0001,
+      "loss": 0.586,
+      "step": 490
+    },
+    {
+      "epoch": 1.4204545454545454,
+      "grad_norm": 18.75411605834961,
+      "learning_rate": 0.0001,
+      "loss": 0.565,
+      "step": 500
+    },
+    {
+      "epoch": 1.4488636363636362,
+      "grad_norm": 21.52752113342285,
+      "learning_rate": 0.0001,
+      "loss": 0.5923,
+      "step": 510
+    },
+    {
+      "epoch": 1.4772727272727273,
+      "grad_norm": 21.464553833007812,
+      "learning_rate": 0.0001,
+      "loss": 0.5726,
+      "step": 520
+    },
+    {
+      "epoch": 1.5056818181818183,
+      "grad_norm": 19.978652954101562,
+      "learning_rate": 0.0001,
+      "loss": 0.5734,
+      "step": 530
+    },
+    {
+      "epoch": 1.5340909090909092,
+      "grad_norm": 21.40723419189453,
+      "learning_rate": 0.0001,
+      "loss": 0.549,
+      "step": 540
+    },
+    {
+      "epoch": 1.5625,
+      "grad_norm": 18.181068420410156,
+      "learning_rate": 0.0001,
+      "loss": 0.5337,
+      "step": 550
+    },
+    {
+      "epoch": 1.5909090909090908,
+      "grad_norm": 16.99245262145996,
+      "learning_rate": 0.0001,
+      "loss": 0.528,
+      "step": 560
+    },
+    {
+      "epoch": 1.6193181818181817,
+      "grad_norm": 18.75094223022461,
+      "learning_rate": 0.0001,
+      "loss": 0.5316,
+      "step": 570
+    },
+    {
+      "epoch": 1.6477272727272727,
+      "grad_norm": 23.37386703491211,
+      "learning_rate": 0.0001,
+      "loss": 0.5116,
+      "step": 580
+    },
+    {
+      "epoch": 1.6761363636363638,
+      "grad_norm": 20.599090576171875,
+      "learning_rate": 0.0001,
+      "loss": 0.5135,
+      "step": 590
+    },
+    {
+      "epoch": 1.7045454545454546,
+      "grad_norm": 19.43827247619629,
+      "learning_rate": 0.0001,
+      "loss": 0.4833,
+      "step": 600
+    },
+    {
+      "epoch": 1.7329545454545454,
+      "grad_norm": 16.123802185058594,
+      "learning_rate": 0.0001,
+      "loss": 0.486,
+      "step": 610
+    },
+    {
+      "epoch": 1.7613636363636362,
+      "grad_norm": 18.867568969726562,
+      "learning_rate": 0.0001,
+      "loss": 0.47,
+      "step": 620
+    },
+    {
+      "epoch": 1.7897727272727273,
+      "grad_norm": 16.580411911010742,
+      "learning_rate": 0.0001,
+      "loss": 0.4803,
+      "step": 630
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 16.470767974853516,
+      "learning_rate": 0.0001,
+      "loss": 0.4763,
+      "step": 640
+    },
+    {
+      "epoch": 1.8465909090909092,
+      "grad_norm": 15.845813751220703,
+      "learning_rate": 0.0001,
+      "loss": 0.4796,
+      "step": 650
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 16.313016891479492,
+      "learning_rate": 0.0001,
+      "loss": 0.4726,
+      "step": 660
+    },
+    {
+      "epoch": 1.9034090909090908,
+      "grad_norm": 16.114171981811523,
+      "learning_rate": 0.0001,
+      "loss": 0.4657,
+      "step": 670
+    },
+    {
+      "epoch": 1.9318181818181817,
+      "grad_norm": 18.153635025024414,
+      "learning_rate": 0.0001,
+      "loss": 0.4656,
+      "step": 680
+    },
+    {
+      "epoch": 1.9602272727272727,
+      "grad_norm": 19.012916564941406,
+      "learning_rate": 0.0001,
+      "loss": 0.4543,
+      "step": 690
+    },
+    {
+      "epoch": 1.9886363636363638,
+      "grad_norm": 15.661040306091309,
+      "learning_rate": 0.0001,
+      "loss": 0.4451,
+      "step": 700
+    },
+    {
+      "epoch": 2.0170454545454546,
+      "grad_norm": 20.665252685546875,
+      "learning_rate": 0.0001,
+      "loss": 0.4542,
+      "step": 710
+    },
+    {
+      "epoch": 2.0454545454545454,
+      "grad_norm": 18.220745086669922,
+      "learning_rate": 0.0001,
+      "loss": 0.4615,
+      "step": 720
+    },
+    {
+      "epoch": 2.0738636363636362,
+      "grad_norm": 15.511617660522461,
+      "learning_rate": 0.0001,
+      "loss": 0.4379,
+      "step": 730
+    },
+    {
+      "epoch": 2.102272727272727,
+      "grad_norm": 16.05436897277832,
+      "learning_rate": 0.0001,
+      "loss": 0.4457,
+      "step": 740
+    },
+    {
+      "epoch": 2.1306818181818183,
+      "grad_norm": 14.067421913146973,
+      "learning_rate": 0.0001,
+      "loss": 0.4404,
+      "step": 750
+    },
+    {
+      "epoch": 2.159090909090909,
+      "grad_norm": 17.595314025878906,
+      "learning_rate": 0.0001,
+      "loss": 0.4359,
+      "step": 760
+    },
+    {
+      "epoch": 2.1875,
+      "grad_norm": 14.292813301086426,
+      "learning_rate": 0.0001,
+      "loss": 0.4265,
+      "step": 770
+    },
+    {
+      "epoch": 2.215909090909091,
+      "grad_norm": 13.254941940307617,
+      "learning_rate": 0.0001,
+      "loss": 0.4282,
+      "step": 780
+    },
+    {
+      "epoch": 2.2443181818181817,
+      "grad_norm": 14.131694793701172,
+      "learning_rate": 0.0001,
+      "loss": 0.4154,
+      "step": 790
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 16.197458267211914,
+      "learning_rate": 0.0001,
+      "loss": 0.4271,
+      "step": 800
+    },
+    {
+      "epoch": 2.3011363636363638,
+      "grad_norm": 13.791603088378906,
+      "learning_rate": 0.0001,
+      "loss": 0.4159,
+      "step": 810
+    },
+    {
+      "epoch": 2.3295454545454546,
+      "grad_norm": 15.197473526000977,
+      "learning_rate": 0.0001,
+      "loss": 0.4011,
+      "step": 820
+    },
+    {
+      "epoch": 2.3579545454545454,
+      "grad_norm": 16.548952102661133,
+      "learning_rate": 0.0001,
+      "loss": 0.4153,
+      "step": 830
+    },
+    {
+      "epoch": 2.3863636363636362,
+      "grad_norm": 15.021014213562012,
+      "learning_rate": 0.0001,
+      "loss": 0.4077,
+      "step": 840
+    },
+    {
+      "epoch": 2.4147727272727275,
+      "grad_norm": 11.394856452941895,
+      "learning_rate": 0.0001,
+      "loss": 0.412,
+      "step": 850
+    },
+    {
+      "epoch": 2.4431818181818183,
+      "grad_norm": 14.868868827819824,
+      "learning_rate": 0.0001,
+      "loss": 0.3898,
+      "step": 860
+    },
+    {
+      "epoch": 2.471590909090909,
+      "grad_norm": 15.362899780273438,
+      "learning_rate": 0.0001,
+      "loss": 0.3985,
+      "step": 870
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 12.927398681640625,
+      "learning_rate": 0.0001,
+      "loss": 0.3969,
+      "step": 880
+    },
+    {
+      "epoch": 2.528409090909091,
+      "grad_norm": 12.813764572143555,
+      "learning_rate": 0.0001,
+      "loss": 0.3808,
+      "step": 890
+    },
+    {
+      "epoch": 2.5568181818181817,
+      "grad_norm": 14.54391860961914,
+      "learning_rate": 0.0001,
+      "loss": 0.3786,
+      "step": 900
+    },
+    {
+      "epoch": 2.5852272727272725,
+      "grad_norm": 12.791834831237793,
+      "learning_rate": 0.0001,
+      "loss": 0.3891,
+      "step": 910
+    },
+    {
+      "epoch": 2.6136363636363638,
+      "grad_norm": 12.401715278625488,
+      "learning_rate": 0.0001,
+      "loss": 0.3903,
+      "step": 920
+    },
+    {
+      "epoch": 2.6420454545454546,
+      "grad_norm": 12.847018241882324,
+      "learning_rate": 0.0001,
+      "loss": 0.3741,
+      "step": 930
+    },
+    {
+      "epoch": 2.6704545454545454,
+      "grad_norm": 14.203393936157227,
+      "learning_rate": 0.0001,
+      "loss": 0.3827,
+      "step": 940
+    },
+    {
+      "epoch": 2.6988636363636362,
+      "grad_norm": 14.513806343078613,
+      "learning_rate": 0.0001,
+      "loss": 0.3848,
+      "step": 950
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 15.528099060058594,
+      "learning_rate": 0.0001,
+      "loss": 0.3861,
+      "step": 960
+    },
+    {
+      "epoch": 2.7556818181818183,
+      "grad_norm": 16.94673728942871,
+      "learning_rate": 0.0001,
+      "loss": 0.3796,
+      "step": 970
+    },
+    {
+      "epoch": 2.784090909090909,
+      "grad_norm": 14.714554786682129,
+      "learning_rate": 0.0001,
+      "loss": 0.3683,
+      "step": 980
+    },
+    {
+      "epoch": 2.8125,
+      "grad_norm": 13.82036304473877,
+      "learning_rate": 0.0001,
+      "loss": 0.3751,
+      "step": 990
+    },
+    {
+      "epoch": 2.840909090909091,
+      "grad_norm": 13.940563201904297,
+      "learning_rate": 0.0001,
+      "loss": 0.3719,
+      "step": 1000
+    },
+    {
+      "epoch": 2.8693181818181817,
+      "grad_norm": 13.515235900878906,
+      "learning_rate": 0.0001,
+      "loss": 0.3761,
+      "step": 1010
+    },
+    {
+      "epoch": 2.8977272727272725,
+      "grad_norm": 14.955562591552734,
+      "learning_rate": 0.0001,
+      "loss": 0.3576,
+      "step": 1020
+    },
+    {
+      "epoch": 2.9261363636363638,
+      "grad_norm": 11.280851364135742,
+      "learning_rate": 0.0001,
+      "loss": 0.3614,
+      "step": 1030
+    },
+    {
+      "epoch": 2.9545454545454546,
+      "grad_norm": 12.09704303741455,
+      "learning_rate": 0.0001,
+      "loss": 0.3661,
+      "step": 1040
+    },
+    {
+      "epoch": 2.9829545454545454,
+      "grad_norm": 14.392845153808594,
+      "learning_rate": 0.0001,
+      "loss": 0.3722,
+      "step": 1050
+    },
+    {
+      "epoch": 3.0113636363636362,
+      "grad_norm": 13.309704780578613,
+      "learning_rate": 0.0001,
+      "loss": 0.3595,
+      "step": 1060
+    },
+    {
+      "epoch": 3.039772727272727,
+      "grad_norm": 15.42646312713623,
+      "learning_rate": 0.0001,
+      "loss": 0.3427,
+      "step": 1070
+    },
+    {
+      "epoch": 3.0681818181818183,
+      "grad_norm": 10.911493301391602,
+      "learning_rate": 0.0001,
+      "loss": 0.3517,
+      "step": 1080
+    },
+    {
+      "epoch": 3.096590909090909,
+      "grad_norm": 12.293902397155762,
+      "learning_rate": 0.0001,
+      "loss": 0.3398,
+      "step": 1090
+    },
+    {
+      "epoch": 3.125,
+      "grad_norm": 13.060087203979492,
+      "learning_rate": 0.0001,
+      "loss": 0.3491,
+      "step": 1100
+    },
+    {
+      "epoch": 3.153409090909091,
+      "grad_norm": 11.007771492004395,
+      "learning_rate": 0.0001,
+      "loss": 0.3566,
+      "step": 1110
+    },
+    {
+      "epoch": 3.1818181818181817,
+      "grad_norm": 13.680668830871582,
+      "learning_rate": 0.0001,
+      "loss": 0.341,
+      "step": 1120
+    },
+    {
+      "epoch": 3.210227272727273,
+      "grad_norm": 9.585054397583008,
+      "learning_rate": 0.0001,
+      "loss": 0.3582,
+      "step": 1130
+    },
+    {
+      "epoch": 3.2386363636363638,
+      "grad_norm": 12.668915748596191,
+      "learning_rate": 0.0001,
+      "loss": 0.351,
+      "step": 1140
+    },
+    {
+      "epoch": 3.2670454545454546,
+      "grad_norm": 12.355649948120117,
+      "learning_rate": 0.0001,
+      "loss": 0.3489,
+      "step": 1150
+    },
+    {
+      "epoch": 3.2954545454545454,
+      "grad_norm": 11.911877632141113,
+      "learning_rate": 0.0001,
+      "loss": 0.3348,
+      "step": 1160
+    },
+    {
+      "epoch": 3.3238636363636362,
+      "grad_norm": 11.719152450561523,
+      "learning_rate": 0.0001,
+      "loss": 0.3364,
+      "step": 1170
+    },
+    {
+      "epoch": 3.3522727272727275,
+      "grad_norm": 10.242898941040039,
+      "learning_rate": 0.0001,
+      "loss": 0.3264,
+      "step": 1180
+    },
+    {
+      "epoch": 3.3806818181818183,
+      "grad_norm": 9.679462432861328,
+      "learning_rate": 0.0001,
+      "loss": 0.3251,
+      "step": 1190
+    },
+    {
+      "epoch": 3.409090909090909,
+      "grad_norm": 10.055230140686035,
+      "learning_rate": 0.0001,
+      "loss": 0.3275,
+      "step": 1200
+    },
+    {
+      "epoch": 3.4375,
+      "grad_norm": 8.881628036499023,
+      "learning_rate": 0.0001,
+      "loss": 0.325,
+      "step": 1210
+    },
+    {
+      "epoch": 3.465909090909091,
+      "grad_norm": 11.237067222595215,
+      "learning_rate": 0.0001,
+      "loss": 0.3497,
+      "step": 1220
+    },
+    {
+      "epoch": 3.4943181818181817,
+      "grad_norm": 12.494010925292969,
+      "learning_rate": 0.0001,
+      "loss": 0.334,
+      "step": 1230
+    },
+    {
+      "epoch": 3.5227272727272725,
+      "grad_norm": 11.651463508605957,
+      "learning_rate": 0.0001,
+      "loss": 0.3291,
+      "step": 1240
+    },
+    {
+      "epoch": 3.5511363636363638,
+      "grad_norm": 11.586341857910156,
+      "learning_rate": 0.0001,
+      "loss": 0.3236,
+      "step": 1250
+    },
+    {
+      "epoch": 3.5795454545454546,
+      "grad_norm": 10.351299285888672,
+      "learning_rate": 0.0001,
+      "loss": 0.3314,
+      "step": 1260
+    },
+    {
+      "epoch": 3.6079545454545454,
+      "grad_norm": 11.262849807739258,
+      "learning_rate": 0.0001,
+      "loss": 0.3375,
+      "step": 1270
+    },
+    {
+      "epoch": 3.6363636363636362,
+      "grad_norm": 10.904534339904785,
+      "learning_rate": 0.0001,
+      "loss": 0.3281,
+      "step": 1280
+    },
+    {
+      "epoch": 3.6647727272727275,
+      "grad_norm": 8.963489532470703,
+      "learning_rate": 0.0001,
+      "loss": 0.3298,
+      "step": 1290
+    },
+    {
+      "epoch": 3.6931818181818183,
+      "grad_norm": 10.735923767089844,
+      "learning_rate": 0.0001,
+      "loss": 0.3272,
+      "step": 1300
+    },
+    {
+      "epoch": 3.721590909090909,
+      "grad_norm": 12.867420196533203,
+      "learning_rate": 0.0001,
+      "loss": 0.3234,
+      "step": 1310
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 11.347630500793457,
+      "learning_rate": 0.0001,
+      "loss": 0.3209,
+      "step": 1320
+    },
+    {
+      "epoch": 3.778409090909091,
+      "grad_norm": 10.211435317993164,
+      "learning_rate": 0.0001,
+      "loss": 0.3173,
+      "step": 1330
+    },
+    {
+      "epoch": 3.8068181818181817,
+      "grad_norm": 8.992242813110352,
+      "learning_rate": 0.0001,
+      "loss": 0.3199,
+      "step": 1340
+    },
+    {
+      "epoch": 3.8352272727272725,
+      "grad_norm": 9.036025047302246,
+      "learning_rate": 0.0001,
+      "loss": 0.3124,
+      "step": 1350
+    },
+    {
+      "epoch": 3.8636363636363638,
+      "grad_norm": 11.357304573059082,
+      "learning_rate": 0.0001,
+      "loss": 0.3256,
+      "step": 1360
+    },
+    {
+      "epoch": 3.8920454545454546,
+      "grad_norm": 12.447697639465332,
+      "learning_rate": 0.0001,
+      "loss": 0.3234,
+      "step": 1370
+    },
+    {
+      "epoch": 3.9204545454545454,
+      "grad_norm": 10.388401985168457,
+      "learning_rate": 0.0001,
+      "loss": 0.3134,
+      "step": 1380
+    },
+    {
+      "epoch": 3.9488636363636362,
+      "grad_norm": 10.227154731750488,
+      "learning_rate": 0.0001,
+      "loss": 0.3036,
+      "step": 1390
+    },
+    {
+      "epoch": 3.9772727272727275,
+      "grad_norm": 11.007589340209961,
+      "learning_rate": 0.0001,
+      "loss": 0.3087,
+      "step": 1400
+    },
+    {
+      "epoch": 4.005681818181818,
+      "grad_norm": 9.138888359069824,
+      "learning_rate": 0.0001,
+      "loss": 0.3109,
+      "step": 1410
+    },
+    {
+      "epoch": 4.034090909090909,
+      "grad_norm": 9.967912673950195,
+      "learning_rate": 0.0001,
+      "loss": 0.307,
+      "step": 1420
+    },
+    {
+      "epoch": 4.0625,
+      "grad_norm": 10.28461742401123,
+      "learning_rate": 0.0001,
+      "loss": 0.2973,
+      "step": 1430
+    },
+    {
+      "epoch": 4.090909090909091,
+      "grad_norm": 9.785955429077148,
+      "learning_rate": 0.0001,
+      "loss": 0.3004,
+      "step": 1440
+    },
+    {
+      "epoch": 4.119318181818182,
+      "grad_norm": 10.850992202758789,
+      "learning_rate": 0.0001,
+      "loss": 0.3005,
+      "step": 1450
+    },
+    {
+      "epoch": 4.1477272727272725,
+      "grad_norm": 12.91825008392334,
+      "learning_rate": 0.0001,
+      "loss": 0.3005,
+      "step": 1460
+    },
+    {
+      "epoch": 4.176136363636363,
+      "grad_norm": 11.555893898010254,
+      "learning_rate": 0.0001,
+      "loss": 0.3107,
+      "step": 1470
+    },
+    {
+      "epoch": 4.204545454545454,
+      "grad_norm": 10.81035327911377,
+      "learning_rate": 0.0001,
+      "loss": 0.2977,
+      "step": 1480
+    },
+    {
+      "epoch": 4.232954545454546,
+      "grad_norm": 11.161906242370605,
+      "learning_rate": 0.0001,
+      "loss": 0.2962,
+      "step": 1490
+    },
+    {
+      "epoch": 4.261363636363637,
+      "grad_norm": 12.040653228759766,
+      "learning_rate": 0.0001,
+      "loss": 0.3025,
+      "step": 1500
+    },
+    {
+      "epoch": 4.2897727272727275,
+      "grad_norm": 9.015117645263672,
+      "learning_rate": 0.0001,
+      "loss": 0.292,
+      "step": 1510
+    },
+    {
+      "epoch": 4.318181818181818,
+      "grad_norm": 10.438865661621094,
+      "learning_rate": 0.0001,
+      "loss": 0.2881,
+      "step": 1520
+    },
+    {
+      "epoch": 4.346590909090909,
+      "grad_norm": 10.363481521606445,
+      "learning_rate": 0.0001,
+      "loss": 0.2919,
+      "step": 1530
+    },
+    {
+      "epoch": 4.375,
+      "grad_norm": 10.898138999938965,
+      "learning_rate": 0.0001,
+      "loss": 0.3035,
+      "step": 1540
+    },
+    {
+      "epoch": 4.403409090909091,
+      "grad_norm": 9.264910697937012,
+      "learning_rate": 0.0001,
+      "loss": 0.2866,
+      "step": 1550
+    },
+    {
+      "epoch": 4.431818181818182,
+      "grad_norm": 9.535000801086426,
+      "learning_rate": 0.0001,
+      "loss": 0.2833,
+      "step": 1560
+    },
+    {
+      "epoch": 4.4602272727272725,
+      "grad_norm": 10.703611373901367,
+      "learning_rate": 0.0001,
+      "loss": 0.2977,
+      "step": 1570
+    },
+    {
+      "epoch": 4.488636363636363,
+      "grad_norm": 8.830336570739746,
+      "learning_rate": 0.0001,
+      "loss": 0.2822,
+      "step": 1580
+    },
+    {
+      "epoch": 4.517045454545455,
+      "grad_norm": 9.80781364440918,
+      "learning_rate": 0.0001,
+      "loss": 0.2897,
+      "step": 1590
+    },
+    {
+      "epoch": 4.545454545454545,
+      "grad_norm": 9.538243293762207,
+      "learning_rate": 0.0001,
+      "loss": 0.2893,
+      "step": 1600
+    },
+    {
+      "epoch": 4.573863636363637,
+      "grad_norm": 7.803942680358887,
+      "learning_rate": 0.0001,
+      "loss": 0.2898,
+      "step": 1610
+    },
+    {
+      "epoch": 4.6022727272727275,
+      "grad_norm": 9.329748153686523,
+      "learning_rate": 0.0001,
+      "loss": 0.288,
+      "step": 1620
+    },
+    {
+      "epoch": 4.630681818181818,
+      "grad_norm": 9.706318855285645,
+      "learning_rate": 0.0001,
+      "loss": 0.2824,
+      "step": 1630
+    },
+    {
+      "epoch": 4.659090909090909,
+      "grad_norm": 8.528480529785156,
+      "learning_rate": 0.0001,
+      "loss": 0.2969,
+      "step": 1640
+    },
+    {
+      "epoch": 4.6875,
+      "grad_norm": 8.045533180236816,
+      "learning_rate": 0.0001,
+      "loss": 0.2995,
+      "step": 1650
+    },
+    {
+      "epoch": 4.715909090909091,
+      "grad_norm": 8.474618911743164,
+      "learning_rate": 0.0001,
+      "loss": 0.2906,
+      "step": 1660
+    },
+    {
+      "epoch": 4.744318181818182,
+      "grad_norm": 9.250617027282715,
+      "learning_rate": 0.0001,
+      "loss": 0.2915,
+      "step": 1670
+    },
+    {
+      "epoch": 4.7727272727272725,
+      "grad_norm": 8.805644989013672,
+      "learning_rate": 0.0001,
+      "loss": 0.2835,
+      "step": 1680
+    },
+    {
+      "epoch": 4.801136363636363,
+      "grad_norm": 8.365328788757324,
+      "learning_rate": 0.0001,
+      "loss": 0.2872,
+      "step": 1690
+    },
+    {
+      "epoch": 4.829545454545455,
+      "grad_norm": 9.336677551269531,
+      "learning_rate": 0.0001,
+      "loss": 0.2822,
+      "step": 1700
+    },
+    {
+      "epoch": 4.857954545454545,
+      "grad_norm": 8.8048095703125,
+      "learning_rate": 0.0001,
+      "loss": 0.2802,
+      "step": 1710
+    },
+    {
+      "epoch": 4.886363636363637,
+      "grad_norm": 10.265268325805664,
+      "learning_rate": 0.0001,
+      "loss": 0.2718,
+      "step": 1720
+    },
+    {
+      "epoch": 4.9147727272727275,
+      "grad_norm": 9.204639434814453,
+      "learning_rate": 0.0001,
+      "loss": 0.2874,
+      "step": 1730
+    },
+    {
+      "epoch": 4.943181818181818,
+      "grad_norm": 8.516647338867188,
+      "learning_rate": 0.0001,
+      "loss": 0.2913,
+      "step": 1740
+    },
+    {
+      "epoch": 4.971590909090909,
+      "grad_norm": 7.545566082000732,
+      "learning_rate": 0.0001,
+      "loss": 0.2799,
+      "step": 1750
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 9.611028671264648,
+      "learning_rate": 0.0001,
+      "loss": 0.2816,
+      "step": 1760
+    },
+    {
+      "epoch": 5.028409090909091,
+      "grad_norm": 7.730203151702881,
+      "learning_rate": 0.0001,
+      "loss": 0.278,
+      "step": 1770
+    },
+    {
+      "epoch": 5.056818181818182,
+      "grad_norm": 9.771706581115723,
+      "learning_rate": 0.0001,
+      "loss": 0.2764,
+      "step": 1780
+    },
+    {
+      "epoch": 5.0852272727272725,
+      "grad_norm": 8.7466402053833,
+      "learning_rate": 0.0001,
+      "loss": 0.2823,
+      "step": 1790
+    },
+    {
+      "epoch": 5.113636363636363,
+      "grad_norm": 9.843619346618652,
+      "learning_rate": 0.0001,
+      "loss": 0.2654,
+      "step": 1800
+    },
+    {
+      "epoch": 5.142045454545454,
+      "grad_norm": 8.296882629394531,
+      "learning_rate": 0.0001,
+      "loss": 0.2691,
+      "step": 1810
+    },
+    {
+      "epoch": 5.170454545454546,
+      "grad_norm": 8.18472957611084,
+      "learning_rate": 0.0001,
+      "loss": 0.2644,
+      "step": 1820
+    },
+    {
+      "epoch": 5.198863636363637,
+      "grad_norm": 8.96210765838623,
+      "learning_rate": 0.0001,
+      "loss": 0.2688,
+      "step": 1830
+    },
+    {
+      "epoch": 5.2272727272727275,
+      "grad_norm": 9.177153587341309,
+      "learning_rate": 0.0001,
+      "loss": 0.2683,
+      "step": 1840
+    },
+    {
+      "epoch": 5.255681818181818,
+      "grad_norm": 7.267095565795898,
+      "learning_rate": 0.0001,
+      "loss": 0.2673,
+      "step": 1850
+    },
+    {
+      "epoch": 5.284090909090909,
+      "grad_norm": 8.78824520111084,
+      "learning_rate": 0.0001,
+      "loss": 0.2629,
+      "step": 1860
+    },
+    {
+      "epoch": 5.3125,
+      "grad_norm": 8.33309268951416,
+      "learning_rate": 0.0001,
+      "loss": 0.2635,
+      "step": 1870
+    },
+    {
+      "epoch": 5.340909090909091,
+      "grad_norm": 9.574383735656738,
+      "learning_rate": 0.0001,
+      "loss": 0.2563,
+      "step": 1880
+    },
+    {
+      "epoch": 5.369318181818182,
+      "grad_norm": 7.813918590545654,
+      "learning_rate": 0.0001,
+      "loss": 0.2467,
+      "step": 1890
+    },
+    {
+      "epoch": 5.3977272727272725,
+      "grad_norm": 9.375533103942871,
+      "learning_rate": 0.0001,
+      "loss": 0.262,
+      "step": 1900
+    },
+    {
+      "epoch": 5.426136363636363,
+      "grad_norm": 9.987363815307617,
+      "learning_rate": 0.0001,
+      "loss": 0.2618,
+      "step": 1910
+    },
+    {
+      "epoch": 5.454545454545454,
+      "grad_norm": 10.02425479888916,
+      "learning_rate": 0.0001,
+      "loss": 0.2635,
+      "step": 1920
+    },
+    {
+      "epoch": 5.482954545454546,
+      "grad_norm": 9.342535972595215,
+      "learning_rate": 0.0001,
+      "loss": 0.2497,
+      "step": 1930
+    },
+    {
+      "epoch": 5.511363636363637,
+      "grad_norm": 9.32978343963623,
+      "learning_rate": 0.0001,
+      "loss": 0.2592,
+      "step": 1940
+    },
+    {
+      "epoch": 5.5397727272727275,
+      "grad_norm": 7.348328113555908,
+      "learning_rate": 0.0001,
+      "loss": 0.2592,
+      "step": 1950
+    },
+    {
+      "epoch": 5.568181818181818,
+      "grad_norm": 8.86340045928955,
+      "learning_rate": 0.0001,
+      "loss": 0.2541,
+      "step": 1960
+    },
+    {
+      "epoch": 5.596590909090909,
+      "grad_norm": 8.326016426086426,
+      "learning_rate": 0.0001,
+      "loss": 0.2663,
+      "step": 1970
+    },
+    {
+      "epoch": 5.625,
+      "grad_norm": 8.392045021057129,
+      "learning_rate": 0.0001,
+      "loss": 0.2682,
+      "step": 1980
+    },
+    {
+      "epoch": 5.653409090909091,
+      "grad_norm": 8.57619571685791,
+      "learning_rate": 0.0001,
+      "loss": 0.2596,
+      "step": 1990
+    },
+    {
+      "epoch": 5.681818181818182,
+      "grad_norm": 7.7515058517456055,
+      "learning_rate": 0.0001,
+      "loss": 0.2408,
+      "step": 2000
+    },
+    {
+      "epoch": 5.7102272727272725,
+      "grad_norm": 8.581171989440918,
+      "learning_rate": 0.0001,
+      "loss": 0.2492,
+      "step": 2010
+    },
+    {
+      "epoch": 5.738636363636363,
+      "grad_norm": 8.195562362670898,
+      "learning_rate": 0.0001,
+      "loss": 0.2497,
+      "step": 2020
+    },
+    {
+      "epoch": 5.767045454545455,
+      "grad_norm": 7.793923854827881,
+      "learning_rate": 0.0001,
+      "loss": 0.2415,
+      "step": 2030
+    },
+    {
+      "epoch": 5.795454545454545,
+      "grad_norm": 7.39900016784668,
+      "learning_rate": 0.0001,
+      "loss": 0.2436,
+      "step": 2040
+    },
+    {
+      "epoch": 5.823863636363637,
+      "grad_norm": 8.420592308044434,
+      "learning_rate": 0.0001,
+      "loss": 0.2523,
+      "step": 2050
+    },
+    {
+      "epoch": 5.8522727272727275,
+      "grad_norm": 8.713873863220215,
+      "learning_rate": 0.0001,
+      "loss": 0.2623,
+      "step": 2060
+    },
+    {
+      "epoch": 5.880681818181818,
+      "grad_norm": 9.038602828979492,
+      "learning_rate": 0.0001,
+      "loss": 0.2543,
+      "step": 2070
+    },
+    {
+      "epoch": 5.909090909090909,
+      "grad_norm": 8.842888832092285,
+      "learning_rate": 0.0001,
+      "loss": 0.2539,
+      "step": 2080
+    },
+    {
+      "epoch": 5.9375,
+      "grad_norm": 9.234344482421875,
+      "learning_rate": 0.0001,
+      "loss": 0.2393,
+      "step": 2090
+    },
+    {
+      "epoch": 5.965909090909091,
+      "grad_norm": 7.840005874633789,
+      "learning_rate": 0.0001,
+      "loss": 0.2418,
+      "step": 2100
+    },
+    {
+      "epoch": 5.994318181818182,
+      "grad_norm": 8.143929481506348,
+      "learning_rate": 0.0001,
+      "loss": 0.2418,
+      "step": 2110
+    },
+    {
+      "epoch": 6.0227272727272725,
+      "grad_norm": 7.842228412628174,
+      "learning_rate": 0.0001,
+      "loss": 0.2343,
+      "step": 2120
+    },
+    {
+      "epoch": 6.051136363636363,
+      "grad_norm": 6.924618721008301,
+      "learning_rate": 0.0001,
+      "loss": 0.2437,
+      "step": 2130
+    },
+    {
+      "epoch": 6.079545454545454,
+      "grad_norm": 7.25029993057251,
+      "learning_rate": 0.0001,
+      "loss": 0.2463,
+      "step": 2140
+    },
+    {
+      "epoch": 6.107954545454546,
+      "grad_norm": 8.335989952087402,
+      "learning_rate": 0.0001,
+      "loss": 0.2473,
+      "step": 2150
+    },
+    {
+      "epoch": 6.136363636363637,
+      "grad_norm": 6.865011215209961,
+      "learning_rate": 0.0001,
+      "loss": 0.2384,
+      "step": 2160
+    },
+    {
+      "epoch": 6.1647727272727275,
+      "grad_norm": 8.29775619506836,
+      "learning_rate": 0.0001,
+      "loss": 0.2382,
+      "step": 2170
+    },
+    {
+      "epoch": 6.193181818181818,
+      "grad_norm": 7.266998767852783,
+      "learning_rate": 0.0001,
+      "loss": 0.2383,
+      "step": 2180
+    },
+    {
+      "epoch": 6.221590909090909,
+      "grad_norm": 7.584468364715576,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 2190
+    },
+    {
+      "epoch": 6.25,
+      "grad_norm": 6.939903259277344,
+      "learning_rate": 0.0001,
+      "loss": 0.2418,
+      "step": 2200
+    },
+    {
+      "epoch": 6.278409090909091,
+      "grad_norm": 6.492012023925781,
+      "learning_rate": 0.0001,
+      "loss": 0.2358,
+      "step": 2210
+    },
+    {
+      "epoch": 6.306818181818182,
+      "grad_norm": 7.337180137634277,
+      "learning_rate": 0.0001,
+      "loss": 0.2336,
+      "step": 2220
+    },
+    {
+      "epoch": 6.3352272727272725,
+      "grad_norm": 8.410757064819336,
+      "learning_rate": 0.0001,
+      "loss": 0.2292,
+      "step": 2230
+    },
+    {
+      "epoch": 6.363636363636363,
+      "grad_norm": 7.204639911651611,
+      "learning_rate": 0.0001,
+      "loss": 0.2321,
+      "step": 2240
+    },
+    {
+      "epoch": 6.392045454545454,
+      "grad_norm": 7.258450508117676,
+      "learning_rate": 0.0001,
+      "loss": 0.2356,
+      "step": 2250
+    },
+    {
+      "epoch": 6.420454545454546,
+      "grad_norm": 8.304643630981445,
+      "learning_rate": 0.0001,
+      "loss": 0.2347,
+      "step": 2260
+    },
+    {
+      "epoch": 6.448863636363637,
+      "grad_norm": 6.700302600860596,
+      "learning_rate": 0.0001,
+      "loss": 0.2309,
+      "step": 2270
+    },
+    {
+      "epoch": 6.4772727272727275,
+      "grad_norm": 7.752438545227051,
+      "learning_rate": 0.0001,
+      "loss": 0.2225,
+      "step": 2280
+    },
+    {
+      "epoch": 6.505681818181818,
+      "grad_norm": 7.962435245513916,
+      "learning_rate": 0.0001,
+      "loss": 0.2247,
+      "step": 2290
+    },
+    {
+      "epoch": 6.534090909090909,
+      "grad_norm": 8.655714988708496,
+      "learning_rate": 0.0001,
+      "loss": 0.2299,
+      "step": 2300
+    },
+    {
+      "epoch": 6.5625,
+      "grad_norm": 6.5540771484375,
+      "learning_rate": 0.0001,
+      "loss": 0.2243,
+      "step": 2310
+    },
+    {
+      "epoch": 6.590909090909091,
+      "grad_norm": 7.325479507446289,
+      "learning_rate": 0.0001,
+      "loss": 0.2209,
+      "step": 2320
+    },
+    {
+      "epoch": 6.619318181818182,
+      "grad_norm": 7.687260150909424,
+      "learning_rate": 0.0001,
+      "loss": 0.2219,
+      "step": 2330
+    },
+    {
+      "epoch": 6.6477272727272725,
+      "grad_norm": 8.61622428894043,
+      "learning_rate": 0.0001,
+      "loss": 0.2177,
+      "step": 2340
+    },
+    {
+      "epoch": 6.676136363636363,
+      "grad_norm": 7.550006866455078,
+      "learning_rate": 0.0001,
+      "loss": 0.2309,
+      "step": 2350
+    },
+    {
+      "epoch": 6.704545454545455,
+      "grad_norm": 7.1695685386657715,
+      "learning_rate": 0.0001,
+      "loss": 0.2278,
+      "step": 2360
+    },
+    {
+      "epoch": 6.732954545454545,
+      "grad_norm": 8.121203422546387,
+      "learning_rate": 0.0001,
+      "loss": 0.2192,
+      "step": 2370
+    },
+    {
+      "epoch": 6.761363636363637,
+      "grad_norm": 9.03805923461914,
+      "learning_rate": 0.0001,
+      "loss": 0.2298,
+      "step": 2380
+    },
+    {
+      "epoch": 6.7897727272727275,
+      "grad_norm": 7.608403205871582,
+      "learning_rate": 0.0001,
+      "loss": 0.23,
+      "step": 2390
+    },
+    {
+      "epoch": 6.818181818181818,
+      "grad_norm": 7.704319953918457,
+      "learning_rate": 0.0001,
+      "loss": 0.228,
+      "step": 2400
+    },
+    {
+      "epoch": 6.846590909090909,
+      "grad_norm": 6.52188777923584,
+      "learning_rate": 0.0001,
+      "loss": 0.2206,
+      "step": 2410
+    },
+    {
+      "epoch": 6.875,
+      "grad_norm": 6.7469635009765625,
+      "learning_rate": 0.0001,
+      "loss": 0.2283,
+      "step": 2420
+    },
+    {
+      "epoch": 6.903409090909091,
+      "grad_norm": 6.883518218994141,
+      "learning_rate": 0.0001,
+      "loss": 0.2123,
+      "step": 2430
+    },
+    {
+      "epoch": 6.931818181818182,
+      "grad_norm": 7.054996013641357,
+      "learning_rate": 0.0001,
+      "loss": 0.2237,
+      "step": 2440
+    },
+    {
+      "epoch": 6.9602272727272725,
+      "grad_norm": 7.665782451629639,
+      "learning_rate": 0.0001,
+      "loss": 0.2202,
+      "step": 2450
+    },
+    {
+      "epoch": 6.988636363636363,
+      "grad_norm": 8.317813873291016,
+      "learning_rate": 0.0001,
+      "loss": 0.2252,
+      "step": 2460
+    },
+    {
+      "epoch": 7.017045454545454,
+      "grad_norm": 7.469433307647705,
+      "learning_rate": 0.0001,
+      "loss": 0.2244,
+      "step": 2470
+    },
+    {
+      "epoch": 7.045454545454546,
+      "grad_norm": 6.86864709854126,
+      "learning_rate": 0.0001,
+      "loss": 0.2132,
+      "step": 2480
+    },
+    {
+      "epoch": 7.073863636363637,
+      "grad_norm": 7.019229412078857,
+      "learning_rate": 0.0001,
+      "loss": 0.2191,
+      "step": 2490
+    },
+    {
+      "epoch": 7.1022727272727275,
+      "grad_norm": 6.337296962738037,
+      "learning_rate": 0.0001,
+      "loss": 0.2135,
+      "step": 2500
+    },
+    {
+      "epoch": 7.130681818181818,
+      "grad_norm": 7.91449499130249,
+      "learning_rate": 0.0001,
+      "loss": 0.2132,
+      "step": 2510
+    },
+    {
+      "epoch": 7.159090909090909,
+      "grad_norm": 7.0960283279418945,
+      "learning_rate": 0.0001,
+      "loss": 0.2058,
+      "step": 2520
+    },
+    {
+      "epoch": 7.1875,
+      "grad_norm": 6.885858058929443,
+      "learning_rate": 0.0001,
+      "loss": 0.205,
+      "step": 2530
+    },
+    {
+      "epoch": 7.215909090909091,
+      "grad_norm": 7.231472015380859,
+      "learning_rate": 0.0001,
+      "loss": 0.2125,
+      "step": 2540
+    },
+    {
+      "epoch": 7.244318181818182,
+      "grad_norm": 6.965603351593018,
+      "learning_rate": 0.0001,
+      "loss": 0.209,
+      "step": 2550
+    },
+    {
+      "epoch": 7.2727272727272725,
+      "grad_norm": 7.230012893676758,
+      "learning_rate": 0.0001,
+      "loss": 0.2062,
+      "step": 2560
+    },
+    {
+      "epoch": 7.301136363636363,
+      "grad_norm": 6.389279842376709,
+      "learning_rate": 0.0001,
+      "loss": 0.2033,
+      "step": 2570
+    },
+    {
+      "epoch": 7.329545454545454,
+      "grad_norm": 6.917042255401611,
+      "learning_rate": 0.0001,
+      "loss": 0.2089,
+      "step": 2580
+    },
+    {
+      "epoch": 7.357954545454546,
+      "grad_norm": 6.476625919342041,
+      "learning_rate": 0.0001,
+      "loss": 0.203,
+      "step": 2590
+    },
+    {
+      "epoch": 7.386363636363637,
+      "grad_norm": 6.501523494720459,
+      "learning_rate": 0.0001,
+      "loss": 0.2034,
+      "step": 2600
+    },
+    {
+      "epoch": 7.4147727272727275,
+      "grad_norm": 4.998976230621338,
+      "learning_rate": 0.0001,
+      "loss": 0.2078,
+      "step": 2610
+    },
+    {
+      "epoch": 7.443181818181818,
+      "grad_norm": 5.617987632751465,
+      "learning_rate": 0.0001,
+      "loss": 0.2065,
+      "step": 2620
+    },
+    {
+      "epoch": 7.471590909090909,
+      "grad_norm": 7.509957313537598,
+      "learning_rate": 0.0001,
+      "loss": 0.2069,
+      "step": 2630
+    },
+    {
+      "epoch": 7.5,
+      "grad_norm": 6.23819637298584,
+      "learning_rate": 0.0001,
+      "loss": 0.2067,
+      "step": 2640
+    },
+    {
+      "epoch": 7.528409090909091,
+      "grad_norm": 6.4593119621276855,
+      "learning_rate": 0.0001,
+      "loss": 0.2081,
+      "step": 2650
+    },
+    {
+      "epoch": 7.556818181818182,
+      "grad_norm": 7.757627964019775,
+      "learning_rate": 0.0001,
+      "loss": 0.2045,
+      "step": 2660
+    },
+    {
+      "epoch": 7.5852272727272725,
+      "grad_norm": 7.729194641113281,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 2670
+    },
+    {
+      "epoch": 7.613636363636363,
+      "grad_norm": 6.746730804443359,
+      "learning_rate": 0.0001,
+      "loss": 0.2055,
+      "step": 2680
+    },
+    {
+      "epoch": 7.642045454545455,
+      "grad_norm": 6.76716947555542,
+      "learning_rate": 0.0001,
+      "loss": 0.2029,
+      "step": 2690
+    },
+    {
+      "epoch": 7.670454545454545,
+      "grad_norm": 6.230428695678711,
+      "learning_rate": 0.0001,
+      "loss": 0.2086,
+      "step": 2700
+    },
+    {
+      "epoch": 7.698863636363637,
+      "grad_norm": 6.170040607452393,
+      "learning_rate": 0.0001,
+      "loss": 0.2088,
+      "step": 2710
+    },
+    {
+      "epoch": 7.7272727272727275,
+      "grad_norm": 6.0955491065979,
+      "learning_rate": 0.0001,
+      "loss": 0.1942,
+      "step": 2720
+    },
+    {
+      "epoch": 7.755681818181818,
+      "grad_norm": 6.41675329208374,
+      "learning_rate": 0.0001,
+      "loss": 0.198,
+      "step": 2730
+    },
+    {
+      "epoch": 7.784090909090909,
+      "grad_norm": 8.517492294311523,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 2740
+    },
+    {
+      "epoch": 7.8125,
+      "grad_norm": 6.808162689208984,
+      "learning_rate": 0.0001,
+      "loss": 0.2017,
+      "step": 2750
+    },
+    {
+      "epoch": 7.840909090909091,
+      "grad_norm": 6.75582218170166,
+      "learning_rate": 0.0001,
+      "loss": 0.2036,
+      "step": 2760
+    },
+    {
+      "epoch": 7.869318181818182,
+      "grad_norm": 6.981121063232422,
+      "learning_rate": 0.0001,
+      "loss": 0.2075,
+      "step": 2770
+    },
+    {
+      "epoch": 7.8977272727272725,
+      "grad_norm": 5.264399528503418,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 2780
+    },
+    {
+      "epoch": 7.926136363636363,
+      "grad_norm": 6.4845781326293945,
+      "learning_rate": 0.0001,
+      "loss": 0.2064,
+      "step": 2790
+    },
+    {
+      "epoch": 7.954545454545455,
+      "grad_norm": 7.397743225097656,
+      "learning_rate": 0.0001,
+      "loss": 0.2047,
+      "step": 2800
+    },
+    {
+      "epoch": 7.982954545454545,
+      "grad_norm": 8.000630378723145,
+      "learning_rate": 0.0001,
+      "loss": 0.2044,
+      "step": 2810
+    },
+    {
+      "epoch": 8.011363636363637,
+      "grad_norm": 6.957930088043213,
+      "learning_rate": 0.0001,
+      "loss": 0.2003,
+      "step": 2820
+    },
+    {
+      "epoch": 8.039772727272727,
+      "grad_norm": 6.850410461425781,
+      "learning_rate": 0.0001,
+      "loss": 0.1958,
+      "step": 2830
+    },
+    {
+      "epoch": 8.068181818181818,
+      "grad_norm": 6.901455402374268,
+      "learning_rate": 0.0001,
+      "loss": 0.2006,
+      "step": 2840
+    },
+    {
+      "epoch": 8.096590909090908,
+      "grad_norm": 5.5034871101379395,
+      "learning_rate": 0.0001,
+      "loss": 0.1876,
+      "step": 2850
+    },
+    {
+      "epoch": 8.125,
+      "grad_norm": 6.498006343841553,
+      "learning_rate": 0.0001,
+      "loss": 0.191,
+      "step": 2860
+    },
+    {
+      "epoch": 8.153409090909092,
+      "grad_norm": 5.930977821350098,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 2870
+    },
+    {
+      "epoch": 8.181818181818182,
+      "grad_norm": 6.002486705780029,
+      "learning_rate": 0.0001,
+      "loss": 0.1952,
+      "step": 2880
+    },
+    {
+      "epoch": 8.210227272727273,
+      "grad_norm": 5.113884925842285,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 2890
+    },
+    {
+      "epoch": 8.238636363636363,
+      "grad_norm": 6.802750587463379,
+      "learning_rate": 0.0001,
+      "loss": 0.1887,
+      "step": 2900
+    },
+    {
+      "epoch": 8.267045454545455,
+      "grad_norm": 5.978296756744385,
+      "learning_rate": 0.0001,
+      "loss": 0.1909,
+      "step": 2910
+    },
+    {
+      "epoch": 8.295454545454545,
+      "grad_norm": 7.176412105560303,
+      "learning_rate": 0.0001,
+      "loss": 0.1913,
+      "step": 2920
+    },
+    {
+      "epoch": 8.323863636363637,
+      "grad_norm": 6.965484619140625,
+      "learning_rate": 0.0001,
+      "loss": 0.1854,
+      "step": 2930
+    },
+    {
+      "epoch": 8.352272727272727,
+      "grad_norm": 5.903598785400391,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 2940
+    },
+    {
+      "epoch": 8.380681818181818,
+      "grad_norm": 6.34436559677124,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 2950
+    },
+    {
+      "epoch": 8.409090909090908,
+      "grad_norm": 5.903111934661865,
+      "learning_rate": 0.0001,
+      "loss": 0.1851,
+      "step": 2960
+    },
+    {
+      "epoch": 8.4375,
+      "grad_norm": 5.883657455444336,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 2970
+    },
+    {
+      "epoch": 8.465909090909092,
+      "grad_norm": 5.767624378204346,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 2980
+    },
+    {
+      "epoch": 8.494318181818182,
+      "grad_norm": 5.390651226043701,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 2990
+    },
+    {
+      "epoch": 8.522727272727273,
+      "grad_norm": 5.619853496551514,
+      "learning_rate": 0.0001,
+      "loss": 0.1804,
+      "step": 3000
+    },
+    {
+      "epoch": 8.551136363636363,
+      "grad_norm": 6.3636932373046875,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 3010
+    },
+    {
+      "epoch": 8.579545454545455,
+      "grad_norm": 6.031747341156006,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 3020
+    },
+    {
+      "epoch": 8.607954545454545,
+      "grad_norm": 5.940463066101074,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 3030
+    },
+    {
+      "epoch": 8.636363636363637,
+      "grad_norm": 5.887471675872803,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 3040
+    },
+    {
+      "epoch": 8.664772727272727,
+      "grad_norm": 5.479545593261719,
+      "learning_rate": 0.0001,
+      "loss": 0.1778,
+      "step": 3050
+    },
+    {
+      "epoch": 8.693181818181818,
+      "grad_norm": 6.690113544464111,
+      "learning_rate": 0.0001,
+      "loss": 0.1786,
+      "step": 3060
+    },
+    {
+      "epoch": 8.721590909090908,
+      "grad_norm": 5.396069049835205,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 3070
+    },
+    {
+      "epoch": 8.75,
+      "grad_norm": 5.759469509124756,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 3080
+    },
+    {
+      "epoch": 8.778409090909092,
+      "grad_norm": 4.685205459594727,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 3090
+    },
+    {
+      "epoch": 8.806818181818182,
+      "grad_norm": 5.000999927520752,
+      "learning_rate": 0.0001,
+      "loss": 0.1808,
+      "step": 3100
+    },
+    {
+      "epoch": 8.835227272727273,
+      "grad_norm": 5.158972263336182,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 3110
+    },
+    {
+      "epoch": 8.863636363636363,
+      "grad_norm": 5.847781658172607,
+      "learning_rate": 0.0001,
+      "loss": 0.1851,
+      "step": 3120
+    },
+    {
+      "epoch": 8.892045454545455,
+      "grad_norm": 4.9706645011901855,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 3130
+    },
+    {
+      "epoch": 8.920454545454545,
+      "grad_norm": 5.0156660079956055,
+      "learning_rate": 0.0001,
+      "loss": 0.1825,
+      "step": 3140
+    },
+    {
+      "epoch": 8.948863636363637,
+      "grad_norm": 5.2722344398498535,
+      "learning_rate": 0.0001,
+      "loss": 0.1794,
+      "step": 3150
+    },
+    {
+      "epoch": 8.977272727272727,
+      "grad_norm": 4.946606159210205,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 3160
+    },
+    {
+      "epoch": 9.005681818181818,
+      "grad_norm": 5.0111846923828125,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 3170
+    },
+    {
+      "epoch": 9.034090909090908,
+      "grad_norm": 4.587785243988037,
+      "learning_rate": 0.0001,
+      "loss": 0.1725,
+      "step": 3180
+    },
+    {
+      "epoch": 9.0625,
+      "grad_norm": 4.933738708496094,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 3190
+    },
+    {
+      "epoch": 9.090909090909092,
+      "grad_norm": 5.0207037925720215,
+      "learning_rate": 0.0001,
+      "loss": 0.1804,
+      "step": 3200
+    },
+    {
+      "epoch": 9.119318181818182,
+      "grad_norm": 6.469820022583008,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 3210
+    },
+    {
+      "epoch": 9.147727272727273,
+      "grad_norm": 5.247611999511719,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 3220
+    },
+    {
+      "epoch": 9.176136363636363,
+      "grad_norm": 4.957090854644775,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 3230
+    },
+    {
+      "epoch": 9.204545454545455,
+      "grad_norm": 4.917489051818848,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 3240
+    },
+    {
+      "epoch": 9.232954545454545,
+      "grad_norm": 6.98730993270874,
+      "learning_rate": 0.0001,
+      "loss": 0.1773,
+      "step": 3250
+    },
+    {
+      "epoch": 9.261363636363637,
+      "grad_norm": 5.937990665435791,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 3260
+    },
+    {
+      "epoch": 9.289772727272727,
+      "grad_norm": 6.112240791320801,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 3270
+    },
+    {
+      "epoch": 9.318181818181818,
+      "grad_norm": 5.8593878746032715,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 3280
+    },
+    {
+      "epoch": 9.346590909090908,
+      "grad_norm": 6.075056552886963,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 3290
+    },
+    {
+      "epoch": 9.375,
+      "grad_norm": 5.816572666168213,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 3300
+    },
+    {
+      "epoch": 9.403409090909092,
+      "grad_norm": 6.339922904968262,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 3310
+    },
+    {
+      "epoch": 9.431818181818182,
+      "grad_norm": 5.111523628234863,
+      "learning_rate": 0.0001,
+      "loss": 0.1618,
+      "step": 3320
+    },
+    {
+      "epoch": 9.460227272727273,
+      "grad_norm": 5.104013442993164,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 3330
+    },
+    {
+      "epoch": 9.488636363636363,
+      "grad_norm": 4.7531280517578125,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 3340
+    },
+    {
+      "epoch": 9.517045454545455,
+      "grad_norm": 4.486930847167969,
+      "learning_rate": 0.0001,
+      "loss": 0.1729,
+      "step": 3350
+    },
+    {
+      "epoch": 9.545454545454545,
+      "grad_norm": 5.003032684326172,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 3360
+    },
+    {
+      "epoch": 9.573863636363637,
+      "grad_norm": 5.644103050231934,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 3370
+    },
+    {
+      "epoch": 9.602272727272727,
+      "grad_norm": 5.101214408874512,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 3380
+    },
+    {
+      "epoch": 9.630681818181818,
+      "grad_norm": 5.529112815856934,
+      "learning_rate": 0.0001,
+      "loss": 0.1715,
+      "step": 3390
+    },
+    {
+      "epoch": 9.659090909090908,
+      "grad_norm": 5.411925792694092,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 3400
+    },
+    {
+      "epoch": 9.6875,
+      "grad_norm": 5.155153751373291,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 3410
+    },
+    {
+      "epoch": 9.715909090909092,
+      "grad_norm": 4.77042293548584,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 3420
+    },
+    {
+      "epoch": 9.744318181818182,
+      "grad_norm": 4.622435092926025,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 3430
+    },
+    {
+      "epoch": 9.772727272727273,
+      "grad_norm": 5.802976131439209,
+      "learning_rate": 0.0001,
+      "loss": 0.1712,
+      "step": 3440
+    },
+    {
+      "epoch": 9.801136363636363,
+      "grad_norm": 4.810296058654785,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 3450
+    },
+    {
+      "epoch": 9.829545454545455,
+      "grad_norm": 5.124487400054932,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 3460
+    },
+    {
+      "epoch": 9.857954545454545,
+      "grad_norm": 5.081210136413574,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 3470
+    },
+    {
+      "epoch": 9.886363636363637,
+      "grad_norm": 5.038453578948975,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 3480
+    },
+    {
+      "epoch": 9.914772727272727,
+      "grad_norm": 4.524289608001709,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 3490
+    },
+    {
+      "epoch": 9.943181818181818,
+      "grad_norm": 5.175899505615234,
+      "learning_rate": 0.0001,
+      "loss": 0.1601,
+      "step": 3500
+    },
+    {
+      "epoch": 9.971590909090908,
+      "grad_norm": 4.064411640167236,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 3510
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 4.967013835906982,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 3520
+    },
+    {
+      "epoch": 10.028409090909092,
+      "grad_norm": 5.4418535232543945,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 3530
+    },
+    {
+      "epoch": 10.056818181818182,
+      "grad_norm": 5.082826614379883,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 3540
+    },
+    {
+      "epoch": 10.085227272727273,
+      "grad_norm": 4.592067241668701,
+      "learning_rate": 0.0001,
+      "loss": 0.1623,
+      "step": 3550
+    },
+    {
+      "epoch": 10.113636363636363,
+      "grad_norm": 5.288888931274414,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 3560
+    },
+    {
+      "epoch": 10.142045454545455,
+      "grad_norm": 5.104770660400391,
+      "learning_rate": 0.0001,
+      "loss": 0.1593,
+      "step": 3570
+    },
+    {
+      "epoch": 10.170454545454545,
+      "grad_norm": 4.773959159851074,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 3580
+    },
+    {
+      "epoch": 10.198863636363637,
+      "grad_norm": 4.410947799682617,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 3590
+    },
+    {
+      "epoch": 10.227272727272727,
+      "grad_norm": 4.374294281005859,
+      "learning_rate": 0.0001,
+      "loss": 0.158,
+      "step": 3600
+    },
+    {
+      "epoch": 10.255681818181818,
+      "grad_norm": 4.402506351470947,
+      "learning_rate": 0.0001,
+      "loss": 0.1567,
+      "step": 3610
+    },
+    {
+      "epoch": 10.284090909090908,
+      "grad_norm": 5.090147495269775,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 3620
+    },
+    {
+      "epoch": 10.3125,
+      "grad_norm": 5.5478081703186035,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 3630
+    },
+    {
+      "epoch": 10.340909090909092,
+      "grad_norm": 5.645622730255127,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 3640
+    },
+    {
+      "epoch": 10.369318181818182,
+      "grad_norm": 4.826333999633789,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 3650
+    },
+    {
+      "epoch": 10.397727272727273,
+      "grad_norm": 5.210224628448486,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 3660
+    },
+    {
+      "epoch": 10.426136363636363,
+      "grad_norm": 3.516092300415039,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 3670
+    },
+    {
+      "epoch": 10.454545454545455,
+      "grad_norm": 4.710558891296387,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 3680
+    },
+    {
+      "epoch": 10.482954545454545,
+      "grad_norm": 4.940939903259277,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 3690
+    },
+    {
+      "epoch": 10.511363636363637,
+      "grad_norm": 4.353475093841553,
+      "learning_rate": 0.0001,
+      "loss": 0.1537,
+      "step": 3700
+    },
+    {
+      "epoch": 10.539772727272727,
+      "grad_norm": 3.7736759185791016,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 3710
+    },
+    {
+      "epoch": 10.568181818181818,
+      "grad_norm": 4.482377529144287,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 3720
+    },
+    {
+      "epoch": 10.596590909090908,
+      "grad_norm": 4.80997896194458,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 3730
+    },
+    {
+      "epoch": 10.625,
+      "grad_norm": 6.185352802276611,
+      "learning_rate": 0.0001,
+      "loss": 0.1555,
+      "step": 3740
+    },
+    {
+      "epoch": 10.653409090909092,
+      "grad_norm": 5.383978366851807,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 3750
+    },
+    {
+      "epoch": 10.681818181818182,
+      "grad_norm": 6.075902938842773,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 3760
+    },
+    {
+      "epoch": 10.710227272727273,
+      "grad_norm": 5.537624835968018,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 3770
+    },
+    {
+      "epoch": 10.738636363636363,
+      "grad_norm": 4.914467811584473,
+      "learning_rate": 0.0001,
+      "loss": 0.1555,
+      "step": 3780
+    },
+    {
+      "epoch": 10.767045454545455,
+      "grad_norm": 4.567920684814453,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 3790
+    },
+    {
+      "epoch": 10.795454545454545,
+      "grad_norm": 4.5670390129089355,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 3800
+    },
+    {
+      "epoch": 10.823863636363637,
+      "grad_norm": 3.629544734954834,
+      "learning_rate": 0.0001,
+      "loss": 0.1502,
+      "step": 3810
+    },
+    {
+      "epoch": 10.852272727272727,
+      "grad_norm": 4.088180065155029,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 3820
+    },
+    {
+      "epoch": 10.880681818181818,
+      "grad_norm": 4.8524017333984375,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 3830
+    },
+    {
+      "epoch": 10.909090909090908,
+      "grad_norm": 5.3502888679504395,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 3840
+    },
+    {
+      "epoch": 10.9375,
+      "grad_norm": 4.959495544433594,
+      "learning_rate": 0.0001,
+      "loss": 0.1549,
+      "step": 3850
+    },
+    {
+      "epoch": 10.965909090909092,
+      "grad_norm": 4.991962432861328,
+      "learning_rate": 0.0001,
+      "loss": 0.1504,
+      "step": 3860
+    },
+    {
+      "epoch": 10.994318181818182,
+      "grad_norm": 4.054560661315918,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 3870
+    },
+    {
+      "epoch": 11.022727272727273,
+      "grad_norm": 3.9775209426879883,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 3880
+    },
+    {
+      "epoch": 11.051136363636363,
+      "grad_norm": 4.538222312927246,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 3890
+    },
+    {
+      "epoch": 11.079545454545455,
+      "grad_norm": 5.487000465393066,
+      "learning_rate": 0.0001,
+      "loss": 0.15,
+      "step": 3900
+    },
+    {
+      "epoch": 11.107954545454545,
+      "grad_norm": 5.862754821777344,
+      "learning_rate": 0.0001,
+      "loss": 0.1493,
+      "step": 3910
+    },
+    {
+      "epoch": 11.136363636363637,
+      "grad_norm": 4.4752302169799805,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 3920
+    },
+    {
+      "epoch": 11.164772727272727,
+      "grad_norm": 4.51123571395874,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 3930
+    },
+    {
+      "epoch": 11.193181818181818,
+      "grad_norm": 4.44078254699707,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 3940
+    },
+    {
+      "epoch": 11.221590909090908,
+      "grad_norm": 4.542746067047119,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 3950
+    },
+    {
+      "epoch": 11.25,
+      "grad_norm": 4.513556003570557,
+      "learning_rate": 0.0001,
+      "loss": 0.1456,
+      "step": 3960
+    },
+    {
+      "epoch": 11.278409090909092,
+      "grad_norm": 5.227005958557129,
+      "learning_rate": 0.0001,
+      "loss": 0.146,
+      "step": 3970
+    },
+    {
+      "epoch": 11.306818181818182,
+      "grad_norm": 4.134369850158691,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 3980
+    },
+    {
+      "epoch": 11.335227272727273,
+      "grad_norm": 5.030073642730713,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 3990
+    },
+    {
+      "epoch": 11.363636363636363,
+      "grad_norm": 4.397629737854004,
+      "learning_rate": 0.0001,
+      "loss": 0.1462,
+      "step": 4000
+    },
+    {
+      "epoch": 11.392045454545455,
+      "grad_norm": 4.636000633239746,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 4010
+    },
+    {
+      "epoch": 11.420454545454545,
+      "grad_norm": 4.899885177612305,
+      "learning_rate": 0.0001,
+      "loss": 0.1445,
+      "step": 4020
+    },
+    {
+      "epoch": 11.448863636363637,
+      "grad_norm": 4.209653377532959,
+      "learning_rate": 0.0001,
+      "loss": 0.1517,
+      "step": 4030
+    },
+    {
+      "epoch": 11.477272727272727,
+      "grad_norm": 4.315791606903076,
+      "learning_rate": 0.0001,
+      "loss": 0.1423,
+      "step": 4040
+    },
+    {
+      "epoch": 11.505681818181818,
+      "grad_norm": 4.065213203430176,
+      "learning_rate": 0.0001,
+      "loss": 0.1429,
+      "step": 4050
+    },
+    {
+      "epoch": 11.534090909090908,
+      "grad_norm": 4.354069709777832,
+      "learning_rate": 0.0001,
+      "loss": 0.1452,
+      "step": 4060
+    },
+    {
+      "epoch": 11.5625,
+      "grad_norm": 4.485837459564209,
+      "learning_rate": 0.0001,
+      "loss": 0.1429,
+      "step": 4070
+    },
+    {
+      "epoch": 11.590909090909092,
+      "grad_norm": 4.509272575378418,
+      "learning_rate": 0.0001,
+      "loss": 0.1437,
+      "step": 4080
+    },
+    {
+      "epoch": 11.619318181818182,
+      "grad_norm": 4.269772052764893,
+      "learning_rate": 0.0001,
+      "loss": 0.1468,
+      "step": 4090
+    },
+    {
+      "epoch": 11.647727272727273,
+      "grad_norm": 4.422598361968994,
+      "learning_rate": 0.0001,
+      "loss": 0.151,
+      "step": 4100
+    },
+    {
+      "epoch": 11.676136363636363,
+      "grad_norm": 4.730630874633789,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 4110
+    },
+    {
+      "epoch": 11.704545454545455,
+      "grad_norm": 5.042013645172119,
+      "learning_rate": 0.0001,
+      "loss": 0.1476,
+      "step": 4120
+    },
+    {
+      "epoch": 11.732954545454545,
+      "grad_norm": 4.182816982269287,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 4130
+    },
+    {
+      "epoch": 11.761363636363637,
+      "grad_norm": 4.254685401916504,
+      "learning_rate": 0.0001,
+      "loss": 0.1371,
+      "step": 4140
+    },
+    {
+      "epoch": 11.789772727272727,
+      "grad_norm": 4.958248138427734,
+      "learning_rate": 0.0001,
+      "loss": 0.1465,
+      "step": 4150
+    },
+    {
+      "epoch": 11.818181818181818,
+      "grad_norm": 4.743212699890137,
+      "learning_rate": 0.0001,
+      "loss": 0.1448,
+      "step": 4160
+    },
+    {
+      "epoch": 11.846590909090908,
+      "grad_norm": 4.2032084465026855,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 4170
+    },
+    {
+      "epoch": 11.875,
+      "grad_norm": 4.244325637817383,
+      "learning_rate": 0.0001,
+      "loss": 0.1397,
+      "step": 4180
+    },
+    {
+      "epoch": 11.903409090909092,
+      "grad_norm": 3.134256362915039,
+      "learning_rate": 0.0001,
+      "loss": 0.1443,
+      "step": 4190
+    },
+    {
+      "epoch": 11.931818181818182,
+      "grad_norm": 4.238053321838379,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 4200
+    },
+    {
+      "epoch": 11.960227272727273,
+      "grad_norm": 4.34376335144043,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 4210
+    },
+    {
+      "epoch": 11.988636363636363,
+      "grad_norm": 3.7817201614379883,
+      "learning_rate": 0.0001,
+      "loss": 0.1441,
+      "step": 4220
+    },
+    {
+      "epoch": 12.017045454545455,
+      "grad_norm": 3.5958733558654785,
+      "learning_rate": 0.0001,
+      "loss": 0.1373,
+      "step": 4230
+    },
+    {
+      "epoch": 12.045454545454545,
+      "grad_norm": 4.10888147354126,
+      "learning_rate": 0.0001,
+      "loss": 0.1405,
+      "step": 4240
+    },
+    {
+      "epoch": 12.073863636363637,
+      "grad_norm": 3.802342176437378,
+      "learning_rate": 0.0001,
+      "loss": 0.1393,
+      "step": 4250
+    },
+    {
+      "epoch": 12.102272727272727,
+      "grad_norm": 4.85184907913208,
+      "learning_rate": 0.0001,
+      "loss": 0.1359,
+      "step": 4260
+    },
+    {
+      "epoch": 12.130681818181818,
+      "grad_norm": 4.548974514007568,
+      "learning_rate": 0.0001,
+      "loss": 0.1402,
+      "step": 4270
+    },
+    {
+      "epoch": 12.159090909090908,
+      "grad_norm": 4.047370433807373,
+      "learning_rate": 0.0001,
+      "loss": 0.143,
+      "step": 4280
+    },
+    {
+      "epoch": 12.1875,
+      "grad_norm": 4.97476863861084,
+      "learning_rate": 0.0001,
+      "loss": 0.1439,
+      "step": 4290
+    },
+    {
+      "epoch": 12.215909090909092,
+      "grad_norm": 4.076110363006592,
+      "learning_rate": 0.0001,
+      "loss": 0.145,
+      "step": 4300
+    },
+    {
+      "epoch": 12.244318181818182,
+      "grad_norm": 4.098419189453125,
+      "learning_rate": 0.0001,
+      "loss": 0.1423,
+      "step": 4310
+    },
+    {
+      "epoch": 12.272727272727273,
+      "grad_norm": 3.961846351623535,
+      "learning_rate": 0.0001,
+      "loss": 0.1369,
+      "step": 4320
+    },
+    {
+      "epoch": 12.301136363636363,
+      "grad_norm": 4.079448223114014,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 4330
+    },
+    {
+      "epoch": 12.329545454545455,
+      "grad_norm": 3.375678777694702,
+      "learning_rate": 0.0001,
+      "loss": 0.1368,
+      "step": 4340
+    },
+    {
+      "epoch": 12.357954545454545,
+      "grad_norm": 3.7309460639953613,
+      "learning_rate": 0.0001,
+      "loss": 0.1338,
+      "step": 4350
+    },
+    {
+      "epoch": 12.386363636363637,
+      "grad_norm": 4.20289421081543,
+      "learning_rate": 0.0001,
+      "loss": 0.1429,
+      "step": 4360
+    },
+    {
+      "epoch": 12.414772727272727,
+      "grad_norm": 4.175302982330322,
+      "learning_rate": 0.0001,
+      "loss": 0.139,
+      "step": 4370
+    },
+    {
+      "epoch": 12.443181818181818,
+      "grad_norm": 3.7921714782714844,
+      "learning_rate": 0.0001,
+      "loss": 0.1408,
+      "step": 4380
+    },
+    {
+      "epoch": 12.471590909090908,
+      "grad_norm": 4.009100437164307,
+      "learning_rate": 0.0001,
+      "loss": 0.1432,
+      "step": 4390
+    },
+    {
+      "epoch": 12.5,
+      "grad_norm": 3.71403431892395,
+      "learning_rate": 0.0001,
+      "loss": 0.1381,
+      "step": 4400
+    },
+    {
+      "epoch": 12.528409090909092,
+      "grad_norm": 4.153659820556641,
+      "learning_rate": 0.0001,
+      "loss": 0.1362,
+      "step": 4410
+    },
+    {
+      "epoch": 12.556818181818182,
+      "grad_norm": 3.8363094329833984,
+      "learning_rate": 0.0001,
+      "loss": 0.1434,
+      "step": 4420
+    },
+    {
+      "epoch": 12.585227272727273,
+      "grad_norm": 3.9091579914093018,
+      "learning_rate": 0.0001,
+      "loss": 0.1372,
+      "step": 4430
+    },
+    {
+      "epoch": 12.613636363636363,
+      "grad_norm": 4.5517578125,
+      "learning_rate": 0.0001,
+      "loss": 0.1388,
+      "step": 4440
+    },
+    {
+      "epoch": 12.642045454545455,
+      "grad_norm": 3.891643762588501,
+      "learning_rate": 0.0001,
+      "loss": 0.1361,
+      "step": 4450
+    },
+    {
+      "epoch": 12.670454545454545,
+      "grad_norm": 3.9435248374938965,
+      "learning_rate": 0.0001,
+      "loss": 0.1417,
+      "step": 4460
+    },
+    {
+      "epoch": 12.698863636363637,
+      "grad_norm": 3.625453472137451,
+      "learning_rate": 0.0001,
+      "loss": 0.1392,
+      "step": 4470
+    },
+    {
+      "epoch": 12.727272727272727,
+      "grad_norm": 4.054428577423096,
+      "learning_rate": 0.0001,
+      "loss": 0.1335,
+      "step": 4480
+    },
+    {
+      "epoch": 12.755681818181818,
+      "grad_norm": 4.017980098724365,
+      "learning_rate": 0.0001,
+      "loss": 0.1409,
+      "step": 4490
+    },
+    {
+      "epoch": 12.784090909090908,
+      "grad_norm": 3.3853940963745117,
+      "learning_rate": 0.0001,
+      "loss": 0.1407,
+      "step": 4500
+    },
+    {
+      "epoch": 12.8125,
+      "grad_norm": 3.403177261352539,
+      "learning_rate": 0.0001,
+      "loss": 0.1344,
+      "step": 4510
+    },
+    {
+      "epoch": 12.840909090909092,
+      "grad_norm": 3.364267349243164,
+      "learning_rate": 0.0001,
+      "loss": 0.1379,
+      "step": 4520
+    },
+    {
+      "epoch": 12.869318181818182,
+      "grad_norm": 4.48183012008667,
+      "learning_rate": 0.0001,
+      "loss": 0.1407,
+      "step": 4530
+    },
+    {
+      "epoch": 12.897727272727273,
+      "grad_norm": 3.5637905597686768,
+      "learning_rate": 0.0001,
+      "loss": 0.1404,
+      "step": 4540
+    },
+    {
+      "epoch": 12.926136363636363,
+      "grad_norm": 3.4277963638305664,
+      "learning_rate": 0.0001,
+      "loss": 0.1357,
+      "step": 4550
+    },
+    {
+      "epoch": 12.954545454545455,
+      "grad_norm": 3.4155449867248535,
+      "learning_rate": 0.0001,
+      "loss": 0.1385,
+      "step": 4560
+    },
+    {
+      "epoch": 12.982954545454545,
+      "grad_norm": 3.1836628913879395,
+      "learning_rate": 0.0001,
+      "loss": 0.1381,
+      "step": 4570
+    },
+    {
+      "epoch": 13.011363636363637,
+      "grad_norm": 4.119326591491699,
+      "learning_rate": 0.0001,
+      "loss": 0.1311,
+      "step": 4580
+    },
+    {
+      "epoch": 13.039772727272727,
+      "grad_norm": 3.324186086654663,
+      "learning_rate": 0.0001,
+      "loss": 0.1341,
+      "step": 4590
+    },
+    {
+      "epoch": 13.068181818181818,
+      "grad_norm": 3.69582200050354,
+      "learning_rate": 0.0001,
+      "loss": 0.1373,
+      "step": 4600
+    },
+    {
+      "epoch": 13.096590909090908,
+      "grad_norm": 3.6252574920654297,
+      "learning_rate": 0.0001,
+      "loss": 0.1368,
+      "step": 4610
+    },
+    {
+      "epoch": 13.125,
+      "grad_norm": 3.5859949588775635,
+      "learning_rate": 0.0001,
+      "loss": 0.1303,
+      "step": 4620
+    },
+    {
+      "epoch": 13.153409090909092,
+      "grad_norm": 4.536507606506348,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 4630
+    },
+    {
+      "epoch": 13.181818181818182,
+      "grad_norm": 3.7678303718566895,
+      "learning_rate": 0.0001,
+      "loss": 0.1376,
+      "step": 4640
+    },
+    {
+      "epoch": 13.210227272727273,
+      "grad_norm": 3.8305280208587646,
+      "learning_rate": 0.0001,
+      "loss": 0.1299,
+      "step": 4650
+    },
+    {
+      "epoch": 13.238636363636363,
+      "grad_norm": 4.209882736206055,
+      "learning_rate": 0.0001,
+      "loss": 0.1366,
+      "step": 4660
+    },
+    {
+      "epoch": 13.267045454545455,
+      "grad_norm": 3.751279354095459,
+      "learning_rate": 0.0001,
+      "loss": 0.1395,
+      "step": 4670
+    },
+    {
+      "epoch": 13.295454545454545,
+      "grad_norm": 3.758382558822632,
+      "learning_rate": 0.0001,
+      "loss": 0.1371,
+      "step": 4680
+    },
+    {
+      "epoch": 13.323863636363637,
+      "grad_norm": 4.068879127502441,
+      "learning_rate": 0.0001,
+      "loss": 0.1335,
+      "step": 4690
+    },
+    {
+      "epoch": 13.352272727272727,
+      "grad_norm": 4.470997333526611,
+      "learning_rate": 0.0001,
+      "loss": 0.1349,
+      "step": 4700
+    },
+    {
+      "epoch": 13.380681818181818,
+      "grad_norm": 3.5465259552001953,
+      "learning_rate": 0.0001,
+      "loss": 0.1336,
+      "step": 4710
+    },
+    {
+      "epoch": 13.409090909090908,
+      "grad_norm": 3.6585092544555664,
+      "learning_rate": 0.0001,
+      "loss": 0.1279,
+      "step": 4720
+    },
+    {
+      "epoch": 13.4375,
+      "grad_norm": 3.6728506088256836,
+      "learning_rate": 0.0001,
+      "loss": 0.1289,
+      "step": 4730
+    },
+    {
+      "epoch": 13.465909090909092,
+      "grad_norm": 3.1070103645324707,
+      "learning_rate": 0.0001,
+      "loss": 0.1293,
+      "step": 4740
+    },
+    {
+      "epoch": 13.494318181818182,
+      "grad_norm": 2.9372332096099854,
+      "learning_rate": 0.0001,
+      "loss": 0.1329,
+      "step": 4750
+    },
+    {
+      "epoch": 13.522727272727273,
+      "grad_norm": 3.2514431476593018,
+      "learning_rate": 0.0001,
+      "loss": 0.1287,
+      "step": 4760
+    },
+    {
+      "epoch": 13.551136363636363,
+      "grad_norm": 2.84192156791687,
+      "learning_rate": 0.0001,
+      "loss": 0.1372,
+      "step": 4770
+    },
+    {
+      "epoch": 13.579545454545455,
+      "grad_norm": 3.847137928009033,
+      "learning_rate": 0.0001,
+      "loss": 0.1321,
+      "step": 4780
+    },
+    {
+      "epoch": 13.607954545454545,
+      "grad_norm": 3.7606923580169678,
+      "learning_rate": 0.0001,
+      "loss": 0.1337,
+      "step": 4790
+    },
+    {
+      "epoch": 13.636363636363637,
+      "grad_norm": 3.415740966796875,
+      "learning_rate": 0.0001,
+      "loss": 0.1327,
+      "step": 4800
+    },
+    {
+      "epoch": 13.664772727272727,
+      "grad_norm": 3.71706485748291,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 4810
+    },
+    {
+      "epoch": 13.693181818181818,
+      "grad_norm": 3.2357699871063232,
+      "learning_rate": 0.0001,
+      "loss": 0.1308,
+      "step": 4820
+    },
+    {
+      "epoch": 13.721590909090908,
+      "grad_norm": 3.241356372833252,
+      "learning_rate": 0.0001,
+      "loss": 0.1369,
+      "step": 4830
+    },
+    {
+      "epoch": 13.75,
+      "grad_norm": 3.0397732257843018,
+      "learning_rate": 0.0001,
+      "loss": 0.1356,
+      "step": 4840
+    },
+    {
+      "epoch": 13.778409090909092,
+      "grad_norm": 3.939297914505005,
+      "learning_rate": 0.0001,
+      "loss": 0.1342,
+      "step": 4850
+    },
+    {
+      "epoch": 13.806818181818182,
+      "grad_norm": 3.530168294906616,
+      "learning_rate": 0.0001,
+      "loss": 0.1345,
+      "step": 4860
+    },
+    {
+      "epoch": 13.835227272727273,
+      "grad_norm": 3.2555956840515137,
+      "learning_rate": 0.0001,
+      "loss": 0.1304,
+      "step": 4870
+    },
+    {
+      "epoch": 13.863636363636363,
+      "grad_norm": 3.490713357925415,
+      "learning_rate": 0.0001,
+      "loss": 0.132,
+      "step": 4880
+    },
+    {
+      "epoch": 13.892045454545455,
+      "grad_norm": 3.034759521484375,
+      "learning_rate": 0.0001,
+      "loss": 0.1287,
+      "step": 4890
+    },
+    {
+      "epoch": 13.920454545454545,
+      "grad_norm": 3.2557218074798584,
+      "learning_rate": 0.0001,
+      "loss": 0.1358,
+      "step": 4900
+    },
+    {
+      "epoch": 13.948863636363637,
+      "grad_norm": 3.692721128463745,
+      "learning_rate": 0.0001,
+      "loss": 0.1305,
+      "step": 4910
+    },
+    {
+      "epoch": 13.977272727272727,
+      "grad_norm": 3.3548946380615234,
+      "learning_rate": 0.0001,
+      "loss": 0.1306,
+      "step": 4920
+    },
+    {
+      "epoch": 14.005681818181818,
+      "grad_norm": 2.9304184913635254,
+      "learning_rate": 0.0001,
+      "loss": 0.1268,
+      "step": 4930
+    },
+    {
+      "epoch": 14.034090909090908,
+      "grad_norm": 2.7205934524536133,
+      "learning_rate": 0.0001,
+      "loss": 0.1267,
+      "step": 4940
+    },
+    {
+      "epoch": 14.0625,
+      "grad_norm": 3.1881885528564453,
+      "learning_rate": 0.0001,
+      "loss": 0.1292,
+      "step": 4950
+    },
+    {
+      "epoch": 14.090909090909092,
+      "grad_norm": 2.813159465789795,
+      "learning_rate": 0.0001,
+      "loss": 0.1281,
+      "step": 4960
+    },
+    {
+      "epoch": 14.119318181818182,
+      "grad_norm": 3.351205348968506,
+      "learning_rate": 0.0001,
+      "loss": 0.1336,
+      "step": 4970
+    },
+    {
+      "epoch": 14.147727272727273,
+      "grad_norm": 3.1499414443969727,
+      "learning_rate": 0.0001,
+      "loss": 0.1327,
+      "step": 4980
+    },
+    {
+      "epoch": 14.176136363636363,
+      "grad_norm": 3.1446123123168945,
+      "learning_rate": 0.0001,
+      "loss": 0.1296,
+      "step": 4990
+    },
+    {
+      "epoch": 14.204545454545455,
+      "grad_norm": 4.177588939666748,
+      "learning_rate": 0.0001,
+      "loss": 0.1269,
+      "step": 5000
+    },
+    {
+      "epoch": 14.232954545454545,
+      "grad_norm": 3.2882914543151855,
+      "learning_rate": 0.0001,
+      "loss": 0.1294,
+      "step": 5010
+    },
+    {
+      "epoch": 14.261363636363637,
+      "grad_norm": 3.151151657104492,
+      "learning_rate": 0.0001,
+      "loss": 0.128,
+      "step": 5020
+    },
+    {
+      "epoch": 14.289772727272727,
+      "grad_norm": 3.507800340652466,
+      "learning_rate": 0.0001,
+      "loss": 0.1306,
+      "step": 5030
+    },
+    {
+      "epoch": 14.318181818181818,
+      "grad_norm": 3.266287088394165,
+      "learning_rate": 0.0001,
+      "loss": 0.13,
+      "step": 5040
+    },
+    {
+      "epoch": 14.346590909090908,
+      "grad_norm": 3.7392666339874268,
+      "learning_rate": 0.0001,
+      "loss": 0.1246,
+      "step": 5050
+    },
+    {
+      "epoch": 14.375,
+      "grad_norm": 3.385209083557129,
+      "learning_rate": 0.0001,
+      "loss": 0.1206,
+      "step": 5060
+    },
+    {
+      "epoch": 14.403409090909092,
+      "grad_norm": 3.0839014053344727,
+      "learning_rate": 0.0001,
+      "loss": 0.119,
+      "step": 5070
+    },
+    {
+      "epoch": 14.431818181818182,
+      "grad_norm": 2.9895691871643066,
+      "learning_rate": 0.0001,
+      "loss": 0.13,
+      "step": 5080
+    },
+    {
+      "epoch": 14.460227272727273,
+      "grad_norm": 3.4198343753814697,
+      "learning_rate": 0.0001,
+      "loss": 0.1325,
+      "step": 5090
+    },
+    {
+      "epoch": 14.488636363636363,
+      "grad_norm": 3.2257754802703857,
+      "learning_rate": 0.0001,
+      "loss": 0.13,
+      "step": 5100
+    },
+    {
+      "epoch": 14.517045454545455,
+      "grad_norm": 2.9251694679260254,
+      "learning_rate": 0.0001,
+      "loss": 0.1247,
+      "step": 5110
+    },
+    {
+      "epoch": 14.545454545454545,
+      "grad_norm": 3.3132123947143555,
+      "learning_rate": 0.0001,
+      "loss": 0.1246,
+      "step": 5120
+    },
+    {
+      "epoch": 14.573863636363637,
+      "grad_norm": 3.5017828941345215,
+      "learning_rate": 0.0001,
+      "loss": 0.1265,
+      "step": 5130
+    },
+    {
+      "epoch": 14.602272727272727,
+      "grad_norm": 3.087315559387207,
+      "learning_rate": 0.0001,
+      "loss": 0.1275,
+      "step": 5140
+    },
+    {
+      "epoch": 14.630681818181818,
+      "grad_norm": 2.8191609382629395,
+      "learning_rate": 0.0001,
+      "loss": 0.1278,
+      "step": 5150
+    },
+    {
+      "epoch": 14.659090909090908,
+      "grad_norm": 3.038038492202759,
+      "learning_rate": 0.0001,
+      "loss": 0.1197,
+      "step": 5160
+    },
+    {
+      "epoch": 14.6875,
+      "grad_norm": 2.9609692096710205,
+      "learning_rate": 0.0001,
+      "loss": 0.1212,
+      "step": 5170
+    },
+    {
+      "epoch": 14.715909090909092,
+      "grad_norm": 3.029618263244629,
+      "learning_rate": 0.0001,
+      "loss": 0.1235,
+      "step": 5180
+    },
+    {
+      "epoch": 14.744318181818182,
+      "grad_norm": 2.6114909648895264,
+      "learning_rate": 0.0001,
+      "loss": 0.1286,
+      "step": 5190
+    },
+    {
+      "epoch": 14.772727272727273,
+      "grad_norm": 2.887552261352539,
+      "learning_rate": 0.0001,
+      "loss": 0.126,
+      "step": 5200
+    },
+    {
+      "epoch": 14.801136363636363,
+      "grad_norm": 3.0050230026245117,
+      "learning_rate": 0.0001,
+      "loss": 0.1266,
+      "step": 5210
+    },
+    {
+      "epoch": 14.829545454545455,
+      "grad_norm": 3.215804100036621,
+      "learning_rate": 0.0001,
+      "loss": 0.1281,
+      "step": 5220
+    },
+    {
+      "epoch": 14.857954545454545,
+      "grad_norm": 3.709592819213867,
+      "learning_rate": 0.0001,
+      "loss": 0.1325,
+      "step": 5230
+    },
+    {
+      "epoch": 14.886363636363637,
+      "grad_norm": 3.143139600753784,
+      "learning_rate": 0.0001,
+      "loss": 0.1323,
+      "step": 5240
+    },
+    {
+      "epoch": 14.914772727272727,
+      "grad_norm": 3.504509925842285,
+      "learning_rate": 0.0001,
+      "loss": 0.1201,
+      "step": 5250
+    },
+    {
+      "epoch": 14.943181818181818,
+      "grad_norm": 3.8694465160369873,
+      "learning_rate": 0.0001,
+      "loss": 0.1277,
+      "step": 5260
+    },
+    {
+      "epoch": 14.971590909090908,
+      "grad_norm": 3.4749040603637695,
+      "learning_rate": 0.0001,
+      "loss": 0.1272,
+      "step": 5270
+    },
+    {
+      "epoch": 15.0,
+      "grad_norm": 3.4868409633636475,
+      "learning_rate": 0.0001,
+      "loss": 0.1205,
+      "step": 5280
+    },
+    {
+      "epoch": 15.028409090909092,
+      "grad_norm": 2.9750540256500244,
+      "learning_rate": 0.0001,
+      "loss": 0.126,
+      "step": 5290
+    },
+    {
+      "epoch": 15.056818181818182,
+      "grad_norm": 3.6922764778137207,
+      "learning_rate": 0.0001,
+      "loss": 0.1256,
+      "step": 5300
+    },
+    {
+      "epoch": 15.085227272727273,
+      "grad_norm": 2.812814712524414,
+      "learning_rate": 0.0001,
+      "loss": 0.1181,
+      "step": 5310
+    },
+    {
+      "epoch": 15.113636363636363,
+      "grad_norm": 3.3117034435272217,
+      "learning_rate": 0.0001,
+      "loss": 0.1226,
+      "step": 5320
+    },
+    {
+      "epoch": 15.142045454545455,
+      "grad_norm": 3.519850492477417,
+      "learning_rate": 0.0001,
+      "loss": 0.123,
+      "step": 5330
+    },
+    {
+      "epoch": 15.170454545454545,
+      "grad_norm": 3.4698708057403564,
+      "learning_rate": 0.0001,
+      "loss": 0.1176,
+      "step": 5340
+    },
+    {
+      "epoch": 15.198863636363637,
+      "grad_norm": 3.4124035835266113,
+      "learning_rate": 0.0001,
+      "loss": 0.12,
+      "step": 5350
+    },
+    {
+      "epoch": 15.227272727272727,
+      "grad_norm": 3.1546342372894287,
+      "learning_rate": 0.0001,
+      "loss": 0.1215,
+      "step": 5360
+    },
+    {
+      "epoch": 15.255681818181818,
+      "grad_norm": 3.2864038944244385,
+      "learning_rate": 0.0001,
+      "loss": 0.1178,
+      "step": 5370
+    },
+    {
+      "epoch": 15.284090909090908,
+      "grad_norm": 3.288776397705078,
+      "learning_rate": 0.0001,
+      "loss": 0.1235,
+      "step": 5380
+    },
+    {
+      "epoch": 15.3125,
+      "grad_norm": 3.0721805095672607,
+      "learning_rate": 0.0001,
+      "loss": 0.1166,
+      "step": 5390
+    },
+    {
+      "epoch": 15.340909090909092,
+      "grad_norm": 2.994493246078491,
+      "learning_rate": 0.0001,
+      "loss": 0.119,
+      "step": 5400
+    },
+    {
+      "epoch": 15.369318181818182,
+      "grad_norm": 3.0647480487823486,
+      "learning_rate": 0.0001,
+      "loss": 0.1199,
+      "step": 5410
+    },
+    {
+      "epoch": 15.397727272727273,
+      "grad_norm": 3.24365496635437,
+      "learning_rate": 0.0001,
+      "loss": 0.1185,
+      "step": 5420
+    },
+    {
+      "epoch": 15.426136363636363,
+      "grad_norm": 2.872796058654785,
+      "learning_rate": 0.0001,
+      "loss": 0.1239,
+      "step": 5430
+    },
+    {
+      "epoch": 15.454545454545455,
+      "grad_norm": 3.0730972290039062,
+      "learning_rate": 0.0001,
+      "loss": 0.118,
+      "step": 5440
+    },
+    {
+      "epoch": 15.482954545454545,
+      "grad_norm": 3.010819911956787,
+      "learning_rate": 0.0001,
+      "loss": 0.1282,
+      "step": 5450
+    },
+    {
+      "epoch": 15.511363636363637,
+      "grad_norm": 3.111093521118164,
+      "learning_rate": 0.0001,
+      "loss": 0.1203,
+      "step": 5460
+    },
+    {
+      "epoch": 15.539772727272727,
+      "grad_norm": 2.3254058361053467,
+      "learning_rate": 0.0001,
+      "loss": 0.1196,
+      "step": 5470
+    },
+    {
+      "epoch": 15.568181818181818,
+      "grad_norm": 2.7858808040618896,
+      "learning_rate": 0.0001,
+      "loss": 0.1157,
+      "step": 5480
+    },
+    {
+      "epoch": 15.596590909090908,
+      "grad_norm": 2.8064205646514893,
+      "learning_rate": 0.0001,
+      "loss": 0.1241,
+      "step": 5490
+    },
+    {
+      "epoch": 15.625,
+      "grad_norm": 3.140082597732544,
+      "learning_rate": 0.0001,
+      "loss": 0.123,
+      "step": 5500
+    },
+    {
+      "epoch": 15.653409090909092,
+      "grad_norm": 3.064652681350708,
+      "learning_rate": 0.0001,
+      "loss": 0.1263,
+      "step": 5510
+    },
+    {
+      "epoch": 15.681818181818182,
+      "grad_norm": 3.274289131164551,
+      "learning_rate": 0.0001,
+      "loss": 0.1239,
+      "step": 5520
+    },
+    {
+      "epoch": 15.710227272727273,
+      "grad_norm": 2.9138309955596924,
+      "learning_rate": 0.0001,
+      "loss": 0.1232,
+      "step": 5530
+    },
+    {
+      "epoch": 15.738636363636363,
+      "grad_norm": 2.9141759872436523,
+      "learning_rate": 0.0001,
+      "loss": 0.1222,
+      "step": 5540
+    },
+    {
+      "epoch": 15.767045454545455,
+      "grad_norm": 2.755699396133423,
+      "learning_rate": 0.0001,
+      "loss": 0.1209,
+      "step": 5550
+    },
+    {
+      "epoch": 15.795454545454545,
+      "grad_norm": 2.7435543537139893,
+      "learning_rate": 0.0001,
+      "loss": 0.1205,
+      "step": 5560
+    },
+    {
+      "epoch": 15.823863636363637,
+      "grad_norm": 2.868746519088745,
+      "learning_rate": 0.0001,
+      "loss": 0.1188,
+      "step": 5570
+    },
+    {
+      "epoch": 15.852272727272727,
+      "grad_norm": 2.853201389312744,
+      "learning_rate": 0.0001,
+      "loss": 0.1154,
+      "step": 5580
+    },
+    {
+      "epoch": 15.880681818181818,
+      "grad_norm": 3.2404487133026123,
+      "learning_rate": 0.0001,
+      "loss": 0.1174,
+      "step": 5590
+    },
+    {
+      "epoch": 15.909090909090908,
+      "grad_norm": 3.210789203643799,
+      "learning_rate": 0.0001,
+      "loss": 0.1204,
+      "step": 5600
+    },
+    {
+      "epoch": 15.9375,
+      "grad_norm": 2.9118998050689697,
+      "learning_rate": 0.0001,
+      "loss": 0.1183,
+      "step": 5610
+    },
+    {
+      "epoch": 15.965909090909092,
+      "grad_norm": 2.8953421115875244,
+      "learning_rate": 0.0001,
+      "loss": 0.1182,
+      "step": 5620
+    },
+    {
+      "epoch": 15.994318181818182,
+      "grad_norm": 2.942523241043091,
+      "learning_rate": 0.0001,
+      "loss": 0.1195,
+      "step": 5630
+    },
+    {
+      "epoch": 16.022727272727273,
+      "grad_norm": 3.2362887859344482,
+      "learning_rate": 0.0001,
+      "loss": 0.1157,
+      "step": 5640
+    },
+    {
+      "epoch": 16.051136363636363,
+      "grad_norm": 2.438734292984009,
+      "learning_rate": 0.0001,
+      "loss": 0.1124,
+      "step": 5650
+    },
+    {
+      "epoch": 16.079545454545453,
+      "grad_norm": 2.5169425010681152,
+      "learning_rate": 0.0001,
+      "loss": 0.1113,
+      "step": 5660
+    },
+    {
+      "epoch": 16.107954545454547,
+      "grad_norm": 2.644383668899536,
+      "learning_rate": 0.0001,
+      "loss": 0.1165,
+      "step": 5670
+    },
+    {
+      "epoch": 16.136363636363637,
+      "grad_norm": 3.3263514041900635,
+      "learning_rate": 0.0001,
+      "loss": 0.1221,
+      "step": 5680
+    },
+    {
+      "epoch": 16.164772727272727,
+      "grad_norm": 2.8352041244506836,
+      "learning_rate": 0.0001,
+      "loss": 0.1138,
+      "step": 5690
+    },
+    {
+      "epoch": 16.193181818181817,
+      "grad_norm": 3.1213154792785645,
+      "learning_rate": 0.0001,
+      "loss": 0.1164,
+      "step": 5700
+    },
+    {
+      "epoch": 16.22159090909091,
+      "grad_norm": 3.123992681503296,
+      "learning_rate": 0.0001,
+      "loss": 0.1163,
+      "step": 5710
+    },
+    {
+      "epoch": 16.25,
+      "grad_norm": 3.104673385620117,
+      "learning_rate": 0.0001,
+      "loss": 0.1154,
+      "step": 5720
+    },
+    {
+      "epoch": 16.27840909090909,
+      "grad_norm": 3.469186544418335,
+      "learning_rate": 0.0001,
+      "loss": 0.1137,
+      "step": 5730
+    },
+    {
+      "epoch": 16.306818181818183,
+      "grad_norm": 3.1163649559020996,
+      "learning_rate": 0.0001,
+      "loss": 0.1164,
+      "step": 5740
+    },
+    {
+      "epoch": 16.335227272727273,
+      "grad_norm": 2.9757080078125,
+      "learning_rate": 0.0001,
+      "loss": 0.1215,
+      "step": 5750
+    },
+    {
+      "epoch": 16.363636363636363,
+      "grad_norm": 3.346102237701416,
+      "learning_rate": 0.0001,
+      "loss": 0.114,
+      "step": 5760
+    },
+    {
+      "epoch": 16.392045454545453,
+      "grad_norm": 3.4598140716552734,
+      "learning_rate": 0.0001,
+      "loss": 0.1134,
+      "step": 5770
+    },
+    {
+      "epoch": 16.420454545454547,
+      "grad_norm": 2.8395731449127197,
+      "learning_rate": 0.0001,
+      "loss": 0.1136,
+      "step": 5780
+    },
+    {
+      "epoch": 16.448863636363637,
+      "grad_norm": 2.390820026397705,
+      "learning_rate": 0.0001,
+      "loss": 0.115,
+      "step": 5790
+    },
+    {
+      "epoch": 16.477272727272727,
+      "grad_norm": 3.3408634662628174,
+      "learning_rate": 0.0001,
+      "loss": 0.1191,
+      "step": 5800
+    },
+    {
+      "epoch": 16.505681818181817,
+      "grad_norm": 2.721245050430298,
+      "learning_rate": 0.0001,
+      "loss": 0.1187,
+      "step": 5810
+    },
+    {
+      "epoch": 16.53409090909091,
+      "grad_norm": 3.057680130004883,
+      "learning_rate": 0.0001,
+      "loss": 0.1169,
+      "step": 5820
+    },
+    {
+      "epoch": 16.5625,
+      "grad_norm": 2.8173437118530273,
+      "learning_rate": 0.0001,
+      "loss": 0.1178,
+      "step": 5830
+    },
+    {
+      "epoch": 16.59090909090909,
+      "grad_norm": 2.824174404144287,
+      "learning_rate": 0.0001,
+      "loss": 0.1137,
+      "step": 5840
+    },
+    {
+      "epoch": 16.619318181818183,
+      "grad_norm": 2.9729907512664795,
+      "learning_rate": 0.0001,
+      "loss": 0.1149,
+      "step": 5850
+    },
+    {
+      "epoch": 16.647727272727273,
+      "grad_norm": 2.893472909927368,
+      "learning_rate": 0.0001,
+      "loss": 0.1147,
+      "step": 5860
+    },
+    {
+      "epoch": 16.676136363636363,
+      "grad_norm": 2.6419155597686768,
+      "learning_rate": 0.0001,
+      "loss": 0.1166,
+      "step": 5870
+    },
+    {
+      "epoch": 16.704545454545453,
+      "grad_norm": 2.341890811920166,
+      "learning_rate": 0.0001,
+      "loss": 0.1148,
+      "step": 5880
+    },
+    {
+      "epoch": 16.732954545454547,
+      "grad_norm": 2.980921506881714,
+      "learning_rate": 0.0001,
+      "loss": 0.1134,
+      "step": 5890
+    },
+    {
+      "epoch": 16.761363636363637,
+      "grad_norm": 2.975208044052124,
+      "learning_rate": 0.0001,
+      "loss": 0.1146,
+      "step": 5900
+    },
+    {
+      "epoch": 16.789772727272727,
+      "grad_norm": 2.528339147567749,
+      "learning_rate": 0.0001,
+      "loss": 0.1155,
+      "step": 5910
+    },
+    {
+      "epoch": 16.818181818181817,
+      "grad_norm": 2.539898633956909,
+      "learning_rate": 0.0001,
+      "loss": 0.1137,
+      "step": 5920
+    },
+    {
+      "epoch": 16.84659090909091,
+      "grad_norm": 2.4367032051086426,
+      "learning_rate": 0.0001,
+      "loss": 0.1175,
+      "step": 5930
+    },
+    {
+      "epoch": 16.875,
+      "grad_norm": 2.5197834968566895,
+      "learning_rate": 0.0001,
+      "loss": 0.1158,
+      "step": 5940
+    },
+    {
+      "epoch": 16.90340909090909,
+      "grad_norm": 2.4279847145080566,
+      "learning_rate": 0.0001,
+      "loss": 0.1113,
+      "step": 5950
+    },
+    {
+      "epoch": 16.931818181818183,
+      "grad_norm": 3.2526116371154785,
+      "learning_rate": 0.0001,
+      "loss": 0.1129,
+      "step": 5960
+    },
+    {
+      "epoch": 16.960227272727273,
+      "grad_norm": 2.5634706020355225,
+      "learning_rate": 0.0001,
+      "loss": 0.1146,
+      "step": 5970
+    },
+    {
+      "epoch": 16.988636363636363,
+      "grad_norm": 2.89918851852417,
+      "learning_rate": 0.0001,
+      "loss": 0.1147,
+      "step": 5980
+    },
+    {
+      "epoch": 17.017045454545453,
+      "grad_norm": 2.8295469284057617,
+      "learning_rate": 0.0001,
+      "loss": 0.114,
+      "step": 5990
+    },
+    {
+      "epoch": 17.045454545454547,
+      "grad_norm": 2.8802335262298584,
+      "learning_rate": 0.0001,
+      "loss": 0.1125,
+      "step": 6000
+    },
+    {
+      "epoch": 17.073863636363637,
+      "grad_norm": 2.6155662536621094,
+      "learning_rate": 0.0001,
+      "loss": 0.1158,
+      "step": 6010
+    },
+    {
+      "epoch": 17.102272727272727,
+      "grad_norm": 2.791156530380249,
+      "learning_rate": 0.0001,
+      "loss": 0.1147,
+      "step": 6020
+    },
+    {
+      "epoch": 17.130681818181817,
+      "grad_norm": 2.7444076538085938,
+      "learning_rate": 0.0001,
+      "loss": 0.1172,
+      "step": 6030
+    },
+    {
+      "epoch": 17.15909090909091,
+      "grad_norm": 3.0765230655670166,
+      "learning_rate": 0.0001,
+      "loss": 0.1114,
+      "step": 6040
+    },
+    {
+      "epoch": 17.1875,
+      "grad_norm": 3.4001102447509766,
+      "learning_rate": 0.0001,
+      "loss": 0.1156,
+      "step": 6050
+    },
+    {
+      "epoch": 17.21590909090909,
+      "grad_norm": 2.574037790298462,
+      "learning_rate": 0.0001,
+      "loss": 0.114,
+      "step": 6060
+    },
+    {
+      "epoch": 17.244318181818183,
+      "grad_norm": 2.428994655609131,
+      "learning_rate": 0.0001,
+      "loss": 0.1059,
+      "step": 6070
+    },
+    {
+      "epoch": 17.272727272727273,
+      "grad_norm": 2.552593469619751,
+      "learning_rate": 0.0001,
+      "loss": 0.1132,
+      "step": 6080
+    },
+    {
+      "epoch": 17.301136363636363,
+      "grad_norm": 2.748263359069824,
+      "learning_rate": 0.0001,
+      "loss": 0.1149,
+      "step": 6090
+    },
+    {
+      "epoch": 17.329545454545453,
+      "grad_norm": 2.565458059310913,
+      "learning_rate": 0.0001,
+      "loss": 0.1105,
+      "step": 6100
+    },
+    {
+      "epoch": 17.357954545454547,
+      "grad_norm": 3.3726043701171875,
+      "learning_rate": 0.0001,
+      "loss": 0.1108,
+      "step": 6110
+    },
+    {
+      "epoch": 17.386363636363637,
+      "grad_norm": 2.640763282775879,
+      "learning_rate": 0.0001,
+      "loss": 0.1049,
+      "step": 6120
+    },
+    {
+      "epoch": 17.414772727272727,
+      "grad_norm": 2.3288469314575195,
+      "learning_rate": 0.0001,
+      "loss": 0.1073,
+      "step": 6130
+    },
+    {
+      "epoch": 17.443181818181817,
+      "grad_norm": 2.47501802444458,
+      "learning_rate": 0.0001,
+      "loss": 0.1097,
+      "step": 6140
+    },
+    {
+      "epoch": 17.47159090909091,
+      "grad_norm": 2.443122148513794,
+      "learning_rate": 0.0001,
+      "loss": 0.1167,
+      "step": 6150
+    },
+    {
+      "epoch": 17.5,
+      "grad_norm": 3.0210578441619873,
+      "learning_rate": 0.0001,
+      "loss": 0.1185,
+      "step": 6160
+    },
+    {
+      "epoch": 17.52840909090909,
+      "grad_norm": 2.7917838096618652,
+      "learning_rate": 0.0001,
+      "loss": 0.1176,
+      "step": 6170
+    },
+    {
+      "epoch": 17.556818181818183,
+      "grad_norm": 2.502795457839966,
+      "learning_rate": 0.0001,
+      "loss": 0.1101,
+      "step": 6180
+    },
+    {
+      "epoch": 17.585227272727273,
+      "grad_norm": 2.6011240482330322,
+      "learning_rate": 0.0001,
+      "loss": 0.111,
+      "step": 6190
+    },
+    {
+      "epoch": 17.613636363636363,
+      "grad_norm": 2.917656183242798,
+      "learning_rate": 0.0001,
+      "loss": 0.1093,
+      "step": 6200
+    },
+    {
+      "epoch": 17.642045454545453,
+      "grad_norm": 2.369063138961792,
+      "learning_rate": 0.0001,
+      "loss": 0.1116,
+      "step": 6210
+    },
+    {
+      "epoch": 17.670454545454547,
+      "grad_norm": 2.5128045082092285,
+      "learning_rate": 0.0001,
+      "loss": 0.1111,
+      "step": 6220
+    },
+    {
+      "epoch": 17.698863636363637,
+      "grad_norm": 3.0461254119873047,
+      "learning_rate": 0.0001,
+      "loss": 0.1147,
+      "step": 6230
+    },
+    {
+      "epoch": 17.727272727272727,
+      "grad_norm": 2.1202504634857178,
+      "learning_rate": 0.0001,
+      "loss": 0.1107,
+      "step": 6240
+    },
+    {
+      "epoch": 17.755681818181817,
+      "grad_norm": 2.3112752437591553,
+      "learning_rate": 0.0001,
+      "loss": 0.1086,
+      "step": 6250
+    },
+    {
+      "epoch": 17.78409090909091,
+      "grad_norm": 2.8188629150390625,
+      "learning_rate": 0.0001,
+      "loss": 0.121,
+      "step": 6260
+    },
+    {
+      "epoch": 17.8125,
+      "grad_norm": 2.883798599243164,
+      "learning_rate": 0.0001,
+      "loss": 0.1088,
+      "step": 6270
+    },
+    {
+      "epoch": 17.84090909090909,
+      "grad_norm": 2.619675397872925,
+      "learning_rate": 0.0001,
+      "loss": 0.1121,
+      "step": 6280
+    },
+    {
+      "epoch": 17.869318181818183,
+      "grad_norm": 2.5900354385375977,
+      "learning_rate": 0.0001,
+      "loss": 0.111,
+      "step": 6290
+    },
+    {
+      "epoch": 17.897727272727273,
+      "grad_norm": 3.2702383995056152,
+      "learning_rate": 0.0001,
+      "loss": 0.1123,
+      "step": 6300
+    },
+    {
+      "epoch": 17.926136363636363,
+      "grad_norm": 2.632286548614502,
+      "learning_rate": 0.0001,
+      "loss": 0.1096,
+      "step": 6310
+    },
+    {
+      "epoch": 17.954545454545453,
+      "grad_norm": 2.598470687866211,
+      "learning_rate": 0.0001,
+      "loss": 0.109,
+      "step": 6320
+    },
+    {
+      "epoch": 17.982954545454547,
+      "grad_norm": 2.3819477558135986,
+      "learning_rate": 0.0001,
+      "loss": 0.1118,
+      "step": 6330
+    },
+    {
+      "epoch": 18.011363636363637,
+      "grad_norm": 2.4921939373016357,
+      "learning_rate": 0.0001,
+      "loss": 0.1116,
+      "step": 6340
+    },
+    {
+      "epoch": 18.039772727272727,
+      "grad_norm": 2.820632219314575,
+      "learning_rate": 0.0001,
+      "loss": 0.1052,
+      "step": 6350
+    },
+    {
+      "epoch": 18.068181818181817,
+      "grad_norm": 2.769113063812256,
+      "learning_rate": 0.0001,
+      "loss": 0.1141,
+      "step": 6360
+    },
+    {
+      "epoch": 18.09659090909091,
+      "grad_norm": 2.58843731880188,
+      "learning_rate": 0.0001,
+      "loss": 0.1133,
+      "step": 6370
+    },
+    {
+      "epoch": 18.125,
+      "grad_norm": 1.958970069885254,
+      "learning_rate": 0.0001,
+      "loss": 0.114,
+      "step": 6380
+    },
+    {
+      "epoch": 18.15340909090909,
+      "grad_norm": 2.466975688934326,
+      "learning_rate": 0.0001,
+      "loss": 0.1071,
+      "step": 6390
+    },
+    {
+      "epoch": 18.181818181818183,
+      "grad_norm": 2.1818594932556152,
+      "learning_rate": 0.0001,
+      "loss": 0.1095,
+      "step": 6400
+    },
+    {
+      "epoch": 18.210227272727273,
+      "grad_norm": 2.3512721061706543,
+      "learning_rate": 0.0001,
+      "loss": 0.114,
+      "step": 6410
+    },
+    {
+      "epoch": 18.238636363636363,
+      "grad_norm": 2.3737564086914062,
+      "learning_rate": 0.0001,
+      "loss": 0.1113,
+      "step": 6420
+    },
+    {
+      "epoch": 18.267045454545453,
+      "grad_norm": 2.4189605712890625,
+      "learning_rate": 0.0001,
+      "loss": 0.1117,
+      "step": 6430
+    },
+    {
+      "epoch": 18.295454545454547,
+      "grad_norm": 2.6895744800567627,
+      "learning_rate": 0.0001,
+      "loss": 0.1122,
+      "step": 6440
+    },
+    {
+      "epoch": 18.323863636363637,
+      "grad_norm": 2.517216920852661,
+      "learning_rate": 0.0001,
+      "loss": 0.1132,
+      "step": 6450
+    },
+    {
+      "epoch": 18.352272727272727,
+      "grad_norm": 2.251976251602173,
+      "learning_rate": 0.0001,
+      "loss": 0.1058,
+      "step": 6460
+    },
+    {
+      "epoch": 18.380681818181817,
+      "grad_norm": 2.233076333999634,
+      "learning_rate": 0.0001,
+      "loss": 0.1123,
+      "step": 6470
+    },
+    {
+      "epoch": 18.40909090909091,
+      "grad_norm": 2.5515904426574707,
+      "learning_rate": 0.0001,
+      "loss": 0.1107,
+      "step": 6480
+    },
+    {
+      "epoch": 18.4375,
+      "grad_norm": 2.60249662399292,
+      "learning_rate": 0.0001,
+      "loss": 0.1089,
+      "step": 6490
+    },
+    {
+      "epoch": 18.46590909090909,
+      "grad_norm": 2.640946388244629,
+      "learning_rate": 0.0001,
+      "loss": 0.1107,
+      "step": 6500
+    },
+    {
+      "epoch": 18.494318181818183,
+      "grad_norm": 2.663269281387329,
+      "learning_rate": 0.0001,
+      "loss": 0.1085,
+      "step": 6510
+    },
+    {
+      "epoch": 18.522727272727273,
+      "grad_norm": 2.4106152057647705,
+      "learning_rate": 0.0001,
+      "loss": 0.1098,
+      "step": 6520
+    },
+    {
+      "epoch": 18.551136363636363,
+      "grad_norm": 2.3369884490966797,
+      "learning_rate": 0.0001,
+      "loss": 0.1096,
+      "step": 6530
+    },
+    {
+      "epoch": 18.579545454545453,
+      "grad_norm": 2.346574544906616,
+      "learning_rate": 0.0001,
+      "loss": 0.104,
+      "step": 6540
+    },
+    {
+      "epoch": 18.607954545454547,
+      "grad_norm": 2.2791831493377686,
+      "learning_rate": 0.0001,
+      "loss": 0.1105,
+      "step": 6550
+    },
+    {
+      "epoch": 18.636363636363637,
+      "grad_norm": 2.364088773727417,
+      "learning_rate": 0.0001,
+      "loss": 0.1113,
+      "step": 6560
+    },
+    {
+      "epoch": 18.664772727272727,
+      "grad_norm": 2.5654282569885254,
+      "learning_rate": 0.0001,
+      "loss": 0.1067,
+      "step": 6570
+    },
+    {
+      "epoch": 18.693181818181817,
+      "grad_norm": 2.577658176422119,
+      "learning_rate": 0.0001,
+      "loss": 0.1117,
+      "step": 6580
+    },
+    {
+      "epoch": 18.72159090909091,
+      "grad_norm": 2.4139275550842285,
+      "learning_rate": 0.0001,
+      "loss": 0.1043,
+      "step": 6590
+    },
+    {
+      "epoch": 18.75,
+      "grad_norm": 2.5888614654541016,
+      "learning_rate": 0.0001,
+      "loss": 0.1082,
+      "step": 6600
+    },
+    {
+      "epoch": 18.77840909090909,
+      "grad_norm": 2.3184974193573,
+      "learning_rate": 0.0001,
+      "loss": 0.1083,
+      "step": 6610
+    },
+    {
+      "epoch": 18.806818181818183,
+      "grad_norm": 2.522383451461792,
+      "learning_rate": 0.0001,
+      "loss": 0.1083,
+      "step": 6620
+    },
+    {
+      "epoch": 18.835227272727273,
+      "grad_norm": 2.2055583000183105,
+      "learning_rate": 0.0001,
+      "loss": 0.1076,
+      "step": 6630
+    },
+    {
+      "epoch": 18.863636363636363,
+      "grad_norm": 2.58622145652771,
+      "learning_rate": 0.0001,
+      "loss": 0.1087,
+      "step": 6640
+    },
+    {
+      "epoch": 18.892045454545453,
+      "grad_norm": 2.3860034942626953,
+      "learning_rate": 0.0001,
+      "loss": 0.1108,
+      "step": 6650
+    },
+    {
+      "epoch": 18.920454545454547,
+      "grad_norm": 2.6108431816101074,
+      "learning_rate": 0.0001,
+      "loss": 0.1089,
+      "step": 6660
+    },
+    {
+      "epoch": 18.948863636363637,
+      "grad_norm": 2.1429636478424072,
+      "learning_rate": 0.0001,
+      "loss": 0.1127,
+      "step": 6670
+    },
+    {
+      "epoch": 18.977272727272727,
+      "grad_norm": 2.1483328342437744,
+      "learning_rate": 0.0001,
+      "loss": 0.107,
+      "step": 6680
+    },
+    {
+      "epoch": 19.005681818181817,
+      "grad_norm": 2.524930477142334,
+      "learning_rate": 0.0001,
+      "loss": 0.1176,
+      "step": 6690
+    },
+    {
+      "epoch": 19.03409090909091,
+      "grad_norm": 2.55420184135437,
+      "learning_rate": 0.0001,
+      "loss": 0.1134,
+      "step": 6700
+    },
+    {
+      "epoch": 19.0625,
+      "grad_norm": 2.4037156105041504,
+      "learning_rate": 0.0001,
+      "loss": 0.1121,
+      "step": 6710
+    },
+    {
+      "epoch": 19.09090909090909,
+      "grad_norm": 2.545936107635498,
+      "learning_rate": 0.0001,
+      "loss": 0.1038,
+      "step": 6720
+    },
+    {
+      "epoch": 19.119318181818183,
+      "grad_norm": 2.2953386306762695,
+      "learning_rate": 0.0001,
+      "loss": 0.1045,
+      "step": 6730
+    },
+    {
+      "epoch": 19.147727272727273,
+      "grad_norm": 2.4761712551116943,
+      "learning_rate": 0.0001,
+      "loss": 0.1022,
+      "step": 6740
+    },
+    {
+      "epoch": 19.176136363636363,
+      "grad_norm": 2.2314484119415283,
+      "learning_rate": 0.0001,
+      "loss": 0.1048,
+      "step": 6750
+    },
+    {
+      "epoch": 19.204545454545453,
+      "grad_norm": 2.306548833847046,
+      "learning_rate": 0.0001,
+      "loss": 0.1068,
+      "step": 6760
+    },
+    {
+      "epoch": 19.232954545454547,
+      "grad_norm": 2.563133955001831,
+      "learning_rate": 0.0001,
+      "loss": 0.1051,
+      "step": 6770
+    },
+    {
+      "epoch": 19.261363636363637,
+      "grad_norm": 2.306220531463623,
+      "learning_rate": 0.0001,
+      "loss": 0.1028,
+      "step": 6780
+    },
+    {
+      "epoch": 19.289772727272727,
+      "grad_norm": 2.3580806255340576,
+      "learning_rate": 0.0001,
+      "loss": 0.1026,
+      "step": 6790
+    },
+    {
+      "epoch": 19.318181818181817,
+      "grad_norm": 2.317422866821289,
+      "learning_rate": 0.0001,
+      "loss": 0.1045,
+      "step": 6800
+    },
+    {
+      "epoch": 19.34659090909091,
+      "grad_norm": 2.4174487590789795,
+      "learning_rate": 0.0001,
+      "loss": 0.1047,
+      "step": 6810
+    },
+    {
+      "epoch": 19.375,
+      "grad_norm": 2.417792558670044,
+      "learning_rate": 0.0001,
+      "loss": 0.1062,
+      "step": 6820
+    },
+    {
+      "epoch": 19.40340909090909,
+      "grad_norm": 2.08555269241333,
+      "learning_rate": 0.0001,
+      "loss": 0.1075,
+      "step": 6830
+    },
+    {
+      "epoch": 19.431818181818183,
+      "grad_norm": 2.052635431289673,
+      "learning_rate": 0.0001,
+      "loss": 0.1086,
+      "step": 6840
+    },
+    {
+      "epoch": 19.460227272727273,
+      "grad_norm": 2.216602325439453,
+      "learning_rate": 0.0001,
+      "loss": 0.1077,
+      "step": 6850
+    },
+    {
+      "epoch": 19.488636363636363,
+      "grad_norm": 2.393385410308838,
+      "learning_rate": 0.0001,
+      "loss": 0.1084,
+      "step": 6860
+    },
+    {
+      "epoch": 19.517045454545453,
+      "grad_norm": 2.2516062259674072,
+      "learning_rate": 0.0001,
+      "loss": 0.1081,
+      "step": 6870
+    },
+    {
+      "epoch": 19.545454545454547,
+      "grad_norm": 2.2450714111328125,
+      "learning_rate": 0.0001,
+      "loss": 0.1057,
+      "step": 6880
+    },
+    {
+      "epoch": 19.573863636363637,
+      "grad_norm": 2.2236733436584473,
+      "learning_rate": 0.0001,
+      "loss": 0.1057,
+      "step": 6890
+    },
+    {
+      "epoch": 19.602272727272727,
+      "grad_norm": 2.0041747093200684,
+      "learning_rate": 0.0001,
+      "loss": 0.1092,
+      "step": 6900
+    },
+    {
+      "epoch": 19.630681818181817,
+      "grad_norm": 2.264723300933838,
+      "learning_rate": 0.0001,
+      "loss": 0.107,
+      "step": 6910
+    },
+    {
+      "epoch": 19.65909090909091,
+      "grad_norm": 2.467823028564453,
+      "learning_rate": 0.0001,
+      "loss": 0.1042,
+      "step": 6920
+    },
+    {
+      "epoch": 19.6875,
+      "grad_norm": 2.2700631618499756,
+      "learning_rate": 0.0001,
+      "loss": 0.1037,
+      "step": 6930
+    },
+    {
+      "epoch": 19.71590909090909,
+      "grad_norm": 2.2037792205810547,
+      "learning_rate": 0.0001,
+      "loss": 0.1012,
+      "step": 6940
+    },
+    {
+      "epoch": 19.744318181818183,
+      "grad_norm": 2.0393118858337402,
+      "learning_rate": 0.0001,
+      "loss": 0.1018,
+      "step": 6950
+    },
+    {
+      "epoch": 19.772727272727273,
+      "grad_norm": 2.0741636753082275,
+      "learning_rate": 0.0001,
+      "loss": 0.1095,
+      "step": 6960
+    },
+    {
+      "epoch": 19.801136363636363,
+      "grad_norm": 1.9817142486572266,
+      "learning_rate": 0.0001,
+      "loss": 0.102,
+      "step": 6970
+    },
+    {
+      "epoch": 19.829545454545453,
+      "grad_norm": 2.222271680831909,
+      "learning_rate": 0.0001,
+      "loss": 0.1096,
+      "step": 6980
+    },
+    {
+      "epoch": 19.857954545454547,
+      "grad_norm": 2.0376973152160645,
+      "learning_rate": 0.0001,
+      "loss": 0.1086,
+      "step": 6990
+    },
+    {
+      "epoch": 19.886363636363637,
+      "grad_norm": 1.9794195890426636,
+      "learning_rate": 0.0001,
+      "loss": 0.106,
+      "step": 7000
+    },
+    {
+      "epoch": 19.914772727272727,
+      "grad_norm": 2.154062032699585,
+      "learning_rate": 0.0001,
+      "loss": 0.1117,
+      "step": 7010
+    },
+    {
+      "epoch": 19.943181818181817,
+      "grad_norm": 2.5424580574035645,
+      "learning_rate": 0.0001,
+      "loss": 0.1079,
+      "step": 7020
+    },
+    {
+      "epoch": 19.97159090909091,
+      "grad_norm": 2.4422006607055664,
+      "learning_rate": 0.0001,
+      "loss": 0.1005,
+      "step": 7030
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 2.433610200881958,
+      "learning_rate": 0.0001,
+      "loss": 0.1051,
+      "step": 7040
+    },
+    {
+      "epoch": 20.02840909090909,
+      "grad_norm": 2.1430225372314453,
+      "learning_rate": 0.0001,
+      "loss": 0.1003,
+      "step": 7050
+    },
+    {
+      "epoch": 20.056818181818183,
+      "grad_norm": 2.1949706077575684,
+      "learning_rate": 0.0001,
+      "loss": 0.0975,
+      "step": 7060
+    },
+    {
+      "epoch": 20.085227272727273,
+      "grad_norm": 2.3351924419403076,
+      "learning_rate": 0.0001,
+      "loss": 0.1076,
+      "step": 7070
+    },
+    {
+      "epoch": 20.113636363636363,
+      "grad_norm": 2.2610702514648438,
+      "learning_rate": 0.0001,
+      "loss": 0.1052,
+      "step": 7080
+    },
+    {
+      "epoch": 20.142045454545453,
+      "grad_norm": 2.0343222618103027,
+      "learning_rate": 0.0001,
+      "loss": 0.105,
+      "step": 7090
+    },
+    {
+      "epoch": 20.170454545454547,
+      "grad_norm": 2.3669273853302,
+      "learning_rate": 0.0001,
+      "loss": 0.1037,
+      "step": 7100
+    },
+    {
+      "epoch": 20.198863636363637,
+      "grad_norm": 2.224647283554077,
+      "learning_rate": 0.0001,
+      "loss": 0.1007,
+      "step": 7110
+    },
+    {
+      "epoch": 20.227272727272727,
+      "grad_norm": 2.3760414123535156,
+      "learning_rate": 0.0001,
+      "loss": 0.1087,
+      "step": 7120
+    },
+    {
+      "epoch": 20.255681818181817,
+      "grad_norm": 2.272942543029785,
+      "learning_rate": 0.0001,
+      "loss": 0.1051,
+      "step": 7130
+    },
+    {
+      "epoch": 20.28409090909091,
+      "grad_norm": 2.159137487411499,
+      "learning_rate": 0.0001,
+      "loss": 0.1053,
+      "step": 7140
+    },
+    {
+      "epoch": 20.3125,
+      "grad_norm": 2.2997374534606934,
+      "learning_rate": 0.0001,
+      "loss": 0.1046,
+      "step": 7150
+    },
+    {
+      "epoch": 20.34090909090909,
+      "grad_norm": 2.431882381439209,
+      "learning_rate": 0.0001,
+      "loss": 0.1038,
+      "step": 7160
+    },
+    {
+      "epoch": 20.369318181818183,
+      "grad_norm": 2.410435676574707,
+      "learning_rate": 0.0001,
+      "loss": 0.1018,
+      "step": 7170
+    },
+    {
+      "epoch": 20.397727272727273,
+      "grad_norm": 2.3939435482025146,
+      "learning_rate": 0.0001,
+      "loss": 0.102,
+      "step": 7180
+    },
+    {
+      "epoch": 20.426136363636363,
+      "grad_norm": 2.132279396057129,
+      "learning_rate": 0.0001,
+      "loss": 0.1034,
+      "step": 7190
+    },
+    {
+      "epoch": 20.454545454545453,
+      "grad_norm": 2.2916312217712402,
+      "learning_rate": 0.0001,
+      "loss": 0.1052,
+      "step": 7200
+    },
+    {
+      "epoch": 20.482954545454547,
+      "grad_norm": 2.5798282623291016,
+      "learning_rate": 0.0001,
+      "loss": 0.1026,
+      "step": 7210
+    },
+    {
+      "epoch": 20.511363636363637,
+      "grad_norm": 2.2447385787963867,
+      "learning_rate": 0.0001,
+      "loss": 0.0993,
+      "step": 7220
+    },
+    {
+      "epoch": 20.539772727272727,
+      "grad_norm": 2.2323153018951416,
+      "learning_rate": 0.0001,
+      "loss": 0.1009,
+      "step": 7230
+    },
+    {
+      "epoch": 20.568181818181817,
+      "grad_norm": 1.8763328790664673,
+      "learning_rate": 0.0001,
+      "loss": 0.1053,
+      "step": 7240
+    },
+    {
+      "epoch": 20.59659090909091,
+      "grad_norm": 1.501619815826416,
+      "learning_rate": 0.0001,
+      "loss": 0.1027,
+      "step": 7250
+    },
+    {
+      "epoch": 20.625,
+      "grad_norm": 1.9412376880645752,
+      "learning_rate": 0.0001,
+      "loss": 0.1056,
+      "step": 7260
+    },
+    {
+      "epoch": 20.65340909090909,
+      "grad_norm": 2.07375431060791,
+      "learning_rate": 0.0001,
+      "loss": 0.1047,
+      "step": 7270
+    },
+    {
+      "epoch": 20.681818181818183,
+      "grad_norm": 2.0877957344055176,
+      "learning_rate": 0.0001,
+      "loss": 0.1035,
+      "step": 7280
+    },
+    {
+      "epoch": 20.710227272727273,
+      "grad_norm": 2.4032297134399414,
+      "learning_rate": 0.0001,
+      "loss": 0.1039,
+      "step": 7290
+    },
+    {
+      "epoch": 20.738636363636363,
+      "grad_norm": 2.4037721157073975,
+      "learning_rate": 0.0001,
+      "loss": 0.0988,
+      "step": 7300
+    },
+    {
+      "epoch": 20.767045454545453,
+      "grad_norm": 1.8980406522750854,
+      "learning_rate": 0.0001,
+      "loss": 0.1024,
+      "step": 7310
+    },
+    {
+      "epoch": 20.795454545454547,
+      "grad_norm": 2.298187732696533,
+      "learning_rate": 0.0001,
+      "loss": 0.1065,
+      "step": 7320
+    },
+    {
+      "epoch": 20.823863636363637,
+      "grad_norm": 2.2373411655426025,
+      "learning_rate": 0.0001,
+      "loss": 0.1107,
+      "step": 7330
+    },
+    {
+      "epoch": 20.852272727272727,
+      "grad_norm": 2.2414472103118896,
+      "learning_rate": 0.0001,
+      "loss": 0.1063,
+      "step": 7340
+    },
+    {
+      "epoch": 20.880681818181817,
+      "grad_norm": 1.9661855697631836,
+      "learning_rate": 0.0001,
+      "loss": 0.1055,
+      "step": 7350
+    },
+    {
+      "epoch": 20.90909090909091,
+      "grad_norm": 1.9864373207092285,
+      "learning_rate": 0.0001,
+      "loss": 0.1013,
+      "step": 7360
+    },
+    {
+      "epoch": 20.9375,
+      "grad_norm": 2.239394187927246,
+      "learning_rate": 0.0001,
+      "loss": 0.1028,
+      "step": 7370
+    },
+    {
+      "epoch": 20.96590909090909,
+      "grad_norm": 1.8729074001312256,
+      "learning_rate": 0.0001,
+      "loss": 0.1045,
+      "step": 7380
+    },
+    {
+      "epoch": 20.994318181818183,
+      "grad_norm": 1.9113003015518188,
+      "learning_rate": 0.0001,
+      "loss": 0.1061,
+      "step": 7390
+    },
+    {
+      "epoch": 21.022727272727273,
+      "grad_norm": 1.733733892440796,
+      "learning_rate": 0.0001,
+      "loss": 0.1029,
+      "step": 7400
+    },
+    {
+      "epoch": 21.051136363636363,
+      "grad_norm": 1.908080816268921,
+      "learning_rate": 0.0001,
+      "loss": 0.1055,
+      "step": 7410
+    },
+    {
+      "epoch": 21.079545454545453,
+      "grad_norm": 2.046468496322632,
+      "learning_rate": 0.0001,
+      "loss": 0.103,
+      "step": 7420
+    },
+    {
+      "epoch": 21.107954545454547,
+      "grad_norm": 2.1874492168426514,
+      "learning_rate": 0.0001,
+      "loss": 0.1023,
+      "step": 7430
+    },
+    {
+      "epoch": 21.136363636363637,
+      "grad_norm": 1.7166926860809326,
+      "learning_rate": 0.0001,
+      "loss": 0.1023,
+      "step": 7440
+    },
+    {
+      "epoch": 21.164772727272727,
+      "grad_norm": 2.2620322704315186,
+      "learning_rate": 0.0001,
+      "loss": 0.1019,
+      "step": 7450
+    },
+    {
+      "epoch": 21.193181818181817,
+      "grad_norm": 2.283912181854248,
+      "learning_rate": 0.0001,
+      "loss": 0.1044,
+      "step": 7460
+    },
+    {
+      "epoch": 21.22159090909091,
+      "grad_norm": 2.0486154556274414,
+      "learning_rate": 0.0001,
+      "loss": 0.1014,
+      "step": 7470
+    },
+    {
+      "epoch": 21.25,
+      "grad_norm": 1.8538813591003418,
+      "learning_rate": 0.0001,
+      "loss": 0.1014,
+      "step": 7480
+    },
+    {
+      "epoch": 21.27840909090909,
+      "grad_norm": 1.9765312671661377,
+      "learning_rate": 0.0001,
+      "loss": 0.0973,
+      "step": 7490
+    },
+    {
+      "epoch": 21.306818181818183,
+      "grad_norm": 1.8284136056900024,
+      "learning_rate": 0.0001,
+      "loss": 0.1008,
+      "step": 7500
+    },
+    {
+      "epoch": 21.335227272727273,
+      "grad_norm": 1.8823169469833374,
+      "learning_rate": 0.0001,
+      "loss": 0.1045,
+      "step": 7510
+    },
+    {
+      "epoch": 21.363636363636363,
+      "grad_norm": 1.8679107427597046,
+      "learning_rate": 0.0001,
+      "loss": 0.0994,
+      "step": 7520
+    },
+    {
+      "epoch": 21.392045454545453,
+      "grad_norm": 2.5383877754211426,
+      "learning_rate": 0.0001,
+      "loss": 0.1018,
+      "step": 7530
+    },
+    {
+      "epoch": 21.420454545454547,
+      "grad_norm": 1.8846964836120605,
+      "learning_rate": 0.0001,
+      "loss": 0.1014,
+      "step": 7540
+    },
+    {
+      "epoch": 21.448863636363637,
+      "grad_norm": 2.0011348724365234,
+      "learning_rate": 0.0001,
+      "loss": 0.101,
+      "step": 7550
+    },
+    {
+      "epoch": 21.477272727272727,
+      "grad_norm": 2.220806360244751,
+      "learning_rate": 0.0001,
+      "loss": 0.0957,
+      "step": 7560
+    },
+    {
+      "epoch": 21.505681818181817,
+      "grad_norm": 2.1716644763946533,
+      "learning_rate": 0.0001,
+      "loss": 0.1043,
+      "step": 7570
+    },
+    {
+      "epoch": 21.53409090909091,
+      "grad_norm": 2.017302989959717,
+      "learning_rate": 0.0001,
+      "loss": 0.101,
+      "step": 7580
+    },
+    {
+      "epoch": 21.5625,
+      "grad_norm": 1.7568871974945068,
+      "learning_rate": 0.0001,
+      "loss": 0.101,
+      "step": 7590
+    },
+    {
+      "epoch": 21.59090909090909,
+      "grad_norm": 1.4694595336914062,
+      "learning_rate": 0.0001,
+      "loss": 0.0992,
+      "step": 7600
+    },
+    {
+      "epoch": 21.619318181818183,
+      "grad_norm": 1.6587533950805664,
+      "learning_rate": 0.0001,
+      "loss": 0.0947,
+      "step": 7610
+    },
+    {
+      "epoch": 21.647727272727273,
+      "grad_norm": 1.5999675989151,
+      "learning_rate": 0.0001,
+      "loss": 0.0947,
+      "step": 7620
+    },
+    {
+      "epoch": 21.676136363636363,
+      "grad_norm": 1.8176852464675903,
+      "learning_rate": 0.0001,
+      "loss": 0.1049,
+      "step": 7630
+    },
+    {
+      "epoch": 21.704545454545453,
+      "grad_norm": 1.8811140060424805,
+      "learning_rate": 0.0001,
+      "loss": 0.1026,
+      "step": 7640
+    },
+    {
+      "epoch": 21.732954545454547,
+      "grad_norm": 2.0048694610595703,
+      "learning_rate": 0.0001,
+      "loss": 0.106,
+      "step": 7650
+    },
+    {
+      "epoch": 21.761363636363637,
+      "grad_norm": 2.050529956817627,
+      "learning_rate": 0.0001,
+      "loss": 0.102,
+      "step": 7660
+    },
+    {
+      "epoch": 21.789772727272727,
+      "grad_norm": 2.1366333961486816,
+      "learning_rate": 0.0001,
+      "loss": 0.1052,
+      "step": 7670
+    },
+    {
+      "epoch": 21.818181818181817,
+      "grad_norm": 2.113969564437866,
+      "learning_rate": 0.0001,
+      "loss": 0.1044,
+      "step": 7680
+    },
+    {
+      "epoch": 21.84659090909091,
+      "grad_norm": 1.9199646711349487,
+      "learning_rate": 0.0001,
+      "loss": 0.1019,
+      "step": 7690
+    },
+    {
+      "epoch": 21.875,
+      "grad_norm": 2.162484884262085,
+      "learning_rate": 0.0001,
+      "loss": 0.0991,
+      "step": 7700
+    },
+    {
+      "epoch": 21.90340909090909,
+      "grad_norm": 2.1541666984558105,
+      "learning_rate": 0.0001,
+      "loss": 0.0977,
+      "step": 7710
+    },
+    {
+      "epoch": 21.931818181818183,
+      "grad_norm": 2.0925753116607666,
+      "learning_rate": 0.0001,
+      "loss": 0.0971,
+      "step": 7720
+    },
+    {
+      "epoch": 21.960227272727273,
+      "grad_norm": 2.3108863830566406,
+      "learning_rate": 0.0001,
+      "loss": 0.0937,
+      "step": 7730
+    },
+    {
+      "epoch": 21.988636363636363,
+      "grad_norm": 2.105069398880005,
+      "learning_rate": 0.0001,
+      "loss": 0.0945,
+      "step": 7740
+    },
+    {
+      "epoch": 22.017045454545453,
+      "grad_norm": 2.304133892059326,
+      "learning_rate": 0.0001,
+      "loss": 0.0999,
+      "step": 7750
+    },
+    {
+      "epoch": 22.045454545454547,
+      "grad_norm": 1.9112257957458496,
+      "learning_rate": 0.0001,
+      "loss": 0.0944,
+      "step": 7760
+    },
+    {
+      "epoch": 22.073863636363637,
+      "grad_norm": 1.9586460590362549,
+      "learning_rate": 0.0001,
+      "loss": 0.0978,
+      "step": 7770
+    },
+    {
+      "epoch": 22.102272727272727,
+      "grad_norm": 2.0921013355255127,
+      "learning_rate": 0.0001,
+      "loss": 0.1018,
+      "step": 7780
+    },
+    {
+      "epoch": 22.130681818181817,
+      "grad_norm": 2.0148942470550537,
+      "learning_rate": 0.0001,
+      "loss": 0.0995,
+      "step": 7790
+    },
+    {
+      "epoch": 22.15909090909091,
+      "grad_norm": 1.9254063367843628,
+      "learning_rate": 0.0001,
+      "loss": 0.1026,
+      "step": 7800
+    },
+    {
+      "epoch": 22.1875,
+      "grad_norm": 1.7932246923446655,
+      "learning_rate": 0.0001,
+      "loss": 0.1009,
+      "step": 7810
+    },
+    {
+      "epoch": 22.21590909090909,
+      "grad_norm": 2.0023248195648193,
+      "learning_rate": 0.0001,
+      "loss": 0.1006,
+      "step": 7820
+    },
+    {
+      "epoch": 22.244318181818183,
+      "grad_norm": 2.1428818702697754,
+      "learning_rate": 0.0001,
+      "loss": 0.0965,
+      "step": 7830
+    },
+    {
+      "epoch": 22.272727272727273,
+      "grad_norm": 2.021170139312744,
+      "learning_rate": 0.0001,
+      "loss": 0.1,
+      "step": 7840
+    },
+    {
+      "epoch": 22.301136363636363,
+      "grad_norm": 1.9330023527145386,
+      "learning_rate": 0.0001,
+      "loss": 0.0981,
+      "step": 7850
+    },
+    {
+      "epoch": 22.329545454545453,
+      "grad_norm": 2.1269373893737793,
+      "learning_rate": 0.0001,
+      "loss": 0.096,
+      "step": 7860
+    },
+    {
+      "epoch": 22.357954545454547,
+      "grad_norm": 2.0002894401550293,
+      "learning_rate": 0.0001,
+      "loss": 0.0983,
+      "step": 7870
+    },
+    {
+      "epoch": 22.386363636363637,
+      "grad_norm": 1.7350810766220093,
+      "learning_rate": 0.0001,
+      "loss": 0.0979,
+      "step": 7880
+    },
+    {
+      "epoch": 22.414772727272727,
+      "grad_norm": 2.5721471309661865,
+      "learning_rate": 0.0001,
+      "loss": 0.096,
+      "step": 7890
+    },
+    {
+      "epoch": 22.443181818181817,
+      "grad_norm": 2.0510294437408447,
+      "learning_rate": 0.0001,
+      "loss": 0.0967,
+      "step": 7900
+    },
+    {
+      "epoch": 22.47159090909091,
+      "grad_norm": 1.7467889785766602,
+      "learning_rate": 0.0001,
+      "loss": 0.1012,
+      "step": 7910
+    },
+    {
+      "epoch": 22.5,
+      "grad_norm": 1.881221055984497,
+      "learning_rate": 0.0001,
+      "loss": 0.0977,
+      "step": 7920
+    },
+    {
+      "epoch": 22.52840909090909,
+      "grad_norm": 1.6960866451263428,
+      "learning_rate": 0.0001,
+      "loss": 0.1013,
+      "step": 7930
+    },
+    {
+      "epoch": 22.556818181818183,
+      "grad_norm": 1.9011884927749634,
+      "learning_rate": 0.0001,
+      "loss": 0.0998,
+      "step": 7940
+    },
+    {
+      "epoch": 22.585227272727273,
+      "grad_norm": 2.289515972137451,
+      "learning_rate": 0.0001,
+      "loss": 0.1015,
+      "step": 7950
+    },
+    {
+      "epoch": 22.613636363636363,
+      "grad_norm": 2.148452043533325,
+      "learning_rate": 0.0001,
+      "loss": 0.098,
+      "step": 7960
+    },
+    {
+      "epoch": 22.642045454545453,
+      "grad_norm": 2.1038877964019775,
+      "learning_rate": 0.0001,
+      "loss": 0.0994,
+      "step": 7970
+    },
+    {
+      "epoch": 22.670454545454547,
+      "grad_norm": 2.1400623321533203,
+      "learning_rate": 0.0001,
+      "loss": 0.0948,
+      "step": 7980
+    },
+    {
+      "epoch": 22.698863636363637,
+      "grad_norm": 1.864814043045044,
+      "learning_rate": 0.0001,
+      "loss": 0.0917,
+      "step": 7990
+    },
+    {
+      "epoch": 22.727272727272727,
+      "grad_norm": 1.9961179494857788,
+      "learning_rate": 0.0001,
+      "loss": 0.0962,
+      "step": 8000
+    },
+    {
+      "epoch": 22.755681818181817,
+      "grad_norm": 1.9807307720184326,
+      "learning_rate": 0.0001,
+      "loss": 0.0954,
+      "step": 8010
+    },
+    {
+      "epoch": 22.78409090909091,
+      "grad_norm": 1.8741666078567505,
+      "learning_rate": 0.0001,
+      "loss": 0.0944,
+      "step": 8020
+    },
+    {
+      "epoch": 22.8125,
+      "grad_norm": 2.272317409515381,
+      "learning_rate": 0.0001,
+      "loss": 0.1037,
+      "step": 8030
+    },
+    {
+      "epoch": 22.84090909090909,
+      "grad_norm": 1.9558086395263672,
+      "learning_rate": 0.0001,
+      "loss": 0.0963,
+      "step": 8040
+    },
+    {
+      "epoch": 22.869318181818183,
+      "grad_norm": 2.009176731109619,
+      "learning_rate": 0.0001,
+      "loss": 0.1018,
+      "step": 8050
+    },
+    {
+      "epoch": 22.897727272727273,
+      "grad_norm": 2.2107458114624023,
+      "learning_rate": 0.0001,
+      "loss": 0.1025,
+      "step": 8060
+    },
+    {
+      "epoch": 22.926136363636363,
+      "grad_norm": 2.5531365871429443,
+      "learning_rate": 0.0001,
+      "loss": 0.1018,
+      "step": 8070
+    },
+    {
+      "epoch": 22.954545454545453,
+      "grad_norm": 2.2472565174102783,
+      "learning_rate": 0.0001,
+      "loss": 0.0982,
+      "step": 8080
+    },
+    {
+      "epoch": 22.982954545454547,
+      "grad_norm": 1.920031189918518,
+      "learning_rate": 0.0001,
+      "loss": 0.0943,
+      "step": 8090
+    },
+    {
+      "epoch": 23.011363636363637,
+      "grad_norm": 1.6725058555603027,
+      "learning_rate": 0.0001,
+      "loss": 0.0951,
+      "step": 8100
+    },
+    {
+      "epoch": 23.039772727272727,
+      "grad_norm": 2.308568239212036,
+      "learning_rate": 0.0001,
+      "loss": 0.092,
+      "step": 8110
+    },
+    {
+      "epoch": 23.068181818181817,
+      "grad_norm": 1.8834666013717651,
+      "learning_rate": 0.0001,
+      "loss": 0.0927,
+      "step": 8120
+    },
+    {
+      "epoch": 23.09659090909091,
+      "grad_norm": 1.8598517179489136,
+      "learning_rate": 0.0001,
+      "loss": 0.095,
+      "step": 8130
+    },
+    {
+      "epoch": 23.125,
+      "grad_norm": 2.1915621757507324,
+      "learning_rate": 0.0001,
+      "loss": 0.0929,
+      "step": 8140
+    },
+    {
+      "epoch": 23.15340909090909,
+      "grad_norm": 2.160149335861206,
+      "learning_rate": 0.0001,
+      "loss": 0.0964,
+      "step": 8150
+    },
+    {
+      "epoch": 23.181818181818183,
+      "grad_norm": 1.9698961973190308,
+      "learning_rate": 0.0001,
+      "loss": 0.0996,
+      "step": 8160
+    },
+    {
+      "epoch": 23.210227272727273,
+      "grad_norm": 1.9553509950637817,
+      "learning_rate": 0.0001,
+      "loss": 0.0948,
+      "step": 8170
+    },
+    {
+      "epoch": 23.238636363636363,
+      "grad_norm": 1.9348289966583252,
+      "learning_rate": 0.0001,
+      "loss": 0.0945,
+      "step": 8180
+    },
+    {
+      "epoch": 23.267045454545453,
+      "grad_norm": 2.053300142288208,
+      "learning_rate": 0.0001,
+      "loss": 0.0953,
+      "step": 8190
+    },
+    {
+      "epoch": 23.295454545454547,
+      "grad_norm": 1.8271958827972412,
+      "learning_rate": 0.0001,
+      "loss": 0.092,
+      "step": 8200
+    },
+    {
+      "epoch": 23.323863636363637,
+      "grad_norm": 1.5689889192581177,
+      "learning_rate": 0.0001,
+      "loss": 0.1003,
+      "step": 8210
+    },
+    {
+      "epoch": 23.352272727272727,
+      "grad_norm": 2.1360859870910645,
+      "learning_rate": 0.0001,
+      "loss": 0.0956,
+      "step": 8220
+    },
+    {
+      "epoch": 23.380681818181817,
+      "grad_norm": 1.819110631942749,
+      "learning_rate": 0.0001,
+      "loss": 0.0964,
+      "step": 8230
+    },
+    {
+      "epoch": 23.40909090909091,
+      "grad_norm": 2.107375383377075,
+      "learning_rate": 0.0001,
+      "loss": 0.0967,
+      "step": 8240
+    },
+    {
+      "epoch": 23.4375,
+      "grad_norm": 1.8408470153808594,
+      "learning_rate": 0.0001,
+      "loss": 0.0929,
+      "step": 8250
+    },
+    {
+      "epoch": 23.46590909090909,
+      "grad_norm": 1.9134175777435303,
+      "learning_rate": 0.0001,
+      "loss": 0.0956,
+      "step": 8260
+    },
+    {
+      "epoch": 23.494318181818183,
+      "grad_norm": 1.71891188621521,
+      "learning_rate": 0.0001,
+      "loss": 0.0939,
+      "step": 8270
+    },
+    {
+      "epoch": 23.522727272727273,
+      "grad_norm": 1.5312421321868896,
+      "learning_rate": 0.0001,
+      "loss": 0.0987,
+      "step": 8280
+    },
+    {
+      "epoch": 23.551136363636363,
+      "grad_norm": 1.7557247877120972,
+      "learning_rate": 0.0001,
+      "loss": 0.0975,
+      "step": 8290
+    },
+    {
+      "epoch": 23.579545454545453,
+      "grad_norm": 1.8017261028289795,
+      "learning_rate": 0.0001,
+      "loss": 0.0974,
+      "step": 8300
+    },
+    {
+      "epoch": 23.607954545454547,
+      "grad_norm": 1.5881474018096924,
+      "learning_rate": 0.0001,
+      "loss": 0.0987,
+      "step": 8310
+    },
+    {
+      "epoch": 23.636363636363637,
+      "grad_norm": 1.8395788669586182,
+      "learning_rate": 0.0001,
+      "loss": 0.1008,
+      "step": 8320
+    },
+    {
+      "epoch": 23.664772727272727,
+      "grad_norm": 1.791631817817688,
+      "learning_rate": 0.0001,
+      "loss": 0.0983,
+      "step": 8330
+    },
+    {
+      "epoch": 23.693181818181817,
+      "grad_norm": 1.6137861013412476,
+      "learning_rate": 0.0001,
+      "loss": 0.0982,
+      "step": 8340
+    },
+    {
+      "epoch": 23.72159090909091,
+      "grad_norm": 1.9976779222488403,
+      "learning_rate": 0.0001,
+      "loss": 0.1021,
+      "step": 8350
+    },
+    {
+      "epoch": 23.75,
+      "grad_norm": 1.667160987854004,
+      "learning_rate": 0.0001,
+      "loss": 0.0982,
+      "step": 8360
+    },
+    {
+      "epoch": 23.77840909090909,
+      "grad_norm": 1.5434305667877197,
+      "learning_rate": 0.0001,
+      "loss": 0.0967,
+      "step": 8370
+    },
+    {
+      "epoch": 23.806818181818183,
+      "grad_norm": 1.8221416473388672,
+      "learning_rate": 0.0001,
+      "loss": 0.0971,
+      "step": 8380
+    },
+    {
+      "epoch": 23.835227272727273,
+      "grad_norm": 1.9259772300720215,
+      "learning_rate": 0.0001,
+      "loss": 0.0969,
+      "step": 8390
+    },
+    {
+      "epoch": 23.863636363636363,
+      "grad_norm": 1.9943630695343018,
+      "learning_rate": 0.0001,
+      "loss": 0.0993,
+      "step": 8400
+    },
+    {
+      "epoch": 23.892045454545453,
+      "grad_norm": 1.5301824808120728,
+      "learning_rate": 0.0001,
+      "loss": 0.0944,
+      "step": 8410
+    },
+    {
+      "epoch": 23.920454545454547,
+      "grad_norm": 2.062227964401245,
+      "learning_rate": 0.0001,
+      "loss": 0.0949,
+      "step": 8420
+    },
+    {
+      "epoch": 23.948863636363637,
+      "grad_norm": 1.7410181760787964,
+      "learning_rate": 0.0001,
+      "loss": 0.0975,
+      "step": 8430
+    },
+    {
+      "epoch": 23.977272727272727,
+      "grad_norm": 1.7448116540908813,
+      "learning_rate": 0.0001,
+      "loss": 0.0941,
+      "step": 8440
+    },
+    {
+      "epoch": 24.005681818181817,
+      "grad_norm": 2.3489348888397217,
+      "learning_rate": 0.0001,
+      "loss": 0.0946,
+      "step": 8450
+    },
+    {
+      "epoch": 24.03409090909091,
+      "grad_norm": 2.110835075378418,
+      "learning_rate": 0.0001,
+      "loss": 0.0929,
+      "step": 8460
+    },
+    {
+      "epoch": 24.0625,
+      "grad_norm": 2.4186344146728516,
+      "learning_rate": 0.0001,
+      "loss": 0.0982,
+      "step": 8470
+    },
+    {
+      "epoch": 24.09090909090909,
+      "grad_norm": 1.9502896070480347,
+      "learning_rate": 0.0001,
+      "loss": 0.0969,
+      "step": 8480
+    },
+    {
+      "epoch": 24.119318181818183,
+      "grad_norm": 1.9351022243499756,
+      "learning_rate": 0.0001,
+      "loss": 0.094,
+      "step": 8490
+    },
+    {
+      "epoch": 24.147727272727273,
+      "grad_norm": 1.8484196662902832,
+      "learning_rate": 0.0001,
+      "loss": 0.0935,
+      "step": 8500
+    },
+    {
+      "epoch": 24.176136363636363,
+      "grad_norm": 1.8879474401474,
+      "learning_rate": 0.0001,
+      "loss": 0.0947,
+      "step": 8510
+    },
+    {
+      "epoch": 24.204545454545453,
+      "grad_norm": 2.4570751190185547,
+      "learning_rate": 0.0001,
+      "loss": 0.0918,
+      "step": 8520
+    },
+    {
+      "epoch": 24.232954545454547,
+      "grad_norm": 2.6654608249664307,
+      "learning_rate": 0.0001,
+      "loss": 0.0977,
+      "step": 8530
+    },
+    {
+      "epoch": 24.261363636363637,
+      "grad_norm": 2.244088888168335,
+      "learning_rate": 0.0001,
+      "loss": 0.0935,
+      "step": 8540
+    },
+    {
+      "epoch": 24.289772727272727,
+      "grad_norm": 2.7572576999664307,
+      "learning_rate": 0.0001,
+      "loss": 0.0956,
+      "step": 8550
+    },
+    {
+      "epoch": 24.318181818181817,
+      "grad_norm": 2.1149368286132812,
+      "learning_rate": 0.0001,
+      "loss": 0.0929,
+      "step": 8560
+    },
+    {
+      "epoch": 24.34659090909091,
+      "grad_norm": 1.9651392698287964,
+      "learning_rate": 0.0001,
+      "loss": 0.093,
+      "step": 8570
+    },
+    {
+      "epoch": 24.375,
+      "grad_norm": 2.118886947631836,
+      "learning_rate": 0.0001,
+      "loss": 0.0918,
+      "step": 8580
+    },
+    {
+      "epoch": 24.40340909090909,
+      "grad_norm": 1.857898235321045,
+      "learning_rate": 0.0001,
+      "loss": 0.0912,
+      "step": 8590
+    },
+    {
+      "epoch": 24.431818181818183,
+      "grad_norm": 1.8843599557876587,
+      "learning_rate": 0.0001,
+      "loss": 0.0923,
+      "step": 8600
+    },
+    {
+      "epoch": 24.460227272727273,
+      "grad_norm": 1.8303879499435425,
+      "learning_rate": 0.0001,
+      "loss": 0.0916,
+      "step": 8610
+    },
+    {
+      "epoch": 24.488636363636363,
+      "grad_norm": 2.0222115516662598,
+      "learning_rate": 0.0001,
+      "loss": 0.0917,
+      "step": 8620
+    },
+    {
+      "epoch": 24.517045454545453,
+      "grad_norm": 1.773148536682129,
+      "learning_rate": 0.0001,
+      "loss": 0.0939,
+      "step": 8630
+    },
+    {
+      "epoch": 24.545454545454547,
+      "grad_norm": 1.9186317920684814,
+      "learning_rate": 0.0001,
+      "loss": 0.09,
+      "step": 8640
+    },
+    {
+      "epoch": 24.573863636363637,
+      "grad_norm": 1.938623309135437,
+      "learning_rate": 0.0001,
+      "loss": 0.0935,
+      "step": 8650
+    },
+    {
+      "epoch": 24.602272727272727,
+      "grad_norm": 2.09529972076416,
+      "learning_rate": 0.0001,
+      "loss": 0.0936,
+      "step": 8660
+    },
+    {
+      "epoch": 24.630681818181817,
+      "grad_norm": 2.1375720500946045,
+      "learning_rate": 0.0001,
+      "loss": 0.0901,
+      "step": 8670
+    },
+    {
+      "epoch": 24.65909090909091,
+      "grad_norm": 3.3729183673858643,
+      "learning_rate": 0.0001,
+      "loss": 0.0936,
+      "step": 8680
+    },
+    {
+      "epoch": 24.6875,
+      "grad_norm": 2.765795946121216,
+      "learning_rate": 0.0001,
+      "loss": 0.0942,
+      "step": 8690
+    },
+    {
+      "epoch": 24.71590909090909,
+      "grad_norm": 3.0943186283111572,
+      "learning_rate": 0.0001,
+      "loss": 0.0915,
+      "step": 8700
+    },
+    {
+      "epoch": 24.744318181818183,
+      "grad_norm": 2.4649555683135986,
+      "learning_rate": 0.0001,
+      "loss": 0.0902,
+      "step": 8710
+    },
+    {
+      "epoch": 24.772727272727273,
+      "grad_norm": 2.455824851989746,
+      "learning_rate": 0.0001,
+      "loss": 0.0953,
+      "step": 8720
+    },
+    {
+      "epoch": 24.801136363636363,
+      "grad_norm": 2.0996339321136475,
+      "learning_rate": 0.0001,
+      "loss": 0.0949,
+      "step": 8730
+    },
+    {
+      "epoch": 24.829545454545453,
+      "grad_norm": 2.2499396800994873,
+      "learning_rate": 0.0001,
+      "loss": 0.0915,
+      "step": 8740
+    },
+    {
+      "epoch": 24.857954545454547,
+      "grad_norm": 1.920745611190796,
+      "learning_rate": 0.0001,
+      "loss": 0.09,
+      "step": 8750
+    },
+    {
+      "epoch": 24.886363636363637,
+      "grad_norm": 1.906348705291748,
+      "learning_rate": 0.0001,
+      "loss": 0.0921,
+      "step": 8760
+    },
+    {
+      "epoch": 24.914772727272727,
+      "grad_norm": 1.6197078227996826,
+      "learning_rate": 0.0001,
+      "loss": 0.0889,
+      "step": 8770
+    },
+    {
+      "epoch": 24.943181818181817,
+      "grad_norm": 1.6164134740829468,
+      "learning_rate": 0.0001,
+      "loss": 0.0907,
+      "step": 8780
+    },
+    {
+      "epoch": 24.97159090909091,
+      "grad_norm": 1.7616385221481323,
+      "learning_rate": 0.0001,
+      "loss": 0.0928,
+      "step": 8790
+    },
+    {
+      "epoch": 25.0,
+      "grad_norm": 1.5803983211517334,
+      "learning_rate": 0.0001,
+      "loss": 0.0926,
+      "step": 8800
+    },
+    {
+      "epoch": 25.02840909090909,
+      "grad_norm": 1.6062462329864502,
+      "learning_rate": 0.0001,
+      "loss": 0.0961,
+      "step": 8810
+    },
+    {
+      "epoch": 25.056818181818183,
+      "grad_norm": 1.5102510452270508,
+      "learning_rate": 0.0001,
+      "loss": 0.0917,
+      "step": 8820
+    },
+    {
+      "epoch": 25.085227272727273,
+      "grad_norm": 1.6694464683532715,
+      "learning_rate": 0.0001,
+      "loss": 0.0946,
+      "step": 8830
+    },
+    {
+      "epoch": 25.113636363636363,
+      "grad_norm": 1.6508196592330933,
+      "learning_rate": 0.0001,
+      "loss": 0.0926,
+      "step": 8840
+    },
+    {
+      "epoch": 25.142045454545453,
+      "grad_norm": 1.7550101280212402,
+      "learning_rate": 0.0001,
+      "loss": 0.0912,
+      "step": 8850
+    },
+    {
+      "epoch": 25.170454545454547,
+      "grad_norm": 1.6361439228057861,
+      "learning_rate": 0.0001,
+      "loss": 0.0986,
+      "step": 8860
+    },
+    {
+      "epoch": 25.198863636363637,
+      "grad_norm": 1.810949683189392,
+      "learning_rate": 0.0001,
+      "loss": 0.0952,
+      "step": 8870
+    },
+    {
+      "epoch": 25.227272727272727,
+      "grad_norm": 1.7442113161087036,
+      "learning_rate": 0.0001,
+      "loss": 0.0936,
+      "step": 8880
+    },
+    {
+      "epoch": 25.255681818181817,
+      "grad_norm": 1.9462569952011108,
+      "learning_rate": 0.0001,
+      "loss": 0.0978,
+      "step": 8890
+    },
+    {
+      "epoch": 25.28409090909091,
+      "grad_norm": 1.865937352180481,
+      "learning_rate": 0.0001,
+      "loss": 0.0949,
+      "step": 8900
+    },
+    {
+      "epoch": 25.3125,
+      "grad_norm": 1.5846524238586426,
+      "learning_rate": 0.0001,
+      "loss": 0.0912,
+      "step": 8910
+    },
+    {
+      "epoch": 25.34090909090909,
+      "grad_norm": 1.6086736917495728,
+      "learning_rate": 0.0001,
+      "loss": 0.0945,
+      "step": 8920
+    },
+    {
+      "epoch": 25.369318181818183,
+      "grad_norm": 1.665158748626709,
+      "learning_rate": 0.0001,
+      "loss": 0.0895,
+      "step": 8930
+    },
+    {
+      "epoch": 25.397727272727273,
+      "grad_norm": 1.7580013275146484,
+      "learning_rate": 0.0001,
+      "loss": 0.0888,
+      "step": 8940
+    },
+    {
+      "epoch": 25.426136363636363,
+      "grad_norm": 1.765702247619629,
+      "learning_rate": 0.0001,
+      "loss": 0.0975,
+      "step": 8950
+    },
+    {
+      "epoch": 25.454545454545453,
+      "grad_norm": 1.6462661027908325,
+      "learning_rate": 0.0001,
+      "loss": 0.0956,
+      "step": 8960
+    },
+    {
+      "epoch": 25.482954545454547,
+      "grad_norm": 2.0281505584716797,
+      "learning_rate": 0.0001,
+      "loss": 0.0912,
+      "step": 8970
+    },
+    {
+      "epoch": 25.511363636363637,
+      "grad_norm": 1.7537845373153687,
+      "learning_rate": 0.0001,
+      "loss": 0.0932,
+      "step": 8980
+    },
+    {
+      "epoch": 25.539772727272727,
+      "grad_norm": 1.776159644126892,
+      "learning_rate": 0.0001,
+      "loss": 0.0967,
+      "step": 8990
+    },
+    {
+      "epoch": 25.568181818181817,
+      "grad_norm": 1.6971244812011719,
+      "learning_rate": 0.0001,
+      "loss": 0.0924,
+      "step": 9000
+    },
+    {
+      "epoch": 25.59659090909091,
+      "grad_norm": 1.4512749910354614,
+      "learning_rate": 0.0001,
+      "loss": 0.0897,
+      "step": 9010
+    },
+    {
+      "epoch": 25.625,
+      "grad_norm": 1.7332180738449097,
+      "learning_rate": 0.0001,
+      "loss": 0.0923,
+      "step": 9020
+    },
+    {
+      "epoch": 25.65340909090909,
+      "grad_norm": 1.9260343313217163,
+      "learning_rate": 0.0001,
+      "loss": 0.0945,
+      "step": 9030
+    },
+    {
+      "epoch": 25.681818181818183,
+      "grad_norm": 1.56917405128479,
+      "learning_rate": 0.0001,
+      "loss": 0.0959,
+      "step": 9040
+    },
+    {
+      "epoch": 25.710227272727273,
+      "grad_norm": 1.4904402494430542,
+      "learning_rate": 0.0001,
+      "loss": 0.0961,
+      "step": 9050
+    },
+    {
+      "epoch": 25.738636363636363,
+      "grad_norm": 1.7849690914154053,
+      "learning_rate": 0.0001,
+      "loss": 0.0987,
+      "step": 9060
+    },
+    {
+      "epoch": 25.767045454545453,
+      "grad_norm": 1.63370943069458,
+      "learning_rate": 0.0001,
+      "loss": 0.0994,
+      "step": 9070
+    },
+    {
+      "epoch": 25.795454545454547,
+      "grad_norm": 2.0049540996551514,
+      "learning_rate": 0.0001,
+      "loss": 0.0929,
+      "step": 9080
+    },
+    {
+      "epoch": 25.823863636363637,
+      "grad_norm": 1.876146912574768,
+      "learning_rate": 0.0001,
+      "loss": 0.0967,
+      "step": 9090
+    },
+    {
+      "epoch": 25.852272727272727,
+      "grad_norm": 1.7067279815673828,
+      "learning_rate": 0.0001,
+      "loss": 0.0968,
+      "step": 9100
+    },
+    {
+      "epoch": 25.880681818181817,
+      "grad_norm": 1.7685781717300415,
+      "learning_rate": 0.0001,
+      "loss": 0.093,
+      "step": 9110
+    },
+    {
+      "epoch": 25.90909090909091,
+      "grad_norm": 1.493255376815796,
+      "learning_rate": 0.0001,
+      "loss": 0.0905,
+      "step": 9120
+    },
+    {
+      "epoch": 25.9375,
+      "grad_norm": 1.7333801984786987,
+      "learning_rate": 0.0001,
+      "loss": 0.0947,
+      "step": 9130
+    },
+    {
+      "epoch": 25.96590909090909,
+      "grad_norm": 1.5893990993499756,
+      "learning_rate": 0.0001,
+      "loss": 0.092,
+      "step": 9140
+    },
+    {
+      "epoch": 25.994318181818183,
+      "grad_norm": 1.9104375839233398,
+      "learning_rate": 0.0001,
+      "loss": 0.0963,
+      "step": 9150
+    },
+    {
+      "epoch": 26.022727272727273,
+      "grad_norm": 1.5926457643508911,
+      "learning_rate": 0.0001,
+      "loss": 0.0939,
+      "step": 9160
+    },
+    {
+      "epoch": 26.051136363636363,
+      "grad_norm": 1.5772978067398071,
+      "learning_rate": 0.0001,
+      "loss": 0.0872,
+      "step": 9170
+    },
+    {
+      "epoch": 26.079545454545453,
+      "grad_norm": 1.5457425117492676,
+      "learning_rate": 0.0001,
+      "loss": 0.0927,
+      "step": 9180
+    },
+    {
+      "epoch": 26.107954545454547,
+      "grad_norm": 1.6755262613296509,
+      "learning_rate": 0.0001,
+      "loss": 0.0954,
+      "step": 9190
+    },
+    {
+      "epoch": 26.136363636363637,
+      "grad_norm": 1.461090326309204,
+      "learning_rate": 0.0001,
+      "loss": 0.0936,
+      "step": 9200
+    },
+    {
+      "epoch": 26.164772727272727,
+      "grad_norm": 1.4528868198394775,
+      "learning_rate": 0.0001,
+      "loss": 0.0937,
+      "step": 9210
+    },
+    {
+      "epoch": 26.193181818181817,
+      "grad_norm": 1.5315214395523071,
+      "learning_rate": 0.0001,
+      "loss": 0.1011,
+      "step": 9220
+    },
+    {
+      "epoch": 26.22159090909091,
+      "grad_norm": 1.560640573501587,
+      "learning_rate": 0.0001,
+      "loss": 0.092,
+      "step": 9230
+    },
+    {
+      "epoch": 26.25,
+      "grad_norm": 1.4106330871582031,
+      "learning_rate": 0.0001,
+      "loss": 0.0971,
+      "step": 9240
+    },
+    {
+      "epoch": 26.27840909090909,
+      "grad_norm": 1.4047380685806274,
+      "learning_rate": 0.0001,
+      "loss": 0.0924,
+      "step": 9250
+    },
+    {
+      "epoch": 26.306818181818183,
+      "grad_norm": 1.4862667322158813,
+      "learning_rate": 0.0001,
+      "loss": 0.0979,
+      "step": 9260
+    },
+    {
+      "epoch": 26.335227272727273,
+      "grad_norm": 1.5880588293075562,
+      "learning_rate": 0.0001,
+      "loss": 0.0934,
+      "step": 9270
+    },
+    {
+      "epoch": 26.363636363636363,
+      "grad_norm": 1.428328514099121,
+      "learning_rate": 0.0001,
+      "loss": 0.0971,
+      "step": 9280
+    },
+    {
+      "epoch": 26.392045454545453,
+      "grad_norm": 1.5301146507263184,
+      "learning_rate": 0.0001,
+      "loss": 0.095,
+      "step": 9290
+    },
+    {
+      "epoch": 26.420454545454547,
+      "grad_norm": 1.6134599447250366,
+      "learning_rate": 0.0001,
+      "loss": 0.0975,
+      "step": 9300
+    },
+    {
+      "epoch": 26.448863636363637,
+      "grad_norm": 1.497191071510315,
+      "learning_rate": 0.0001,
+      "loss": 0.0948,
+      "step": 9310
+    },
+    {
+      "epoch": 26.477272727272727,
+      "grad_norm": 1.7432132959365845,
+      "learning_rate": 0.0001,
+      "loss": 0.0956,
+      "step": 9320
+    },
+    {
+      "epoch": 26.505681818181817,
+      "grad_norm": 1.3511826992034912,
+      "learning_rate": 0.0001,
+      "loss": 0.0919,
+      "step": 9330
+    },
+    {
+      "epoch": 26.53409090909091,
+      "grad_norm": 1.5720796585083008,
+      "learning_rate": 0.0001,
+      "loss": 0.0936,
+      "step": 9340
+    },
+    {
+      "epoch": 26.5625,
+      "grad_norm": 1.5396133661270142,
+      "learning_rate": 0.0001,
+      "loss": 0.0959,
+      "step": 9350
+    },
+    {
+      "epoch": 26.59090909090909,
+      "grad_norm": 1.6743911504745483,
+      "learning_rate": 0.0001,
+      "loss": 0.0936,
+      "step": 9360
+    },
+    {
+      "epoch": 26.619318181818183,
+      "grad_norm": 1.85866117477417,
+      "learning_rate": 0.0001,
+      "loss": 0.0995,
+      "step": 9370
+    },
+    {
+      "epoch": 26.647727272727273,
+      "grad_norm": 1.370617151260376,
+      "learning_rate": 0.0001,
+      "loss": 0.0962,
+      "step": 9380
+    },
+    {
+      "epoch": 26.676136363636363,
+      "grad_norm": 1.953228235244751,
+      "learning_rate": 0.0001,
+      "loss": 0.0981,
+      "step": 9390
+    },
+    {
+      "epoch": 26.704545454545453,
+      "grad_norm": 2.1395270824432373,
+      "learning_rate": 0.0001,
+      "loss": 0.094,
+      "step": 9400
+    },
+    {
+      "epoch": 26.732954545454547,
+      "grad_norm": 2.1154062747955322,
+      "learning_rate": 0.0001,
+      "loss": 0.0969,
+      "step": 9410
+    },
+    {
+      "epoch": 26.761363636363637,
+      "grad_norm": 1.7266603708267212,
+      "learning_rate": 0.0001,
+      "loss": 0.094,
+      "step": 9420
+    },
+    {
+      "epoch": 26.789772727272727,
+      "grad_norm": 1.8902325630187988,
+      "learning_rate": 0.0001,
+      "loss": 0.0912,
+      "step": 9430
+    },
+    {
+      "epoch": 26.818181818181817,
+      "grad_norm": 1.7739678621292114,
+      "learning_rate": 0.0001,
+      "loss": 0.0894,
+      "step": 9440
+    },
+    {
+      "epoch": 26.84659090909091,
+      "grad_norm": 1.9259507656097412,
+      "learning_rate": 0.0001,
+      "loss": 0.0915,
+      "step": 9450
+    },
+    {
+      "epoch": 26.875,
+      "grad_norm": 1.898050308227539,
+      "learning_rate": 0.0001,
+      "loss": 0.0892,
+      "step": 9460
+    },
+    {
+      "epoch": 26.90340909090909,
+      "grad_norm": 1.8099193572998047,
+      "learning_rate": 0.0001,
+      "loss": 0.09,
+      "step": 9470
+    },
+    {
+      "epoch": 26.931818181818183,
+      "grad_norm": 1.7650624513626099,
+      "learning_rate": 0.0001,
+      "loss": 0.0917,
+      "step": 9480
+    },
+    {
+      "epoch": 26.960227272727273,
+      "grad_norm": 1.813428521156311,
+      "learning_rate": 0.0001,
+      "loss": 0.085,
+      "step": 9490
+    },
+    {
+      "epoch": 26.988636363636363,
+      "grad_norm": 1.67322838306427,
+      "learning_rate": 0.0001,
+      "loss": 0.0919,
+      "step": 9500
+    },
+    {
+      "epoch": 27.017045454545453,
+      "grad_norm": 1.668229103088379,
+      "learning_rate": 0.0001,
+      "loss": 0.0889,
+      "step": 9510
+    },
+    {
+      "epoch": 27.045454545454547,
+      "grad_norm": 1.6641284227371216,
+      "learning_rate": 0.0001,
+      "loss": 0.0871,
+      "step": 9520
+    },
+    {
+      "epoch": 27.073863636363637,
+      "grad_norm": 1.5563348531723022,
+      "learning_rate": 0.0001,
+      "loss": 0.0856,
+      "step": 9530
+    },
+    {
+      "epoch": 27.102272727272727,
+      "grad_norm": 1.72633957862854,
+      "learning_rate": 0.0001,
+      "loss": 0.0898,
+      "step": 9540
+    },
+    {
+      "epoch": 27.130681818181817,
+      "grad_norm": 1.5098172426223755,
+      "learning_rate": 0.0001,
+      "loss": 0.0893,
+      "step": 9550
+    },
+    {
+      "epoch": 27.15909090909091,
+      "grad_norm": 1.7383455038070679,
+      "learning_rate": 0.0001,
+      "loss": 0.0859,
+      "step": 9560
+    },
+    {
+      "epoch": 27.1875,
+      "grad_norm": 1.732844352722168,
+      "learning_rate": 0.0001,
+      "loss": 0.0954,
+      "step": 9570
+    },
+    {
+      "epoch": 27.21590909090909,
+      "grad_norm": 1.5359463691711426,
+      "learning_rate": 0.0001,
+      "loss": 0.09,
+      "step": 9580
+    },
+    {
+      "epoch": 27.244318181818183,
+      "grad_norm": 1.5415890216827393,
+      "learning_rate": 0.0001,
+      "loss": 0.0947,
+      "step": 9590
+    },
+    {
+      "epoch": 27.272727272727273,
+      "grad_norm": 1.702905297279358,
+      "learning_rate": 0.0001,
+      "loss": 0.0868,
+      "step": 9600
+    },
+    {
+      "epoch": 27.301136363636363,
+      "grad_norm": 1.5514419078826904,
+      "learning_rate": 0.0001,
+      "loss": 0.0931,
+      "step": 9610
+    },
+    {
+      "epoch": 27.329545454545453,
+      "grad_norm": 1.6951237916946411,
+      "learning_rate": 0.0001,
+      "loss": 0.0946,
+      "step": 9620
+    },
+    {
+      "epoch": 27.357954545454547,
+      "grad_norm": 1.726776123046875,
+      "learning_rate": 0.0001,
+      "loss": 0.0914,
+      "step": 9630
+    },
+    {
+      "epoch": 27.386363636363637,
+      "grad_norm": 1.5069643259048462,
+      "learning_rate": 0.0001,
+      "loss": 0.0955,
+      "step": 9640
+    },
+    {
+      "epoch": 27.414772727272727,
+      "grad_norm": 1.468774676322937,
+      "learning_rate": 0.0001,
+      "loss": 0.0895,
+      "step": 9650
+    },
+    {
+      "epoch": 27.443181818181817,
+      "grad_norm": 1.7724437713623047,
+      "learning_rate": 0.0001,
+      "loss": 0.0914,
+      "step": 9660
+    },
+    {
+      "epoch": 27.47159090909091,
+      "grad_norm": 1.3602412939071655,
+      "learning_rate": 0.0001,
+      "loss": 0.0851,
+      "step": 9670
+    },
+    {
+      "epoch": 27.5,
+      "grad_norm": 1.5905205011367798,
+      "learning_rate": 0.0001,
+      "loss": 0.0892,
+      "step": 9680
+    },
+    {
+      "epoch": 27.52840909090909,
+      "grad_norm": 1.389020562171936,
+      "learning_rate": 0.0001,
+      "loss": 0.0882,
+      "step": 9690
+    },
+    {
+      "epoch": 27.556818181818183,
+      "grad_norm": 1.7958135604858398,
+      "learning_rate": 0.0001,
+      "loss": 0.0896,
+      "step": 9700
+    },
+    {
+      "epoch": 27.585227272727273,
+      "grad_norm": 1.629370093345642,
+      "learning_rate": 0.0001,
+      "loss": 0.0945,
+      "step": 9710
+    },
+    {
+      "epoch": 27.613636363636363,
+      "grad_norm": 1.8372656106948853,
+      "learning_rate": 0.0001,
+      "loss": 0.0917,
+      "step": 9720
+    },
+    {
+      "epoch": 27.642045454545453,
+      "grad_norm": 2.0454485416412354,
+      "learning_rate": 0.0001,
+      "loss": 0.0911,
+      "step": 9730
+    },
+    {
+      "epoch": 27.670454545454547,
+      "grad_norm": 1.712260365486145,
+      "learning_rate": 0.0001,
+      "loss": 0.0974,
+      "step": 9740
+    },
+    {
+      "epoch": 27.698863636363637,
+      "grad_norm": 1.8884317874908447,
+      "learning_rate": 0.0001,
+      "loss": 0.0891,
+      "step": 9750
+    },
+    {
+      "epoch": 27.727272727272727,
+      "grad_norm": 1.5852235555648804,
+      "learning_rate": 0.0001,
+      "loss": 0.0898,
+      "step": 9760
+    },
+    {
+      "epoch": 27.755681818181817,
+      "grad_norm": 1.4623692035675049,
+      "learning_rate": 0.0001,
+      "loss": 0.0866,
+      "step": 9770
+    },
+    {
+      "epoch": 27.78409090909091,
+      "grad_norm": 1.6662063598632812,
+      "learning_rate": 0.0001,
+      "loss": 0.0875,
+      "step": 9780
+    },
+    {
+      "epoch": 27.8125,
+      "grad_norm": 1.5903691053390503,
+      "learning_rate": 0.0001,
+      "loss": 0.0898,
+      "step": 9790
+    },
+    {
+      "epoch": 27.84090909090909,
+      "grad_norm": 1.575780987739563,
+      "learning_rate": 0.0001,
+      "loss": 0.0918,
+      "step": 9800
+    },
+    {
+      "epoch": 27.869318181818183,
+      "grad_norm": 1.4641938209533691,
+      "learning_rate": 0.0001,
+      "loss": 0.0865,
+      "step": 9810
+    },
+    {
+      "epoch": 27.897727272727273,
+      "grad_norm": 1.3061418533325195,
+      "learning_rate": 0.0001,
+      "loss": 0.0837,
+      "step": 9820
+    },
+    {
+      "epoch": 27.926136363636363,
+      "grad_norm": 1.5644803047180176,
+      "learning_rate": 0.0001,
+      "loss": 0.0894,
+      "step": 9830
+    },
+    {
+      "epoch": 27.954545454545453,
+      "grad_norm": 1.9139891862869263,
+      "learning_rate": 0.0001,
+      "loss": 0.0925,
+      "step": 9840
+    },
+    {
+      "epoch": 27.982954545454547,
+      "grad_norm": 1.4878668785095215,
+      "learning_rate": 0.0001,
+      "loss": 0.0892,
+      "step": 9850
+    },
+    {
+      "epoch": 28.011363636363637,
+      "grad_norm": 1.41493821144104,
+      "learning_rate": 0.0001,
+      "loss": 0.0914,
+      "step": 9860
+    },
+    {
+      "epoch": 28.039772727272727,
+      "grad_norm": 1.5278867483139038,
+      "learning_rate": 0.0001,
+      "loss": 0.0875,
+      "step": 9870
+    },
+    {
+      "epoch": 28.068181818181817,
+      "grad_norm": 1.3877768516540527,
+      "learning_rate": 0.0001,
+      "loss": 0.0914,
+      "step": 9880
+    },
+    {
+      "epoch": 28.09659090909091,
+      "grad_norm": 1.3210440874099731,
+      "learning_rate": 0.0001,
+      "loss": 0.0854,
+      "step": 9890
+    },
+    {
+      "epoch": 28.125,
+      "grad_norm": 1.3031365871429443,
+      "learning_rate": 0.0001,
+      "loss": 0.0854,
+      "step": 9900
+    },
+    {
+      "epoch": 28.15340909090909,
+      "grad_norm": 1.334957242012024,
+      "learning_rate": 0.0001,
+      "loss": 0.0926,
+      "step": 9910
+    },
+    {
+      "epoch": 28.181818181818183,
+      "grad_norm": 1.3106921911239624,
+      "learning_rate": 0.0001,
+      "loss": 0.0938,
+      "step": 9920
+    },
+    {
+      "epoch": 28.210227272727273,
+      "grad_norm": 1.9076578617095947,
+      "learning_rate": 0.0001,
+      "loss": 0.0869,
+      "step": 9930
+    },
+    {
+      "epoch": 28.238636363636363,
+      "grad_norm": 1.8114360570907593,
+      "learning_rate": 0.0001,
+      "loss": 0.0923,
+      "step": 9940
+    },
+    {
+      "epoch": 28.267045454545453,
+      "grad_norm": 1.475202202796936,
+      "learning_rate": 0.0001,
+      "loss": 0.0942,
+      "step": 9950
+    },
+    {
+      "epoch": 28.295454545454547,
+      "grad_norm": 1.6145161390304565,
+      "learning_rate": 0.0001,
+      "loss": 0.0842,
+      "step": 9960
+    },
+    {
+      "epoch": 28.323863636363637,
+      "grad_norm": 1.5492805242538452,
+      "learning_rate": 0.0001,
+      "loss": 0.0866,
+      "step": 9970
+    },
+    {
+      "epoch": 28.352272727272727,
+      "grad_norm": 1.6605106592178345,
+      "learning_rate": 0.0001,
+      "loss": 0.0917,
+      "step": 9980
+    },
+    {
+      "epoch": 28.380681818181817,
+      "grad_norm": 1.6075584888458252,
+      "learning_rate": 0.0001,
+      "loss": 0.0888,
+      "step": 9990
+    },
+    {
+      "epoch": 28.40909090909091,
+      "grad_norm": 1.3253341913223267,
+      "learning_rate": 0.0001,
+      "loss": 0.0883,
+      "step": 10000
+    },
+    {
+      "epoch": 28.4375,
+      "grad_norm": 1.5245485305786133,
+      "learning_rate": 0.0001,
+      "loss": 0.0899,
+      "step": 10010
+    },
+    {
+      "epoch": 28.46590909090909,
+      "grad_norm": 1.7123736143112183,
+      "learning_rate": 0.0001,
+      "loss": 0.0868,
+      "step": 10020
+    },
+    {
+      "epoch": 28.494318181818183,
+      "grad_norm": 1.572593092918396,
+      "learning_rate": 0.0001,
+      "loss": 0.0865,
+      "step": 10030
+    },
+    {
+      "epoch": 28.522727272727273,
+      "grad_norm": 1.693306565284729,
+      "learning_rate": 0.0001,
+      "loss": 0.0904,
+      "step": 10040
+    },
+    {
+      "epoch": 28.551136363636363,
+      "grad_norm": 1.8397178649902344,
+      "learning_rate": 0.0001,
+      "loss": 0.0898,
+      "step": 10050
+    },
+    {
+      "epoch": 28.579545454545453,
+      "grad_norm": 1.6443665027618408,
+      "learning_rate": 0.0001,
+      "loss": 0.0911,
+      "step": 10060
+    },
+    {
+      "epoch": 28.607954545454547,
+      "grad_norm": 1.5029046535491943,
+      "learning_rate": 0.0001,
+      "loss": 0.0918,
+      "step": 10070
+    },
+    {
+      "epoch": 28.636363636363637,
+      "grad_norm": 1.4545835256576538,
+      "learning_rate": 0.0001,
+      "loss": 0.0851,
+      "step": 10080
+    },
+    {
+      "epoch": 28.664772727272727,
+      "grad_norm": 1.6282371282577515,
+      "learning_rate": 0.0001,
+      "loss": 0.0879,
+      "step": 10090
+    },
+    {
+      "epoch": 28.693181818181817,
+      "grad_norm": 1.659529209136963,
+      "learning_rate": 0.0001,
+      "loss": 0.0881,
+      "step": 10100
+    },
+    {
+      "epoch": 28.72159090909091,
+      "grad_norm": 1.797834873199463,
+      "learning_rate": 0.0001,
+      "loss": 0.0919,
+      "step": 10110
+    },
+    {
+      "epoch": 28.75,
+      "grad_norm": 1.6592271327972412,
+      "learning_rate": 0.0001,
+      "loss": 0.0833,
+      "step": 10120
+    },
+    {
+      "epoch": 28.77840909090909,
+      "grad_norm": 1.4718973636627197,
+      "learning_rate": 0.0001,
+      "loss": 0.0918,
+      "step": 10130
+    },
+    {
+      "epoch": 28.806818181818183,
+      "grad_norm": 1.2769532203674316,
+      "learning_rate": 0.0001,
+      "loss": 0.0915,
+      "step": 10140
+    },
+    {
+      "epoch": 28.835227272727273,
+      "grad_norm": 1.3063241243362427,
+      "learning_rate": 0.0001,
+      "loss": 0.0856,
+      "step": 10150
+    },
+    {
+      "epoch": 28.863636363636363,
+      "grad_norm": 1.497151494026184,
+      "learning_rate": 0.0001,
+      "loss": 0.0927,
+      "step": 10160
+    },
+    {
+      "epoch": 28.892045454545453,
+      "grad_norm": 1.538161277770996,
+      "learning_rate": 0.0001,
+      "loss": 0.0849,
+      "step": 10170
+    },
+    {
+      "epoch": 28.920454545454547,
+      "grad_norm": 1.5118201971054077,
+      "learning_rate": 0.0001,
+      "loss": 0.0891,
+      "step": 10180
+    },
+    {
+      "epoch": 28.948863636363637,
+      "grad_norm": 1.5277782678604126,
+      "learning_rate": 0.0001,
+      "loss": 0.0872,
+      "step": 10190
+    },
+    {
+      "epoch": 28.977272727272727,
+      "grad_norm": 1.4347714185714722,
+      "learning_rate": 0.0001,
+      "loss": 0.0868,
+      "step": 10200
+    },
+    {
+      "epoch": 29.005681818181817,
+      "grad_norm": 1.3337539434432983,
+      "learning_rate": 0.0001,
+      "loss": 0.0872,
+      "step": 10210
+    },
+    {
+      "epoch": 29.03409090909091,
+      "grad_norm": 1.6862537860870361,
+      "learning_rate": 0.0001,
+      "loss": 0.0841,
+      "step": 10220
+    },
+    {
+      "epoch": 29.0625,
+      "grad_norm": 1.4856092929840088,
+      "learning_rate": 0.0001,
+      "loss": 0.0844,
+      "step": 10230
+    },
+    {
+      "epoch": 29.09090909090909,
+      "grad_norm": 1.6301956176757812,
+      "learning_rate": 0.0001,
+      "loss": 0.0855,
+      "step": 10240
+    },
+    {
+      "epoch": 29.119318181818183,
+      "grad_norm": 1.4501256942749023,
+      "learning_rate": 0.0001,
+      "loss": 0.0866,
+      "step": 10250
+    },
+    {
+      "epoch": 29.147727272727273,
+      "grad_norm": 1.4976732730865479,
+      "learning_rate": 0.0001,
+      "loss": 0.0871,
+      "step": 10260
+    },
+    {
+      "epoch": 29.176136363636363,
+      "grad_norm": 1.5367155075073242,
+      "learning_rate": 0.0001,
+      "loss": 0.083,
+      "step": 10270
+    },
+    {
+      "epoch": 29.204545454545453,
+      "grad_norm": 1.4822138547897339,
+      "learning_rate": 0.0001,
+      "loss": 0.0877,
+      "step": 10280
+    },
+    {
+      "epoch": 29.232954545454547,
+      "grad_norm": 1.3127697706222534,
+      "learning_rate": 0.0001,
+      "loss": 0.0813,
+      "step": 10290
+    },
+    {
+      "epoch": 29.261363636363637,
+      "grad_norm": 1.4777271747589111,
+      "learning_rate": 0.0001,
+      "loss": 0.0886,
+      "step": 10300
+    },
+    {
+      "epoch": 29.289772727272727,
+      "grad_norm": 1.4947670698165894,
+      "learning_rate": 0.0001,
+      "loss": 0.0887,
+      "step": 10310
+    },
+    {
+      "epoch": 29.318181818181817,
+      "grad_norm": 1.7451188564300537,
+      "learning_rate": 0.0001,
+      "loss": 0.0892,
+      "step": 10320
+    },
+    {
+      "epoch": 29.34659090909091,
+      "grad_norm": 1.5838991403579712,
+      "learning_rate": 0.0001,
+      "loss": 0.0867,
+      "step": 10330
+    },
+    {
+      "epoch": 29.375,
+      "grad_norm": 1.4703574180603027,
+      "learning_rate": 0.0001,
+      "loss": 0.0879,
+      "step": 10340
+    },
+    {
+      "epoch": 29.40340909090909,
+      "grad_norm": 1.335748553276062,
+      "learning_rate": 0.0001,
+      "loss": 0.0838,
+      "step": 10350
+    },
+    {
+      "epoch": 29.431818181818183,
+      "grad_norm": 1.5957832336425781,
+      "learning_rate": 0.0001,
+      "loss": 0.0849,
+      "step": 10360
+    },
+    {
+      "epoch": 29.460227272727273,
+      "grad_norm": 1.7073551416397095,
+      "learning_rate": 0.0001,
+      "loss": 0.0866,
+      "step": 10370
+    },
+    {
+      "epoch": 29.488636363636363,
+      "grad_norm": 1.4526339769363403,
+      "learning_rate": 0.0001,
+      "loss": 0.0864,
+      "step": 10380
+    },
+    {
+      "epoch": 29.517045454545453,
+      "grad_norm": 1.439193606376648,
+      "learning_rate": 0.0001,
+      "loss": 0.0876,
+      "step": 10390
+    },
+    {
+      "epoch": 29.545454545454547,
+      "grad_norm": 1.460564136505127,
+      "learning_rate": 0.0001,
+      "loss": 0.0871,
+      "step": 10400
+    },
+    {
+      "epoch": 29.573863636363637,
+      "grad_norm": 1.3847678899765015,
+      "learning_rate": 0.0001,
+      "loss": 0.0813,
+      "step": 10410
+    },
+    {
+      "epoch": 29.602272727272727,
+      "grad_norm": 1.4355100393295288,
+      "learning_rate": 0.0001,
+      "loss": 0.084,
+      "step": 10420
+    },
+    {
+      "epoch": 29.630681818181817,
+      "grad_norm": 1.5265635251998901,
+      "learning_rate": 0.0001,
+      "loss": 0.085,
+      "step": 10430
+    },
+    {
+      "epoch": 29.65909090909091,
+      "grad_norm": 1.5522079467773438,
+      "learning_rate": 0.0001,
+      "loss": 0.0898,
+      "step": 10440
+    },
+    {
+      "epoch": 29.6875,
+      "grad_norm": 1.5721166133880615,
+      "learning_rate": 0.0001,
+      "loss": 0.0882,
+      "step": 10450
+    },
+    {
+      "epoch": 29.71590909090909,
+      "grad_norm": 1.7917033433914185,
+      "learning_rate": 0.0001,
+      "loss": 0.0908,
+      "step": 10460
+    },
+    {
+      "epoch": 29.744318181818183,
+      "grad_norm": 1.5711758136749268,
+      "learning_rate": 0.0001,
+      "loss": 0.085,
+      "step": 10470
+    },
+    {
+      "epoch": 29.772727272727273,
+      "grad_norm": 1.7206867933273315,
+      "learning_rate": 0.0001,
+      "loss": 0.0861,
+      "step": 10480
+    },
+    {
+      "epoch": 29.801136363636363,
+      "grad_norm": 2.0375149250030518,
+      "learning_rate": 0.0001,
+      "loss": 0.0908,
+      "step": 10490
+    },
+    {
+      "epoch": 29.829545454545453,
+      "grad_norm": 1.7295266389846802,
+      "learning_rate": 0.0001,
+      "loss": 0.0853,
+      "step": 10500
+    },
+    {
+      "epoch": 29.857954545454547,
+      "grad_norm": 1.5999189615249634,
+      "learning_rate": 0.0001,
+      "loss": 0.0831,
+      "step": 10510
+    },
+    {
+      "epoch": 29.886363636363637,
+      "grad_norm": 1.6992350816726685,
+      "learning_rate": 0.0001,
+      "loss": 0.0891,
+      "step": 10520
+    },
+    {
+      "epoch": 29.914772727272727,
+      "grad_norm": 1.819216012954712,
+      "learning_rate": 0.0001,
+      "loss": 0.0801,
+      "step": 10530
+    },
+    {
+      "epoch": 29.943181818181817,
+      "grad_norm": 1.4884485006332397,
+      "learning_rate": 0.0001,
+      "loss": 0.0881,
+      "step": 10540
+    },
+    {
+      "epoch": 29.97159090909091,
+      "grad_norm": 1.6694735288619995,
+      "learning_rate": 0.0001,
+      "loss": 0.0832,
+      "step": 10550
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 1.6155649423599243,
+      "learning_rate": 0.0001,
+      "loss": 0.0838,
+      "step": 10560
+    },
+    {
+      "epoch": 30.02840909090909,
+      "grad_norm": 1.3682494163513184,
+      "learning_rate": 0.0001,
+      "loss": 0.081,
+      "step": 10570
+    },
+    {
+      "epoch": 30.056818181818183,
+      "grad_norm": 1.2710999250411987,
+      "learning_rate": 0.0001,
+      "loss": 0.0797,
+      "step": 10580
+    },
+    {
+      "epoch": 30.085227272727273,
+      "grad_norm": 1.2718095779418945,
+      "learning_rate": 0.0001,
+      "loss": 0.0838,
+      "step": 10590
+    },
+    {
+      "epoch": 30.113636363636363,
+      "grad_norm": 1.4790953397750854,
+      "learning_rate": 0.0001,
+      "loss": 0.0832,
+      "step": 10600
+    },
+    {
+      "epoch": 30.142045454545453,
+      "grad_norm": 4.220450401306152,
+      "learning_rate": 0.0001,
+      "loss": 0.0899,
+      "step": 10610
+    },
+    {
+      "epoch": 30.170454545454547,
+      "grad_norm": 3.7375528812408447,
+      "learning_rate": 0.0001,
+      "loss": 0.0921,
+      "step": 10620
+    },
+    {
+      "epoch": 30.198863636363637,
+      "grad_norm": 2.229771614074707,
+      "learning_rate": 0.0001,
+      "loss": 0.0843,
+      "step": 10630
+    },
+    {
+      "epoch": 30.227272727272727,
+      "grad_norm": 1.9127345085144043,
+      "learning_rate": 0.0001,
+      "loss": 0.0843,
+      "step": 10640
+    },
+    {
+      "epoch": 30.255681818181817,
+      "grad_norm": 1.7945572137832642,
+      "learning_rate": 0.0001,
+      "loss": 0.0855,
+      "step": 10650
+    },
+    {
+      "epoch": 30.28409090909091,
+      "grad_norm": 2.2238516807556152,
+      "learning_rate": 0.0001,
+      "loss": 0.08,
+      "step": 10660
+    },
+    {
+      "epoch": 30.3125,
+      "grad_norm": 2.0408074855804443,
+      "learning_rate": 0.0001,
+      "loss": 0.0812,
+      "step": 10670
+    },
+    {
+      "epoch": 30.34090909090909,
+      "grad_norm": 1.6794369220733643,
+      "learning_rate": 0.0001,
+      "loss": 0.0808,
+      "step": 10680
+    },
+    {
+      "epoch": 30.369318181818183,
+      "grad_norm": 1.8030424118041992,
+      "learning_rate": 0.0001,
+      "loss": 0.0819,
+      "step": 10690
+    },
+    {
+      "epoch": 30.397727272727273,
+      "grad_norm": 1.8233095407485962,
+      "learning_rate": 0.0001,
+      "loss": 0.0819,
+      "step": 10700
+    },
+    {
+      "epoch": 30.426136363636363,
+      "grad_norm": 1.6274789571762085,
+      "learning_rate": 0.0001,
+      "loss": 0.0806,
+      "step": 10710
+    },
+    {
+      "epoch": 30.454545454545453,
+      "grad_norm": 1.6231640577316284,
+      "learning_rate": 0.0001,
+      "loss": 0.0777,
+      "step": 10720
+    },
+    {
+      "epoch": 30.482954545454547,
+      "grad_norm": 1.6162152290344238,
+      "learning_rate": 0.0001,
+      "loss": 0.0864,
+      "step": 10730
+    },
+    {
+      "epoch": 30.511363636363637,
+      "grad_norm": 1.9388537406921387,
+      "learning_rate": 0.0001,
+      "loss": 0.0833,
+      "step": 10740
+    },
+    {
+      "epoch": 30.539772727272727,
+      "grad_norm": 1.3377724885940552,
+      "learning_rate": 0.0001,
+      "loss": 0.0836,
+      "step": 10750
+    },
+    {
+      "epoch": 30.568181818181817,
+      "grad_norm": 1.4415621757507324,
+      "learning_rate": 0.0001,
+      "loss": 0.0805,
+      "step": 10760
+    },
+    {
+      "epoch": 30.59659090909091,
+      "grad_norm": 1.4234329462051392,
+      "learning_rate": 0.0001,
+      "loss": 0.088,
+      "step": 10770
+    },
+    {
+      "epoch": 30.625,
+      "grad_norm": 1.5712944269180298,
+      "learning_rate": 0.0001,
+      "loss": 0.0866,
+      "step": 10780
+    },
+    {
+      "epoch": 30.65340909090909,
+      "grad_norm": 1.3370726108551025,
+      "learning_rate": 0.0001,
+      "loss": 0.086,
+      "step": 10790
+    },
+    {
+      "epoch": 30.681818181818183,
+      "grad_norm": 1.976441502571106,
+      "learning_rate": 0.0001,
+      "loss": 0.0894,
+      "step": 10800
+    },
+    {
+      "epoch": 30.710227272727273,
+      "grad_norm": 1.6814191341400146,
+      "learning_rate": 0.0001,
+      "loss": 0.083,
+      "step": 10810
+    },
+    {
+      "epoch": 30.738636363636363,
+      "grad_norm": 1.406453013420105,
+      "learning_rate": 0.0001,
+      "loss": 0.087,
+      "step": 10820
+    },
+    {
+      "epoch": 30.767045454545453,
+      "grad_norm": 1.406148910522461,
+      "learning_rate": 0.0001,
+      "loss": 0.0879,
+      "step": 10830
+    },
+    {
+      "epoch": 30.795454545454547,
+      "grad_norm": 1.2990154027938843,
+      "learning_rate": 0.0001,
+      "loss": 0.0826,
+      "step": 10840
+    },
+    {
+      "epoch": 30.823863636363637,
+      "grad_norm": 1.3759732246398926,
+      "learning_rate": 0.0001,
+      "loss": 0.0826,
+      "step": 10850
+    },
+    {
+      "epoch": 30.852272727272727,
+      "grad_norm": 1.6295632123947144,
+      "learning_rate": 0.0001,
+      "loss": 0.086,
+      "step": 10860
+    },
+    {
+      "epoch": 30.880681818181817,
+      "grad_norm": 1.4913337230682373,
+      "learning_rate": 0.0001,
+      "loss": 0.0879,
+      "step": 10870
+    },
+    {
+      "epoch": 30.90909090909091,
+      "grad_norm": 1.6488226652145386,
+      "learning_rate": 0.0001,
+      "loss": 0.0865,
+      "step": 10880
+    },
+    {
+      "epoch": 30.9375,
+      "grad_norm": 1.4675461053848267,
+      "learning_rate": 0.0001,
+      "loss": 0.0819,
+      "step": 10890
+    },
+    {
+      "epoch": 30.96590909090909,
+      "grad_norm": 1.4237877130508423,
+      "learning_rate": 0.0001,
+      "loss": 0.0878,
+      "step": 10900
+    },
+    {
+      "epoch": 30.994318181818183,
+      "grad_norm": 1.673284888267517,
+      "learning_rate": 0.0001,
+      "loss": 0.0876,
+      "step": 10910
+    },
+    {
+      "epoch": 31.022727272727273,
+      "grad_norm": 1.6682919263839722,
+      "learning_rate": 0.0001,
+      "loss": 0.0815,
+      "step": 10920
+    },
+    {
+      "epoch": 31.051136363636363,
+      "grad_norm": 1.567307949066162,
+      "learning_rate": 0.0001,
+      "loss": 0.0914,
+      "step": 10930
+    },
+    {
+      "epoch": 31.079545454545453,
+      "grad_norm": 1.646187424659729,
+      "learning_rate": 0.0001,
+      "loss": 0.0891,
+      "step": 10940
+    },
+    {
+      "epoch": 31.107954545454547,
+      "grad_norm": 1.3649544715881348,
+      "learning_rate": 0.0001,
+      "loss": 0.0822,
+      "step": 10950
+    },
+    {
+      "epoch": 31.136363636363637,
+      "grad_norm": 1.5282686948776245,
+      "learning_rate": 0.0001,
+      "loss": 0.0854,
+      "step": 10960
+    },
+    {
+      "epoch": 31.164772727272727,
+      "grad_norm": 1.5806162357330322,
+      "learning_rate": 0.0001,
+      "loss": 0.0835,
+      "step": 10970
+    },
+    {
+      "epoch": 31.193181818181817,
+      "grad_norm": 1.44295334815979,
+      "learning_rate": 0.0001,
+      "loss": 0.0872,
+      "step": 10980
+    },
+    {
+      "epoch": 31.22159090909091,
+      "grad_norm": 1.354772925376892,
+      "learning_rate": 0.0001,
+      "loss": 0.0843,
+      "step": 10990
+    },
+    {
+      "epoch": 31.25,
+      "grad_norm": 1.0891101360321045,
+      "learning_rate": 0.0001,
+      "loss": 0.0873,
+      "step": 11000
+    },
+    {
+      "epoch": 31.27840909090909,
+      "grad_norm": 1.6048698425292969,
+      "learning_rate": 0.0001,
+      "loss": 0.0872,
+      "step": 11010
+    },
+    {
+      "epoch": 31.306818181818183,
+      "grad_norm": 1.6361032724380493,
+      "learning_rate": 0.0001,
+      "loss": 0.084,
+      "step": 11020
+    },
+    {
+      "epoch": 31.335227272727273,
+      "grad_norm": 1.4586684703826904,
+      "learning_rate": 0.0001,
+      "loss": 0.0855,
+      "step": 11030
+    },
+    {
+      "epoch": 31.363636363636363,
+      "grad_norm": 1.440597653388977,
+      "learning_rate": 0.0001,
+      "loss": 0.0847,
+      "step": 11040
+    },
+    {
+      "epoch": 31.392045454545453,
+      "grad_norm": 1.579797387123108,
+      "learning_rate": 0.0001,
+      "loss": 0.085,
+      "step": 11050
+    },
+    {
+      "epoch": 31.420454545454547,
+      "grad_norm": 1.5352915525436401,
+      "learning_rate": 0.0001,
+      "loss": 0.0847,
+      "step": 11060
+    },
+    {
+      "epoch": 31.448863636363637,
+      "grad_norm": 1.5840104818344116,
+      "learning_rate": 0.0001,
+      "loss": 0.0853,
+      "step": 11070
+    },
+    {
+      "epoch": 31.477272727272727,
+      "grad_norm": 1.4220658540725708,
+      "learning_rate": 0.0001,
+      "loss": 0.0834,
+      "step": 11080
+    },
+    {
+      "epoch": 31.505681818181817,
+      "grad_norm": 1.3997197151184082,
+      "learning_rate": 0.0001,
+      "loss": 0.0874,
+      "step": 11090
+    },
+    {
+      "epoch": 31.53409090909091,
+      "grad_norm": 1.4210138320922852,
+      "learning_rate": 0.0001,
+      "loss": 0.0884,
+      "step": 11100
+    },
+    {
+      "epoch": 31.5625,
+      "grad_norm": 1.6847612857818604,
+      "learning_rate": 0.0001,
+      "loss": 0.0881,
+      "step": 11110
+    },
+    {
+      "epoch": 31.59090909090909,
+      "grad_norm": 1.4961141347885132,
+      "learning_rate": 0.0001,
+      "loss": 0.0891,
+      "step": 11120
+    },
+    {
+      "epoch": 31.619318181818183,
+      "grad_norm": 1.3087717294692993,
+      "learning_rate": 0.0001,
+      "loss": 0.0857,
+      "step": 11130
+    },
+    {
+      "epoch": 31.647727272727273,
+      "grad_norm": 1.3386777639389038,
+      "learning_rate": 0.0001,
+      "loss": 0.0828,
+      "step": 11140
+    },
+    {
+      "epoch": 31.676136363636363,
+      "grad_norm": 1.3143402338027954,
+      "learning_rate": 0.0001,
+      "loss": 0.0859,
+      "step": 11150
+    },
+    {
+      "epoch": 31.704545454545453,
+      "grad_norm": 2.002627372741699,
+      "learning_rate": 0.0001,
+      "loss": 0.0895,
+      "step": 11160
+    },
+    {
+      "epoch": 31.732954545454547,
+      "grad_norm": 1.8589037656784058,
+      "learning_rate": 0.0001,
+      "loss": 0.0922,
+      "step": 11170
+    },
+    {
+      "epoch": 31.761363636363637,
+      "grad_norm": 1.910962700843811,
+      "learning_rate": 0.0001,
+      "loss": 0.0927,
+      "step": 11180
+    },
+    {
+      "epoch": 31.789772727272727,
+      "grad_norm": 1.631377100944519,
+      "learning_rate": 0.0001,
+      "loss": 0.091,
+      "step": 11190
+    },
+    {
+      "epoch": 31.818181818181817,
+      "grad_norm": 1.6555157899856567,
+      "learning_rate": 0.0001,
+      "loss": 0.0869,
+      "step": 11200
+    },
+    {
+      "epoch": 31.84659090909091,
+      "grad_norm": 1.6346405744552612,
+      "learning_rate": 0.0001,
+      "loss": 0.0883,
+      "step": 11210
+    },
+    {
+      "epoch": 31.875,
+      "grad_norm": 1.2686203718185425,
+      "learning_rate": 0.0001,
+      "loss": 0.0856,
+      "step": 11220
+    },
+    {
+      "epoch": 31.90340909090909,
+      "grad_norm": 1.062166690826416,
+      "learning_rate": 0.0001,
+      "loss": 0.0823,
+      "step": 11230
+    },
+    {
+      "epoch": 31.931818181818183,
+      "grad_norm": 1.0907399654388428,
+      "learning_rate": 0.0001,
+      "loss": 0.0825,
+      "step": 11240
+    },
+    {
+      "epoch": 31.960227272727273,
+      "grad_norm": 1.4261188507080078,
+      "learning_rate": 0.0001,
+      "loss": 0.0857,
+      "step": 11250
+    },
+    {
+      "epoch": 31.988636363636363,
+      "grad_norm": 1.3329896926879883,
+      "learning_rate": 0.0001,
+      "loss": 0.083,
+      "step": 11260
+    },
+    {
+      "epoch": 32.01704545454545,
+      "grad_norm": 1.3007625341415405,
+      "learning_rate": 0.0001,
+      "loss": 0.0843,
+      "step": 11270
+    },
+    {
+      "epoch": 32.04545454545455,
+      "grad_norm": 1.3520420789718628,
+      "learning_rate": 0.0001,
+      "loss": 0.0834,
+      "step": 11280
+    },
+    {
+      "epoch": 32.07386363636363,
+      "grad_norm": 1.2438174486160278,
+      "learning_rate": 0.0001,
+      "loss": 0.0852,
+      "step": 11290
+    },
+    {
+      "epoch": 32.10227272727273,
+      "grad_norm": 1.422757863998413,
+      "learning_rate": 0.0001,
+      "loss": 0.0834,
+      "step": 11300
+    },
+    {
+      "epoch": 32.13068181818182,
+      "grad_norm": 1.2228397130966187,
+      "learning_rate": 0.0001,
+      "loss": 0.0852,
+      "step": 11310
+    },
+    {
+      "epoch": 32.15909090909091,
+      "grad_norm": 1.2606850862503052,
+      "learning_rate": 0.0001,
+      "loss": 0.0853,
+      "step": 11320
+    },
+    {
+      "epoch": 32.1875,
+      "grad_norm": 1.2866079807281494,
+      "learning_rate": 0.0001,
+      "loss": 0.0834,
+      "step": 11330
+    },
+    {
+      "epoch": 32.21590909090909,
+      "grad_norm": 1.2392500638961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0808,
+      "step": 11340
+    },
+    {
+      "epoch": 32.24431818181818,
+      "grad_norm": 1.3232767581939697,
+      "learning_rate": 0.0001,
+      "loss": 0.0842,
+      "step": 11350
+    },
+    {
+      "epoch": 32.27272727272727,
+      "grad_norm": 1.2873938083648682,
+      "learning_rate": 0.0001,
+      "loss": 0.086,
+      "step": 11360
+    },
+    {
+      "epoch": 32.30113636363637,
+      "grad_norm": 1.2557556629180908,
+      "learning_rate": 0.0001,
+      "loss": 0.0828,
+      "step": 11370
+    },
+    {
+      "epoch": 32.32954545454545,
+      "grad_norm": 1.2549748420715332,
+      "learning_rate": 0.0001,
+      "loss": 0.0877,
+      "step": 11380
+    },
+    {
+      "epoch": 32.35795454545455,
+      "grad_norm": 1.199981689453125,
+      "learning_rate": 0.0001,
+      "loss": 0.0876,
+      "step": 11390
+    },
+    {
+      "epoch": 32.38636363636363,
+      "grad_norm": 1.204467535018921,
+      "learning_rate": 0.0001,
+      "loss": 0.0804,
+      "step": 11400
+    },
+    {
+      "epoch": 32.41477272727273,
+      "grad_norm": 1.5204373598098755,
+      "learning_rate": 0.0001,
+      "loss": 0.087,
+      "step": 11410
+    },
+    {
+      "epoch": 32.44318181818182,
+      "grad_norm": 1.773655652999878,
+      "learning_rate": 0.0001,
+      "loss": 0.0824,
+      "step": 11420
+    },
+    {
+      "epoch": 32.47159090909091,
+      "grad_norm": 2.06709885597229,
+      "learning_rate": 0.0001,
+      "loss": 0.0848,
+      "step": 11430
+    },
+    {
+      "epoch": 32.5,
+      "grad_norm": 1.8769580125808716,
+      "learning_rate": 0.0001,
+      "loss": 0.0848,
+      "step": 11440
+    },
+    {
+      "epoch": 32.52840909090909,
+      "grad_norm": 1.8511193990707397,
+      "learning_rate": 0.0001,
+      "loss": 0.0834,
+      "step": 11450
+    },
+    {
+      "epoch": 32.55681818181818,
+      "grad_norm": 1.5107983350753784,
+      "learning_rate": 0.0001,
+      "loss": 0.0855,
+      "step": 11460
+    },
+    {
+      "epoch": 32.58522727272727,
+      "grad_norm": 1.7288358211517334,
+      "learning_rate": 0.0001,
+      "loss": 0.0806,
+      "step": 11470
+    },
+    {
+      "epoch": 32.61363636363637,
+      "grad_norm": 1.7605435848236084,
+      "learning_rate": 0.0001,
+      "loss": 0.0795,
+      "step": 11480
+    },
+    {
+      "epoch": 32.64204545454545,
+      "grad_norm": 1.3793346881866455,
+      "learning_rate": 0.0001,
+      "loss": 0.0768,
+      "step": 11490
+    },
+    {
+      "epoch": 32.67045454545455,
+      "grad_norm": 1.468401551246643,
+      "learning_rate": 0.0001,
+      "loss": 0.0773,
+      "step": 11500
+    },
+    {
+      "epoch": 32.69886363636363,
+      "grad_norm": 1.4680513143539429,
+      "learning_rate": 0.0001,
+      "loss": 0.0832,
+      "step": 11510
+    },
+    {
+      "epoch": 32.72727272727273,
+      "grad_norm": 1.3982374668121338,
+      "learning_rate": 0.0001,
+      "loss": 0.081,
+      "step": 11520
+    },
+    {
+      "epoch": 32.75568181818182,
+      "grad_norm": 1.5383310317993164,
+      "learning_rate": 0.0001,
+      "loss": 0.0809,
+      "step": 11530
+    },
+    {
+      "epoch": 32.78409090909091,
+      "grad_norm": 1.2121058702468872,
+      "learning_rate": 0.0001,
+      "loss": 0.078,
+      "step": 11540
+    },
+    {
+      "epoch": 32.8125,
+      "grad_norm": 1.4773236513137817,
+      "learning_rate": 0.0001,
+      "loss": 0.0792,
+      "step": 11550
+    },
+    {
+      "epoch": 32.84090909090909,
+      "grad_norm": 1.153468370437622,
+      "learning_rate": 0.0001,
+      "loss": 0.0786,
+      "step": 11560
+    },
+    {
+      "epoch": 32.86931818181818,
+      "grad_norm": 1.1868444681167603,
+      "learning_rate": 0.0001,
+      "loss": 0.0839,
+      "step": 11570
+    },
+    {
+      "epoch": 32.89772727272727,
+      "grad_norm": 1.2777554988861084,
+      "learning_rate": 0.0001,
+      "loss": 0.0803,
+      "step": 11580
+    },
+    {
+      "epoch": 32.92613636363637,
+      "grad_norm": 1.643078088760376,
+      "learning_rate": 0.0001,
+      "loss": 0.0798,
+      "step": 11590
+    },
+    {
+      "epoch": 32.95454545454545,
+      "grad_norm": 1.6660960912704468,
+      "learning_rate": 0.0001,
+      "loss": 0.0804,
+      "step": 11600
+    },
+    {
+      "epoch": 32.98295454545455,
+      "grad_norm": 1.773538589477539,
+      "learning_rate": 0.0001,
+      "loss": 0.0779,
+      "step": 11610
+    },
+    {
+      "epoch": 33.01136363636363,
+      "grad_norm": 1.5937694311141968,
+      "learning_rate": 0.0001,
+      "loss": 0.0801,
+      "step": 11620
+    },
+    {
+      "epoch": 33.03977272727273,
+      "grad_norm": 1.380580186843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0763,
+      "step": 11630
+    },
+    {
+      "epoch": 33.06818181818182,
+      "grad_norm": 1.3506697416305542,
+      "learning_rate": 0.0001,
+      "loss": 0.0765,
+      "step": 11640
+    },
+    {
+      "epoch": 33.09659090909091,
+      "grad_norm": 1.3326584100723267,
+      "learning_rate": 0.0001,
+      "loss": 0.0845,
+      "step": 11650
+    },
+    {
+      "epoch": 33.125,
+      "grad_norm": 1.3809016942977905,
+      "learning_rate": 0.0001,
+      "loss": 0.0786,
+      "step": 11660
+    },
+    {
+      "epoch": 33.15340909090909,
+      "grad_norm": 1.5046674013137817,
+      "learning_rate": 0.0001,
+      "loss": 0.0781,
+      "step": 11670
+    },
+    {
+      "epoch": 33.18181818181818,
+      "grad_norm": 1.373900294303894,
+      "learning_rate": 0.0001,
+      "loss": 0.0794,
+      "step": 11680
+    },
+    {
+      "epoch": 33.21022727272727,
+      "grad_norm": 1.6719146966934204,
+      "learning_rate": 0.0001,
+      "loss": 0.0828,
+      "step": 11690
+    },
+    {
+      "epoch": 33.23863636363637,
+      "grad_norm": 1.2766826152801514,
+      "learning_rate": 0.0001,
+      "loss": 0.0841,
+      "step": 11700
+    },
+    {
+      "epoch": 33.26704545454545,
+      "grad_norm": 1.2881532907485962,
+      "learning_rate": 0.0001,
+      "loss": 0.0781,
+      "step": 11710
+    },
+    {
+      "epoch": 33.29545454545455,
+      "grad_norm": 1.5469038486480713,
+      "learning_rate": 0.0001,
+      "loss": 0.0814,
+      "step": 11720
+    },
+    {
+      "epoch": 33.32386363636363,
+      "grad_norm": 1.404578685760498,
+      "learning_rate": 0.0001,
+      "loss": 0.0818,
+      "step": 11730
+    },
+    {
+      "epoch": 33.35227272727273,
+      "grad_norm": 1.504791021347046,
+      "learning_rate": 0.0001,
+      "loss": 0.0773,
+      "step": 11740
+    },
+    {
+      "epoch": 33.38068181818182,
+      "grad_norm": 1.8719531297683716,
+      "learning_rate": 0.0001,
+      "loss": 0.0797,
+      "step": 11750
+    },
+    {
+      "epoch": 33.40909090909091,
+      "grad_norm": 1.418943166732788,
+      "learning_rate": 0.0001,
+      "loss": 0.0768,
+      "step": 11760
+    },
+    {
+      "epoch": 33.4375,
+      "grad_norm": 1.6536691188812256,
+      "learning_rate": 0.0001,
+      "loss": 0.0772,
+      "step": 11770
+    },
+    {
+      "epoch": 33.46590909090909,
+      "grad_norm": 1.4639066457748413,
+      "learning_rate": 0.0001,
+      "loss": 0.076,
+      "step": 11780
+    },
+    {
+      "epoch": 33.49431818181818,
+      "grad_norm": 1.4900221824645996,
+      "learning_rate": 0.0001,
+      "loss": 0.0816,
+      "step": 11790
+    },
+    {
+      "epoch": 33.52272727272727,
+      "grad_norm": 1.6789690256118774,
+      "learning_rate": 0.0001,
+      "loss": 0.0807,
+      "step": 11800
+    },
+    {
+      "epoch": 33.55113636363637,
+      "grad_norm": 1.3068170547485352,
+      "learning_rate": 0.0001,
+      "loss": 0.083,
+      "step": 11810
+    },
+    {
+      "epoch": 33.57954545454545,
+      "grad_norm": 1.1812000274658203,
+      "learning_rate": 0.0001,
+      "loss": 0.0844,
+      "step": 11820
+    },
+    {
+      "epoch": 33.60795454545455,
+      "grad_norm": 1.303970456123352,
+      "learning_rate": 0.0001,
+      "loss": 0.077,
+      "step": 11830
+    },
+    {
+      "epoch": 33.63636363636363,
+      "grad_norm": 1.3309065103530884,
+      "learning_rate": 0.0001,
+      "loss": 0.08,
+      "step": 11840
+    },
+    {
+      "epoch": 33.66477272727273,
+      "grad_norm": 1.4080289602279663,
+      "learning_rate": 0.0001,
+      "loss": 0.0801,
+      "step": 11850
+    },
+    {
+      "epoch": 33.69318181818182,
+      "grad_norm": 1.5647964477539062,
+      "learning_rate": 0.0001,
+      "loss": 0.0822,
+      "step": 11860
+    },
+    {
+      "epoch": 33.72159090909091,
+      "grad_norm": 1.4176783561706543,
+      "learning_rate": 0.0001,
+      "loss": 0.0827,
+      "step": 11870
+    },
+    {
+      "epoch": 33.75,
+      "grad_norm": 1.288572072982788,
+      "learning_rate": 0.0001,
+      "loss": 0.079,
+      "step": 11880
+    },
+    {
+      "epoch": 33.77840909090909,
+      "grad_norm": 1.22144615650177,
+      "learning_rate": 0.0001,
+      "loss": 0.0835,
+      "step": 11890
+    },
+    {
+      "epoch": 33.80681818181818,
+      "grad_norm": 1.6118239164352417,
+      "learning_rate": 0.0001,
+      "loss": 0.0823,
+      "step": 11900
+    },
+    {
+      "epoch": 33.83522727272727,
+      "grad_norm": 1.3921666145324707,
+      "learning_rate": 0.0001,
+      "loss": 0.0801,
+      "step": 11910
+    },
+    {
+      "epoch": 33.86363636363637,
+      "grad_norm": 1.2439217567443848,
+      "learning_rate": 0.0001,
+      "loss": 0.081,
+      "step": 11920
+    },
+    {
+      "epoch": 33.89204545454545,
+      "grad_norm": 1.4157015085220337,
+      "learning_rate": 0.0001,
+      "loss": 0.0792,
+      "step": 11930
+    },
+    {
+      "epoch": 33.92045454545455,
+      "grad_norm": 1.2521430253982544,
+      "learning_rate": 0.0001,
+      "loss": 0.0815,
+      "step": 11940
+    },
+    {
+      "epoch": 33.94886363636363,
+      "grad_norm": 1.3754600286483765,
+      "learning_rate": 0.0001,
+      "loss": 0.0816,
+      "step": 11950
+    },
+    {
+      "epoch": 33.97727272727273,
+      "grad_norm": 1.2032493352890015,
+      "learning_rate": 0.0001,
+      "loss": 0.0788,
+      "step": 11960
+    },
+    {
+      "epoch": 34.00568181818182,
+      "grad_norm": 1.2101504802703857,
+      "learning_rate": 0.0001,
+      "loss": 0.0782,
+      "step": 11970
+    },
+    {
+      "epoch": 34.03409090909091,
+      "grad_norm": 1.4837169647216797,
+      "learning_rate": 0.0001,
+      "loss": 0.0785,
+      "step": 11980
+    },
+    {
+      "epoch": 34.0625,
+      "grad_norm": 1.3385684490203857,
+      "learning_rate": 0.0001,
+      "loss": 0.079,
+      "step": 11990
+    },
+    {
+      "epoch": 34.09090909090909,
+      "grad_norm": 1.264683723449707,
+      "learning_rate": 0.0001,
+      "loss": 0.0773,
+      "step": 12000
+    },
+    {
+      "epoch": 34.11931818181818,
+      "grad_norm": 1.3990920782089233,
+      "learning_rate": 0.0001,
+      "loss": 0.076,
+      "step": 12010
+    },
+    {
+      "epoch": 34.14772727272727,
+      "grad_norm": 1.4375253915786743,
+      "learning_rate": 0.0001,
+      "loss": 0.0832,
+      "step": 12020
+    },
+    {
+      "epoch": 34.17613636363637,
+      "grad_norm": 1.555679202079773,
+      "learning_rate": 0.0001,
+      "loss": 0.0784,
+      "step": 12030
+    },
+    {
+      "epoch": 34.20454545454545,
+      "grad_norm": 1.1317201852798462,
+      "learning_rate": 0.0001,
+      "loss": 0.0825,
+      "step": 12040
+    },
+    {
+      "epoch": 34.23295454545455,
+      "grad_norm": 1.1169489622116089,
+      "learning_rate": 0.0001,
+      "loss": 0.0768,
+      "step": 12050
+    },
+    {
+      "epoch": 34.26136363636363,
+      "grad_norm": 1.3788570165634155,
+      "learning_rate": 0.0001,
+      "loss": 0.0773,
+      "step": 12060
+    },
+    {
+      "epoch": 34.28977272727273,
+      "grad_norm": 1.4683411121368408,
+      "learning_rate": 0.0001,
+      "loss": 0.0808,
+      "step": 12070
+    },
+    {
+      "epoch": 34.31818181818182,
+      "grad_norm": 1.2734488248825073,
+      "learning_rate": 0.0001,
+      "loss": 0.0799,
+      "step": 12080
+    },
+    {
+      "epoch": 34.34659090909091,
+      "grad_norm": 1.175338625907898,
+      "learning_rate": 0.0001,
+      "loss": 0.086,
+      "step": 12090
+    },
+    {
+      "epoch": 34.375,
+      "grad_norm": 1.2136415243148804,
+      "learning_rate": 0.0001,
+      "loss": 0.0827,
+      "step": 12100
+    },
+    {
+      "epoch": 34.40340909090909,
+      "grad_norm": 1.1225967407226562,
+      "learning_rate": 0.0001,
+      "loss": 0.0801,
+      "step": 12110
+    },
+    {
+      "epoch": 34.43181818181818,
+      "grad_norm": 1.2012512683868408,
+      "learning_rate": 0.0001,
+      "loss": 0.0796,
+      "step": 12120
+    },
+    {
+      "epoch": 34.46022727272727,
+      "grad_norm": 1.3884018659591675,
+      "learning_rate": 0.0001,
+      "loss": 0.0818,
+      "step": 12130
+    },
+    {
+      "epoch": 34.48863636363637,
+      "grad_norm": 1.3851127624511719,
+      "learning_rate": 0.0001,
+      "loss": 0.0788,
+      "step": 12140
+    },
+    {
+      "epoch": 34.51704545454545,
+      "grad_norm": 1.442125678062439,
+      "learning_rate": 0.0001,
+      "loss": 0.078,
+      "step": 12150
+    },
+    {
+      "epoch": 34.54545454545455,
+      "grad_norm": 1.3751837015151978,
+      "learning_rate": 0.0001,
+      "loss": 0.0795,
+      "step": 12160
+    },
+    {
+      "epoch": 34.57386363636363,
+      "grad_norm": 1.5753248929977417,
+      "learning_rate": 0.0001,
+      "loss": 0.078,
+      "step": 12170
+    },
+    {
+      "epoch": 34.60227272727273,
+      "grad_norm": 1.5038411617279053,
+      "learning_rate": 0.0001,
+      "loss": 0.0804,
+      "step": 12180
+    },
+    {
+      "epoch": 34.63068181818182,
+      "grad_norm": 1.1918764114379883,
+      "learning_rate": 0.0001,
+      "loss": 0.0789,
+      "step": 12190
+    },
+    {
+      "epoch": 34.65909090909091,
+      "grad_norm": 1.28202223777771,
+      "learning_rate": 0.0001,
+      "loss": 0.0805,
+      "step": 12200
+    },
+    {
+      "epoch": 34.6875,
+      "grad_norm": 1.1731418371200562,
+      "learning_rate": 0.0001,
+      "loss": 0.0844,
+      "step": 12210
+    },
+    {
+      "epoch": 34.71590909090909,
+      "grad_norm": 1.4400376081466675,
+      "learning_rate": 0.0001,
+      "loss": 0.0793,
+      "step": 12220
+    },
+    {
+      "epoch": 34.74431818181818,
+      "grad_norm": 1.279740810394287,
+      "learning_rate": 0.0001,
+      "loss": 0.0837,
+      "step": 12230
+    },
+    {
+      "epoch": 34.77272727272727,
+      "grad_norm": 1.084643840789795,
+      "learning_rate": 0.0001,
+      "loss": 0.0785,
+      "step": 12240
+    },
+    {
+      "epoch": 34.80113636363637,
+      "grad_norm": 0.9900233745574951,
+      "learning_rate": 0.0001,
+      "loss": 0.0761,
+      "step": 12250
+    },
+    {
+      "epoch": 34.82954545454545,
+      "grad_norm": 1.505629539489746,
+      "learning_rate": 0.0001,
+      "loss": 0.0811,
+      "step": 12260
+    },
+    {
+      "epoch": 34.85795454545455,
+      "grad_norm": 1.2708748579025269,
+      "learning_rate": 0.0001,
+      "loss": 0.0805,
+      "step": 12270
+    },
+    {
+      "epoch": 34.88636363636363,
+      "grad_norm": 1.3224505186080933,
+      "learning_rate": 0.0001,
+      "loss": 0.0797,
+      "step": 12280
+    },
+    {
+      "epoch": 34.91477272727273,
+      "grad_norm": 1.1570143699645996,
+      "learning_rate": 0.0001,
+      "loss": 0.08,
+      "step": 12290
+    },
+    {
+      "epoch": 34.94318181818182,
+      "grad_norm": 1.260339379310608,
+      "learning_rate": 0.0001,
+      "loss": 0.0788,
+      "step": 12300
+    },
+    {
+      "epoch": 34.97159090909091,
+      "grad_norm": 1.1624943017959595,
+      "learning_rate": 0.0001,
+      "loss": 0.0794,
+      "step": 12310
+    },
+    {
+      "epoch": 35.0,
+      "grad_norm": 1.2895053625106812,
+      "learning_rate": 0.0001,
+      "loss": 0.0772,
+      "step": 12320
+    },
+    {
+      "epoch": 35.02840909090909,
+      "grad_norm": 1.2539187669754028,
+      "learning_rate": 0.0001,
+      "loss": 0.0751,
+      "step": 12330
+    },
+    {
+      "epoch": 35.05681818181818,
+      "grad_norm": 1.303320288658142,
+      "learning_rate": 0.0001,
+      "loss": 0.0853,
+      "step": 12340
+    },
+    {
+      "epoch": 35.08522727272727,
+      "grad_norm": 1.1098580360412598,
+      "learning_rate": 0.0001,
+      "loss": 0.082,
+      "step": 12350
+    },
+    {
+      "epoch": 35.11363636363637,
+      "grad_norm": 1.5412824153900146,
+      "learning_rate": 0.0001,
+      "loss": 0.0832,
+      "step": 12360
+    },
+    {
+      "epoch": 35.14204545454545,
+      "grad_norm": 1.3514188528060913,
+      "learning_rate": 0.0001,
+      "loss": 0.0761,
+      "step": 12370
+    },
+    {
+      "epoch": 35.17045454545455,
+      "grad_norm": 1.380109190940857,
+      "learning_rate": 0.0001,
+      "loss": 0.0747,
+      "step": 12380
+    },
+    {
+      "epoch": 35.19886363636363,
+      "grad_norm": 1.1679573059082031,
+      "learning_rate": 0.0001,
+      "loss": 0.0797,
+      "step": 12390
+    },
+    {
+      "epoch": 35.22727272727273,
+      "grad_norm": 1.2729599475860596,
+      "learning_rate": 0.0001,
+      "loss": 0.0795,
+      "step": 12400
+    },
+    {
+      "epoch": 35.25568181818182,
+      "grad_norm": 1.273162841796875,
+      "learning_rate": 0.0001,
+      "loss": 0.0801,
+      "step": 12410
+    },
+    {
+      "epoch": 35.28409090909091,
+      "grad_norm": 1.4882563352584839,
+      "learning_rate": 0.0001,
+      "loss": 0.0781,
+      "step": 12420
+    },
+    {
+      "epoch": 35.3125,
+      "grad_norm": 1.2322689294815063,
+      "learning_rate": 0.0001,
+      "loss": 0.082,
+      "step": 12430
+    },
+    {
+      "epoch": 35.34090909090909,
+      "grad_norm": 1.324062705039978,
+      "learning_rate": 0.0001,
+      "loss": 0.0762,
+      "step": 12440
+    },
+    {
+      "epoch": 35.36931818181818,
+      "grad_norm": 1.236574411392212,
+      "learning_rate": 0.0001,
+      "loss": 0.0766,
+      "step": 12450
+    },
+    {
+      "epoch": 35.39772727272727,
+      "grad_norm": 1.17245614528656,
+      "learning_rate": 0.0001,
+      "loss": 0.0785,
+      "step": 12460
+    },
+    {
+      "epoch": 35.42613636363637,
+      "grad_norm": 1.337213397026062,
+      "learning_rate": 0.0001,
+      "loss": 0.0811,
+      "step": 12470
+    },
+    {
+      "epoch": 35.45454545454545,
+      "grad_norm": 1.299586534500122,
+      "learning_rate": 0.0001,
+      "loss": 0.0821,
+      "step": 12480
+    },
+    {
+      "epoch": 35.48295454545455,
+      "grad_norm": 1.2936147451400757,
+      "learning_rate": 0.0001,
+      "loss": 0.0799,
+      "step": 12490
+    },
+    {
+      "epoch": 35.51136363636363,
+      "grad_norm": 1.203636646270752,
+      "learning_rate": 0.0001,
+      "loss": 0.0826,
+      "step": 12500
+    },
+    {
+      "epoch": 35.53977272727273,
+      "grad_norm": 1.260736346244812,
+      "learning_rate": 0.0001,
+      "loss": 0.0754,
+      "step": 12510
+    },
+    {
+      "epoch": 35.56818181818182,
+      "grad_norm": 1.2215540409088135,
+      "learning_rate": 0.0001,
+      "loss": 0.0806,
+      "step": 12520
+    },
+    {
+      "epoch": 35.59659090909091,
+      "grad_norm": 1.176795244216919,
+      "learning_rate": 0.0001,
+      "loss": 0.0741,
+      "step": 12530
+    },
+    {
+      "epoch": 35.625,
+      "grad_norm": 1.3341177701950073,
+      "learning_rate": 0.0001,
+      "loss": 0.0804,
+      "step": 12540
+    },
+    {
+      "epoch": 35.65340909090909,
+      "grad_norm": 1.2871227264404297,
+      "learning_rate": 0.0001,
+      "loss": 0.0839,
+      "step": 12550
+    },
+    {
+      "epoch": 35.68181818181818,
+      "grad_norm": 1.2276510000228882,
+      "learning_rate": 0.0001,
+      "loss": 0.0824,
+      "step": 12560
+    },
+    {
+      "epoch": 35.71022727272727,
+      "grad_norm": 1.1885565519332886,
+      "learning_rate": 0.0001,
+      "loss": 0.0794,
+      "step": 12570
+    },
+    {
+      "epoch": 35.73863636363637,
+      "grad_norm": 1.3643691539764404,
+      "learning_rate": 0.0001,
+      "loss": 0.0829,
+      "step": 12580
+    },
+    {
+      "epoch": 35.76704545454545,
+      "grad_norm": 1.1791919469833374,
+      "learning_rate": 0.0001,
+      "loss": 0.0798,
+      "step": 12590
+    },
+    {
+      "epoch": 35.79545454545455,
+      "grad_norm": 1.14838707447052,
+      "learning_rate": 0.0001,
+      "loss": 0.077,
+      "step": 12600
+    },
+    {
+      "epoch": 35.82386363636363,
+      "grad_norm": 1.418837308883667,
+      "learning_rate": 0.0001,
+      "loss": 0.0773,
+      "step": 12610
+    },
+    {
+      "epoch": 35.85227272727273,
+      "grad_norm": 1.3161529302597046,
+      "learning_rate": 0.0001,
+      "loss": 0.0827,
+      "step": 12620
+    },
+    {
+      "epoch": 35.88068181818182,
+      "grad_norm": 1.7161197662353516,
+      "learning_rate": 0.0001,
+      "loss": 0.0823,
+      "step": 12630
+    },
+    {
+      "epoch": 35.90909090909091,
+      "grad_norm": 1.4447532892227173,
+      "learning_rate": 0.0001,
+      "loss": 0.0795,
+      "step": 12640
+    },
+    {
+      "epoch": 35.9375,
+      "grad_norm": 1.743769884109497,
+      "learning_rate": 0.0001,
+      "loss": 0.0742,
+      "step": 12650
+    },
+    {
+      "epoch": 35.96590909090909,
+      "grad_norm": 1.4263979196548462,
+      "learning_rate": 0.0001,
+      "loss": 0.0773,
+      "step": 12660
+    },
+    {
+      "epoch": 35.99431818181818,
+      "grad_norm": 1.431891679763794,
+      "learning_rate": 0.0001,
+      "loss": 0.0765,
+      "step": 12670
+    },
+    {
+      "epoch": 36.02272727272727,
+      "grad_norm": 1.2072255611419678,
+      "learning_rate": 0.0001,
+      "loss": 0.0764,
+      "step": 12680
+    },
+    {
+      "epoch": 36.05113636363637,
+      "grad_norm": 1.2662785053253174,
+      "learning_rate": 0.0001,
+      "loss": 0.0763,
+      "step": 12690
+    },
+    {
+      "epoch": 36.07954545454545,
+      "grad_norm": 1.184961199760437,
+      "learning_rate": 0.0001,
+      "loss": 0.0696,
+      "step": 12700
+    },
+    {
+      "epoch": 36.10795454545455,
+      "grad_norm": 1.1873055696487427,
+      "learning_rate": 0.0001,
+      "loss": 0.0758,
+      "step": 12710
+    },
+    {
+      "epoch": 36.13636363636363,
+      "grad_norm": 1.1388927698135376,
+      "learning_rate": 0.0001,
+      "loss": 0.0731,
+      "step": 12720
+    },
+    {
+      "epoch": 36.16477272727273,
+      "grad_norm": 1.415749430656433,
+      "learning_rate": 0.0001,
+      "loss": 0.0774,
+      "step": 12730
+    },
+    {
+      "epoch": 36.19318181818182,
+      "grad_norm": 1.6537916660308838,
+      "learning_rate": 0.0001,
+      "loss": 0.0806,
+      "step": 12740
+    },
+    {
+      "epoch": 36.22159090909091,
+      "grad_norm": 1.479551076889038,
+      "learning_rate": 0.0001,
+      "loss": 0.0787,
+      "step": 12750
+    },
+    {
+      "epoch": 36.25,
+      "grad_norm": 1.4956623315811157,
+      "learning_rate": 0.0001,
+      "loss": 0.0746,
+      "step": 12760
+    },
+    {
+      "epoch": 36.27840909090909,
+      "grad_norm": 1.4210376739501953,
+      "learning_rate": 0.0001,
+      "loss": 0.0791,
+      "step": 12770
+    },
+    {
+      "epoch": 36.30681818181818,
+      "grad_norm": 1.4137691259384155,
+      "learning_rate": 0.0001,
+      "loss": 0.0746,
+      "step": 12780
+    },
+    {
+      "epoch": 36.33522727272727,
+      "grad_norm": 1.5666520595550537,
+      "learning_rate": 0.0001,
+      "loss": 0.0795,
+      "step": 12790
+    },
+    {
+      "epoch": 36.36363636363637,
+      "grad_norm": 2.1862380504608154,
+      "learning_rate": 0.0001,
+      "loss": 0.0799,
+      "step": 12800
+    },
+    {
+      "epoch": 36.39204545454545,
+      "grad_norm": 2.0382354259490967,
+      "learning_rate": 0.0001,
+      "loss": 0.076,
+      "step": 12810
+    },
+    {
+      "epoch": 36.42045454545455,
+      "grad_norm": 1.9271392822265625,
+      "learning_rate": 0.0001,
+      "loss": 0.0719,
+      "step": 12820
+    },
+    {
+      "epoch": 36.44886363636363,
+      "grad_norm": 1.600040078163147,
+      "learning_rate": 0.0001,
+      "loss": 0.0778,
+      "step": 12830
+    },
+    {
+      "epoch": 36.47727272727273,
+      "grad_norm": 1.6347086429595947,
+      "learning_rate": 0.0001,
+      "loss": 0.071,
+      "step": 12840
+    },
+    {
+      "epoch": 36.50568181818182,
+      "grad_norm": 1.4493471384048462,
+      "learning_rate": 0.0001,
+      "loss": 0.0753,
+      "step": 12850
+    },
+    {
+      "epoch": 36.53409090909091,
+      "grad_norm": 1.5854036808013916,
+      "learning_rate": 0.0001,
+      "loss": 0.0737,
+      "step": 12860
+    },
+    {
+      "epoch": 36.5625,
+      "grad_norm": 1.3838077783584595,
+      "learning_rate": 0.0001,
+      "loss": 0.0744,
+      "step": 12870
+    },
+    {
+      "epoch": 36.59090909090909,
+      "grad_norm": 1.5723717212677002,
+      "learning_rate": 0.0001,
+      "loss": 0.0716,
+      "step": 12880
+    },
+    {
+      "epoch": 36.61931818181818,
+      "grad_norm": 1.6842889785766602,
+      "learning_rate": 0.0001,
+      "loss": 0.0732,
+      "step": 12890
+    },
+    {
+      "epoch": 36.64772727272727,
+      "grad_norm": 1.209652066230774,
+      "learning_rate": 0.0001,
+      "loss": 0.0741,
+      "step": 12900
+    },
+    {
+      "epoch": 36.67613636363637,
+      "grad_norm": 1.4279730319976807,
+      "learning_rate": 0.0001,
+      "loss": 0.0762,
+      "step": 12910
+    },
+    {
+      "epoch": 36.70454545454545,
+      "grad_norm": 1.6031663417816162,
+      "learning_rate": 0.0001,
+      "loss": 0.0701,
+      "step": 12920
+    },
+    {
+      "epoch": 36.73295454545455,
+      "grad_norm": 1.3902431726455688,
+      "learning_rate": 0.0001,
+      "loss": 0.0785,
+      "step": 12930
+    },
+    {
+      "epoch": 36.76136363636363,
+      "grad_norm": 1.3266302347183228,
+      "learning_rate": 0.0001,
+      "loss": 0.0758,
+      "step": 12940
+    },
+    {
+      "epoch": 36.78977272727273,
+      "grad_norm": 1.4028958082199097,
+      "learning_rate": 0.0001,
+      "loss": 0.0741,
+      "step": 12950
+    },
+    {
+      "epoch": 36.81818181818182,
+      "grad_norm": 1.494147539138794,
+      "learning_rate": 0.0001,
+      "loss": 0.0716,
+      "step": 12960
+    },
+    {
+      "epoch": 36.84659090909091,
+      "grad_norm": 1.4022266864776611,
+      "learning_rate": 0.0001,
+      "loss": 0.0751,
+      "step": 12970
+    },
+    {
+      "epoch": 36.875,
+      "grad_norm": 1.1766167879104614,
+      "learning_rate": 0.0001,
+      "loss": 0.0718,
+      "step": 12980
+    },
+    {
+      "epoch": 36.90340909090909,
+      "grad_norm": 1.346309781074524,
+      "learning_rate": 0.0001,
+      "loss": 0.0773,
+      "step": 12990
+    },
+    {
+      "epoch": 36.93181818181818,
+      "grad_norm": 1.2325224876403809,
+      "learning_rate": 0.0001,
+      "loss": 0.0771,
+      "step": 13000
+    },
+    {
+      "epoch": 36.96022727272727,
+      "grad_norm": 1.4607832431793213,
+      "learning_rate": 0.0001,
+      "loss": 0.0731,
+      "step": 13010
+    },
+    {
+      "epoch": 36.98863636363637,
+      "grad_norm": 1.426300287246704,
+      "learning_rate": 0.0001,
+      "loss": 0.0744,
+      "step": 13020
+    },
+    {
+      "epoch": 37.01704545454545,
+      "grad_norm": 1.3669778108596802,
+      "learning_rate": 0.0001,
+      "loss": 0.0717,
+      "step": 13030
+    },
+    {
+      "epoch": 37.04545454545455,
+      "grad_norm": 1.3466182947158813,
+      "learning_rate": 0.0001,
+      "loss": 0.0708,
+      "step": 13040
+    },
+    {
+      "epoch": 37.07386363636363,
+      "grad_norm": 1.3738116025924683,
+      "learning_rate": 0.0001,
+      "loss": 0.0792,
+      "step": 13050
+    },
+    {
+      "epoch": 37.10227272727273,
+      "grad_norm": 1.414943814277649,
+      "learning_rate": 0.0001,
+      "loss": 0.0772,
+      "step": 13060
+    },
+    {
+      "epoch": 37.13068181818182,
+      "grad_norm": 1.4304829835891724,
+      "learning_rate": 0.0001,
+      "loss": 0.0728,
+      "step": 13070
+    },
+    {
+      "epoch": 37.15909090909091,
+      "grad_norm": 1.280889630317688,
+      "learning_rate": 0.0001,
+      "loss": 0.076,
+      "step": 13080
+    },
+    {
+      "epoch": 37.1875,
+      "grad_norm": 1.4901535511016846,
+      "learning_rate": 0.0001,
+      "loss": 0.0743,
+      "step": 13090
+    },
+    {
+      "epoch": 37.21590909090909,
+      "grad_norm": 1.2257989645004272,
+      "learning_rate": 0.0001,
+      "loss": 0.0739,
+      "step": 13100
+    },
+    {
+      "epoch": 37.24431818181818,
+      "grad_norm": 1.345968246459961,
+      "learning_rate": 0.0001,
+      "loss": 0.0746,
+      "step": 13110
+    },
+    {
+      "epoch": 37.27272727272727,
+      "grad_norm": 1.5769586563110352,
+      "learning_rate": 0.0001,
+      "loss": 0.0776,
+      "step": 13120
+    },
+    {
+      "epoch": 37.30113636363637,
+      "grad_norm": 1.3803303241729736,
+      "learning_rate": 0.0001,
+      "loss": 0.0747,
+      "step": 13130
+    },
+    {
+      "epoch": 37.32954545454545,
+      "grad_norm": 1.2624835968017578,
+      "learning_rate": 0.0001,
+      "loss": 0.0758,
+      "step": 13140
+    },
+    {
+      "epoch": 37.35795454545455,
+      "grad_norm": 1.2229900360107422,
+      "learning_rate": 0.0001,
+      "loss": 0.0739,
+      "step": 13150
+    },
+    {
+      "epoch": 37.38636363636363,
+      "grad_norm": 1.0312384366989136,
+      "learning_rate": 0.0001,
+      "loss": 0.0748,
+      "step": 13160
+    },
+    {
+      "epoch": 37.41477272727273,
+      "grad_norm": 1.174302339553833,
+      "learning_rate": 0.0001,
+      "loss": 0.0771,
+      "step": 13170
+    },
+    {
+      "epoch": 37.44318181818182,
+      "grad_norm": 1.4073001146316528,
+      "learning_rate": 0.0001,
+      "loss": 0.0804,
+      "step": 13180
+    },
+    {
+      "epoch": 37.47159090909091,
+      "grad_norm": 1.2546253204345703,
+      "learning_rate": 0.0001,
+      "loss": 0.0738,
+      "step": 13190
+    },
+    {
+      "epoch": 37.5,
+      "grad_norm": 1.0951519012451172,
+      "learning_rate": 0.0001,
+      "loss": 0.0778,
+      "step": 13200
+    },
+    {
+      "epoch": 37.52840909090909,
+      "grad_norm": 1.2326979637145996,
+      "learning_rate": 0.0001,
+      "loss": 0.0758,
+      "step": 13210
+    },
+    {
+      "epoch": 37.55681818181818,
+      "grad_norm": 1.4816584587097168,
+      "learning_rate": 0.0001,
+      "loss": 0.0742,
+      "step": 13220
+    },
+    {
+      "epoch": 37.58522727272727,
+      "grad_norm": 1.2732244729995728,
+      "learning_rate": 0.0001,
+      "loss": 0.0792,
+      "step": 13230
+    },
+    {
+      "epoch": 37.61363636363637,
+      "grad_norm": 1.2015769481658936,
+      "learning_rate": 0.0001,
+      "loss": 0.0774,
+      "step": 13240
+    },
+    {
+      "epoch": 37.64204545454545,
+      "grad_norm": 1.4944400787353516,
+      "learning_rate": 0.0001,
+      "loss": 0.0764,
+      "step": 13250
+    },
+    {
+      "epoch": 37.67045454545455,
+      "grad_norm": 1.596579670906067,
+      "learning_rate": 0.0001,
+      "loss": 0.0765,
+      "step": 13260
+    },
+    {
+      "epoch": 37.69886363636363,
+      "grad_norm": 1.1385072469711304,
+      "learning_rate": 0.0001,
+      "loss": 0.0729,
+      "step": 13270
+    },
+    {
+      "epoch": 37.72727272727273,
+      "grad_norm": 1.2156968116760254,
+      "learning_rate": 0.0001,
+      "loss": 0.0782,
+      "step": 13280
+    },
+    {
+      "epoch": 37.75568181818182,
+      "grad_norm": 1.4938963651657104,
+      "learning_rate": 0.0001,
+      "loss": 0.0718,
+      "step": 13290
+    },
+    {
+      "epoch": 37.78409090909091,
+      "grad_norm": 1.3732664585113525,
+      "learning_rate": 0.0001,
+      "loss": 0.0763,
+      "step": 13300
+    },
+    {
+      "epoch": 37.8125,
+      "grad_norm": 1.195371389389038,
+      "learning_rate": 0.0001,
+      "loss": 0.0767,
+      "step": 13310
+    },
+    {
+      "epoch": 37.84090909090909,
+      "grad_norm": 1.272255301475525,
+      "learning_rate": 0.0001,
+      "loss": 0.074,
+      "step": 13320
+    },
+    {
+      "epoch": 37.86931818181818,
+      "grad_norm": 1.451314926147461,
+      "learning_rate": 0.0001,
+      "loss": 0.0764,
+      "step": 13330
+    },
+    {
+      "epoch": 37.89772727272727,
+      "grad_norm": 1.2322098016738892,
+      "learning_rate": 0.0001,
+      "loss": 0.0804,
+      "step": 13340
+    },
+    {
+      "epoch": 37.92613636363637,
+      "grad_norm": 1.3791121244430542,
+      "learning_rate": 0.0001,
+      "loss": 0.07,
+      "step": 13350
+    },
+    {
+      "epoch": 37.95454545454545,
+      "grad_norm": 1.0808604955673218,
+      "learning_rate": 0.0001,
+      "loss": 0.073,
+      "step": 13360
+    },
+    {
+      "epoch": 37.98295454545455,
+      "grad_norm": 1.199265480041504,
+      "learning_rate": 0.0001,
+      "loss": 0.0705,
+      "step": 13370
+    },
+    {
+      "epoch": 38.01136363636363,
+      "grad_norm": 1.1241381168365479,
+      "learning_rate": 0.0001,
+      "loss": 0.0733,
+      "step": 13380
+    },
+    {
+      "epoch": 38.03977272727273,
+      "grad_norm": 1.279880404472351,
+      "learning_rate": 0.0001,
+      "loss": 0.0782,
+      "step": 13390
+    },
+    {
+      "epoch": 38.06818181818182,
+      "grad_norm": 1.2983338832855225,
+      "learning_rate": 0.0001,
+      "loss": 0.073,
+      "step": 13400
+    },
+    {
+      "epoch": 38.09659090909091,
+      "grad_norm": 1.2902085781097412,
+      "learning_rate": 0.0001,
+      "loss": 0.0778,
+      "step": 13410
+    },
+    {
+      "epoch": 38.125,
+      "grad_norm": 1.2779277563095093,
+      "learning_rate": 0.0001,
+      "loss": 0.0726,
+      "step": 13420
+    },
+    {
+      "epoch": 38.15340909090909,
+      "grad_norm": 1.4885528087615967,
+      "learning_rate": 0.0001,
+      "loss": 0.0762,
+      "step": 13430
+    },
+    {
+      "epoch": 38.18181818181818,
+      "grad_norm": 1.3437947034835815,
+      "learning_rate": 0.0001,
+      "loss": 0.0772,
+      "step": 13440
+    },
+    {
+      "epoch": 38.21022727272727,
+      "grad_norm": 1.2803330421447754,
+      "learning_rate": 0.0001,
+      "loss": 0.0773,
+      "step": 13450
+    },
+    {
+      "epoch": 38.23863636363637,
+      "grad_norm": 1.2626174688339233,
+      "learning_rate": 0.0001,
+      "loss": 0.0757,
+      "step": 13460
+    },
+    {
+      "epoch": 38.26704545454545,
+      "grad_norm": 1.3316041231155396,
+      "learning_rate": 0.0001,
+      "loss": 0.0755,
+      "step": 13470
+    },
+    {
+      "epoch": 38.29545454545455,
+      "grad_norm": 1.2017104625701904,
+      "learning_rate": 0.0001,
+      "loss": 0.0737,
+      "step": 13480
+    },
+    {
+      "epoch": 38.32386363636363,
+      "grad_norm": 1.158381462097168,
+      "learning_rate": 0.0001,
+      "loss": 0.0785,
+      "step": 13490
+    },
+    {
+      "epoch": 38.35227272727273,
+      "grad_norm": 1.5186028480529785,
+      "learning_rate": 0.0001,
+      "loss": 0.0765,
+      "step": 13500
+    },
+    {
+      "epoch": 38.38068181818182,
+      "grad_norm": 1.3031560182571411,
+      "learning_rate": 0.0001,
+      "loss": 0.0772,
+      "step": 13510
+    },
+    {
+      "epoch": 38.40909090909091,
+      "grad_norm": 1.0976595878601074,
+      "learning_rate": 0.0001,
+      "loss": 0.077,
+      "step": 13520
+    },
+    {
+      "epoch": 38.4375,
+      "grad_norm": 1.3816660642623901,
+      "learning_rate": 0.0001,
+      "loss": 0.0767,
+      "step": 13530
+    },
+    {
+      "epoch": 38.46590909090909,
+      "grad_norm": 1.3313119411468506,
+      "learning_rate": 0.0001,
+      "loss": 0.0759,
+      "step": 13540
+    },
+    {
+      "epoch": 38.49431818181818,
+      "grad_norm": 1.124053716659546,
+      "learning_rate": 0.0001,
+      "loss": 0.0761,
+      "step": 13550
+    },
+    {
+      "epoch": 38.52272727272727,
+      "grad_norm": 1.1821684837341309,
+      "learning_rate": 0.0001,
+      "loss": 0.0777,
+      "step": 13560
+    },
+    {
+      "epoch": 38.55113636363637,
+      "grad_norm": 1.1764349937438965,
+      "learning_rate": 0.0001,
+      "loss": 0.0755,
+      "step": 13570
+    },
+    {
+      "epoch": 38.57954545454545,
+      "grad_norm": 1.2892837524414062,
+      "learning_rate": 0.0001,
+      "loss": 0.0766,
+      "step": 13580
+    },
+    {
+      "epoch": 38.60795454545455,
+      "grad_norm": 1.2852972745895386,
+      "learning_rate": 0.0001,
+      "loss": 0.0749,
+      "step": 13590
+    },
+    {
+      "epoch": 38.63636363636363,
+      "grad_norm": 1.0687012672424316,
+      "learning_rate": 0.0001,
+      "loss": 0.0715,
+      "step": 13600
+    },
+    {
+      "epoch": 38.66477272727273,
+      "grad_norm": 1.088240385055542,
+      "learning_rate": 0.0001,
+      "loss": 0.0722,
+      "step": 13610
+    },
+    {
+      "epoch": 38.69318181818182,
+      "grad_norm": 1.0786799192428589,
+      "learning_rate": 0.0001,
+      "loss": 0.0729,
+      "step": 13620
+    },
+    {
+      "epoch": 38.72159090909091,
+      "grad_norm": 1.2037075757980347,
+      "learning_rate": 0.0001,
+      "loss": 0.0734,
+      "step": 13630
+    },
+    {
+      "epoch": 38.75,
+      "grad_norm": 1.2549363374710083,
+      "learning_rate": 0.0001,
+      "loss": 0.0737,
+      "step": 13640
+    },
+    {
+      "epoch": 38.77840909090909,
+      "grad_norm": 1.194278597831726,
+      "learning_rate": 0.0001,
+      "loss": 0.0733,
+      "step": 13650
+    },
+    {
+      "epoch": 38.80681818181818,
+      "grad_norm": 1.5739684104919434,
+      "learning_rate": 0.0001,
+      "loss": 0.0714,
+      "step": 13660
+    },
+    {
+      "epoch": 38.83522727272727,
+      "grad_norm": 1.3770755529403687,
+      "learning_rate": 0.0001,
+      "loss": 0.0728,
+      "step": 13670
+    },
+    {
+      "epoch": 38.86363636363637,
+      "grad_norm": 1.294053316116333,
+      "learning_rate": 0.0001,
+      "loss": 0.0678,
+      "step": 13680
+    },
+    {
+      "epoch": 38.89204545454545,
+      "grad_norm": 1.2223044633865356,
+      "learning_rate": 0.0001,
+      "loss": 0.0733,
+      "step": 13690
+    },
+    {
+      "epoch": 38.92045454545455,
+      "grad_norm": 1.3834986686706543,
+      "learning_rate": 0.0001,
+      "loss": 0.0716,
+      "step": 13700
+    },
+    {
+      "epoch": 38.94886363636363,
+      "grad_norm": 1.2326815128326416,
+      "learning_rate": 0.0001,
+      "loss": 0.0719,
+      "step": 13710
+    },
+    {
+      "epoch": 38.97727272727273,
+      "grad_norm": 1.0974726676940918,
+      "learning_rate": 0.0001,
+      "loss": 0.0731,
+      "step": 13720
+    },
+    {
+      "epoch": 39.00568181818182,
+      "grad_norm": 1.2831435203552246,
+      "learning_rate": 0.0001,
+      "loss": 0.0749,
+      "step": 13730
+    },
+    {
+      "epoch": 39.03409090909091,
+      "grad_norm": 1.2879631519317627,
+      "learning_rate": 0.0001,
+      "loss": 0.0756,
+      "step": 13740
+    },
+    {
+      "epoch": 39.0625,
+      "grad_norm": 1.3753108978271484,
+      "learning_rate": 0.0001,
+      "loss": 0.0723,
+      "step": 13750
+    },
+    {
+      "epoch": 39.09090909090909,
+      "grad_norm": 1.29164719581604,
+      "learning_rate": 0.0001,
+      "loss": 0.0722,
+      "step": 13760
+    },
+    {
+      "epoch": 39.11931818181818,
+      "grad_norm": 1.0802414417266846,
+      "learning_rate": 0.0001,
+      "loss": 0.0749,
+      "step": 13770
+    },
+    {
+      "epoch": 39.14772727272727,
+      "grad_norm": 1.1304676532745361,
+      "learning_rate": 0.0001,
+      "loss": 0.0735,
+      "step": 13780
+    },
+    {
+      "epoch": 39.17613636363637,
+      "grad_norm": 1.0044450759887695,
+      "learning_rate": 0.0001,
+      "loss": 0.0731,
+      "step": 13790
+    },
+    {
+      "epoch": 39.20454545454545,
+      "grad_norm": 0.9919958114624023,
+      "learning_rate": 0.0001,
+      "loss": 0.0731,
+      "step": 13800
+    },
+    {
+      "epoch": 39.23295454545455,
+      "grad_norm": 1.1686891317367554,
+      "learning_rate": 0.0001,
+      "loss": 0.0733,
+      "step": 13810
+    },
+    {
+      "epoch": 39.26136363636363,
+      "grad_norm": 1.145544409751892,
+      "learning_rate": 0.0001,
+      "loss": 0.0703,
+      "step": 13820
+    },
+    {
+      "epoch": 39.28977272727273,
+      "grad_norm": 1.1628680229187012,
+      "learning_rate": 0.0001,
+      "loss": 0.0739,
+      "step": 13830
+    },
+    {
+      "epoch": 39.31818181818182,
+      "grad_norm": 1.199570655822754,
+      "learning_rate": 0.0001,
+      "loss": 0.0734,
+      "step": 13840
+    },
+    {
+      "epoch": 39.34659090909091,
+      "grad_norm": 1.2358002662658691,
+      "learning_rate": 0.0001,
+      "loss": 0.0733,
+      "step": 13850
+    },
+    {
+      "epoch": 39.375,
+      "grad_norm": 1.3280445337295532,
+      "learning_rate": 0.0001,
+      "loss": 0.0711,
+      "step": 13860
+    },
+    {
+      "epoch": 39.40340909090909,
+      "grad_norm": 1.103401780128479,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "step": 13870
+    },
+    {
+      "epoch": 39.43181818181818,
+      "grad_norm": 1.4654299020767212,
+      "learning_rate": 0.0001,
+      "loss": 0.0724,
+      "step": 13880
+    },
+    {
+      "epoch": 39.46022727272727,
+      "grad_norm": 1.3928581476211548,
+      "learning_rate": 0.0001,
+      "loss": 0.0728,
+      "step": 13890
+    },
+    {
+      "epoch": 39.48863636363637,
+      "grad_norm": 1.2623183727264404,
+      "learning_rate": 0.0001,
+      "loss": 0.0761,
+      "step": 13900
+    },
+    {
+      "epoch": 39.51704545454545,
+      "grad_norm": 1.0778528451919556,
+      "learning_rate": 0.0001,
+      "loss": 0.0699,
+      "step": 13910
+    },
+    {
+      "epoch": 39.54545454545455,
+      "grad_norm": 1.1330665349960327,
+      "learning_rate": 0.0001,
+      "loss": 0.0777,
+      "step": 13920
+    },
+    {
+      "epoch": 39.57386363636363,
+      "grad_norm": 1.1069144010543823,
+      "learning_rate": 0.0001,
+      "loss": 0.0757,
+      "step": 13930
+    },
+    {
+      "epoch": 39.60227272727273,
+      "grad_norm": 1.1391063928604126,
+      "learning_rate": 0.0001,
+      "loss": 0.0715,
+      "step": 13940
+    },
+    {
+      "epoch": 39.63068181818182,
+      "grad_norm": 1.3045668601989746,
+      "learning_rate": 0.0001,
+      "loss": 0.0744,
+      "step": 13950
+    },
+    {
+      "epoch": 39.65909090909091,
+      "grad_norm": 1.3357367515563965,
+      "learning_rate": 0.0001,
+      "loss": 0.0769,
+      "step": 13960
+    },
+    {
+      "epoch": 39.6875,
+      "grad_norm": 1.3155001401901245,
+      "learning_rate": 0.0001,
+      "loss": 0.0743,
+      "step": 13970
+    },
+    {
+      "epoch": 39.71590909090909,
+      "grad_norm": 1.215240716934204,
+      "learning_rate": 0.0001,
+      "loss": 0.0778,
+      "step": 13980
+    },
+    {
+      "epoch": 39.74431818181818,
+      "grad_norm": 1.0790923833847046,
+      "learning_rate": 0.0001,
+      "loss": 0.0748,
+      "step": 13990
+    },
+    {
+      "epoch": 39.77272727272727,
+      "grad_norm": 1.0089366436004639,
+      "learning_rate": 0.0001,
+      "loss": 0.0738,
+      "step": 14000
+    },
+    {
+      "epoch": 39.80113636363637,
+      "grad_norm": 1.0045474767684937,
+      "learning_rate": 0.0001,
+      "loss": 0.0748,
+      "step": 14010
+    },
+    {
+      "epoch": 39.82954545454545,
+      "grad_norm": 1.0148450136184692,
+      "learning_rate": 0.0001,
+      "loss": 0.0752,
+      "step": 14020
+    },
+    {
+      "epoch": 39.85795454545455,
+      "grad_norm": 1.1197142601013184,
+      "learning_rate": 0.0001,
+      "loss": 0.0762,
+      "step": 14030
+    },
+    {
+      "epoch": 39.88636363636363,
+      "grad_norm": 1.324987530708313,
+      "learning_rate": 0.0001,
+      "loss": 0.075,
+      "step": 14040
+    },
+    {
+      "epoch": 39.91477272727273,
+      "grad_norm": 1.3428856134414673,
+      "learning_rate": 0.0001,
+      "loss": 0.0791,
+      "step": 14050
+    },
+    {
+      "epoch": 39.94318181818182,
+      "grad_norm": 1.025665521621704,
+      "learning_rate": 0.0001,
+      "loss": 0.0761,
+      "step": 14060
+    },
+    {
+      "epoch": 39.97159090909091,
+      "grad_norm": 1.2127065658569336,
+      "learning_rate": 0.0001,
+      "loss": 0.0739,
+      "step": 14070
+    },
+    {
+      "epoch": 40.0,
+      "grad_norm": 1.304028868675232,
+      "learning_rate": 0.0001,
+      "loss": 0.0708,
+      "step": 14080
+    },
+    {
+      "epoch": 40.02840909090909,
+      "grad_norm": 1.194968819618225,
+      "learning_rate": 0.0001,
+      "loss": 0.0732,
+      "step": 14090
+    },
+    {
+      "epoch": 40.05681818181818,
+      "grad_norm": 1.2401859760284424,
+      "learning_rate": 0.0001,
+      "loss": 0.0718,
+      "step": 14100
+    },
+    {
+      "epoch": 40.08522727272727,
+      "grad_norm": 1.2328866720199585,
+      "learning_rate": 0.0001,
+      "loss": 0.0711,
+      "step": 14110
+    },
+    {
+      "epoch": 40.11363636363637,
+      "grad_norm": 1.178868055343628,
+      "learning_rate": 0.0001,
+      "loss": 0.0729,
+      "step": 14120
+    },
+    {
+      "epoch": 40.14204545454545,
+      "grad_norm": 1.2818002700805664,
+      "learning_rate": 0.0001,
+      "loss": 0.0656,
+      "step": 14130
+    },
+    {
+      "epoch": 40.17045454545455,
+      "grad_norm": 1.167518138885498,
+      "learning_rate": 0.0001,
+      "loss": 0.0708,
+      "step": 14140
+    },
+    {
+      "epoch": 40.19886363636363,
+      "grad_norm": 1.28043794631958,
+      "learning_rate": 0.0001,
+      "loss": 0.0727,
+      "step": 14150
+    },
+    {
+      "epoch": 40.22727272727273,
+      "grad_norm": 0.9946736097335815,
+      "learning_rate": 0.0001,
+      "loss": 0.0725,
+      "step": 14160
+    },
+    {
+      "epoch": 40.25568181818182,
+      "grad_norm": 1.186169981956482,
+      "learning_rate": 0.0001,
+      "loss": 0.0716,
+      "step": 14170
+    },
+    {
+      "epoch": 40.28409090909091,
+      "grad_norm": 1.138846755027771,
+      "learning_rate": 0.0001,
+      "loss": 0.0744,
+      "step": 14180
+    },
+    {
+      "epoch": 40.3125,
+      "grad_norm": 1.1993027925491333,
+      "learning_rate": 0.0001,
+      "loss": 0.0741,
+      "step": 14190
+    },
+    {
+      "epoch": 40.34090909090909,
+      "grad_norm": 1.4275126457214355,
+      "learning_rate": 0.0001,
+      "loss": 0.0724,
+      "step": 14200
+    },
+    {
+      "epoch": 40.36931818181818,
+      "grad_norm": 1.4426121711730957,
+      "learning_rate": 0.0001,
+      "loss": 0.0728,
+      "step": 14210
+    },
+    {
+      "epoch": 40.39772727272727,
+      "grad_norm": 1.7519258260726929,
+      "learning_rate": 0.0001,
+      "loss": 0.0723,
+      "step": 14220
+    },
+    {
+      "epoch": 40.42613636363637,
+      "grad_norm": 1.821484923362732,
+      "learning_rate": 0.0001,
+      "loss": 0.0727,
+      "step": 14230
+    },
+    {
+      "epoch": 40.45454545454545,
+      "grad_norm": 1.4972963333129883,
+      "learning_rate": 0.0001,
+      "loss": 0.0699,
+      "step": 14240
+    },
+    {
+      "epoch": 40.48295454545455,
+      "grad_norm": 1.267853021621704,
+      "learning_rate": 0.0001,
+      "loss": 0.0722,
+      "step": 14250
+    },
+    {
+      "epoch": 40.51136363636363,
+      "grad_norm": 1.0927573442459106,
+      "learning_rate": 0.0001,
+      "loss": 0.069,
+      "step": 14260
+    },
+    {
+      "epoch": 40.53977272727273,
+      "grad_norm": 1.4752814769744873,
+      "learning_rate": 0.0001,
+      "loss": 0.0672,
+      "step": 14270
+    },
+    {
+      "epoch": 40.56818181818182,
+      "grad_norm": 1.8464324474334717,
+      "learning_rate": 0.0001,
+      "loss": 0.0738,
+      "step": 14280
+    },
+    {
+      "epoch": 40.59659090909091,
+      "grad_norm": 1.4506888389587402,
+      "learning_rate": 0.0001,
+      "loss": 0.0693,
+      "step": 14290
+    },
+    {
+      "epoch": 40.625,
+      "grad_norm": 1.7628158330917358,
+      "learning_rate": 0.0001,
+      "loss": 0.068,
+      "step": 14300
+    },
+    {
+      "epoch": 40.65340909090909,
+      "grad_norm": 1.4462136030197144,
+      "learning_rate": 0.0001,
+      "loss": 0.0713,
+      "step": 14310
+    },
+    {
+      "epoch": 40.68181818181818,
+      "grad_norm": 1.3682395219802856,
+      "learning_rate": 0.0001,
+      "loss": 0.0688,
+      "step": 14320
+    },
+    {
+      "epoch": 40.71022727272727,
+      "grad_norm": 1.5475159883499146,
+      "learning_rate": 0.0001,
+      "loss": 0.0689,
+      "step": 14330
+    },
+    {
+      "epoch": 40.73863636363637,
+      "grad_norm": 1.2970659732818604,
+      "learning_rate": 0.0001,
+      "loss": 0.071,
+      "step": 14340
+    },
+    {
+      "epoch": 40.76704545454545,
+      "grad_norm": 1.2972766160964966,
+      "learning_rate": 0.0001,
+      "loss": 0.0691,
+      "step": 14350
+    },
+    {
+      "epoch": 40.79545454545455,
+      "grad_norm": 1.5972353219985962,
+      "learning_rate": 0.0001,
+      "loss": 0.0668,
+      "step": 14360
+    },
+    {
+      "epoch": 40.82386363636363,
+      "grad_norm": 1.2204009294509888,
+      "learning_rate": 0.0001,
+      "loss": 0.0688,
+      "step": 14370
+    },
+    {
+      "epoch": 40.85227272727273,
+      "grad_norm": 1.4080297946929932,
+      "learning_rate": 0.0001,
+      "loss": 0.0724,
+      "step": 14380
+    },
+    {
+      "epoch": 40.88068181818182,
+      "grad_norm": 1.5412365198135376,
+      "learning_rate": 0.0001,
+      "loss": 0.0688,
+      "step": 14390
+    },
+    {
+      "epoch": 40.90909090909091,
+      "grad_norm": 1.3308732509613037,
+      "learning_rate": 0.0001,
+      "loss": 0.0703,
+      "step": 14400
+    },
+    {
+      "epoch": 40.9375,
+      "grad_norm": 1.2614033222198486,
+      "learning_rate": 0.0001,
+      "loss": 0.0676,
+      "step": 14410
+    },
+    {
+      "epoch": 40.96590909090909,
+      "grad_norm": 1.3623310327529907,
+      "learning_rate": 0.0001,
+      "loss": 0.0691,
+      "step": 14420
+    },
+    {
+      "epoch": 40.99431818181818,
+      "grad_norm": 1.3249821662902832,
+      "learning_rate": 0.0001,
+      "loss": 0.0709,
+      "step": 14430
+    },
+    {
+      "epoch": 41.02272727272727,
+      "grad_norm": 1.224489450454712,
+      "learning_rate": 0.0001,
+      "loss": 0.0677,
+      "step": 14440
+    },
+    {
+      "epoch": 41.05113636363637,
+      "grad_norm": 1.2260276079177856,
+      "learning_rate": 0.0001,
+      "loss": 0.0687,
+      "step": 14450
+    },
+    {
+      "epoch": 41.07954545454545,
+      "grad_norm": 1.0387647151947021,
+      "learning_rate": 0.0001,
+      "loss": 0.0671,
+      "step": 14460
+    },
+    {
+      "epoch": 41.10795454545455,
+      "grad_norm": 1.3038042783737183,
+      "learning_rate": 0.0001,
+      "loss": 0.0709,
+      "step": 14470
+    },
+    {
+      "epoch": 41.13636363636363,
+      "grad_norm": 1.170926570892334,
+      "learning_rate": 0.0001,
+      "loss": 0.078,
+      "step": 14480
+    },
+    {
+      "epoch": 41.16477272727273,
+      "grad_norm": 1.118465781211853,
+      "learning_rate": 0.0001,
+      "loss": 0.072,
+      "step": 14490
+    },
+    {
+      "epoch": 41.19318181818182,
+      "grad_norm": 1.1143956184387207,
+      "learning_rate": 0.0001,
+      "loss": 0.0716,
+      "step": 14500
+    },
+    {
+      "epoch": 41.22159090909091,
+      "grad_norm": 1.1609805822372437,
+      "learning_rate": 0.0001,
+      "loss": 0.0724,
+      "step": 14510
+    },
+    {
+      "epoch": 41.25,
+      "grad_norm": 1.0995590686798096,
+      "learning_rate": 0.0001,
+      "loss": 0.0727,
+      "step": 14520
+    },
+    {
+      "epoch": 41.27840909090909,
+      "grad_norm": 1.0863865613937378,
+      "learning_rate": 0.0001,
+      "loss": 0.0706,
+      "step": 14530
+    },
+    {
+      "epoch": 41.30681818181818,
+      "grad_norm": 1.0013092756271362,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "step": 14540
+    },
+    {
+      "epoch": 41.33522727272727,
+      "grad_norm": 1.1944328546524048,
+      "learning_rate": 0.0001,
+      "loss": 0.0739,
+      "step": 14550
+    },
+    {
+      "epoch": 41.36363636363637,
+      "grad_norm": 1.015170931816101,
+      "learning_rate": 0.0001,
+      "loss": 0.0699,
+      "step": 14560
+    },
+    {
+      "epoch": 41.39204545454545,
+      "grad_norm": 1.0812183618545532,
+      "learning_rate": 0.0001,
+      "loss": 0.0672,
+      "step": 14570
+    },
+    {
+      "epoch": 41.42045454545455,
+      "grad_norm": 1.2042665481567383,
+      "learning_rate": 0.0001,
+      "loss": 0.0695,
+      "step": 14580
+    },
+    {
+      "epoch": 41.44886363636363,
+      "grad_norm": 1.3115397691726685,
+      "learning_rate": 0.0001,
+      "loss": 0.0704,
+      "step": 14590
+    },
+    {
+      "epoch": 41.47727272727273,
+      "grad_norm": 1.325830101966858,
+      "learning_rate": 0.0001,
+      "loss": 0.0737,
+      "step": 14600
+    },
+    {
+      "epoch": 41.50568181818182,
+      "grad_norm": 1.1970694065093994,
+      "learning_rate": 0.0001,
+      "loss": 0.0721,
+      "step": 14610
+    },
+    {
+      "epoch": 41.53409090909091,
+      "grad_norm": 1.0794974565505981,
+      "learning_rate": 0.0001,
+      "loss": 0.0714,
+      "step": 14620
+    },
+    {
+      "epoch": 41.5625,
+      "grad_norm": 0.8236928582191467,
+      "learning_rate": 0.0001,
+      "loss": 0.0741,
+      "step": 14630
+    },
+    {
+      "epoch": 41.59090909090909,
+      "grad_norm": 0.8976597785949707,
+      "learning_rate": 0.0001,
+      "loss": 0.0713,
+      "step": 14640
+    },
+    {
+      "epoch": 41.61931818181818,
+      "grad_norm": 0.84346604347229,
+      "learning_rate": 0.0001,
+      "loss": 0.071,
+      "step": 14650
+    },
+    {
+      "epoch": 41.64772727272727,
+      "grad_norm": 0.9527440667152405,
+      "learning_rate": 0.0001,
+      "loss": 0.0691,
+      "step": 14660
+    },
+    {
+      "epoch": 41.67613636363637,
+      "grad_norm": 1.0900448560714722,
+      "learning_rate": 0.0001,
+      "loss": 0.0708,
+      "step": 14670
+    },
+    {
+      "epoch": 41.70454545454545,
+      "grad_norm": 1.057309627532959,
+      "learning_rate": 0.0001,
+      "loss": 0.0743,
+      "step": 14680
+    },
+    {
+      "epoch": 41.73295454545455,
+      "grad_norm": 1.161765456199646,
+      "learning_rate": 0.0001,
+      "loss": 0.0693,
+      "step": 14690
+    },
+    {
+      "epoch": 41.76136363636363,
+      "grad_norm": 1.0732311010360718,
+      "learning_rate": 0.0001,
+      "loss": 0.0744,
+      "step": 14700
+    },
+    {
+      "epoch": 41.78977272727273,
+      "grad_norm": 0.9801560044288635,
+      "learning_rate": 0.0001,
+      "loss": 0.0713,
+      "step": 14710
+    },
+    {
+      "epoch": 41.81818181818182,
+      "grad_norm": 0.952893853187561,
+      "learning_rate": 0.0001,
+      "loss": 0.0705,
+      "step": 14720
+    },
+    {
+      "epoch": 41.84659090909091,
+      "grad_norm": 1.247277021408081,
+      "learning_rate": 0.0001,
+      "loss": 0.0722,
+      "step": 14730
+    },
+    {
+      "epoch": 41.875,
+      "grad_norm": 1.1207205057144165,
+      "learning_rate": 0.0001,
+      "loss": 0.0691,
+      "step": 14740
+    },
+    {
+      "epoch": 41.90340909090909,
+      "grad_norm": 1.397698163986206,
+      "learning_rate": 0.0001,
+      "loss": 0.0718,
+      "step": 14750
+    },
+    {
+      "epoch": 41.93181818181818,
+      "grad_norm": 1.3861267566680908,
+      "learning_rate": 0.0001,
+      "loss": 0.0726,
+      "step": 14760
+    },
+    {
+      "epoch": 41.96022727272727,
+      "grad_norm": 1.3560404777526855,
+      "learning_rate": 0.0001,
+      "loss": 0.0725,
+      "step": 14770
+    },
+    {
+      "epoch": 41.98863636363637,
+      "grad_norm": 1.32721745967865,
+      "learning_rate": 0.0001,
+      "loss": 0.071,
+      "step": 14780
+    },
+    {
+      "epoch": 42.01704545454545,
+      "grad_norm": 1.1854987144470215,
+      "learning_rate": 0.0001,
+      "loss": 0.0712,
+      "step": 14790
+    },
+    {
+      "epoch": 42.04545454545455,
+      "grad_norm": 1.1440140008926392,
+      "learning_rate": 0.0001,
+      "loss": 0.0718,
+      "step": 14800
+    },
+    {
+      "epoch": 42.07386363636363,
+      "grad_norm": 1.064915657043457,
+      "learning_rate": 0.0001,
+      "loss": 0.0687,
+      "step": 14810
+    },
+    {
+      "epoch": 42.10227272727273,
+      "grad_norm": 1.279728889465332,
+      "learning_rate": 0.0001,
+      "loss": 0.0681,
+      "step": 14820
+    },
+    {
+      "epoch": 42.13068181818182,
+      "grad_norm": 1.2349141836166382,
+      "learning_rate": 0.0001,
+      "loss": 0.0651,
+      "step": 14830
+    },
+    {
+      "epoch": 42.15909090909091,
+      "grad_norm": 1.2165275812149048,
+      "learning_rate": 0.0001,
+      "loss": 0.0667,
+      "step": 14840
+    },
+    {
+      "epoch": 42.1875,
+      "grad_norm": 1.3267191648483276,
+      "learning_rate": 0.0001,
+      "loss": 0.071,
+      "step": 14850
+    },
+    {
+      "epoch": 42.21590909090909,
+      "grad_norm": 0.9966840147972107,
+      "learning_rate": 0.0001,
+      "loss": 0.0703,
+      "step": 14860
+    },
+    {
+      "epoch": 42.24431818181818,
+      "grad_norm": 1.1279515027999878,
+      "learning_rate": 0.0001,
+      "loss": 0.068,
+      "step": 14870
+    },
+    {
+      "epoch": 42.27272727272727,
+      "grad_norm": 1.3340145349502563,
+      "learning_rate": 0.0001,
+      "loss": 0.0705,
+      "step": 14880
+    },
+    {
+      "epoch": 42.30113636363637,
+      "grad_norm": 1.193154215812683,
+      "learning_rate": 0.0001,
+      "loss": 0.0686,
+      "step": 14890
+    },
+    {
+      "epoch": 42.32954545454545,
+      "grad_norm": 1.1050076484680176,
+      "learning_rate": 0.0001,
+      "loss": 0.0664,
+      "step": 14900
+    },
+    {
+      "epoch": 42.35795454545455,
+      "grad_norm": 1.1995725631713867,
+      "learning_rate": 0.0001,
+      "loss": 0.0677,
+      "step": 14910
+    },
+    {
+      "epoch": 42.38636363636363,
+      "grad_norm": 1.0086230039596558,
+      "learning_rate": 0.0001,
+      "loss": 0.0665,
+      "step": 14920
+    },
+    {
+      "epoch": 42.41477272727273,
+      "grad_norm": 0.8045913577079773,
+      "learning_rate": 0.0001,
+      "loss": 0.0673,
+      "step": 14930
+    },
+    {
+      "epoch": 42.44318181818182,
+      "grad_norm": 0.9728057980537415,
+      "learning_rate": 0.0001,
+      "loss": 0.0712,
+      "step": 14940
+    },
+    {
+      "epoch": 42.47159090909091,
+      "grad_norm": 1.2215827703475952,
+      "learning_rate": 0.0001,
+      "loss": 0.0702,
+      "step": 14950
+    },
+    {
+      "epoch": 42.5,
+      "grad_norm": 1.3105379343032837,
+      "learning_rate": 0.0001,
+      "loss": 0.0745,
+      "step": 14960
+    },
+    {
+      "epoch": 42.52840909090909,
+      "grad_norm": 1.5222160816192627,
+      "learning_rate": 0.0001,
+      "loss": 0.0719,
+      "step": 14970
+    },
+    {
+      "epoch": 42.55681818181818,
+      "grad_norm": 1.5078243017196655,
+      "learning_rate": 0.0001,
+      "loss": 0.0722,
+      "step": 14980
+    },
+    {
+      "epoch": 42.58522727272727,
+      "grad_norm": 1.5456411838531494,
+      "learning_rate": 0.0001,
+      "loss": 0.0701,
+      "step": 14990
+    },
+    {
+      "epoch": 42.61363636363637,
+      "grad_norm": 1.329354166984558,
+      "learning_rate": 0.0001,
+      "loss": 0.0727,
+      "step": 15000
+    },
+    {
+      "epoch": 42.64204545454545,
+      "grad_norm": 1.4910674095153809,
+      "learning_rate": 0.0001,
+      "loss": 0.0684,
+      "step": 15010
+    },
+    {
+      "epoch": 42.67045454545455,
+      "grad_norm": 1.3281280994415283,
+      "learning_rate": 0.0001,
+      "loss": 0.0707,
+      "step": 15020
+    },
+    {
+      "epoch": 42.69886363636363,
+      "grad_norm": 1.312171220779419,
+      "learning_rate": 0.0001,
+      "loss": 0.0683,
+      "step": 15030
+    },
+    {
+      "epoch": 42.72727272727273,
+      "grad_norm": 1.8369287252426147,
+      "learning_rate": 0.0001,
+      "loss": 0.0695,
+      "step": 15040
+    },
+    {
+      "epoch": 42.75568181818182,
+      "grad_norm": 1.3710136413574219,
+      "learning_rate": 0.0001,
+      "loss": 0.0684,
+      "step": 15050
+    },
+    {
+      "epoch": 42.78409090909091,
+      "grad_norm": 1.343110203742981,
+      "learning_rate": 0.0001,
+      "loss": 0.0685,
+      "step": 15060
+    },
+    {
+      "epoch": 42.8125,
+      "grad_norm": 1.2409188747406006,
+      "learning_rate": 0.0001,
+      "loss": 0.0666,
+      "step": 15070
+    },
+    {
+      "epoch": 42.84090909090909,
+      "grad_norm": 1.375400185585022,
+      "learning_rate": 0.0001,
+      "loss": 0.0678,
+      "step": 15080
+    },
+    {
+      "epoch": 42.86931818181818,
+      "grad_norm": 1.4058459997177124,
+      "learning_rate": 0.0001,
+      "loss": 0.0686,
+      "step": 15090
+    },
+    {
+      "epoch": 42.89772727272727,
+      "grad_norm": 1.4581125974655151,
+      "learning_rate": 0.0001,
+      "loss": 0.0726,
+      "step": 15100
+    },
+    {
+      "epoch": 42.92613636363637,
+      "grad_norm": 1.129081130027771,
+      "learning_rate": 0.0001,
+      "loss": 0.0705,
+      "step": 15110
+    },
+    {
+      "epoch": 42.95454545454545,
+      "grad_norm": 1.0302661657333374,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "step": 15120
+    },
+    {
+      "epoch": 42.98295454545455,
+      "grad_norm": 0.974215030670166,
+      "learning_rate": 0.0001,
+      "loss": 0.0671,
+      "step": 15130
+    },
+    {
+      "epoch": 43.01136363636363,
+      "grad_norm": 1.0873677730560303,
+      "learning_rate": 0.0001,
+      "loss": 0.0674,
+      "step": 15140
+    },
+    {
+      "epoch": 43.03977272727273,
+      "grad_norm": 1.3052396774291992,
+      "learning_rate": 0.0001,
+      "loss": 0.069,
+      "step": 15150
+    },
+    {
+      "epoch": 43.06818181818182,
+      "grad_norm": 1.1269701719284058,
+      "learning_rate": 0.0001,
+      "loss": 0.0685,
+      "step": 15160
+    },
+    {
+      "epoch": 43.09659090909091,
+      "grad_norm": 1.0936193466186523,
+      "learning_rate": 0.0001,
+      "loss": 0.0674,
+      "step": 15170
+    },
+    {
+      "epoch": 43.125,
+      "grad_norm": 1.168410301208496,
+      "learning_rate": 0.0001,
+      "loss": 0.0705,
+      "step": 15180
+    },
+    {
+      "epoch": 43.15340909090909,
+      "grad_norm": 1.151530385017395,
+      "learning_rate": 0.0001,
+      "loss": 0.0662,
+      "step": 15190
+    },
+    {
+      "epoch": 43.18181818181818,
+      "grad_norm": 1.0683995485305786,
+      "learning_rate": 0.0001,
+      "loss": 0.07,
+      "step": 15200
+    },
+    {
+      "epoch": 43.21022727272727,
+      "grad_norm": 1.1150685548782349,
+      "learning_rate": 0.0001,
+      "loss": 0.07,
+      "step": 15210
+    },
+    {
+      "epoch": 43.23863636363637,
+      "grad_norm": 1.1715824604034424,
+      "learning_rate": 0.0001,
+      "loss": 0.0693,
+      "step": 15220
+    },
+    {
+      "epoch": 43.26704545454545,
+      "grad_norm": 1.2279553413391113,
+      "learning_rate": 0.0001,
+      "loss": 0.0663,
+      "step": 15230
+    },
+    {
+      "epoch": 43.29545454545455,
+      "grad_norm": 1.020858883857727,
+      "learning_rate": 0.0001,
+      "loss": 0.0716,
+      "step": 15240
+    },
+    {
+      "epoch": 43.32386363636363,
+      "grad_norm": 1.4411025047302246,
+      "learning_rate": 0.0001,
+      "loss": 0.0724,
+      "step": 15250
+    },
+    {
+      "epoch": 43.35227272727273,
+      "grad_norm": 1.0164995193481445,
+      "learning_rate": 0.0001,
+      "loss": 0.0726,
+      "step": 15260
+    },
+    {
+      "epoch": 43.38068181818182,
+      "grad_norm": 1.092726469039917,
+      "learning_rate": 0.0001,
+      "loss": 0.0701,
+      "step": 15270
+    },
+    {
+      "epoch": 43.40909090909091,
+      "grad_norm": 0.9105169773101807,
+      "learning_rate": 0.0001,
+      "loss": 0.0713,
+      "step": 15280
+    },
+    {
+      "epoch": 43.4375,
+      "grad_norm": 1.0798728466033936,
+      "learning_rate": 0.0001,
+      "loss": 0.0711,
+      "step": 15290
+    },
+    {
+      "epoch": 43.46590909090909,
+      "grad_norm": 1.08077871799469,
+      "learning_rate": 0.0001,
+      "loss": 0.0703,
+      "step": 15300
+    },
+    {
+      "epoch": 43.49431818181818,
+      "grad_norm": 1.2410343885421753,
+      "learning_rate": 0.0001,
+      "loss": 0.0712,
+      "step": 15310
+    },
+    {
+      "epoch": 43.52272727272727,
+      "grad_norm": 1.0531221628189087,
+      "learning_rate": 0.0001,
+      "loss": 0.0677,
+      "step": 15320
+    },
+    {
+      "epoch": 43.55113636363637,
+      "grad_norm": 1.0954484939575195,
+      "learning_rate": 0.0001,
+      "loss": 0.0706,
+      "step": 15330
+    },
+    {
+      "epoch": 43.57954545454545,
+      "grad_norm": 1.155243158340454,
+      "learning_rate": 0.0001,
+      "loss": 0.0712,
+      "step": 15340
+    },
+    {
+      "epoch": 43.60795454545455,
+      "grad_norm": 0.9618707299232483,
+      "learning_rate": 0.0001,
+      "loss": 0.0708,
+      "step": 15350
+    },
+    {
+      "epoch": 43.63636363636363,
+      "grad_norm": 1.0885404348373413,
+      "learning_rate": 0.0001,
+      "loss": 0.0645,
+      "step": 15360
+    },
+    {
+      "epoch": 43.66477272727273,
+      "grad_norm": 1.2146936655044556,
+      "learning_rate": 0.0001,
+      "loss": 0.0696,
+      "step": 15370
+    },
+    {
+      "epoch": 43.69318181818182,
+      "grad_norm": 1.3160111904144287,
+      "learning_rate": 0.0001,
+      "loss": 0.0703,
+      "step": 15380
+    },
+    {
+      "epoch": 43.72159090909091,
+      "grad_norm": 1.0108137130737305,
+      "learning_rate": 0.0001,
+      "loss": 0.0734,
+      "step": 15390
+    },
+    {
+      "epoch": 43.75,
+      "grad_norm": 1.0981762409210205,
+      "learning_rate": 0.0001,
+      "loss": 0.0683,
+      "step": 15400
+    },
+    {
+      "epoch": 43.77840909090909,
+      "grad_norm": 1.1093978881835938,
+      "learning_rate": 0.0001,
+      "loss": 0.0691,
+      "step": 15410
+    },
+    {
+      "epoch": 43.80681818181818,
+      "grad_norm": 1.1320979595184326,
+      "learning_rate": 0.0001,
+      "loss": 0.0704,
+      "step": 15420
+    },
+    {
+      "epoch": 43.83522727272727,
+      "grad_norm": 1.089289665222168,
+      "learning_rate": 0.0001,
+      "loss": 0.069,
+      "step": 15430
+    },
+    {
+      "epoch": 43.86363636363637,
+      "grad_norm": 1.138124704360962,
+      "learning_rate": 0.0001,
+      "loss": 0.0681,
+      "step": 15440
+    },
+    {
+      "epoch": 43.89204545454545,
+      "grad_norm": 1.080330491065979,
+      "learning_rate": 0.0001,
+      "loss": 0.0678,
+      "step": 15450
+    },
+    {
+      "epoch": 43.92045454545455,
+      "grad_norm": 0.9879652857780457,
+      "learning_rate": 0.0001,
+      "loss": 0.0728,
+      "step": 15460
+    },
+    {
+      "epoch": 43.94886363636363,
+      "grad_norm": 1.2704243659973145,
+      "learning_rate": 0.0001,
+      "loss": 0.0671,
+      "step": 15470
+    },
+    {
+      "epoch": 43.97727272727273,
+      "grad_norm": 1.0896133184432983,
+      "learning_rate": 0.0001,
+      "loss": 0.0709,
+      "step": 15480
+    },
+    {
+      "epoch": 44.00568181818182,
+      "grad_norm": 1.3182461261749268,
+      "learning_rate": 0.0001,
+      "loss": 0.0697,
+      "step": 15490
+    },
+    {
+      "epoch": 44.03409090909091,
+      "grad_norm": 1.1766080856323242,
+      "learning_rate": 0.0001,
+      "loss": 0.0683,
+      "step": 15500
+    },
+    {
+      "epoch": 44.0625,
+      "grad_norm": 1.1963214874267578,
+      "learning_rate": 0.0001,
+      "loss": 0.0729,
+      "step": 15510
+    },
+    {
+      "epoch": 44.09090909090909,
+      "grad_norm": 1.32454252243042,
+      "learning_rate": 0.0001,
+      "loss": 0.0702,
+      "step": 15520
+    },
+    {
+      "epoch": 44.11931818181818,
+      "grad_norm": 0.9734973907470703,
+      "learning_rate": 0.0001,
+      "loss": 0.0722,
+      "step": 15530
+    },
+    {
+      "epoch": 44.14772727272727,
+      "grad_norm": 0.9227120280265808,
+      "learning_rate": 0.0001,
+      "loss": 0.0679,
+      "step": 15540
+    },
+    {
+      "epoch": 44.17613636363637,
+      "grad_norm": 1.0899708271026611,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "step": 15550
+    },
+    {
+      "epoch": 44.20454545454545,
+      "grad_norm": 0.9477924704551697,
+      "learning_rate": 0.0001,
+      "loss": 0.0739,
+      "step": 15560
+    },
+    {
+      "epoch": 44.23295454545455,
+      "grad_norm": 0.9881532788276672,
+      "learning_rate": 0.0001,
+      "loss": 0.0682,
+      "step": 15570
+    },
+    {
+      "epoch": 44.26136363636363,
+      "grad_norm": 1.1461341381072998,
+      "learning_rate": 0.0001,
+      "loss": 0.0682,
+      "step": 15580
+    },
+    {
+      "epoch": 44.28977272727273,
+      "grad_norm": 1.1203874349594116,
+      "learning_rate": 0.0001,
+      "loss": 0.0668,
+      "step": 15590
+    },
+    {
+      "epoch": 44.31818181818182,
+      "grad_norm": 1.113659381866455,
+      "learning_rate": 0.0001,
+      "loss": 0.0714,
+      "step": 15600
+    },
+    {
+      "epoch": 44.34659090909091,
+      "grad_norm": 0.9644593596458435,
+      "learning_rate": 0.0001,
+      "loss": 0.0665,
+      "step": 15610
+    },
+    {
+      "epoch": 44.375,
+      "grad_norm": 1.232541561126709,
+      "learning_rate": 0.0001,
+      "loss": 0.0693,
+      "step": 15620
+    },
+    {
+      "epoch": 44.40340909090909,
+      "grad_norm": 1.1204942464828491,
+      "learning_rate": 0.0001,
+      "loss": 0.0671,
+      "step": 15630
+    },
+    {
+      "epoch": 44.43181818181818,
+      "grad_norm": 1.2313846349716187,
+      "learning_rate": 0.0001,
+      "loss": 0.0711,
+      "step": 15640
+    },
+    {
+      "epoch": 44.46022727272727,
+      "grad_norm": 1.1166387796401978,
+      "learning_rate": 0.0001,
+      "loss": 0.0648,
+      "step": 15650
+    },
+    {
+      "epoch": 44.48863636363637,
+      "grad_norm": 1.1295506954193115,
+      "learning_rate": 0.0001,
+      "loss": 0.0706,
+      "step": 15660
+    },
+    {
+      "epoch": 44.51704545454545,
+      "grad_norm": 1.103320598602295,
+      "learning_rate": 0.0001,
+      "loss": 0.0692,
+      "step": 15670
+    },
+    {
+      "epoch": 44.54545454545455,
+      "grad_norm": 0.8729053139686584,
+      "learning_rate": 0.0001,
+      "loss": 0.0701,
+      "step": 15680
+    },
+    {
+      "epoch": 44.57386363636363,
+      "grad_norm": 0.8826537728309631,
+      "learning_rate": 0.0001,
+      "loss": 0.0729,
+      "step": 15690
+    },
+    {
+      "epoch": 44.60227272727273,
+      "grad_norm": 1.1283091306686401,
+      "learning_rate": 0.0001,
+      "loss": 0.0691,
+      "step": 15700
+    },
+    {
+      "epoch": 44.63068181818182,
+      "grad_norm": 1.0607900619506836,
+      "learning_rate": 0.0001,
+      "loss": 0.0719,
+      "step": 15710
+    },
+    {
+      "epoch": 44.65909090909091,
+      "grad_norm": 0.8510501384735107,
+      "learning_rate": 0.0001,
+      "loss": 0.0731,
+      "step": 15720
+    },
+    {
+      "epoch": 44.6875,
+      "grad_norm": 0.9762911796569824,
+      "learning_rate": 0.0001,
+      "loss": 0.0685,
+      "step": 15730
+    },
+    {
+      "epoch": 44.71590909090909,
+      "grad_norm": 0.9683955907821655,
+      "learning_rate": 0.0001,
+      "loss": 0.0732,
+      "step": 15740
+    },
+    {
+      "epoch": 44.74431818181818,
+      "grad_norm": 1.0100533962249756,
+      "learning_rate": 0.0001,
+      "loss": 0.0707,
+      "step": 15750
+    },
+    {
+      "epoch": 44.77272727272727,
+      "grad_norm": 0.860872745513916,
+      "learning_rate": 0.0001,
+      "loss": 0.0674,
+      "step": 15760
+    },
+    {
+      "epoch": 44.80113636363637,
+      "grad_norm": 1.065424919128418,
+      "learning_rate": 0.0001,
+      "loss": 0.0706,
+      "step": 15770
+    },
+    {
+      "epoch": 44.82954545454545,
+      "grad_norm": 0.9998086094856262,
+      "learning_rate": 0.0001,
+      "loss": 0.0727,
+      "step": 15780
+    },
+    {
+      "epoch": 44.85795454545455,
+      "grad_norm": 1.1579582691192627,
+      "learning_rate": 0.0001,
+      "loss": 0.0688,
+      "step": 15790
+    },
+    {
+      "epoch": 44.88636363636363,
+      "grad_norm": 1.0913549661636353,
+      "learning_rate": 0.0001,
+      "loss": 0.0713,
+      "step": 15800
+    },
+    {
+      "epoch": 44.91477272727273,
+      "grad_norm": 1.1674264669418335,
+      "learning_rate": 0.0001,
+      "loss": 0.0695,
+      "step": 15810
+    },
+    {
+      "epoch": 44.94318181818182,
+      "grad_norm": 1.1136603355407715,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "step": 15820
+    },
+    {
+      "epoch": 44.97159090909091,
+      "grad_norm": 1.3215253353118896,
+      "learning_rate": 0.0001,
+      "loss": 0.0709,
+      "step": 15830
+    },
+    {
+      "epoch": 45.0,
+      "grad_norm": 1.308078646659851,
+      "learning_rate": 0.0001,
+      "loss": 0.0685,
+      "step": 15840
+    },
+    {
+      "epoch": 45.02840909090909,
+      "grad_norm": 1.174768090248108,
+      "learning_rate": 0.0001,
+      "loss": 0.072,
+      "step": 15850
+    },
+    {
+      "epoch": 45.05681818181818,
+      "grad_norm": 1.1076934337615967,
+      "learning_rate": 0.0001,
+      "loss": 0.067,
+      "step": 15860
+    },
+    {
+      "epoch": 45.08522727272727,
+      "grad_norm": 1.202553629875183,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "step": 15870
+    },
+    {
+      "epoch": 45.11363636363637,
+      "grad_norm": 1.2212430238723755,
+      "learning_rate": 0.0001,
+      "loss": 0.0671,
+      "step": 15880
+    },
+    {
+      "epoch": 45.14204545454545,
+      "grad_norm": 1.785838007926941,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "step": 15890
+    },
+    {
+      "epoch": 45.17045454545455,
+      "grad_norm": 1.2640763521194458,
+      "learning_rate": 0.0001,
+      "loss": 0.0717,
+      "step": 15900
+    },
+    {
+      "epoch": 45.19886363636363,
+      "grad_norm": 1.1825248003005981,
+      "learning_rate": 0.0001,
+      "loss": 0.0646,
+      "step": 15910
+    },
+    {
+      "epoch": 45.22727272727273,
+      "grad_norm": 1.1265792846679688,
+      "learning_rate": 0.0001,
+      "loss": 0.0666,
+      "step": 15920
+    },
+    {
+      "epoch": 45.25568181818182,
+      "grad_norm": 1.1798592805862427,
+      "learning_rate": 0.0001,
+      "loss": 0.0645,
+      "step": 15930
+    },
+    {
+      "epoch": 45.28409090909091,
+      "grad_norm": 0.996825635433197,
+      "learning_rate": 0.0001,
+      "loss": 0.0659,
+      "step": 15940
+    },
+    {
+      "epoch": 45.3125,
+      "grad_norm": 1.1232649087905884,
+      "learning_rate": 0.0001,
+      "loss": 0.0699,
+      "step": 15950
+    },
+    {
+      "epoch": 45.34090909090909,
+      "grad_norm": 1.0283252000808716,
+      "learning_rate": 0.0001,
+      "loss": 0.0697,
+      "step": 15960
+    },
+    {
+      "epoch": 45.36931818181818,
+      "grad_norm": 1.0725383758544922,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "step": 15970
+    },
+    {
+      "epoch": 45.39772727272727,
+      "grad_norm": 0.88676917552948,
+      "learning_rate": 0.0001,
+      "loss": 0.068,
+      "step": 15980
+    },
+    {
+      "epoch": 45.42613636363637,
+      "grad_norm": 0.8896072506904602,
+      "learning_rate": 0.0001,
+      "loss": 0.0689,
+      "step": 15990
+    },
+    {
+      "epoch": 45.45454545454545,
+      "grad_norm": 1.008349061012268,
+      "learning_rate": 0.0001,
+      "loss": 0.0664,
+      "step": 16000
+    },
+    {
+      "epoch": 45.48295454545455,
+      "grad_norm": 0.9974130988121033,
+      "learning_rate": 0.0001,
+      "loss": 0.0672,
+      "step": 16010
+    },
+    {
+      "epoch": 45.51136363636363,
+      "grad_norm": 1.0345064401626587,
+      "learning_rate": 0.0001,
+      "loss": 0.0678,
+      "step": 16020
+    },
+    {
+      "epoch": 45.53977272727273,
+      "grad_norm": 1.2052901983261108,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "step": 16030
+    },
+    {
+      "epoch": 45.56818181818182,
+      "grad_norm": 1.0479304790496826,
+      "learning_rate": 0.0001,
+      "loss": 0.0677,
+      "step": 16040
+    },
+    {
+      "epoch": 45.59659090909091,
+      "grad_norm": 1.1516709327697754,
+      "learning_rate": 0.0001,
+      "loss": 0.0724,
+      "step": 16050
+    },
+    {
+      "epoch": 45.625,
+      "grad_norm": 1.3962410688400269,
+      "learning_rate": 0.0001,
+      "loss": 0.0685,
+      "step": 16060
+    },
+    {
+      "epoch": 45.65340909090909,
+      "grad_norm": 1.3850163221359253,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "step": 16070
+    },
+    {
+      "epoch": 45.68181818181818,
+      "grad_norm": 1.6799479722976685,
+      "learning_rate": 0.0001,
+      "loss": 0.069,
+      "step": 16080
+    },
+    {
+      "epoch": 45.71022727272727,
+      "grad_norm": 0.9734259247779846,
+      "learning_rate": 0.0001,
+      "loss": 0.0663,
+      "step": 16090
+    },
+    {
+      "epoch": 45.73863636363637,
+      "grad_norm": 1.6348193883895874,
+      "learning_rate": 0.0001,
+      "loss": 0.0656,
+      "step": 16100
+    },
+    {
+      "epoch": 45.76704545454545,
+      "grad_norm": 1.480636715888977,
+      "learning_rate": 0.0001,
+      "loss": 0.0638,
+      "step": 16110
+    },
+    {
+      "epoch": 45.79545454545455,
+      "grad_norm": 1.3442875146865845,
+      "learning_rate": 0.0001,
+      "loss": 0.0707,
+      "step": 16120
+    },
+    {
+      "epoch": 45.82386363636363,
+      "grad_norm": 1.4665964841842651,
+      "learning_rate": 0.0001,
+      "loss": 0.065,
+      "step": 16130
+    },
+    {
+      "epoch": 45.85227272727273,
+      "grad_norm": 1.2870675325393677,
+      "learning_rate": 0.0001,
+      "loss": 0.0657,
+      "step": 16140
+    },
+    {
+      "epoch": 45.88068181818182,
+      "grad_norm": 1.2831790447235107,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "step": 16150
+    },
+    {
+      "epoch": 45.90909090909091,
+      "grad_norm": 1.2384798526763916,
+      "learning_rate": 0.0001,
+      "loss": 0.0691,
+      "step": 16160
+    },
+    {
+      "epoch": 45.9375,
+      "grad_norm": 1.0735738277435303,
+      "learning_rate": 0.0001,
+      "loss": 0.0671,
+      "step": 16170
+    },
+    {
+      "epoch": 45.96590909090909,
+      "grad_norm": 1.3313257694244385,
+      "learning_rate": 0.0001,
+      "loss": 0.0643,
+      "step": 16180
+    },
+    {
+      "epoch": 45.99431818181818,
+      "grad_norm": 1.170076608657837,
+      "learning_rate": 0.0001,
+      "loss": 0.068,
+      "step": 16190
+    },
+    {
+      "epoch": 46.02272727272727,
+      "grad_norm": 1.383847713470459,
+      "learning_rate": 0.0001,
+      "loss": 0.0648,
+      "step": 16200
+    },
+    {
+      "epoch": 46.05113636363637,
+      "grad_norm": 1.4958339929580688,
+      "learning_rate": 0.0001,
+      "loss": 0.0684,
+      "step": 16210
+    },
+    {
+      "epoch": 46.07954545454545,
+      "grad_norm": 0.9990864992141724,
+      "learning_rate": 0.0001,
+      "loss": 0.0663,
+      "step": 16220
+    },
+    {
+      "epoch": 46.10795454545455,
+      "grad_norm": 1.1924002170562744,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "step": 16230
+    },
+    {
+      "epoch": 46.13636363636363,
+      "grad_norm": 1.2387312650680542,
+      "learning_rate": 0.0001,
+      "loss": 0.0637,
+      "step": 16240
+    },
+    {
+      "epoch": 46.16477272727273,
+      "grad_norm": 1.2327535152435303,
+      "learning_rate": 0.0001,
+      "loss": 0.0668,
+      "step": 16250
+    },
+    {
+      "epoch": 46.19318181818182,
+      "grad_norm": 1.0648465156555176,
+      "learning_rate": 0.0001,
+      "loss": 0.0661,
+      "step": 16260
+    },
+    {
+      "epoch": 46.22159090909091,
+      "grad_norm": 0.9485817551612854,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "step": 16270
+    },
+    {
+      "epoch": 46.25,
+      "grad_norm": 1.2529845237731934,
+      "learning_rate": 0.0001,
+      "loss": 0.0652,
+      "step": 16280
+    },
+    {
+      "epoch": 46.27840909090909,
+      "grad_norm": 1.0558775663375854,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 16290
+    },
+    {
+      "epoch": 46.30681818181818,
+      "grad_norm": 1.1467417478561401,
+      "learning_rate": 0.0001,
+      "loss": 0.0659,
+      "step": 16300
+    },
+    {
+      "epoch": 46.33522727272727,
+      "grad_norm": 1.4327044486999512,
+      "learning_rate": 0.0001,
+      "loss": 0.0664,
+      "step": 16310
+    },
+    {
+      "epoch": 46.36363636363637,
+      "grad_norm": 1.25315260887146,
+      "learning_rate": 0.0001,
+      "loss": 0.0721,
+      "step": 16320
+    },
+    {
+      "epoch": 46.39204545454545,
+      "grad_norm": 1.2470471858978271,
+      "learning_rate": 0.0001,
+      "loss": 0.066,
+      "step": 16330
+    },
+    {
+      "epoch": 46.42045454545455,
+      "grad_norm": 1.2044808864593506,
+      "learning_rate": 0.0001,
+      "loss": 0.0679,
+      "step": 16340
+    },
+    {
+      "epoch": 46.44886363636363,
+      "grad_norm": 1.3889199495315552,
+      "learning_rate": 0.0001,
+      "loss": 0.0621,
+      "step": 16350
+    },
+    {
+      "epoch": 46.47727272727273,
+      "grad_norm": 1.1649527549743652,
+      "learning_rate": 0.0001,
+      "loss": 0.0651,
+      "step": 16360
+    },
+    {
+      "epoch": 46.50568181818182,
+      "grad_norm": 1.0425108671188354,
+      "learning_rate": 0.0001,
+      "loss": 0.0621,
+      "step": 16370
+    },
+    {
+      "epoch": 46.53409090909091,
+      "grad_norm": 1.1113712787628174,
+      "learning_rate": 0.0001,
+      "loss": 0.0634,
+      "step": 16380
+    },
+    {
+      "epoch": 46.5625,
+      "grad_norm": 1.1670324802398682,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 16390
+    },
+    {
+      "epoch": 46.59090909090909,
+      "grad_norm": 1.195335030555725,
+      "learning_rate": 0.0001,
+      "loss": 0.0672,
+      "step": 16400
+    },
+    {
+      "epoch": 46.61931818181818,
+      "grad_norm": 1.1802396774291992,
+      "learning_rate": 0.0001,
+      "loss": 0.0678,
+      "step": 16410
+    },
+    {
+      "epoch": 46.64772727272727,
+      "grad_norm": 1.2869985103607178,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "step": 16420
+    },
+    {
+      "epoch": 46.67613636363637,
+      "grad_norm": 1.2032572031021118,
+      "learning_rate": 0.0001,
+      "loss": 0.0651,
+      "step": 16430
+    },
+    {
+      "epoch": 46.70454545454545,
+      "grad_norm": 1.116356611251831,
+      "learning_rate": 0.0001,
+      "loss": 0.0647,
+      "step": 16440
+    },
+    {
+      "epoch": 46.73295454545455,
+      "grad_norm": 1.3527079820632935,
+      "learning_rate": 0.0001,
+      "loss": 0.0686,
+      "step": 16450
+    },
+    {
+      "epoch": 46.76136363636363,
+      "grad_norm": 1.105823040008545,
+      "learning_rate": 0.0001,
+      "loss": 0.0654,
+      "step": 16460
+    },
+    {
+      "epoch": 46.78977272727273,
+      "grad_norm": 1.1165571212768555,
+      "learning_rate": 0.0001,
+      "loss": 0.0661,
+      "step": 16470
+    },
+    {
+      "epoch": 46.81818181818182,
+      "grad_norm": 1.0524123907089233,
+      "learning_rate": 0.0001,
+      "loss": 0.0662,
+      "step": 16480
+    },
+    {
+      "epoch": 46.84659090909091,
+      "grad_norm": 1.0740056037902832,
+      "learning_rate": 0.0001,
+      "loss": 0.0673,
+      "step": 16490
+    },
+    {
+      "epoch": 46.875,
+      "grad_norm": 1.0902903079986572,
+      "learning_rate": 0.0001,
+      "loss": 0.0713,
+      "step": 16500
+    },
+    {
+      "epoch": 46.90340909090909,
+      "grad_norm": 1.1585133075714111,
+      "learning_rate": 0.0001,
+      "loss": 0.0668,
+      "step": 16510
+    },
+    {
+      "epoch": 46.93181818181818,
+      "grad_norm": 0.8834289312362671,
+      "learning_rate": 0.0001,
+      "loss": 0.0683,
+      "step": 16520
+    },
+    {
+      "epoch": 46.96022727272727,
+      "grad_norm": 1.0335935354232788,
+      "learning_rate": 0.0001,
+      "loss": 0.0684,
+      "step": 16530
+    },
+    {
+      "epoch": 46.98863636363637,
+      "grad_norm": 1.1123617887496948,
+      "learning_rate": 0.0001,
+      "loss": 0.0665,
+      "step": 16540
+    },
+    {
+      "epoch": 47.01704545454545,
+      "grad_norm": 1.012212872505188,
+      "learning_rate": 0.0001,
+      "loss": 0.0663,
+      "step": 16550
+    },
+    {
+      "epoch": 47.04545454545455,
+      "grad_norm": 0.8919275999069214,
+      "learning_rate": 0.0001,
+      "loss": 0.0684,
+      "step": 16560
+    },
+    {
+      "epoch": 47.07386363636363,
+      "grad_norm": 1.0549639463424683,
+      "learning_rate": 0.0001,
+      "loss": 0.0699,
+      "step": 16570
+    },
+    {
+      "epoch": 47.10227272727273,
+      "grad_norm": 1.048153281211853,
+      "learning_rate": 0.0001,
+      "loss": 0.0657,
+      "step": 16580
+    },
+    {
+      "epoch": 47.13068181818182,
+      "grad_norm": 0.9343200325965881,
+      "learning_rate": 0.0001,
+      "loss": 0.0698,
+      "step": 16590
+    },
+    {
+      "epoch": 47.15909090909091,
+      "grad_norm": 0.970174252986908,
+      "learning_rate": 0.0001,
+      "loss": 0.0691,
+      "step": 16600
+    },
+    {
+      "epoch": 47.1875,
+      "grad_norm": 0.8680684566497803,
+      "learning_rate": 0.0001,
+      "loss": 0.0669,
+      "step": 16610
+    },
+    {
+      "epoch": 47.21590909090909,
+      "grad_norm": 1.1044062376022339,
+      "learning_rate": 0.0001,
+      "loss": 0.0678,
+      "step": 16620
+    },
+    {
+      "epoch": 47.24431818181818,
+      "grad_norm": 1.067230224609375,
+      "learning_rate": 0.0001,
+      "loss": 0.0694,
+      "step": 16630
+    },
+    {
+      "epoch": 47.27272727272727,
+      "grad_norm": 1.0942895412445068,
+      "learning_rate": 0.0001,
+      "loss": 0.0677,
+      "step": 16640
+    },
+    {
+      "epoch": 47.30113636363637,
+      "grad_norm": 0.955245852470398,
+      "learning_rate": 0.0001,
+      "loss": 0.0649,
+      "step": 16650
+    },
+    {
+      "epoch": 47.32954545454545,
+      "grad_norm": 0.9594484567642212,
+      "learning_rate": 0.0001,
+      "loss": 0.0675,
+      "step": 16660
+    },
+    {
+      "epoch": 47.35795454545455,
+      "grad_norm": 1.0073819160461426,
+      "learning_rate": 0.0001,
+      "loss": 0.0635,
+      "step": 16670
+    },
+    {
+      "epoch": 47.38636363636363,
+      "grad_norm": 1.0799858570098877,
+      "learning_rate": 0.0001,
+      "loss": 0.0673,
+      "step": 16680
+    },
+    {
+      "epoch": 47.41477272727273,
+      "grad_norm": 1.113233208656311,
+      "learning_rate": 0.0001,
+      "loss": 0.0686,
+      "step": 16690
+    },
+    {
+      "epoch": 47.44318181818182,
+      "grad_norm": 0.9608368277549744,
+      "learning_rate": 0.0001,
+      "loss": 0.0644,
+      "step": 16700
+    },
+    {
+      "epoch": 47.47159090909091,
+      "grad_norm": 0.9679139256477356,
+      "learning_rate": 0.0001,
+      "loss": 0.0648,
+      "step": 16710
+    },
+    {
+      "epoch": 47.5,
+      "grad_norm": 0.9586361050605774,
+      "learning_rate": 0.0001,
+      "loss": 0.0636,
+      "step": 16720
+    },
+    {
+      "epoch": 47.52840909090909,
+      "grad_norm": 0.7961944341659546,
+      "learning_rate": 0.0001,
+      "loss": 0.0649,
+      "step": 16730
+    },
+    {
+      "epoch": 47.55681818181818,
+      "grad_norm": 0.891774594783783,
+      "learning_rate": 0.0001,
+      "loss": 0.0647,
+      "step": 16740
+    },
+    {
+      "epoch": 47.58522727272727,
+      "grad_norm": 1.0160497426986694,
+      "learning_rate": 0.0001,
+      "loss": 0.062,
+      "step": 16750
+    },
+    {
+      "epoch": 47.61363636363637,
+      "grad_norm": 1.0420070886611938,
+      "learning_rate": 0.0001,
+      "loss": 0.0662,
+      "step": 16760
+    },
+    {
+      "epoch": 47.64204545454545,
+      "grad_norm": 1.0268901586532593,
+      "learning_rate": 0.0001,
+      "loss": 0.0671,
+      "step": 16770
+    },
+    {
+      "epoch": 47.67045454545455,
+      "grad_norm": 0.8848260641098022,
+      "learning_rate": 0.0001,
+      "loss": 0.0675,
+      "step": 16780
+    },
+    {
+      "epoch": 47.69886363636363,
+      "grad_norm": 0.9918054938316345,
+      "learning_rate": 0.0001,
+      "loss": 0.0667,
+      "step": 16790
+    },
+    {
+      "epoch": 47.72727272727273,
+      "grad_norm": 0.9512577056884766,
+      "learning_rate": 0.0001,
+      "loss": 0.0699,
+      "step": 16800
+    },
+    {
+      "epoch": 47.75568181818182,
+      "grad_norm": 1.1260731220245361,
+      "learning_rate": 0.0001,
+      "loss": 0.0679,
+      "step": 16810
+    },
+    {
+      "epoch": 47.78409090909091,
+      "grad_norm": 1.003982424736023,
+      "learning_rate": 0.0001,
+      "loss": 0.0644,
+      "step": 16820
+    },
+    {
+      "epoch": 47.8125,
+      "grad_norm": 0.9938884973526001,
+      "learning_rate": 0.0001,
+      "loss": 0.0677,
+      "step": 16830
+    },
+    {
+      "epoch": 47.84090909090909,
+      "grad_norm": 1.222053050994873,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 16840
+    },
+    {
+      "epoch": 47.86931818181818,
+      "grad_norm": 0.9429041147232056,
+      "learning_rate": 0.0001,
+      "loss": 0.0656,
+      "step": 16850
+    },
+    {
+      "epoch": 47.89772727272727,
+      "grad_norm": 1.0304569005966187,
+      "learning_rate": 0.0001,
+      "loss": 0.0632,
+      "step": 16860
+    },
+    {
+      "epoch": 47.92613636363637,
+      "grad_norm": 0.8651162981987,
+      "learning_rate": 0.0001,
+      "loss": 0.0638,
+      "step": 16870
+    },
+    {
+      "epoch": 47.95454545454545,
+      "grad_norm": 1.1727617979049683,
+      "learning_rate": 0.0001,
+      "loss": 0.0641,
+      "step": 16880
+    },
+    {
+      "epoch": 47.98295454545455,
+      "grad_norm": 0.9716474413871765,
+      "learning_rate": 0.0001,
+      "loss": 0.0672,
+      "step": 16890
+    },
+    {
+      "epoch": 48.01136363636363,
+      "grad_norm": 0.9646078944206238,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 16900
+    },
+    {
+      "epoch": 48.03977272727273,
+      "grad_norm": 0.7903724312782288,
+      "learning_rate": 0.0001,
+      "loss": 0.0652,
+      "step": 16910
+    },
+    {
+      "epoch": 48.06818181818182,
+      "grad_norm": 0.7010796070098877,
+      "learning_rate": 0.0001,
+      "loss": 0.065,
+      "step": 16920
+    },
+    {
+      "epoch": 48.09659090909091,
+      "grad_norm": 0.9780798554420471,
+      "learning_rate": 0.0001,
+      "loss": 0.064,
+      "step": 16930
+    },
+    {
+      "epoch": 48.125,
+      "grad_norm": 0.8578932285308838,
+      "learning_rate": 0.0001,
+      "loss": 0.065,
+      "step": 16940
+    },
+    {
+      "epoch": 48.15340909090909,
+      "grad_norm": 1.0814779996871948,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "step": 16950
+    },
+    {
+      "epoch": 48.18181818181818,
+      "grad_norm": 1.1047223806381226,
+      "learning_rate": 0.0001,
+      "loss": 0.0659,
+      "step": 16960
+    },
+    {
+      "epoch": 48.21022727272727,
+      "grad_norm": 1.0128185749053955,
+      "learning_rate": 0.0001,
+      "loss": 0.07,
+      "step": 16970
+    },
+    {
+      "epoch": 48.23863636363637,
+      "grad_norm": 1.191439151763916,
+      "learning_rate": 0.0001,
+      "loss": 0.0683,
+      "step": 16980
+    },
+    {
+      "epoch": 48.26704545454545,
+      "grad_norm": 1.02851140499115,
+      "learning_rate": 0.0001,
+      "loss": 0.0646,
+      "step": 16990
+    },
+    {
+      "epoch": 48.29545454545455,
+      "grad_norm": 0.9317130446434021,
+      "learning_rate": 0.0001,
+      "loss": 0.0666,
+      "step": 17000
+    },
+    {
+      "epoch": 48.32386363636363,
+      "grad_norm": 0.8646169900894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0645,
+      "step": 17010
+    },
+    {
+      "epoch": 48.35227272727273,
+      "grad_norm": 0.8524001240730286,
+      "learning_rate": 0.0001,
+      "loss": 0.0654,
+      "step": 17020
+    },
+    {
+      "epoch": 48.38068181818182,
+      "grad_norm": 0.7672250270843506,
+      "learning_rate": 0.0001,
+      "loss": 0.0676,
+      "step": 17030
+    },
+    {
+      "epoch": 48.40909090909091,
+      "grad_norm": 0.9044290781021118,
+      "learning_rate": 0.0001,
+      "loss": 0.0657,
+      "step": 17040
+    },
+    {
+      "epoch": 48.4375,
+      "grad_norm": 0.8433730602264404,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "step": 17050
+    },
+    {
+      "epoch": 48.46590909090909,
+      "grad_norm": 0.7054641842842102,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 17060
+    },
+    {
+      "epoch": 48.49431818181818,
+      "grad_norm": 0.7396852374076843,
+      "learning_rate": 0.0001,
+      "loss": 0.0654,
+      "step": 17070
+    },
+    {
+      "epoch": 48.52272727272727,
+      "grad_norm": 0.8507287502288818,
+      "learning_rate": 0.0001,
+      "loss": 0.063,
+      "step": 17080
+    },
+    {
+      "epoch": 48.55113636363637,
+      "grad_norm": 0.7744329571723938,
+      "learning_rate": 0.0001,
+      "loss": 0.0652,
+      "step": 17090
+    },
+    {
+      "epoch": 48.57954545454545,
+      "grad_norm": 0.8510982990264893,
+      "learning_rate": 0.0001,
+      "loss": 0.0652,
+      "step": 17100
+    },
+    {
+      "epoch": 48.60795454545455,
+      "grad_norm": 0.9002107977867126,
+      "learning_rate": 0.0001,
+      "loss": 0.0652,
+      "step": 17110
+    },
+    {
+      "epoch": 48.63636363636363,
+      "grad_norm": 1.3234931230545044,
+      "learning_rate": 0.0001,
+      "loss": 0.0644,
+      "step": 17120
+    },
+    {
+      "epoch": 48.66477272727273,
+      "grad_norm": 1.2231804132461548,
+      "learning_rate": 0.0001,
+      "loss": 0.0651,
+      "step": 17130
+    },
+    {
+      "epoch": 48.69318181818182,
+      "grad_norm": 1.0435932874679565,
+      "learning_rate": 0.0001,
+      "loss": 0.063,
+      "step": 17140
+    },
+    {
+      "epoch": 48.72159090909091,
+      "grad_norm": 1.1819216012954712,
+      "learning_rate": 0.0001,
+      "loss": 0.0639,
+      "step": 17150
+    },
+    {
+      "epoch": 48.75,
+      "grad_norm": 1.3706990480422974,
+      "learning_rate": 0.0001,
+      "loss": 0.0649,
+      "step": 17160
+    },
+    {
+      "epoch": 48.77840909090909,
+      "grad_norm": 1.0659633874893188,
+      "learning_rate": 0.0001,
+      "loss": 0.0643,
+      "step": 17170
+    },
+    {
+      "epoch": 48.80681818181818,
+      "grad_norm": 1.1273925304412842,
+      "learning_rate": 0.0001,
+      "loss": 0.0647,
+      "step": 17180
+    },
+    {
+      "epoch": 48.83522727272727,
+      "grad_norm": 1.0904277563095093,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "step": 17190
+    },
+    {
+      "epoch": 48.86363636363637,
+      "grad_norm": 1.4346485137939453,
+      "learning_rate": 0.0001,
+      "loss": 0.0658,
+      "step": 17200
+    },
+    {
+      "epoch": 48.89204545454545,
+      "grad_norm": 1.2752107381820679,
+      "learning_rate": 0.0001,
+      "loss": 0.0666,
+      "step": 17210
+    },
+    {
+      "epoch": 48.92045454545455,
+      "grad_norm": 1.179890751838684,
+      "learning_rate": 0.0001,
+      "loss": 0.0634,
+      "step": 17220
+    },
+    {
+      "epoch": 48.94886363636363,
+      "grad_norm": 1.0924259424209595,
+      "learning_rate": 0.0001,
+      "loss": 0.063,
+      "step": 17230
+    },
+    {
+      "epoch": 48.97727272727273,
+      "grad_norm": 1.2451024055480957,
+      "learning_rate": 0.0001,
+      "loss": 0.063,
+      "step": 17240
+    },
+    {
+      "epoch": 49.00568181818182,
+      "grad_norm": 1.29100501537323,
+      "learning_rate": 0.0001,
+      "loss": 0.0625,
+      "step": 17250
+    },
+    {
+      "epoch": 49.03409090909091,
+      "grad_norm": 1.175595998764038,
+      "learning_rate": 0.0001,
+      "loss": 0.065,
+      "step": 17260
+    },
+    {
+      "epoch": 49.0625,
+      "grad_norm": 1.084436297416687,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 17270
+    },
+    {
+      "epoch": 49.09090909090909,
+      "grad_norm": 1.1387672424316406,
+      "learning_rate": 0.0001,
+      "loss": 0.0597,
+      "step": 17280
+    },
+    {
+      "epoch": 49.11931818181818,
+      "grad_norm": 1.0709339380264282,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 17290
+    },
+    {
+      "epoch": 49.14772727272727,
+      "grad_norm": 1.1278772354125977,
+      "learning_rate": 0.0001,
+      "loss": 0.0592,
+      "step": 17300
+    },
+    {
+      "epoch": 49.17613636363637,
+      "grad_norm": 1.2357215881347656,
+      "learning_rate": 0.0001,
+      "loss": 0.0645,
+      "step": 17310
+    },
+    {
+      "epoch": 49.20454545454545,
+      "grad_norm": 1.3850127458572388,
+      "learning_rate": 0.0001,
+      "loss": 0.0643,
+      "step": 17320
+    },
+    {
+      "epoch": 49.23295454545455,
+      "grad_norm": 1.5667473077774048,
+      "learning_rate": 0.0001,
+      "loss": 0.064,
+      "step": 17330
+    },
+    {
+      "epoch": 49.26136363636363,
+      "grad_norm": 2.0329859256744385,
+      "learning_rate": 0.0001,
+      "loss": 0.0638,
+      "step": 17340
+    },
+    {
+      "epoch": 49.28977272727273,
+      "grad_norm": 1.9868841171264648,
+      "learning_rate": 0.0001,
+      "loss": 0.0619,
+      "step": 17350
+    },
+    {
+      "epoch": 49.31818181818182,
+      "grad_norm": 1.650707721710205,
+      "learning_rate": 0.0001,
+      "loss": 0.0622,
+      "step": 17360
+    },
+    {
+      "epoch": 49.34659090909091,
+      "grad_norm": 1.6646281480789185,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 17370
+    },
+    {
+      "epoch": 49.375,
+      "grad_norm": 1.3677852153778076,
+      "learning_rate": 0.0001,
+      "loss": 0.061,
+      "step": 17380
+    },
+    {
+      "epoch": 49.40340909090909,
+      "grad_norm": 1.33896005153656,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 17390
+    },
+    {
+      "epoch": 49.43181818181818,
+      "grad_norm": 1.3581352233886719,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 17400
+    },
+    {
+      "epoch": 49.46022727272727,
+      "grad_norm": 1.3389812707901,
+      "learning_rate": 0.0001,
+      "loss": 0.0574,
+      "step": 17410
+    },
+    {
+      "epoch": 49.48863636363637,
+      "grad_norm": 1.304565191268921,
+      "learning_rate": 0.0001,
+      "loss": 0.0628,
+      "step": 17420
+    },
+    {
+      "epoch": 49.51704545454545,
+      "grad_norm": 1.9376806020736694,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 17430
+    },
+    {
+      "epoch": 49.54545454545455,
+      "grad_norm": 1.8597931861877441,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 17440
+    },
+    {
+      "epoch": 49.57386363636363,
+      "grad_norm": 1.4572594165802002,
+      "learning_rate": 0.0001,
+      "loss": 0.0615,
+      "step": 17450
+    },
+    {
+      "epoch": 49.60227272727273,
+      "grad_norm": 1.7948307991027832,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 17460
+    },
+    {
+      "epoch": 49.63068181818182,
+      "grad_norm": 1.3871301412582397,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "step": 17470
+    },
+    {
+      "epoch": 49.65909090909091,
+      "grad_norm": 1.322991132736206,
+      "learning_rate": 0.0001,
+      "loss": 0.0601,
+      "step": 17480
+    },
+    {
+      "epoch": 49.6875,
+      "grad_norm": 1.2961491346359253,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 17490
+    },
+    {
+      "epoch": 49.71590909090909,
+      "grad_norm": 1.283707618713379,
+      "learning_rate": 0.0001,
+      "loss": 0.0609,
+      "step": 17500
+    },
+    {
+      "epoch": 49.74431818181818,
+      "grad_norm": 1.142791986465454,
+      "learning_rate": 0.0001,
+      "loss": 0.0609,
+      "step": 17510
+    },
+    {
+      "epoch": 49.77272727272727,
+      "grad_norm": 0.8002589344978333,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "step": 17520
+    },
+    {
+      "epoch": 49.80113636363637,
+      "grad_norm": 0.9746940732002258,
+      "learning_rate": 0.0001,
+      "loss": 0.0597,
+      "step": 17530
+    },
+    {
+      "epoch": 49.82954545454545,
+      "grad_norm": 0.965501606464386,
+      "learning_rate": 0.0001,
+      "loss": 0.0646,
+      "step": 17540
+    },
+    {
+      "epoch": 49.85795454545455,
+      "grad_norm": 1.053093671798706,
+      "learning_rate": 0.0001,
+      "loss": 0.0643,
+      "step": 17550
+    },
+    {
+      "epoch": 49.88636363636363,
+      "grad_norm": 1.2206720113754272,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "step": 17560
+    },
+    {
+      "epoch": 49.91477272727273,
+      "grad_norm": 1.094285488128662,
+      "learning_rate": 0.0001,
+      "loss": 0.0606,
+      "step": 17570
+    },
+    {
+      "epoch": 49.94318181818182,
+      "grad_norm": 1.9030991792678833,
+      "learning_rate": 0.0001,
+      "loss": 0.0679,
+      "step": 17580
+    },
+    {
+      "epoch": 49.97159090909091,
+      "grad_norm": 1.452059030532837,
+      "learning_rate": 0.0001,
+      "loss": 0.0687,
+      "step": 17590
+    },
+    {
+      "epoch": 50.0,
+      "grad_norm": 1.4255893230438232,
+      "learning_rate": 0.0001,
+      "loss": 0.0641,
+      "step": 17600
+    },
+    {
+      "epoch": 50.02840909090909,
+      "grad_norm": 1.547782063484192,
+      "learning_rate": 0.0001,
+      "loss": 0.0607,
+      "step": 17610
+    },
+    {
+      "epoch": 50.05681818181818,
+      "grad_norm": 1.533228874206543,
+      "learning_rate": 0.0001,
+      "loss": 0.0623,
+      "step": 17620
+    },
+    {
+      "epoch": 50.08522727272727,
+      "grad_norm": 2.1276297569274902,
+      "learning_rate": 0.0001,
+      "loss": 0.062,
+      "step": 17630
+    },
+    {
+      "epoch": 50.11363636363637,
+      "grad_norm": 1.5320310592651367,
+      "learning_rate": 0.0001,
+      "loss": 0.059,
+      "step": 17640
+    },
+    {
+      "epoch": 50.14204545454545,
+      "grad_norm": 1.4868521690368652,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 17650
+    },
+    {
+      "epoch": 50.17045454545455,
+      "grad_norm": 1.3425029516220093,
+      "learning_rate": 0.0001,
+      "loss": 0.0625,
+      "step": 17660
+    },
+    {
+      "epoch": 50.19886363636363,
+      "grad_norm": 1.683226466178894,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 17670
+    },
+    {
+      "epoch": 50.22727272727273,
+      "grad_norm": 1.6359002590179443,
+      "learning_rate": 0.0001,
+      "loss": 0.0584,
+      "step": 17680
+    },
+    {
+      "epoch": 50.25568181818182,
+      "grad_norm": 1.1186552047729492,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 17690
+    },
+    {
+      "epoch": 50.28409090909091,
+      "grad_norm": 0.9917884469032288,
+      "learning_rate": 0.0001,
+      "loss": 0.0568,
+      "step": 17700
+    },
+    {
+      "epoch": 50.3125,
+      "grad_norm": 1.3089747428894043,
+      "learning_rate": 0.0001,
+      "loss": 0.0608,
+      "step": 17710
+    },
+    {
+      "epoch": 50.34090909090909,
+      "grad_norm": 1.094618320465088,
+      "learning_rate": 0.0001,
+      "loss": 0.0628,
+      "step": 17720
+    },
+    {
+      "epoch": 50.36931818181818,
+      "grad_norm": 1.0189076662063599,
+      "learning_rate": 0.0001,
+      "loss": 0.061,
+      "step": 17730
+    },
+    {
+      "epoch": 50.39772727272727,
+      "grad_norm": 1.0388256311416626,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 17740
+    },
+    {
+      "epoch": 50.42613636363637,
+      "grad_norm": 1.2349307537078857,
+      "learning_rate": 0.0001,
+      "loss": 0.0614,
+      "step": 17750
+    },
+    {
+      "epoch": 50.45454545454545,
+      "grad_norm": 1.1165496110916138,
+      "learning_rate": 0.0001,
+      "loss": 0.0615,
+      "step": 17760
+    },
+    {
+      "epoch": 50.48295454545455,
+      "grad_norm": 1.3187179565429688,
+      "learning_rate": 0.0001,
+      "loss": 0.064,
+      "step": 17770
+    },
+    {
+      "epoch": 50.51136363636363,
+      "grad_norm": 1.1244080066680908,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 17780
+    },
+    {
+      "epoch": 50.53977272727273,
+      "grad_norm": 1.1014071702957153,
+      "learning_rate": 0.0001,
+      "loss": 0.0612,
+      "step": 17790
+    },
+    {
+      "epoch": 50.56818181818182,
+      "grad_norm": 1.0804203748703003,
+      "learning_rate": 0.0001,
+      "loss": 0.0615,
+      "step": 17800
+    },
+    {
+      "epoch": 50.59659090909091,
+      "grad_norm": 1.1172409057617188,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 17810
+    },
+    {
+      "epoch": 50.625,
+      "grad_norm": 1.2156904935836792,
+      "learning_rate": 0.0001,
+      "loss": 0.0635,
+      "step": 17820
+    },
+    {
+      "epoch": 50.65340909090909,
+      "grad_norm": 1.0518922805786133,
+      "learning_rate": 0.0001,
+      "loss": 0.0649,
+      "step": 17830
+    },
+    {
+      "epoch": 50.68181818181818,
+      "grad_norm": 1.0572881698608398,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "step": 17840
+    },
+    {
+      "epoch": 50.71022727272727,
+      "grad_norm": 0.9975048899650574,
+      "learning_rate": 0.0001,
+      "loss": 0.0637,
+      "step": 17850
+    },
+    {
+      "epoch": 50.73863636363637,
+      "grad_norm": 1.0606894493103027,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "step": 17860
+    },
+    {
+      "epoch": 50.76704545454545,
+      "grad_norm": 1.1085385084152222,
+      "learning_rate": 0.0001,
+      "loss": 0.0638,
+      "step": 17870
+    },
+    {
+      "epoch": 50.79545454545455,
+      "grad_norm": 1.0079888105392456,
+      "learning_rate": 0.0001,
+      "loss": 0.0645,
+      "step": 17880
+    },
+    {
+      "epoch": 50.82386363636363,
+      "grad_norm": 0.9119659662246704,
+      "learning_rate": 0.0001,
+      "loss": 0.0657,
+      "step": 17890
+    },
+    {
+      "epoch": 50.85227272727273,
+      "grad_norm": 1.0529975891113281,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 17900
+    },
+    {
+      "epoch": 50.88068181818182,
+      "grad_norm": 1.101491928100586,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 17910
+    },
+    {
+      "epoch": 50.90909090909091,
+      "grad_norm": 1.049623966217041,
+      "learning_rate": 0.0001,
+      "loss": 0.0659,
+      "step": 17920
+    },
+    {
+      "epoch": 50.9375,
+      "grad_norm": 0.855363130569458,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 17930
+    },
+    {
+      "epoch": 50.96590909090909,
+      "grad_norm": 0.8307511210441589,
+      "learning_rate": 0.0001,
+      "loss": 0.0639,
+      "step": 17940
+    },
+    {
+      "epoch": 50.99431818181818,
+      "grad_norm": 0.9840141534805298,
+      "learning_rate": 0.0001,
+      "loss": 0.0638,
+      "step": 17950
+    },
+    {
+      "epoch": 51.02272727272727,
+      "grad_norm": 0.7411724328994751,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "step": 17960
+    },
+    {
+      "epoch": 51.05113636363637,
+      "grad_norm": 1.1654945611953735,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "step": 17970
+    },
+    {
+      "epoch": 51.07954545454545,
+      "grad_norm": 1.0566688776016235,
+      "learning_rate": 0.0001,
+      "loss": 0.0717,
+      "step": 17980
+    },
+    {
+      "epoch": 51.10795454545455,
+      "grad_norm": 1.1673790216445923,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 17990
+    },
+    {
+      "epoch": 51.13636363636363,
+      "grad_norm": 1.0396006107330322,
+      "learning_rate": 0.0001,
+      "loss": 0.0672,
+      "step": 18000
+    },
+    {
+      "epoch": 51.16477272727273,
+      "grad_norm": 0.9316548705101013,
+      "learning_rate": 0.0001,
+      "loss": 0.0688,
+      "step": 18010
+    },
+    {
+      "epoch": 51.19318181818182,
+      "grad_norm": 1.0011225938796997,
+      "learning_rate": 0.0001,
+      "loss": 0.0623,
+      "step": 18020
+    },
+    {
+      "epoch": 51.22159090909091,
+      "grad_norm": 1.2169394493103027,
+      "learning_rate": 0.0001,
+      "loss": 0.0668,
+      "step": 18030
+    },
+    {
+      "epoch": 51.25,
+      "grad_norm": 1.0183131694793701,
+      "learning_rate": 0.0001,
+      "loss": 0.0637,
+      "step": 18040
+    },
+    {
+      "epoch": 51.27840909090909,
+      "grad_norm": 1.3156449794769287,
+      "learning_rate": 0.0001,
+      "loss": 0.0674,
+      "step": 18050
+    },
+    {
+      "epoch": 51.30681818181818,
+      "grad_norm": 1.2456082105636597,
+      "learning_rate": 0.0001,
+      "loss": 0.0677,
+      "step": 18060
+    },
+    {
+      "epoch": 51.33522727272727,
+      "grad_norm": 1.2302768230438232,
+      "learning_rate": 0.0001,
+      "loss": 0.064,
+      "step": 18070
+    },
+    {
+      "epoch": 51.36363636363637,
+      "grad_norm": 1.3110893964767456,
+      "learning_rate": 0.0001,
+      "loss": 0.0643,
+      "step": 18080
+    },
+    {
+      "epoch": 51.39204545454545,
+      "grad_norm": 1.0676565170288086,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 18090
+    },
+    {
+      "epoch": 51.42045454545455,
+      "grad_norm": 0.8240692019462585,
+      "learning_rate": 0.0001,
+      "loss": 0.0613,
+      "step": 18100
+    },
+    {
+      "epoch": 51.44886363636363,
+      "grad_norm": 1.0472469329833984,
+      "learning_rate": 0.0001,
+      "loss": 0.0635,
+      "step": 18110
+    },
+    {
+      "epoch": 51.47727272727273,
+      "grad_norm": 1.0723049640655518,
+      "learning_rate": 0.0001,
+      "loss": 0.0626,
+      "step": 18120
+    },
+    {
+      "epoch": 51.50568181818182,
+      "grad_norm": 0.8317203521728516,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 18130
+    },
+    {
+      "epoch": 51.53409090909091,
+      "grad_norm": 0.8893545269966125,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 18140
+    },
+    {
+      "epoch": 51.5625,
+      "grad_norm": 1.143960952758789,
+      "learning_rate": 0.0001,
+      "loss": 0.0645,
+      "step": 18150
+    },
+    {
+      "epoch": 51.59090909090909,
+      "grad_norm": 1.1038347482681274,
+      "learning_rate": 0.0001,
+      "loss": 0.0654,
+      "step": 18160
+    },
+    {
+      "epoch": 51.61931818181818,
+      "grad_norm": 0.9441390037536621,
+      "learning_rate": 0.0001,
+      "loss": 0.0614,
+      "step": 18170
+    },
+    {
+      "epoch": 51.64772727272727,
+      "grad_norm": 1.1258492469787598,
+      "learning_rate": 0.0001,
+      "loss": 0.0609,
+      "step": 18180
+    },
+    {
+      "epoch": 51.67613636363637,
+      "grad_norm": 1.1269819736480713,
+      "learning_rate": 0.0001,
+      "loss": 0.0637,
+      "step": 18190
+    },
+    {
+      "epoch": 51.70454545454545,
+      "grad_norm": 0.8500455021858215,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "step": 18200
+    },
+    {
+      "epoch": 51.73295454545455,
+      "grad_norm": 0.8912470936775208,
+      "learning_rate": 0.0001,
+      "loss": 0.0639,
+      "step": 18210
+    },
+    {
+      "epoch": 51.76136363636363,
+      "grad_norm": 1.0278549194335938,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 18220
+    },
+    {
+      "epoch": 51.78977272727273,
+      "grad_norm": 0.9313192367553711,
+      "learning_rate": 0.0001,
+      "loss": 0.0656,
+      "step": 18230
+    },
+    {
+      "epoch": 51.81818181818182,
+      "grad_norm": 0.8399008512496948,
+      "learning_rate": 0.0001,
+      "loss": 0.066,
+      "step": 18240
+    },
+    {
+      "epoch": 51.84659090909091,
+      "grad_norm": 0.8020362257957458,
+      "learning_rate": 0.0001,
+      "loss": 0.0647,
+      "step": 18250
+    },
+    {
+      "epoch": 51.875,
+      "grad_norm": 0.9596895575523376,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 18260
+    },
+    {
+      "epoch": 51.90340909090909,
+      "grad_norm": 0.8371610045433044,
+      "learning_rate": 0.0001,
+      "loss": 0.0669,
+      "step": 18270
+    },
+    {
+      "epoch": 51.93181818181818,
+      "grad_norm": 0.774612307548523,
+      "learning_rate": 0.0001,
+      "loss": 0.0637,
+      "step": 18280
+    },
+    {
+      "epoch": 51.96022727272727,
+      "grad_norm": 0.9038988351821899,
+      "learning_rate": 0.0001,
+      "loss": 0.0662,
+      "step": 18290
+    },
+    {
+      "epoch": 51.98863636363637,
+      "grad_norm": 1.150199055671692,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 18300
+    },
+    {
+      "epoch": 52.01704545454545,
+      "grad_norm": 1.3316466808319092,
+      "learning_rate": 0.0001,
+      "loss": 0.0621,
+      "step": 18310
+    },
+    {
+      "epoch": 52.04545454545455,
+      "grad_norm": 1.0422097444534302,
+      "learning_rate": 0.0001,
+      "loss": 0.0657,
+      "step": 18320
+    },
+    {
+      "epoch": 52.07386363636363,
+      "grad_norm": 1.0511544942855835,
+      "learning_rate": 0.0001,
+      "loss": 0.064,
+      "step": 18330
+    },
+    {
+      "epoch": 52.10227272727273,
+      "grad_norm": 0.9210748672485352,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 18340
+    },
+    {
+      "epoch": 52.13068181818182,
+      "grad_norm": 1.0048185586929321,
+      "learning_rate": 0.0001,
+      "loss": 0.0614,
+      "step": 18350
+    },
+    {
+      "epoch": 52.15909090909091,
+      "grad_norm": 1.0767287015914917,
+      "learning_rate": 0.0001,
+      "loss": 0.0613,
+      "step": 18360
+    },
+    {
+      "epoch": 52.1875,
+      "grad_norm": 1.0105133056640625,
+      "learning_rate": 0.0001,
+      "loss": 0.0615,
+      "step": 18370
+    },
+    {
+      "epoch": 52.21590909090909,
+      "grad_norm": 0.7021766304969788,
+      "learning_rate": 0.0001,
+      "loss": 0.068,
+      "step": 18380
+    },
+    {
+      "epoch": 52.24431818181818,
+      "grad_norm": 0.8372295498847961,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 18390
+    },
+    {
+      "epoch": 52.27272727272727,
+      "grad_norm": 0.7811925411224365,
+      "learning_rate": 0.0001,
+      "loss": 0.0649,
+      "step": 18400
+    },
+    {
+      "epoch": 52.30113636363637,
+      "grad_norm": 0.821349024772644,
+      "learning_rate": 0.0001,
+      "loss": 0.0654,
+      "step": 18410
+    },
+    {
+      "epoch": 52.32954545454545,
+      "grad_norm": 0.7928653359413147,
+      "learning_rate": 0.0001,
+      "loss": 0.0626,
+      "step": 18420
+    },
+    {
+      "epoch": 52.35795454545455,
+      "grad_norm": 0.8362237215042114,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "step": 18430
+    },
+    {
+      "epoch": 52.38636363636363,
+      "grad_norm": 0.9375684261322021,
+      "learning_rate": 0.0001,
+      "loss": 0.0659,
+      "step": 18440
+    },
+    {
+      "epoch": 52.41477272727273,
+      "grad_norm": 0.9780309796333313,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 18450
+    },
+    {
+      "epoch": 52.44318181818182,
+      "grad_norm": 1.062294840812683,
+      "learning_rate": 0.0001,
+      "loss": 0.0665,
+      "step": 18460
+    },
+    {
+      "epoch": 52.47159090909091,
+      "grad_norm": 1.0660057067871094,
+      "learning_rate": 0.0001,
+      "loss": 0.065,
+      "step": 18470
+    },
+    {
+      "epoch": 52.5,
+      "grad_norm": 1.05930495262146,
+      "learning_rate": 0.0001,
+      "loss": 0.0626,
+      "step": 18480
+    },
+    {
+      "epoch": 52.52840909090909,
+      "grad_norm": 0.8140740394592285,
+      "learning_rate": 0.0001,
+      "loss": 0.0615,
+      "step": 18490
+    },
+    {
+      "epoch": 52.55681818181818,
+      "grad_norm": 1.0095051527023315,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 18500
+    },
+    {
+      "epoch": 52.58522727272727,
+      "grad_norm": 0.9089073538780212,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "step": 18510
+    },
+    {
+      "epoch": 52.61363636363637,
+      "grad_norm": 0.8695672154426575,
+      "learning_rate": 0.0001,
+      "loss": 0.0623,
+      "step": 18520
+    },
+    {
+      "epoch": 52.64204545454545,
+      "grad_norm": 1.0662381649017334,
+      "learning_rate": 0.0001,
+      "loss": 0.0639,
+      "step": 18530
+    },
+    {
+      "epoch": 52.67045454545455,
+      "grad_norm": 1.0227075815200806,
+      "learning_rate": 0.0001,
+      "loss": 0.0632,
+      "step": 18540
+    },
+    {
+      "epoch": 52.69886363636363,
+      "grad_norm": 1.003777265548706,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "step": 18550
+    },
+    {
+      "epoch": 52.72727272727273,
+      "grad_norm": 0.9030758738517761,
+      "learning_rate": 0.0001,
+      "loss": 0.0591,
+      "step": 18560
+    },
+    {
+      "epoch": 52.75568181818182,
+      "grad_norm": 1.0003148317337036,
+      "learning_rate": 0.0001,
+      "loss": 0.0632,
+      "step": 18570
+    },
+    {
+      "epoch": 52.78409090909091,
+      "grad_norm": 0.8785012364387512,
+      "learning_rate": 0.0001,
+      "loss": 0.0622,
+      "step": 18580
+    },
+    {
+      "epoch": 52.8125,
+      "grad_norm": 1.1217682361602783,
+      "learning_rate": 0.0001,
+      "loss": 0.0636,
+      "step": 18590
+    },
+    {
+      "epoch": 52.84090909090909,
+      "grad_norm": 1.0676538944244385,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 18600
+    },
+    {
+      "epoch": 52.86931818181818,
+      "grad_norm": 1.2667808532714844,
+      "learning_rate": 0.0001,
+      "loss": 0.0634,
+      "step": 18610
+    },
+    {
+      "epoch": 52.89772727272727,
+      "grad_norm": 1.255224347114563,
+      "learning_rate": 0.0001,
+      "loss": 0.0632,
+      "step": 18620
+    },
+    {
+      "epoch": 52.92613636363637,
+      "grad_norm": 1.0495514869689941,
+      "learning_rate": 0.0001,
+      "loss": 0.0635,
+      "step": 18630
+    },
+    {
+      "epoch": 52.95454545454545,
+      "grad_norm": 1.2018243074417114,
+      "learning_rate": 0.0001,
+      "loss": 0.066,
+      "step": 18640
+    },
+    {
+      "epoch": 52.98295454545455,
+      "grad_norm": 1.1866649389266968,
+      "learning_rate": 0.0001,
+      "loss": 0.0592,
+      "step": 18650
+    },
+    {
+      "epoch": 53.01136363636363,
+      "grad_norm": 1.3282562494277954,
+      "learning_rate": 0.0001,
+      "loss": 0.0644,
+      "step": 18660
+    },
+    {
+      "epoch": 53.03977272727273,
+      "grad_norm": 1.1297610998153687,
+      "learning_rate": 0.0001,
+      "loss": 0.0636,
+      "step": 18670
+    },
+    {
+      "epoch": 53.06818181818182,
+      "grad_norm": 1.1281236410140991,
+      "learning_rate": 0.0001,
+      "loss": 0.0598,
+      "step": 18680
+    },
+    {
+      "epoch": 53.09659090909091,
+      "grad_norm": 1.2281813621520996,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 18690
+    },
+    {
+      "epoch": 53.125,
+      "grad_norm": 1.1154251098632812,
+      "learning_rate": 0.0001,
+      "loss": 0.0613,
+      "step": 18700
+    },
+    {
+      "epoch": 53.15340909090909,
+      "grad_norm": 0.9964898824691772,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "step": 18710
+    },
+    {
+      "epoch": 53.18181818181818,
+      "grad_norm": 1.0880987644195557,
+      "learning_rate": 0.0001,
+      "loss": 0.0582,
+      "step": 18720
+    },
+    {
+      "epoch": 53.21022727272727,
+      "grad_norm": 1.1195552349090576,
+      "learning_rate": 0.0001,
+      "loss": 0.0612,
+      "step": 18730
+    },
+    {
+      "epoch": 53.23863636363637,
+      "grad_norm": 1.1419169902801514,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "step": 18740
+    },
+    {
+      "epoch": 53.26704545454545,
+      "grad_norm": 0.9695098400115967,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 18750
+    },
+    {
+      "epoch": 53.29545454545455,
+      "grad_norm": 1.0621428489685059,
+      "learning_rate": 0.0001,
+      "loss": 0.0632,
+      "step": 18760
+    },
+    {
+      "epoch": 53.32386363636363,
+      "grad_norm": 1.047326922416687,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 18770
+    },
+    {
+      "epoch": 53.35227272727273,
+      "grad_norm": 1.062296748161316,
+      "learning_rate": 0.0001,
+      "loss": 0.0639,
+      "step": 18780
+    },
+    {
+      "epoch": 53.38068181818182,
+      "grad_norm": 1.134915828704834,
+      "learning_rate": 0.0001,
+      "loss": 0.0619,
+      "step": 18790
+    },
+    {
+      "epoch": 53.40909090909091,
+      "grad_norm": 1.040932536125183,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "step": 18800
+    },
+    {
+      "epoch": 53.4375,
+      "grad_norm": 1.0221163034439087,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 18810
+    },
+    {
+      "epoch": 53.46590909090909,
+      "grad_norm": 0.9637789726257324,
+      "learning_rate": 0.0001,
+      "loss": 0.0614,
+      "step": 18820
+    },
+    {
+      "epoch": 53.49431818181818,
+      "grad_norm": 1.0450648069381714,
+      "learning_rate": 0.0001,
+      "loss": 0.06,
+      "step": 18830
+    },
+    {
+      "epoch": 53.52272727272727,
+      "grad_norm": 0.9436495900154114,
+      "learning_rate": 0.0001,
+      "loss": 0.0622,
+      "step": 18840
+    },
+    {
+      "epoch": 53.55113636363637,
+      "grad_norm": 1.0419658422470093,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 18850
+    },
+    {
+      "epoch": 53.57954545454545,
+      "grad_norm": 0.8948044180870056,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 18860
+    },
+    {
+      "epoch": 53.60795454545455,
+      "grad_norm": 1.0394561290740967,
+      "learning_rate": 0.0001,
+      "loss": 0.0623,
+      "step": 18870
+    },
+    {
+      "epoch": 53.63636363636363,
+      "grad_norm": 1.00216543674469,
+      "learning_rate": 0.0001,
+      "loss": 0.0597,
+      "step": 18880
+    },
+    {
+      "epoch": 53.66477272727273,
+      "grad_norm": 0.7522396445274353,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "step": 18890
+    },
+    {
+      "epoch": 53.69318181818182,
+      "grad_norm": 1.162223219871521,
+      "learning_rate": 0.0001,
+      "loss": 0.063,
+      "step": 18900
+    },
+    {
+      "epoch": 53.72159090909091,
+      "grad_norm": 1.0378843545913696,
+      "learning_rate": 0.0001,
+      "loss": 0.0645,
+      "step": 18910
+    },
+    {
+      "epoch": 53.75,
+      "grad_norm": 0.890440821647644,
+      "learning_rate": 0.0001,
+      "loss": 0.0615,
+      "step": 18920
+    },
+    {
+      "epoch": 53.77840909090909,
+      "grad_norm": 0.825039803981781,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 18930
+    },
+    {
+      "epoch": 53.80681818181818,
+      "grad_norm": 0.8410844802856445,
+      "learning_rate": 0.0001,
+      "loss": 0.0608,
+      "step": 18940
+    },
+    {
+      "epoch": 53.83522727272727,
+      "grad_norm": 1.1039787530899048,
+      "learning_rate": 0.0001,
+      "loss": 0.0653,
+      "step": 18950
+    },
+    {
+      "epoch": 53.86363636363637,
+      "grad_norm": 0.9757326245307922,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "step": 18960
+    },
+    {
+      "epoch": 53.89204545454545,
+      "grad_norm": 0.7968357801437378,
+      "learning_rate": 0.0001,
+      "loss": 0.0639,
+      "step": 18970
+    },
+    {
+      "epoch": 53.92045454545455,
+      "grad_norm": 0.9554797410964966,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 18980
+    },
+    {
+      "epoch": 53.94886363636363,
+      "grad_norm": 0.8579493165016174,
+      "learning_rate": 0.0001,
+      "loss": 0.0584,
+      "step": 18990
+    },
+    {
+      "epoch": 53.97727272727273,
+      "grad_norm": 0.841977059841156,
+      "learning_rate": 0.0001,
+      "loss": 0.0638,
+      "step": 19000
+    },
+    {
+      "epoch": 54.00568181818182,
+      "grad_norm": 0.7986201643943787,
+      "learning_rate": 0.0001,
+      "loss": 0.0593,
+      "step": 19010
+    },
+    {
+      "epoch": 54.03409090909091,
+      "grad_norm": 1.3590857982635498,
+      "learning_rate": 0.0001,
+      "loss": 0.0612,
+      "step": 19020
+    },
+    {
+      "epoch": 54.0625,
+      "grad_norm": 0.7293931841850281,
+      "learning_rate": 0.0001,
+      "loss": 0.0606,
+      "step": 19030
+    },
+    {
+      "epoch": 54.09090909090909,
+      "grad_norm": 0.8448790907859802,
+      "learning_rate": 0.0001,
+      "loss": 0.0649,
+      "step": 19040
+    },
+    {
+      "epoch": 54.11931818181818,
+      "grad_norm": 0.9543153047561646,
+      "learning_rate": 0.0001,
+      "loss": 0.0627,
+      "step": 19050
+    },
+    {
+      "epoch": 54.14772727272727,
+      "grad_norm": 0.859277606010437,
+      "learning_rate": 0.0001,
+      "loss": 0.0657,
+      "step": 19060
+    },
+    {
+      "epoch": 54.17613636363637,
+      "grad_norm": 0.9446835517883301,
+      "learning_rate": 0.0001,
+      "loss": 0.0598,
+      "step": 19070
+    },
+    {
+      "epoch": 54.20454545454545,
+      "grad_norm": 0.9382405281066895,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 19080
+    },
+    {
+      "epoch": 54.23295454545455,
+      "grad_norm": 1.0334746837615967,
+      "learning_rate": 0.0001,
+      "loss": 0.0619,
+      "step": 19090
+    },
+    {
+      "epoch": 54.26136363636363,
+      "grad_norm": 1.0028332471847534,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 19100
+    },
+    {
+      "epoch": 54.28977272727273,
+      "grad_norm": 1.056213140487671,
+      "learning_rate": 0.0001,
+      "loss": 0.0626,
+      "step": 19110
+    },
+    {
+      "epoch": 54.31818181818182,
+      "grad_norm": 0.9702300429344177,
+      "learning_rate": 0.0001,
+      "loss": 0.0677,
+      "step": 19120
+    },
+    {
+      "epoch": 54.34659090909091,
+      "grad_norm": 1.210434079170227,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "step": 19130
+    },
+    {
+      "epoch": 54.375,
+      "grad_norm": 1.048459529876709,
+      "learning_rate": 0.0001,
+      "loss": 0.0628,
+      "step": 19140
+    },
+    {
+      "epoch": 54.40340909090909,
+      "grad_norm": 0.9593278765678406,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 19150
+    },
+    {
+      "epoch": 54.43181818181818,
+      "grad_norm": 0.7503321170806885,
+      "learning_rate": 0.0001,
+      "loss": 0.0589,
+      "step": 19160
+    },
+    {
+      "epoch": 54.46022727272727,
+      "grad_norm": 0.8162446618080139,
+      "learning_rate": 0.0001,
+      "loss": 0.0623,
+      "step": 19170
+    },
+    {
+      "epoch": 54.48863636363637,
+      "grad_norm": 0.619260311126709,
+      "learning_rate": 0.0001,
+      "loss": 0.0598,
+      "step": 19180
+    },
+    {
+      "epoch": 54.51704545454545,
+      "grad_norm": 0.6872047781944275,
+      "learning_rate": 0.0001,
+      "loss": 0.0646,
+      "step": 19190
+    },
+    {
+      "epoch": 54.54545454545455,
+      "grad_norm": 0.9017942547798157,
+      "learning_rate": 0.0001,
+      "loss": 0.0642,
+      "step": 19200
+    },
+    {
+      "epoch": 54.57386363636363,
+      "grad_norm": 0.9055486917495728,
+      "learning_rate": 0.0001,
+      "loss": 0.0615,
+      "step": 19210
+    },
+    {
+      "epoch": 54.60227272727273,
+      "grad_norm": 0.9028745889663696,
+      "learning_rate": 0.0001,
+      "loss": 0.0607,
+      "step": 19220
+    },
+    {
+      "epoch": 54.63068181818182,
+      "grad_norm": 0.937928318977356,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "step": 19230
+    },
+    {
+      "epoch": 54.65909090909091,
+      "grad_norm": 0.9727193713188171,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "step": 19240
+    },
+    {
+      "epoch": 54.6875,
+      "grad_norm": 0.8907310366630554,
+      "learning_rate": 0.0001,
+      "loss": 0.0606,
+      "step": 19250
+    },
+    {
+      "epoch": 54.71590909090909,
+      "grad_norm": 0.8963820934295654,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 19260
+    },
+    {
+      "epoch": 54.74431818181818,
+      "grad_norm": 0.966294527053833,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 19270
+    },
+    {
+      "epoch": 54.77272727272727,
+      "grad_norm": 0.8197779059410095,
+      "learning_rate": 0.0001,
+      "loss": 0.0632,
+      "step": 19280
+    },
+    {
+      "epoch": 54.80113636363637,
+      "grad_norm": 0.9076131582260132,
+      "learning_rate": 0.0001,
+      "loss": 0.063,
+      "step": 19290
+    },
+    {
+      "epoch": 54.82954545454545,
+      "grad_norm": 1.0960373878479004,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "step": 19300
+    },
+    {
+      "epoch": 54.85795454545455,
+      "grad_norm": 1.0276180505752563,
+      "learning_rate": 0.0001,
+      "loss": 0.0614,
+      "step": 19310
+    },
+    {
+      "epoch": 54.88636363636363,
+      "grad_norm": 0.8789876699447632,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 19320
+    },
+    {
+      "epoch": 54.91477272727273,
+      "grad_norm": 0.9199723601341248,
+      "learning_rate": 0.0001,
+      "loss": 0.0596,
+      "step": 19330
+    },
+    {
+      "epoch": 54.94318181818182,
+      "grad_norm": 0.8767564296722412,
+      "learning_rate": 0.0001,
+      "loss": 0.061,
+      "step": 19340
+    },
+    {
+      "epoch": 54.97159090909091,
+      "grad_norm": 0.8243430256843567,
+      "learning_rate": 0.0001,
+      "loss": 0.0598,
+      "step": 19350
+    },
+    {
+      "epoch": 55.0,
+      "grad_norm": 0.8050703406333923,
+      "learning_rate": 0.0001,
+      "loss": 0.0614,
+      "step": 19360
+    },
+    {
+      "epoch": 55.02840909090909,
+      "grad_norm": 0.7457884550094604,
+      "learning_rate": 0.0001,
+      "loss": 0.0591,
+      "step": 19370
+    },
+    {
+      "epoch": 55.05681818181818,
+      "grad_norm": 0.7916040420532227,
+      "learning_rate": 0.0001,
+      "loss": 0.0608,
+      "step": 19380
+    },
+    {
+      "epoch": 55.08522727272727,
+      "grad_norm": 1.00780189037323,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 19390
+    },
+    {
+      "epoch": 55.11363636363637,
+      "grad_norm": 1.1122857332229614,
+      "learning_rate": 0.0001,
+      "loss": 0.0638,
+      "step": 19400
+    },
+    {
+      "epoch": 55.14204545454545,
+      "grad_norm": 1.1893644332885742,
+      "learning_rate": 0.0001,
+      "loss": 0.0606,
+      "step": 19410
+    },
+    {
+      "epoch": 55.17045454545455,
+      "grad_norm": 1.3157379627227783,
+      "learning_rate": 0.0001,
+      "loss": 0.0601,
+      "step": 19420
+    },
+    {
+      "epoch": 55.19886363636363,
+      "grad_norm": 1.305027961730957,
+      "learning_rate": 0.0001,
+      "loss": 0.063,
+      "step": 19430
+    },
+    {
+      "epoch": 55.22727272727273,
+      "grad_norm": 1.2850711345672607,
+      "learning_rate": 0.0001,
+      "loss": 0.0601,
+      "step": 19440
+    },
+    {
+      "epoch": 55.25568181818182,
+      "grad_norm": 1.2181235551834106,
+      "learning_rate": 0.0001,
+      "loss": 0.0606,
+      "step": 19450
+    },
+    {
+      "epoch": 55.28409090909091,
+      "grad_norm": 1.2655651569366455,
+      "learning_rate": 0.0001,
+      "loss": 0.0586,
+      "step": 19460
+    },
+    {
+      "epoch": 55.3125,
+      "grad_norm": 1.0834294557571411,
+      "learning_rate": 0.0001,
+      "loss": 0.0592,
+      "step": 19470
+    },
+    {
+      "epoch": 55.34090909090909,
+      "grad_norm": 0.8892400860786438,
+      "learning_rate": 0.0001,
+      "loss": 0.0594,
+      "step": 19480
+    },
+    {
+      "epoch": 55.36931818181818,
+      "grad_norm": 1.0693202018737793,
+      "learning_rate": 0.0001,
+      "loss": 0.0631,
+      "step": 19490
+    },
+    {
+      "epoch": 55.39772727272727,
+      "grad_norm": 0.9103065729141235,
+      "learning_rate": 0.0001,
+      "loss": 0.0599,
+      "step": 19500
+    },
+    {
+      "epoch": 55.42613636363637,
+      "grad_norm": 0.9410889744758606,
+      "learning_rate": 0.0001,
+      "loss": 0.057,
+      "step": 19510
+    },
+    {
+      "epoch": 55.45454545454545,
+      "grad_norm": 0.9349491000175476,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 19520
+    },
+    {
+      "epoch": 55.48295454545455,
+      "grad_norm": 0.7211048007011414,
+      "learning_rate": 0.0001,
+      "loss": 0.0623,
+      "step": 19530
+    },
+    {
+      "epoch": 55.51136363636363,
+      "grad_norm": 1.076320767402649,
+      "learning_rate": 0.0001,
+      "loss": 0.063,
+      "step": 19540
+    },
+    {
+      "epoch": 55.53977272727273,
+      "grad_norm": 1.0220001935958862,
+      "learning_rate": 0.0001,
+      "loss": 0.0641,
+      "step": 19550
+    },
+    {
+      "epoch": 55.56818181818182,
+      "grad_norm": 1.0938565731048584,
+      "learning_rate": 0.0001,
+      "loss": 0.0648,
+      "step": 19560
+    },
+    {
+      "epoch": 55.59659090909091,
+      "grad_norm": 1.207349181175232,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "step": 19570
+    },
+    {
+      "epoch": 55.625,
+      "grad_norm": 0.8213934302330017,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "step": 19580
+    },
+    {
+      "epoch": 55.65340909090909,
+      "grad_norm": 0.9144793748855591,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 19590
+    },
+    {
+      "epoch": 55.68181818181818,
+      "grad_norm": 0.8391266465187073,
+      "learning_rate": 0.0001,
+      "loss": 0.0593,
+      "step": 19600
+    },
+    {
+      "epoch": 55.71022727272727,
+      "grad_norm": 0.8670185804367065,
+      "learning_rate": 0.0001,
+      "loss": 0.0612,
+      "step": 19610
+    },
+    {
+      "epoch": 55.73863636363637,
+      "grad_norm": 0.909506618976593,
+      "learning_rate": 0.0001,
+      "loss": 0.0622,
+      "step": 19620
+    },
+    {
+      "epoch": 55.76704545454545,
+      "grad_norm": 0.746269166469574,
+      "learning_rate": 0.0001,
+      "loss": 0.0613,
+      "step": 19630
+    },
+    {
+      "epoch": 55.79545454545455,
+      "grad_norm": 0.6903102993965149,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 19640
+    },
+    {
+      "epoch": 55.82386363636363,
+      "grad_norm": 0.7787826061248779,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 19650
+    },
+    {
+      "epoch": 55.85227272727273,
+      "grad_norm": 0.8575695753097534,
+      "learning_rate": 0.0001,
+      "loss": 0.061,
+      "step": 19660
+    },
+    {
+      "epoch": 55.88068181818182,
+      "grad_norm": 1.386139154434204,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 19670
+    },
+    {
+      "epoch": 55.90909090909091,
+      "grad_norm": 1.4138883352279663,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "step": 19680
+    },
+    {
+      "epoch": 55.9375,
+      "grad_norm": 1.166518211364746,
+      "learning_rate": 0.0001,
+      "loss": 0.0607,
+      "step": 19690
+    },
+    {
+      "epoch": 55.96590909090909,
+      "grad_norm": 1.1394963264465332,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 19700
+    },
+    {
+      "epoch": 55.99431818181818,
+      "grad_norm": 1.0092238187789917,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 19710
+    },
+    {
+      "epoch": 56.02272727272727,
+      "grad_norm": 1.2652308940887451,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "step": 19720
+    },
+    {
+      "epoch": 56.05113636363637,
+      "grad_norm": 1.0221445560455322,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 19730
+    },
+    {
+      "epoch": 56.07954545454545,
+      "grad_norm": 1.1227253675460815,
+      "learning_rate": 0.0001,
+      "loss": 0.0593,
+      "step": 19740
+    },
+    {
+      "epoch": 56.10795454545455,
+      "grad_norm": 1.105058193206787,
+      "learning_rate": 0.0001,
+      "loss": 0.0584,
+      "step": 19750
+    },
+    {
+      "epoch": 56.13636363636363,
+      "grad_norm": 0.9531204104423523,
+      "learning_rate": 0.0001,
+      "loss": 0.0614,
+      "step": 19760
+    },
+    {
+      "epoch": 56.16477272727273,
+      "grad_norm": 0.7669575214385986,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 19770
+    },
+    {
+      "epoch": 56.19318181818182,
+      "grad_norm": 1.0532715320587158,
+      "learning_rate": 0.0001,
+      "loss": 0.0572,
+      "step": 19780
+    },
+    {
+      "epoch": 56.22159090909091,
+      "grad_norm": 1.0322656631469727,
+      "learning_rate": 0.0001,
+      "loss": 0.0596,
+      "step": 19790
+    },
+    {
+      "epoch": 56.25,
+      "grad_norm": 1.2287739515304565,
+      "learning_rate": 0.0001,
+      "loss": 0.0567,
+      "step": 19800
+    },
+    {
+      "epoch": 56.27840909090909,
+      "grad_norm": 1.2664307355880737,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 19810
+    },
+    {
+      "epoch": 56.30681818181818,
+      "grad_norm": 1.059367060661316,
+      "learning_rate": 0.0001,
+      "loss": 0.0595,
+      "step": 19820
+    },
+    {
+      "epoch": 56.33522727272727,
+      "grad_norm": 0.7184119820594788,
+      "learning_rate": 0.0001,
+      "loss": 0.0607,
+      "step": 19830
+    },
+    {
+      "epoch": 56.36363636363637,
+      "grad_norm": 0.996780276298523,
+      "learning_rate": 0.0001,
+      "loss": 0.0594,
+      "step": 19840
+    },
+    {
+      "epoch": 56.39204545454545,
+      "grad_norm": 0.8815504312515259,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 19850
+    },
+    {
+      "epoch": 56.42045454545455,
+      "grad_norm": 1.0592563152313232,
+      "learning_rate": 0.0001,
+      "loss": 0.0595,
+      "step": 19860
+    },
+    {
+      "epoch": 56.44886363636363,
+      "grad_norm": 0.7666848301887512,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 19870
+    },
+    {
+      "epoch": 56.47727272727273,
+      "grad_norm": 0.7527984976768494,
+      "learning_rate": 0.0001,
+      "loss": 0.061,
+      "step": 19880
+    },
+    {
+      "epoch": 56.50568181818182,
+      "grad_norm": 1.3445789813995361,
+      "learning_rate": 0.0001,
+      "loss": 0.0625,
+      "step": 19890
+    },
+    {
+      "epoch": 56.53409090909091,
+      "grad_norm": 1.4454388618469238,
+      "learning_rate": 0.0001,
+      "loss": 0.0613,
+      "step": 19900
+    },
+    {
+      "epoch": 56.5625,
+      "grad_norm": 1.112709641456604,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 19910
+    },
+    {
+      "epoch": 56.59090909090909,
+      "grad_norm": 1.1458951234817505,
+      "learning_rate": 0.0001,
+      "loss": 0.0586,
+      "step": 19920
+    },
+    {
+      "epoch": 56.61931818181818,
+      "grad_norm": 1.526114821434021,
+      "learning_rate": 0.0001,
+      "loss": 0.059,
+      "step": 19930
+    },
+    {
+      "epoch": 56.64772727272727,
+      "grad_norm": 1.657968282699585,
+      "learning_rate": 0.0001,
+      "loss": 0.0615,
+      "step": 19940
+    },
+    {
+      "epoch": 56.67613636363637,
+      "grad_norm": 1.5407483577728271,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 19950
+    },
+    {
+      "epoch": 56.70454545454545,
+      "grad_norm": 1.186934232711792,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 19960
+    },
+    {
+      "epoch": 56.73295454545455,
+      "grad_norm": 0.9510246515274048,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 19970
+    },
+    {
+      "epoch": 56.76136363636363,
+      "grad_norm": 0.8539232611656189,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 19980
+    },
+    {
+      "epoch": 56.78977272727273,
+      "grad_norm": 1.1108481884002686,
+      "learning_rate": 0.0001,
+      "loss": 0.0539,
+      "step": 19990
+    },
+    {
+      "epoch": 56.81818181818182,
+      "grad_norm": 1.3073898553848267,
+      "learning_rate": 0.0001,
+      "loss": 0.0571,
+      "step": 20000
+    },
+    {
+      "epoch": 56.84659090909091,
+      "grad_norm": 1.0064797401428223,
+      "learning_rate": 0.0001,
+      "loss": 0.0569,
+      "step": 20010
+    },
+    {
+      "epoch": 56.875,
+      "grad_norm": 1.1430519819259644,
+      "learning_rate": 0.0001,
+      "loss": 0.058,
+      "step": 20020
+    },
+    {
+      "epoch": 56.90340909090909,
+      "grad_norm": 1.2745853662490845,
+      "learning_rate": 0.0001,
+      "loss": 0.0581,
+      "step": 20030
+    },
+    {
+      "epoch": 56.93181818181818,
+      "grad_norm": 1.700903296470642,
+      "learning_rate": 0.0001,
+      "loss": 0.0655,
+      "step": 20040
+    },
+    {
+      "epoch": 56.96022727272727,
+      "grad_norm": 1.9346566200256348,
+      "learning_rate": 0.0001,
+      "loss": 0.0595,
+      "step": 20050
+    },
+    {
+      "epoch": 56.98863636363637,
+      "grad_norm": 1.8350406885147095,
+      "learning_rate": 0.0001,
+      "loss": 0.058,
+      "step": 20060
+    },
+    {
+      "epoch": 57.01704545454545,
+      "grad_norm": 1.5323282480239868,
+      "learning_rate": 0.0001,
+      "loss": 0.0602,
+      "step": 20070
+    },
+    {
+      "epoch": 57.04545454545455,
+      "grad_norm": 1.794053077697754,
+      "learning_rate": 0.0001,
+      "loss": 0.0552,
+      "step": 20080
+    },
+    {
+      "epoch": 57.07386363636363,
+      "grad_norm": 2.6057610511779785,
+      "learning_rate": 0.0001,
+      "loss": 0.0596,
+      "step": 20090
+    },
+    {
+      "epoch": 57.10227272727273,
+      "grad_norm": 1.8744699954986572,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "step": 20100
+    },
+    {
+      "epoch": 57.13068181818182,
+      "grad_norm": 2.1725265979766846,
+      "learning_rate": 0.0001,
+      "loss": 0.0573,
+      "step": 20110
+    },
+    {
+      "epoch": 57.15909090909091,
+      "grad_norm": 1.8657861948013306,
+      "learning_rate": 0.0001,
+      "loss": 0.0546,
+      "step": 20120
+    },
+    {
+      "epoch": 57.1875,
+      "grad_norm": 4.145320415496826,
+      "learning_rate": 0.0001,
+      "loss": 0.0647,
+      "step": 20130
+    },
+    {
+      "epoch": 57.21590909090909,
+      "grad_norm": 3.5790581703186035,
+      "learning_rate": 0.0001,
+      "loss": 0.0546,
+      "step": 20140
+    },
+    {
+      "epoch": 57.24431818181818,
+      "grad_norm": 1.8389453887939453,
+      "learning_rate": 0.0001,
+      "loss": 0.0529,
+      "step": 20150
+    },
+    {
+      "epoch": 57.27272727272727,
+      "grad_norm": 1.848724603652954,
+      "learning_rate": 0.0001,
+      "loss": 0.0652,
+      "step": 20160
+    },
+    {
+      "epoch": 57.30113636363637,
+      "grad_norm": 2.9280498027801514,
+      "learning_rate": 0.0001,
+      "loss": 0.0571,
+      "step": 20170
+    },
+    {
+      "epoch": 57.32954545454545,
+      "grad_norm": 3.1636435985565186,
+      "learning_rate": 0.0001,
+      "loss": 0.0539,
+      "step": 20180
+    },
+    {
+      "epoch": 57.35795454545455,
+      "grad_norm": 1.8078699111938477,
+      "learning_rate": 0.0001,
+      "loss": 0.0524,
+      "step": 20190
+    },
+    {
+      "epoch": 57.38636363636363,
+      "grad_norm": 2.467681884765625,
+      "learning_rate": 0.0001,
+      "loss": 0.0533,
+      "step": 20200
+    },
+    {
+      "epoch": 57.41477272727273,
+      "grad_norm": 1.868058204650879,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 20210
+    },
+    {
+      "epoch": 57.44318181818182,
+      "grad_norm": 1.2513569593429565,
+      "learning_rate": 0.0001,
+      "loss": 0.0498,
+      "step": 20220
+    },
+    {
+      "epoch": 57.47159090909091,
+      "grad_norm": 1.9683401584625244,
+      "learning_rate": 0.0001,
+      "loss": 0.0507,
+      "step": 20230
+    },
+    {
+      "epoch": 57.5,
+      "grad_norm": 1.4425514936447144,
+      "learning_rate": 0.0001,
+      "loss": 0.0527,
+      "step": 20240
+    },
+    {
+      "epoch": 57.52840909090909,
+      "grad_norm": 1.2349433898925781,
+      "learning_rate": 0.0001,
+      "loss": 0.058,
+      "step": 20250
+    },
+    {
+      "epoch": 57.55681818181818,
+      "grad_norm": 1.2428669929504395,
+      "learning_rate": 0.0001,
+      "loss": 0.0541,
+      "step": 20260
+    },
+    {
+      "epoch": 57.58522727272727,
+      "grad_norm": 1.4915668964385986,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "step": 20270
+    },
+    {
+      "epoch": 57.61363636363637,
+      "grad_norm": 1.5012083053588867,
+      "learning_rate": 0.0001,
+      "loss": 0.0564,
+      "step": 20280
+    },
+    {
+      "epoch": 57.64204545454545,
+      "grad_norm": 1.287113070487976,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 20290
+    },
+    {
+      "epoch": 57.67045454545455,
+      "grad_norm": 1.545423984527588,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "step": 20300
+    },
+    {
+      "epoch": 57.69886363636363,
+      "grad_norm": 0.989124596118927,
+      "learning_rate": 0.0001,
+      "loss": 0.0579,
+      "step": 20310
+    },
+    {
+      "epoch": 57.72727272727273,
+      "grad_norm": 1.1454150676727295,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 20320
+    },
+    {
+      "epoch": 57.75568181818182,
+      "grad_norm": 1.7752538919448853,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 20330
+    },
+    {
+      "epoch": 57.78409090909091,
+      "grad_norm": 1.3004405498504639,
+      "learning_rate": 0.0001,
+      "loss": 0.0559,
+      "step": 20340
+    },
+    {
+      "epoch": 57.8125,
+      "grad_norm": 1.165421962738037,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 20350
+    },
+    {
+      "epoch": 57.84090909090909,
+      "grad_norm": 1.3419017791748047,
+      "learning_rate": 0.0001,
+      "loss": 0.0553,
+      "step": 20360
+    },
+    {
+      "epoch": 57.86931818181818,
+      "grad_norm": 1.2583837509155273,
+      "learning_rate": 0.0001,
+      "loss": 0.0574,
+      "step": 20370
+    },
+    {
+      "epoch": 57.89772727272727,
+      "grad_norm": 1.1867432594299316,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 20380
+    },
+    {
+      "epoch": 57.92613636363637,
+      "grad_norm": 1.191956639289856,
+      "learning_rate": 0.0001,
+      "loss": 0.0549,
+      "step": 20390
+    },
+    {
+      "epoch": 57.95454545454545,
+      "grad_norm": 1.0556656122207642,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 20400
+    },
+    {
+      "epoch": 57.98295454545455,
+      "grad_norm": 1.2158259153366089,
+      "learning_rate": 0.0001,
+      "loss": 0.0547,
+      "step": 20410
+    },
+    {
+      "epoch": 58.01136363636363,
+      "grad_norm": 0.9880191087722778,
+      "learning_rate": 0.0001,
+      "loss": 0.0576,
+      "step": 20420
+    },
+    {
+      "epoch": 58.03977272727273,
+      "grad_norm": 1.150399923324585,
+      "learning_rate": 0.0001,
+      "loss": 0.0589,
+      "step": 20430
+    },
+    {
+      "epoch": 58.06818181818182,
+      "grad_norm": 0.9008199572563171,
+      "learning_rate": 0.0001,
+      "loss": 0.0567,
+      "step": 20440
+    },
+    {
+      "epoch": 58.09659090909091,
+      "grad_norm": 0.8457942008972168,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 20450
+    },
+    {
+      "epoch": 58.125,
+      "grad_norm": 0.8913554549217224,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 20460
+    },
+    {
+      "epoch": 58.15340909090909,
+      "grad_norm": 1.0136390924453735,
+      "learning_rate": 0.0001,
+      "loss": 0.0612,
+      "step": 20470
+    },
+    {
+      "epoch": 58.18181818181818,
+      "grad_norm": 1.0893242359161377,
+      "learning_rate": 0.0001,
+      "loss": 0.0602,
+      "step": 20480
+    },
+    {
+      "epoch": 58.21022727272727,
+      "grad_norm": 0.8883498311042786,
+      "learning_rate": 0.0001,
+      "loss": 0.0627,
+      "step": 20490
+    },
+    {
+      "epoch": 58.23863636363637,
+      "grad_norm": 0.7249606251716614,
+      "learning_rate": 0.0001,
+      "loss": 0.0637,
+      "step": 20500
+    },
+    {
+      "epoch": 58.26704545454545,
+      "grad_norm": 0.83709716796875,
+      "learning_rate": 0.0001,
+      "loss": 0.0623,
+      "step": 20510
+    },
+    {
+      "epoch": 58.29545454545455,
+      "grad_norm": 0.8498445749282837,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 20520
+    },
+    {
+      "epoch": 58.32386363636363,
+      "grad_norm": 0.8652094602584839,
+      "learning_rate": 0.0001,
+      "loss": 0.0581,
+      "step": 20530
+    },
+    {
+      "epoch": 58.35227272727273,
+      "grad_norm": 0.933368444442749,
+      "learning_rate": 0.0001,
+      "loss": 0.0607,
+      "step": 20540
+    },
+    {
+      "epoch": 58.38068181818182,
+      "grad_norm": 1.022032380104065,
+      "learning_rate": 0.0001,
+      "loss": 0.0581,
+      "step": 20550
+    },
+    {
+      "epoch": 58.40909090909091,
+      "grad_norm": 0.884529173374176,
+      "learning_rate": 0.0001,
+      "loss": 0.0584,
+      "step": 20560
+    },
+    {
+      "epoch": 58.4375,
+      "grad_norm": 0.7771308422088623,
+      "learning_rate": 0.0001,
+      "loss": 0.0612,
+      "step": 20570
+    },
+    {
+      "epoch": 58.46590909090909,
+      "grad_norm": 1.055200457572937,
+      "learning_rate": 0.0001,
+      "loss": 0.0599,
+      "step": 20580
+    },
+    {
+      "epoch": 58.49431818181818,
+      "grad_norm": 0.9750531315803528,
+      "learning_rate": 0.0001,
+      "loss": 0.0598,
+      "step": 20590
+    },
+    {
+      "epoch": 58.52272727272727,
+      "grad_norm": 0.9744542241096497,
+      "learning_rate": 0.0001,
+      "loss": 0.0627,
+      "step": 20600
+    },
+    {
+      "epoch": 58.55113636363637,
+      "grad_norm": 1.1760064363479614,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "step": 20610
+    },
+    {
+      "epoch": 58.57954545454545,
+      "grad_norm": 1.0370769500732422,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 20620
+    },
+    {
+      "epoch": 58.60795454545455,
+      "grad_norm": 1.3704447746276855,
+      "learning_rate": 0.0001,
+      "loss": 0.0599,
+      "step": 20630
+    },
+    {
+      "epoch": 58.63636363636363,
+      "grad_norm": 1.098480463027954,
+      "learning_rate": 0.0001,
+      "loss": 0.0622,
+      "step": 20640
+    },
+    {
+      "epoch": 58.66477272727273,
+      "grad_norm": 1.4485833644866943,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 20650
+    },
+    {
+      "epoch": 58.69318181818182,
+      "grad_norm": 1.1087062358856201,
+      "learning_rate": 0.0001,
+      "loss": 0.0618,
+      "step": 20660
+    },
+    {
+      "epoch": 58.72159090909091,
+      "grad_norm": 1.0735747814178467,
+      "learning_rate": 0.0001,
+      "loss": 0.0629,
+      "step": 20670
+    },
+    {
+      "epoch": 58.75,
+      "grad_norm": 1.2116317749023438,
+      "learning_rate": 0.0001,
+      "loss": 0.0578,
+      "step": 20680
+    },
+    {
+      "epoch": 58.77840909090909,
+      "grad_norm": 0.8660669922828674,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 20690
+    },
+    {
+      "epoch": 58.80681818181818,
+      "grad_norm": 0.9465160369873047,
+      "learning_rate": 0.0001,
+      "loss": 0.0602,
+      "step": 20700
+    },
+    {
+      "epoch": 58.83522727272727,
+      "grad_norm": 1.117857575416565,
+      "learning_rate": 0.0001,
+      "loss": 0.0577,
+      "step": 20710
+    },
+    {
+      "epoch": 58.86363636363637,
+      "grad_norm": 1.4936555624008179,
+      "learning_rate": 0.0001,
+      "loss": 0.064,
+      "step": 20720
+    },
+    {
+      "epoch": 58.89204545454545,
+      "grad_norm": 1.6041721105575562,
+      "learning_rate": 0.0001,
+      "loss": 0.0621,
+      "step": 20730
+    },
+    {
+      "epoch": 58.92045454545455,
+      "grad_norm": 1.4327595233917236,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 20740
+    },
+    {
+      "epoch": 58.94886363636363,
+      "grad_norm": 1.3365850448608398,
+      "learning_rate": 0.0001,
+      "loss": 0.0582,
+      "step": 20750
+    },
+    {
+      "epoch": 58.97727272727273,
+      "grad_norm": 1.3097879886627197,
+      "learning_rate": 0.0001,
+      "loss": 0.0568,
+      "step": 20760
+    },
+    {
+      "epoch": 59.00568181818182,
+      "grad_norm": 1.252744436264038,
+      "learning_rate": 0.0001,
+      "loss": 0.0598,
+      "step": 20770
+    },
+    {
+      "epoch": 59.03409090909091,
+      "grad_norm": 0.9207512140274048,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 20780
+    },
+    {
+      "epoch": 59.0625,
+      "grad_norm": 1.2421956062316895,
+      "learning_rate": 0.0001,
+      "loss": 0.0568,
+      "step": 20790
+    },
+    {
+      "epoch": 59.09090909090909,
+      "grad_norm": 0.8143938183784485,
+      "learning_rate": 0.0001,
+      "loss": 0.0578,
+      "step": 20800
+    },
+    {
+      "epoch": 59.11931818181818,
+      "grad_norm": 0.9497162103652954,
+      "learning_rate": 0.0001,
+      "loss": 0.0563,
+      "step": 20810
+    },
+    {
+      "epoch": 59.14772727272727,
+      "grad_norm": 1.0337462425231934,
+      "learning_rate": 0.0001,
+      "loss": 0.0572,
+      "step": 20820
+    },
+    {
+      "epoch": 59.17613636363637,
+      "grad_norm": 0.9024602770805359,
+      "learning_rate": 0.0001,
+      "loss": 0.0586,
+      "step": 20830
+    },
+    {
+      "epoch": 59.20454545454545,
+      "grad_norm": 0.9900998473167419,
+      "learning_rate": 0.0001,
+      "loss": 0.058,
+      "step": 20840
+    },
+    {
+      "epoch": 59.23295454545455,
+      "grad_norm": 0.9825519919395447,
+      "learning_rate": 0.0001,
+      "loss": 0.0619,
+      "step": 20850
+    },
+    {
+      "epoch": 59.26136363636363,
+      "grad_norm": 0.9931562542915344,
+      "learning_rate": 0.0001,
+      "loss": 0.0606,
+      "step": 20860
+    },
+    {
+      "epoch": 59.28977272727273,
+      "grad_norm": 0.8159868121147156,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 20870
+    },
+    {
+      "epoch": 59.31818181818182,
+      "grad_norm": 1.0349278450012207,
+      "learning_rate": 0.0001,
+      "loss": 0.0622,
+      "step": 20880
+    },
+    {
+      "epoch": 59.34659090909091,
+      "grad_norm": 0.8769015073776245,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 20890
+    },
+    {
+      "epoch": 59.375,
+      "grad_norm": 0.94813472032547,
+      "learning_rate": 0.0001,
+      "loss": 0.0591,
+      "step": 20900
+    },
+    {
+      "epoch": 59.40340909090909,
+      "grad_norm": 1.1130449771881104,
+      "learning_rate": 0.0001,
+      "loss": 0.0608,
+      "step": 20910
+    },
+    {
+      "epoch": 59.43181818181818,
+      "grad_norm": 0.9908705353736877,
+      "learning_rate": 0.0001,
+      "loss": 0.0595,
+      "step": 20920
+    },
+    {
+      "epoch": 59.46022727272727,
+      "grad_norm": 1.1722239255905151,
+      "learning_rate": 0.0001,
+      "loss": 0.0589,
+      "step": 20930
+    },
+    {
+      "epoch": 59.48863636363637,
+      "grad_norm": 1.0646426677703857,
+      "learning_rate": 0.0001,
+      "loss": 0.0617,
+      "step": 20940
+    },
+    {
+      "epoch": 59.51704545454545,
+      "grad_norm": 0.9773575067520142,
+      "learning_rate": 0.0001,
+      "loss": 0.0584,
+      "step": 20950
+    },
+    {
+      "epoch": 59.54545454545455,
+      "grad_norm": 0.9211640954017639,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 20960
+    },
+    {
+      "epoch": 59.57386363636363,
+      "grad_norm": 0.7611501216888428,
+      "learning_rate": 0.0001,
+      "loss": 0.0589,
+      "step": 20970
+    },
+    {
+      "epoch": 59.60227272727273,
+      "grad_norm": 0.877237856388092,
+      "learning_rate": 0.0001,
+      "loss": 0.0615,
+      "step": 20980
+    },
+    {
+      "epoch": 59.63068181818182,
+      "grad_norm": 0.921630322933197,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 20990
+    },
+    {
+      "epoch": 59.65909090909091,
+      "grad_norm": 0.9630839824676514,
+      "learning_rate": 0.0001,
+      "loss": 0.0589,
+      "step": 21000
+    },
+    {
+      "epoch": 59.6875,
+      "grad_norm": 0.9061483144760132,
+      "learning_rate": 0.0001,
+      "loss": 0.0584,
+      "step": 21010
+    },
+    {
+      "epoch": 59.71590909090909,
+      "grad_norm": 0.847222626209259,
+      "learning_rate": 0.0001,
+      "loss": 0.0651,
+      "step": 21020
+    },
+    {
+      "epoch": 59.74431818181818,
+      "grad_norm": 0.8466194868087769,
+      "learning_rate": 0.0001,
+      "loss": 0.0589,
+      "step": 21030
+    },
+    {
+      "epoch": 59.77272727272727,
+      "grad_norm": 1.054270625114441,
+      "learning_rate": 0.0001,
+      "loss": 0.0598,
+      "step": 21040
+    },
+    {
+      "epoch": 59.80113636363637,
+      "grad_norm": 1.2087162733078003,
+      "learning_rate": 0.0001,
+      "loss": 0.0594,
+      "step": 21050
+    },
+    {
+      "epoch": 59.82954545454545,
+      "grad_norm": 1.2614654302597046,
+      "learning_rate": 0.0001,
+      "loss": 0.06,
+      "step": 21060
+    },
+    {
+      "epoch": 59.85795454545455,
+      "grad_norm": 1.2037792205810547,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 21070
+    },
+    {
+      "epoch": 59.88636363636363,
+      "grad_norm": 1.1717537641525269,
+      "learning_rate": 0.0001,
+      "loss": 0.0596,
+      "step": 21080
+    },
+    {
+      "epoch": 59.91477272727273,
+      "grad_norm": 1.085711121559143,
+      "learning_rate": 0.0001,
+      "loss": 0.0579,
+      "step": 21090
+    },
+    {
+      "epoch": 59.94318181818182,
+      "grad_norm": 0.8551573753356934,
+      "learning_rate": 0.0001,
+      "loss": 0.0599,
+      "step": 21100
+    },
+    {
+      "epoch": 59.97159090909091,
+      "grad_norm": 0.7682262063026428,
+      "learning_rate": 0.0001,
+      "loss": 0.0595,
+      "step": 21110
+    },
+    {
+      "epoch": 60.0,
+      "grad_norm": 0.6982107162475586,
+      "learning_rate": 0.0001,
+      "loss": 0.0619,
+      "step": 21120
+    },
+    {
+      "epoch": 60.02840909090909,
+      "grad_norm": 0.9472159147262573,
+      "learning_rate": 0.0001,
+      "loss": 0.0614,
+      "step": 21130
+    },
+    {
+      "epoch": 60.05681818181818,
+      "grad_norm": 0.8628683686256409,
+      "learning_rate": 0.0001,
+      "loss": 0.0609,
+      "step": 21140
+    },
+    {
+      "epoch": 60.08522727272727,
+      "grad_norm": 0.6452422142028809,
+      "learning_rate": 0.0001,
+      "loss": 0.0622,
+      "step": 21150
+    },
+    {
+      "epoch": 60.11363636363637,
+      "grad_norm": 0.6607347726821899,
+      "learning_rate": 0.0001,
+      "loss": 0.0611,
+      "step": 21160
+    },
+    {
+      "epoch": 60.14204545454545,
+      "grad_norm": 0.6297292709350586,
+      "learning_rate": 0.0001,
+      "loss": 0.06,
+      "step": 21170
+    },
+    {
+      "epoch": 60.17045454545455,
+      "grad_norm": 0.705437958240509,
+      "learning_rate": 0.0001,
+      "loss": 0.0598,
+      "step": 21180
+    },
+    {
+      "epoch": 60.19886363636363,
+      "grad_norm": 0.64570552110672,
+      "learning_rate": 0.0001,
+      "loss": 0.0579,
+      "step": 21190
+    },
+    {
+      "epoch": 60.22727272727273,
+      "grad_norm": 0.8154585361480713,
+      "learning_rate": 0.0001,
+      "loss": 0.0557,
+      "step": 21200
+    },
+    {
+      "epoch": 60.25568181818182,
+      "grad_norm": 0.8044834136962891,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 21210
+    },
+    {
+      "epoch": 60.28409090909091,
+      "grad_norm": 0.984665036201477,
+      "learning_rate": 0.0001,
+      "loss": 0.0582,
+      "step": 21220
+    },
+    {
+      "epoch": 60.3125,
+      "grad_norm": 0.8446553349494934,
+      "learning_rate": 0.0001,
+      "loss": 0.0591,
+      "step": 21230
+    },
+    {
+      "epoch": 60.34090909090909,
+      "grad_norm": 0.9261309504508972,
+      "learning_rate": 0.0001,
+      "loss": 0.0607,
+      "step": 21240
+    },
+    {
+      "epoch": 60.36931818181818,
+      "grad_norm": 1.1594133377075195,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 21250
+    },
+    {
+      "epoch": 60.39772727272727,
+      "grad_norm": 1.2541656494140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0568,
+      "step": 21260
+    },
+    {
+      "epoch": 60.42613636363637,
+      "grad_norm": 0.9860923290252686,
+      "learning_rate": 0.0001,
+      "loss": 0.0604,
+      "step": 21270
+    },
+    {
+      "epoch": 60.45454545454545,
+      "grad_norm": 1.032243251800537,
+      "learning_rate": 0.0001,
+      "loss": 0.0572,
+      "step": 21280
+    },
+    {
+      "epoch": 60.48295454545455,
+      "grad_norm": 1.0910956859588623,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 21290
+    },
+    {
+      "epoch": 60.51136363636363,
+      "grad_norm": 1.106114387512207,
+      "learning_rate": 0.0001,
+      "loss": 0.0581,
+      "step": 21300
+    },
+    {
+      "epoch": 60.53977272727273,
+      "grad_norm": 1.733173131942749,
+      "learning_rate": 0.0001,
+      "loss": 0.0624,
+      "step": 21310
+    },
+    {
+      "epoch": 60.56818181818182,
+      "grad_norm": 1.755391001701355,
+      "learning_rate": 0.0001,
+      "loss": 0.0586,
+      "step": 21320
+    },
+    {
+      "epoch": 60.59659090909091,
+      "grad_norm": 1.6064823865890503,
+      "learning_rate": 0.0001,
+      "loss": 0.0564,
+      "step": 21330
+    },
+    {
+      "epoch": 60.625,
+      "grad_norm": 1.0228577852249146,
+      "learning_rate": 0.0001,
+      "loss": 0.0563,
+      "step": 21340
+    },
+    {
+      "epoch": 60.65340909090909,
+      "grad_norm": 1.1767072677612305,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 21350
+    },
+    {
+      "epoch": 60.68181818181818,
+      "grad_norm": 0.9804391264915466,
+      "learning_rate": 0.0001,
+      "loss": 0.0564,
+      "step": 21360
+    },
+    {
+      "epoch": 60.71022727272727,
+      "grad_norm": 1.459820032119751,
+      "learning_rate": 0.0001,
+      "loss": 0.0549,
+      "step": 21370
+    },
+    {
+      "epoch": 60.73863636363637,
+      "grad_norm": 1.2355256080627441,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 21380
+    },
+    {
+      "epoch": 60.76704545454545,
+      "grad_norm": 0.9702253341674805,
+      "learning_rate": 0.0001,
+      "loss": 0.0582,
+      "step": 21390
+    },
+    {
+      "epoch": 60.79545454545455,
+      "grad_norm": 0.8217170834541321,
+      "learning_rate": 0.0001,
+      "loss": 0.0577,
+      "step": 21400
+    },
+    {
+      "epoch": 60.82386363636363,
+      "grad_norm": 1.1219531297683716,
+      "learning_rate": 0.0001,
+      "loss": 0.0571,
+      "step": 21410
+    },
+    {
+      "epoch": 60.85227272727273,
+      "grad_norm": 1.0550838708877563,
+      "learning_rate": 0.0001,
+      "loss": 0.0573,
+      "step": 21420
+    },
+    {
+      "epoch": 60.88068181818182,
+      "grad_norm": 1.1298226118087769,
+      "learning_rate": 0.0001,
+      "loss": 0.0567,
+      "step": 21430
+    },
+    {
+      "epoch": 60.90909090909091,
+      "grad_norm": 1.1814019680023193,
+      "learning_rate": 0.0001,
+      "loss": 0.0602,
+      "step": 21440
+    },
+    {
+      "epoch": 60.9375,
+      "grad_norm": 1.0315929651260376,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 21450
+    },
+    {
+      "epoch": 60.96590909090909,
+      "grad_norm": 1.0429394245147705,
+      "learning_rate": 0.0001,
+      "loss": 0.0576,
+      "step": 21460
+    },
+    {
+      "epoch": 60.99431818181818,
+      "grad_norm": 1.109660029411316,
+      "learning_rate": 0.0001,
+      "loss": 0.0579,
+      "step": 21470
+    },
+    {
+      "epoch": 61.02272727272727,
+      "grad_norm": 0.8597354292869568,
+      "learning_rate": 0.0001,
+      "loss": 0.0553,
+      "step": 21480
+    },
+    {
+      "epoch": 61.05113636363637,
+      "grad_norm": 0.7767676711082458,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 21490
+    },
+    {
+      "epoch": 61.07954545454545,
+      "grad_norm": 0.9464530944824219,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 21500
+    },
+    {
+      "epoch": 61.10795454545455,
+      "grad_norm": 0.8091188073158264,
+      "learning_rate": 0.0001,
+      "loss": 0.0596,
+      "step": 21510
+    },
+    {
+      "epoch": 61.13636363636363,
+      "grad_norm": 0.9647312760353088,
+      "learning_rate": 0.0001,
+      "loss": 0.0556,
+      "step": 21520
+    },
+    {
+      "epoch": 61.16477272727273,
+      "grad_norm": 1.1387494802474976,
+      "learning_rate": 0.0001,
+      "loss": 0.0577,
+      "step": 21530
+    },
+    {
+      "epoch": 61.19318181818182,
+      "grad_norm": 0.8227630853652954,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 21540
+    },
+    {
+      "epoch": 61.22159090909091,
+      "grad_norm": 1.2130613327026367,
+      "learning_rate": 0.0001,
+      "loss": 0.0565,
+      "step": 21550
+    },
+    {
+      "epoch": 61.25,
+      "grad_norm": 1.1565511226654053,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 21560
+    },
+    {
+      "epoch": 61.27840909090909,
+      "grad_norm": 1.049648404121399,
+      "learning_rate": 0.0001,
+      "loss": 0.0556,
+      "step": 21570
+    },
+    {
+      "epoch": 61.30681818181818,
+      "grad_norm": 1.279056429862976,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 21580
+    },
+    {
+      "epoch": 61.33522727272727,
+      "grad_norm": 0.8837600350379944,
+      "learning_rate": 0.0001,
+      "loss": 0.0579,
+      "step": 21590
+    },
+    {
+      "epoch": 61.36363636363637,
+      "grad_norm": 0.9417069554328918,
+      "learning_rate": 0.0001,
+      "loss": 0.0574,
+      "step": 21600
+    },
+    {
+      "epoch": 61.39204545454545,
+      "grad_norm": 0.7844614386558533,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 21610
+    },
+    {
+      "epoch": 61.42045454545455,
+      "grad_norm": 0.9132207632064819,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 21620
+    },
+    {
+      "epoch": 61.44886363636363,
+      "grad_norm": 0.8967658877372742,
+      "learning_rate": 0.0001,
+      "loss": 0.0568,
+      "step": 21630
+    },
+    {
+      "epoch": 61.47727272727273,
+      "grad_norm": 0.8098888397216797,
+      "learning_rate": 0.0001,
+      "loss": 0.0599,
+      "step": 21640
+    },
+    {
+      "epoch": 61.50568181818182,
+      "grad_norm": 0.856517493724823,
+      "learning_rate": 0.0001,
+      "loss": 0.0556,
+      "step": 21650
+    },
+    {
+      "epoch": 61.53409090909091,
+      "grad_norm": 0.7805210947990417,
+      "learning_rate": 0.0001,
+      "loss": 0.0603,
+      "step": 21660
+    },
+    {
+      "epoch": 61.5625,
+      "grad_norm": 0.8382397294044495,
+      "learning_rate": 0.0001,
+      "loss": 0.0578,
+      "step": 21670
+    },
+    {
+      "epoch": 61.59090909090909,
+      "grad_norm": 0.9148212671279907,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 21680
+    },
+    {
+      "epoch": 61.61931818181818,
+      "grad_norm": 0.774455726146698,
+      "learning_rate": 0.0001,
+      "loss": 0.0573,
+      "step": 21690
+    },
+    {
+      "epoch": 61.64772727272727,
+      "grad_norm": 0.6750848889350891,
+      "learning_rate": 0.0001,
+      "loss": 0.0559,
+      "step": 21700
+    },
+    {
+      "epoch": 61.67613636363637,
+      "grad_norm": 0.7105973362922668,
+      "learning_rate": 0.0001,
+      "loss": 0.0583,
+      "step": 21710
+    },
+    {
+      "epoch": 61.70454545454545,
+      "grad_norm": 1.188941240310669,
+      "learning_rate": 0.0001,
+      "loss": 0.0608,
+      "step": 21720
+    },
+    {
+      "epoch": 61.73295454545455,
+      "grad_norm": 1.0456372499465942,
+      "learning_rate": 0.0001,
+      "loss": 0.0599,
+      "step": 21730
+    },
+    {
+      "epoch": 61.76136363636363,
+      "grad_norm": 0.9662376642227173,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 21740
+    },
+    {
+      "epoch": 61.78977272727273,
+      "grad_norm": 1.0228948593139648,
+      "learning_rate": 0.0001,
+      "loss": 0.0577,
+      "step": 21750
+    },
+    {
+      "epoch": 61.81818181818182,
+      "grad_norm": 1.133011817932129,
+      "learning_rate": 0.0001,
+      "loss": 0.0564,
+      "step": 21760
+    },
+    {
+      "epoch": 61.84659090909091,
+      "grad_norm": 1.138669729232788,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 21770
+    },
+    {
+      "epoch": 61.875,
+      "grad_norm": 0.9990003705024719,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "step": 21780
+    },
+    {
+      "epoch": 61.90340909090909,
+      "grad_norm": 0.8538486957550049,
+      "learning_rate": 0.0001,
+      "loss": 0.057,
+      "step": 21790
+    },
+    {
+      "epoch": 61.93181818181818,
+      "grad_norm": 0.9877942204475403,
+      "learning_rate": 0.0001,
+      "loss": 0.0582,
+      "step": 21800
+    },
+    {
+      "epoch": 61.96022727272727,
+      "grad_norm": 0.7974020838737488,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 21810
+    },
+    {
+      "epoch": 61.98863636363637,
+      "grad_norm": 0.9531463980674744,
+      "learning_rate": 0.0001,
+      "loss": 0.0569,
+      "step": 21820
+    },
+    {
+      "epoch": 62.01704545454545,
+      "grad_norm": 0.8213363289833069,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 21830
+    },
+    {
+      "epoch": 62.04545454545455,
+      "grad_norm": 0.7434073686599731,
+      "learning_rate": 0.0001,
+      "loss": 0.0608,
+      "step": 21840
+    },
+    {
+      "epoch": 62.07386363636363,
+      "grad_norm": 1.0179238319396973,
+      "learning_rate": 0.0001,
+      "loss": 0.057,
+      "step": 21850
+    },
+    {
+      "epoch": 62.10227272727273,
+      "grad_norm": 0.8162310719490051,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "step": 21860
+    },
+    {
+      "epoch": 62.13068181818182,
+      "grad_norm": 0.749879777431488,
+      "learning_rate": 0.0001,
+      "loss": 0.0587,
+      "step": 21870
+    },
+    {
+      "epoch": 62.15909090909091,
+      "grad_norm": 0.7732171416282654,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 21880
+    },
+    {
+      "epoch": 62.1875,
+      "grad_norm": 0.7089868783950806,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 21890
+    },
+    {
+      "epoch": 62.21590909090909,
+      "grad_norm": 0.7101325988769531,
+      "learning_rate": 0.0001,
+      "loss": 0.0546,
+      "step": 21900
+    },
+    {
+      "epoch": 62.24431818181818,
+      "grad_norm": 0.7819742560386658,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 21910
+    },
+    {
+      "epoch": 62.27272727272727,
+      "grad_norm": 0.6902415752410889,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 21920
+    },
+    {
+      "epoch": 62.30113636363637,
+      "grad_norm": 0.8869762420654297,
+      "learning_rate": 0.0001,
+      "loss": 0.0582,
+      "step": 21930
+    },
+    {
+      "epoch": 62.32954545454545,
+      "grad_norm": 1.154178500175476,
+      "learning_rate": 0.0001,
+      "loss": 0.0571,
+      "step": 21940
+    },
+    {
+      "epoch": 62.35795454545455,
+      "grad_norm": 0.8612966537475586,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 21950
+    },
+    {
+      "epoch": 62.38636363636363,
+      "grad_norm": 0.8771665692329407,
+      "learning_rate": 0.0001,
+      "loss": 0.0573,
+      "step": 21960
+    },
+    {
+      "epoch": 62.41477272727273,
+      "grad_norm": 0.9387429356575012,
+      "learning_rate": 0.0001,
+      "loss": 0.0555,
+      "step": 21970
+    },
+    {
+      "epoch": 62.44318181818182,
+      "grad_norm": 0.7586554884910583,
+      "learning_rate": 0.0001,
+      "loss": 0.0597,
+      "step": 21980
+    },
+    {
+      "epoch": 62.47159090909091,
+      "grad_norm": 0.809971809387207,
+      "learning_rate": 0.0001,
+      "loss": 0.0588,
+      "step": 21990
+    },
+    {
+      "epoch": 62.5,
+      "grad_norm": 0.9225670099258423,
+      "learning_rate": 0.0001,
+      "loss": 0.0577,
+      "step": 22000
+    },
+    {
+      "epoch": 62.52840909090909,
+      "grad_norm": 0.9746827483177185,
+      "learning_rate": 0.0001,
+      "loss": 0.0559,
+      "step": 22010
+    },
+    {
+      "epoch": 62.55681818181818,
+      "grad_norm": 0.8748590350151062,
+      "learning_rate": 0.0001,
+      "loss": 0.0565,
+      "step": 22020
+    },
+    {
+      "epoch": 62.58522727272727,
+      "grad_norm": 1.0417462587356567,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 22030
+    },
+    {
+      "epoch": 62.61363636363637,
+      "grad_norm": 0.9173468351364136,
+      "learning_rate": 0.0001,
+      "loss": 0.0555,
+      "step": 22040
+    },
+    {
+      "epoch": 62.64204545454545,
+      "grad_norm": 0.9043504595756531,
+      "learning_rate": 0.0001,
+      "loss": 0.0533,
+      "step": 22050
+    },
+    {
+      "epoch": 62.67045454545455,
+      "grad_norm": 0.6783923506736755,
+      "learning_rate": 0.0001,
+      "loss": 0.0563,
+      "step": 22060
+    },
+    {
+      "epoch": 62.69886363636363,
+      "grad_norm": 0.6317957043647766,
+      "learning_rate": 0.0001,
+      "loss": 0.0539,
+      "step": 22070
+    },
+    {
+      "epoch": 62.72727272727273,
+      "grad_norm": 0.7598891258239746,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "step": 22080
+    },
+    {
+      "epoch": 62.75568181818182,
+      "grad_norm": 0.880479633808136,
+      "learning_rate": 0.0001,
+      "loss": 0.0572,
+      "step": 22090
+    },
+    {
+      "epoch": 62.78409090909091,
+      "grad_norm": 0.8354278802871704,
+      "learning_rate": 0.0001,
+      "loss": 0.0565,
+      "step": 22100
+    },
+    {
+      "epoch": 62.8125,
+      "grad_norm": 0.9944140911102295,
+      "learning_rate": 0.0001,
+      "loss": 0.0524,
+      "step": 22110
+    },
+    {
+      "epoch": 62.84090909090909,
+      "grad_norm": 0.9438153505325317,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 22120
+    },
+    {
+      "epoch": 62.86931818181818,
+      "grad_norm": 1.1514655351638794,
+      "learning_rate": 0.0001,
+      "loss": 0.0561,
+      "step": 22130
+    },
+    {
+      "epoch": 62.89772727272727,
+      "grad_norm": 1.0064482688903809,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 22140
+    },
+    {
+      "epoch": 62.92613636363637,
+      "grad_norm": 0.8011857271194458,
+      "learning_rate": 0.0001,
+      "loss": 0.0557,
+      "step": 22150
+    },
+    {
+      "epoch": 62.95454545454545,
+      "grad_norm": 0.7131310105323792,
+      "learning_rate": 0.0001,
+      "loss": 0.057,
+      "step": 22160
+    },
+    {
+      "epoch": 62.98295454545455,
+      "grad_norm": 0.7730833292007446,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 22170
+    },
+    {
+      "epoch": 63.01136363636363,
+      "grad_norm": 0.8114839792251587,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 22180
+    },
+    {
+      "epoch": 63.03977272727273,
+      "grad_norm": 0.714423656463623,
+      "learning_rate": 0.0001,
+      "loss": 0.0564,
+      "step": 22190
+    },
+    {
+      "epoch": 63.06818181818182,
+      "grad_norm": 1.106858730316162,
+      "learning_rate": 0.0001,
+      "loss": 0.0556,
+      "step": 22200
+    },
+    {
+      "epoch": 63.09659090909091,
+      "grad_norm": 1.0788785219192505,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 22210
+    },
+    {
+      "epoch": 63.125,
+      "grad_norm": 1.0309641361236572,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 22220
+    },
+    {
+      "epoch": 63.15340909090909,
+      "grad_norm": 0.9370083808898926,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 22230
+    },
+    {
+      "epoch": 63.18181818181818,
+      "grad_norm": 1.0266667604446411,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "step": 22240
+    },
+    {
+      "epoch": 63.21022727272727,
+      "grad_norm": 1.0693178176879883,
+      "learning_rate": 0.0001,
+      "loss": 0.0543,
+      "step": 22250
+    },
+    {
+      "epoch": 63.23863636363637,
+      "grad_norm": 1.409576416015625,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "step": 22260
+    },
+    {
+      "epoch": 63.26704545454545,
+      "grad_norm": 1.0419577360153198,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 22270
+    },
+    {
+      "epoch": 63.29545454545455,
+      "grad_norm": 0.9042021632194519,
+      "learning_rate": 0.0001,
+      "loss": 0.0537,
+      "step": 22280
+    },
+    {
+      "epoch": 63.32386363636363,
+      "grad_norm": 0.9674776196479797,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 22290
+    },
+    {
+      "epoch": 63.35227272727273,
+      "grad_norm": 1.08954918384552,
+      "learning_rate": 0.0001,
+      "loss": 0.0549,
+      "step": 22300
+    },
+    {
+      "epoch": 63.38068181818182,
+      "grad_norm": 1.0163633823394775,
+      "learning_rate": 0.0001,
+      "loss": 0.0553,
+      "step": 22310
+    },
+    {
+      "epoch": 63.40909090909091,
+      "grad_norm": 1.0497983694076538,
+      "learning_rate": 0.0001,
+      "loss": 0.0553,
+      "step": 22320
+    },
+    {
+      "epoch": 63.4375,
+      "grad_norm": 0.7507213354110718,
+      "learning_rate": 0.0001,
+      "loss": 0.0543,
+      "step": 22330
+    },
+    {
+      "epoch": 63.46590909090909,
+      "grad_norm": 0.9704498648643494,
+      "learning_rate": 0.0001,
+      "loss": 0.0542,
+      "step": 22340
+    },
+    {
+      "epoch": 63.49431818181818,
+      "grad_norm": 0.9345890283584595,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 22350
+    },
+    {
+      "epoch": 63.52272727272727,
+      "grad_norm": 0.8323131799697876,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 22360
+    },
+    {
+      "epoch": 63.55113636363637,
+      "grad_norm": 0.8425998687744141,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 22370
+    },
+    {
+      "epoch": 63.57954545454545,
+      "grad_norm": 0.7790317535400391,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "step": 22380
+    },
+    {
+      "epoch": 63.60795454545455,
+      "grad_norm": 0.679909348487854,
+      "learning_rate": 0.0001,
+      "loss": 0.0564,
+      "step": 22390
+    },
+    {
+      "epoch": 63.63636363636363,
+      "grad_norm": 0.6957236528396606,
+      "learning_rate": 0.0001,
+      "loss": 0.0521,
+      "step": 22400
+    },
+    {
+      "epoch": 63.66477272727273,
+      "grad_norm": 0.7631174325942993,
+      "learning_rate": 0.0001,
+      "loss": 0.0539,
+      "step": 22410
+    },
+    {
+      "epoch": 63.69318181818182,
+      "grad_norm": 0.8418110609054565,
+      "learning_rate": 0.0001,
+      "loss": 0.0587,
+      "step": 22420
+    },
+    {
+      "epoch": 63.72159090909091,
+      "grad_norm": 0.9006642699241638,
+      "learning_rate": 0.0001,
+      "loss": 0.0548,
+      "step": 22430
+    },
+    {
+      "epoch": 63.75,
+      "grad_norm": 1.0244066715240479,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "step": 22440
+    },
+    {
+      "epoch": 63.77840909090909,
+      "grad_norm": 0.8364577293395996,
+      "learning_rate": 0.0001,
+      "loss": 0.0561,
+      "step": 22450
+    },
+    {
+      "epoch": 63.80681818181818,
+      "grad_norm": 1.0565218925476074,
+      "learning_rate": 0.0001,
+      "loss": 0.0575,
+      "step": 22460
+    },
+    {
+      "epoch": 63.83522727272727,
+      "grad_norm": 0.994922935962677,
+      "learning_rate": 0.0001,
+      "loss": 0.0564,
+      "step": 22470
+    },
+    {
+      "epoch": 63.86363636363637,
+      "grad_norm": 1.0308321714401245,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 22480
+    },
+    {
+      "epoch": 63.89204545454545,
+      "grad_norm": 0.9732064604759216,
+      "learning_rate": 0.0001,
+      "loss": 0.0537,
+      "step": 22490
+    },
+    {
+      "epoch": 63.92045454545455,
+      "grad_norm": 0.8068335056304932,
+      "learning_rate": 0.0001,
+      "loss": 0.0563,
+      "step": 22500
+    },
+    {
+      "epoch": 63.94886363636363,
+      "grad_norm": 0.7390735745429993,
+      "learning_rate": 0.0001,
+      "loss": 0.0569,
+      "step": 22510
+    },
+    {
+      "epoch": 63.97727272727273,
+      "grad_norm": 0.7898790240287781,
+      "learning_rate": 0.0001,
+      "loss": 0.0552,
+      "step": 22520
+    },
+    {
+      "epoch": 64.00568181818181,
+      "grad_norm": 0.8491553068161011,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 22530
+    },
+    {
+      "epoch": 64.0340909090909,
+      "grad_norm": 0.7833629846572876,
+      "learning_rate": 0.0001,
+      "loss": 0.0538,
+      "step": 22540
+    },
+    {
+      "epoch": 64.0625,
+      "grad_norm": 0.8554551005363464,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "step": 22550
+    },
+    {
+      "epoch": 64.0909090909091,
+      "grad_norm": 0.8123806715011597,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 22560
+    },
+    {
+      "epoch": 64.11931818181819,
+      "grad_norm": 0.7412775754928589,
+      "learning_rate": 0.0001,
+      "loss": 0.0535,
+      "step": 22570
+    },
+    {
+      "epoch": 64.14772727272727,
+      "grad_norm": 0.6799927949905396,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 22580
+    },
+    {
+      "epoch": 64.17613636363636,
+      "grad_norm": 0.8203033804893494,
+      "learning_rate": 0.0001,
+      "loss": 0.0539,
+      "step": 22590
+    },
+    {
+      "epoch": 64.20454545454545,
+      "grad_norm": 0.7033742666244507,
+      "learning_rate": 0.0001,
+      "loss": 0.0522,
+      "step": 22600
+    },
+    {
+      "epoch": 64.23295454545455,
+      "grad_norm": 0.7215442061424255,
+      "learning_rate": 0.0001,
+      "loss": 0.0542,
+      "step": 22610
+    },
+    {
+      "epoch": 64.26136363636364,
+      "grad_norm": 0.5706157684326172,
+      "learning_rate": 0.0001,
+      "loss": 0.0542,
+      "step": 22620
+    },
+    {
+      "epoch": 64.28977272727273,
+      "grad_norm": 0.6812960505485535,
+      "learning_rate": 0.0001,
+      "loss": 0.0551,
+      "step": 22630
+    },
+    {
+      "epoch": 64.31818181818181,
+      "grad_norm": 0.6466752886772156,
+      "learning_rate": 0.0001,
+      "loss": 0.0539,
+      "step": 22640
+    },
+    {
+      "epoch": 64.3465909090909,
+      "grad_norm": 0.7440110445022583,
+      "learning_rate": 0.0001,
+      "loss": 0.0548,
+      "step": 22650
+    },
+    {
+      "epoch": 64.375,
+      "grad_norm": 0.7839359641075134,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 22660
+    },
+    {
+      "epoch": 64.4034090909091,
+      "grad_norm": 0.8440365195274353,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 22670
+    },
+    {
+      "epoch": 64.43181818181819,
+      "grad_norm": 0.774544358253479,
+      "learning_rate": 0.0001,
+      "loss": 0.0573,
+      "step": 22680
+    },
+    {
+      "epoch": 64.46022727272727,
+      "grad_norm": 0.601563036441803,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "step": 22690
+    },
+    {
+      "epoch": 64.48863636363636,
+      "grad_norm": 0.7574673891067505,
+      "learning_rate": 0.0001,
+      "loss": 0.0535,
+      "step": 22700
+    },
+    {
+      "epoch": 64.51704545454545,
+      "grad_norm": 0.836617648601532,
+      "learning_rate": 0.0001,
+      "loss": 0.0549,
+      "step": 22710
+    },
+    {
+      "epoch": 64.54545454545455,
+      "grad_norm": 1.0453118085861206,
+      "learning_rate": 0.0001,
+      "loss": 0.0548,
+      "step": 22720
+    },
+    {
+      "epoch": 64.57386363636364,
+      "grad_norm": 1.0585159063339233,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "step": 22730
+    },
+    {
+      "epoch": 64.60227272727273,
+      "grad_norm": 1.0894050598144531,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "step": 22740
+    },
+    {
+      "epoch": 64.63068181818181,
+      "grad_norm": 0.9796726107597351,
+      "learning_rate": 0.0001,
+      "loss": 0.0526,
+      "step": 22750
+    },
+    {
+      "epoch": 64.6590909090909,
+      "grad_norm": 0.9019004702568054,
+      "learning_rate": 0.0001,
+      "loss": 0.0568,
+      "step": 22760
+    },
+    {
+      "epoch": 64.6875,
+      "grad_norm": 0.7627422213554382,
+      "learning_rate": 0.0001,
+      "loss": 0.0508,
+      "step": 22770
+    },
+    {
+      "epoch": 64.7159090909091,
+      "grad_norm": 0.7884737849235535,
+      "learning_rate": 0.0001,
+      "loss": 0.0553,
+      "step": 22780
+    },
+    {
+      "epoch": 64.74431818181819,
+      "grad_norm": 0.7071219682693481,
+      "learning_rate": 0.0001,
+      "loss": 0.0547,
+      "step": 22790
+    },
+    {
+      "epoch": 64.77272727272727,
+      "grad_norm": 0.7242072820663452,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 22800
+    },
+    {
+      "epoch": 64.80113636363636,
+      "grad_norm": 0.9985579252243042,
+      "learning_rate": 0.0001,
+      "loss": 0.0542,
+      "step": 22810
+    },
+    {
+      "epoch": 64.82954545454545,
+      "grad_norm": 0.7592743635177612,
+      "learning_rate": 0.0001,
+      "loss": 0.0557,
+      "step": 22820
+    },
+    {
+      "epoch": 64.85795454545455,
+      "grad_norm": 0.6539085507392883,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 22830
+    },
+    {
+      "epoch": 64.88636363636364,
+      "grad_norm": 0.699675977230072,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "step": 22840
+    },
+    {
+      "epoch": 64.91477272727273,
+      "grad_norm": 0.830615222454071,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 22850
+    },
+    {
+      "epoch": 64.94318181818181,
+      "grad_norm": 0.760208785533905,
+      "learning_rate": 0.0001,
+      "loss": 0.0549,
+      "step": 22860
+    },
+    {
+      "epoch": 64.9715909090909,
+      "grad_norm": 0.7713984847068787,
+      "learning_rate": 0.0001,
+      "loss": 0.0542,
+      "step": 22870
+    },
+    {
+      "epoch": 65.0,
+      "grad_norm": 0.7789033055305481,
+      "learning_rate": 0.0001,
+      "loss": 0.0532,
+      "step": 22880
+    },
+    {
+      "epoch": 65.0284090909091,
+      "grad_norm": 0.780392050743103,
+      "learning_rate": 0.0001,
+      "loss": 0.0549,
+      "step": 22890
+    },
+    {
+      "epoch": 65.05681818181819,
+      "grad_norm": 0.7739676237106323,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 22900
+    },
+    {
+      "epoch": 65.08522727272727,
+      "grad_norm": 0.7210514545440674,
+      "learning_rate": 0.0001,
+      "loss": 0.0533,
+      "step": 22910
+    },
+    {
+      "epoch": 65.11363636363636,
+      "grad_norm": 1.0596988201141357,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 22920
+    },
+    {
+      "epoch": 65.14204545454545,
+      "grad_norm": 1.2048275470733643,
+      "learning_rate": 0.0001,
+      "loss": 0.0524,
+      "step": 22930
+    },
+    {
+      "epoch": 65.17045454545455,
+      "grad_norm": 0.8827832937240601,
+      "learning_rate": 0.0001,
+      "loss": 0.0537,
+      "step": 22940
+    },
+    {
+      "epoch": 65.19886363636364,
+      "grad_norm": 0.889045238494873,
+      "learning_rate": 0.0001,
+      "loss": 0.052,
+      "step": 22950
+    },
+    {
+      "epoch": 65.22727272727273,
+      "grad_norm": 1.0528745651245117,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 22960
+    },
+    {
+      "epoch": 65.25568181818181,
+      "grad_norm": 1.0414397716522217,
+      "learning_rate": 0.0001,
+      "loss": 0.0526,
+      "step": 22970
+    },
+    {
+      "epoch": 65.2840909090909,
+      "grad_norm": 1.096603512763977,
+      "learning_rate": 0.0001,
+      "loss": 0.0533,
+      "step": 22980
+    },
+    {
+      "epoch": 65.3125,
+      "grad_norm": 0.8513028025627136,
+      "learning_rate": 0.0001,
+      "loss": 0.0521,
+      "step": 22990
+    },
+    {
+      "epoch": 65.3409090909091,
+      "grad_norm": 1.2057493925094604,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 23000
+    },
+    {
+      "epoch": 65.36931818181819,
+      "grad_norm": 1.0297787189483643,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "step": 23010
+    },
+    {
+      "epoch": 65.39772727272727,
+      "grad_norm": 0.8249045610427856,
+      "learning_rate": 0.0001,
+      "loss": 0.0526,
+      "step": 23020
+    },
+    {
+      "epoch": 65.42613636363636,
+      "grad_norm": 0.949597179889679,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 23030
+    },
+    {
+      "epoch": 65.45454545454545,
+      "grad_norm": 0.7203119397163391,
+      "learning_rate": 0.0001,
+      "loss": 0.0537,
+      "step": 23040
+    },
+    {
+      "epoch": 65.48295454545455,
+      "grad_norm": 0.7339312434196472,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "step": 23050
+    },
+    {
+      "epoch": 65.51136363636364,
+      "grad_norm": 0.716767430305481,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 23060
+    },
+    {
+      "epoch": 65.53977272727273,
+      "grad_norm": 0.8138099312782288,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 23070
+    },
+    {
+      "epoch": 65.56818181818181,
+      "grad_norm": 0.7120342254638672,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 23080
+    },
+    {
+      "epoch": 65.5965909090909,
+      "grad_norm": 0.7150150537490845,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 23090
+    },
+    {
+      "epoch": 65.625,
+      "grad_norm": 0.7718611359596252,
+      "learning_rate": 0.0001,
+      "loss": 0.0556,
+      "step": 23100
+    },
+    {
+      "epoch": 65.6534090909091,
+      "grad_norm": 0.842397153377533,
+      "learning_rate": 0.0001,
+      "loss": 0.0518,
+      "step": 23110
+    },
+    {
+      "epoch": 65.68181818181819,
+      "grad_norm": 0.8018172979354858,
+      "learning_rate": 0.0001,
+      "loss": 0.0526,
+      "step": 23120
+    },
+    {
+      "epoch": 65.71022727272727,
+      "grad_norm": 0.756478488445282,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 23130
+    },
+    {
+      "epoch": 65.73863636363636,
+      "grad_norm": 0.8237600922584534,
+      "learning_rate": 0.0001,
+      "loss": 0.0556,
+      "step": 23140
+    },
+    {
+      "epoch": 65.76704545454545,
+      "grad_norm": 0.6838138103485107,
+      "learning_rate": 0.0001,
+      "loss": 0.0539,
+      "step": 23150
+    },
+    {
+      "epoch": 65.79545454545455,
+      "grad_norm": 0.7186658382415771,
+      "learning_rate": 0.0001,
+      "loss": 0.0543,
+      "step": 23160
+    },
+    {
+      "epoch": 65.82386363636364,
+      "grad_norm": 0.8290245532989502,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "step": 23170
+    },
+    {
+      "epoch": 65.85227272727273,
+      "grad_norm": 0.7229530811309814,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 23180
+    },
+    {
+      "epoch": 65.88068181818181,
+      "grad_norm": 0.6716543436050415,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "step": 23190
+    },
+    {
+      "epoch": 65.9090909090909,
+      "grad_norm": 0.8731271028518677,
+      "learning_rate": 0.0001,
+      "loss": 0.0578,
+      "step": 23200
+    },
+    {
+      "epoch": 65.9375,
+      "grad_norm": 0.8098838329315186,
+      "learning_rate": 0.0001,
+      "loss": 0.0544,
+      "step": 23210
+    },
+    {
+      "epoch": 65.9659090909091,
+      "grad_norm": 1.0541036128997803,
+      "learning_rate": 0.0001,
+      "loss": 0.0548,
+      "step": 23220
+    },
+    {
+      "epoch": 65.99431818181819,
+      "grad_norm": 0.8643235564231873,
+      "learning_rate": 0.0001,
+      "loss": 0.0555,
+      "step": 23230
+    },
+    {
+      "epoch": 66.02272727272727,
+      "grad_norm": 0.8315423130989075,
+      "learning_rate": 0.0001,
+      "loss": 0.0543,
+      "step": 23240
+    },
+    {
+      "epoch": 66.05113636363636,
+      "grad_norm": 0.6959272623062134,
+      "learning_rate": 0.0001,
+      "loss": 0.0541,
+      "step": 23250
+    },
+    {
+      "epoch": 66.07954545454545,
+      "grad_norm": 0.7065873742103577,
+      "learning_rate": 0.0001,
+      "loss": 0.0526,
+      "step": 23260
+    },
+    {
+      "epoch": 66.10795454545455,
+      "grad_norm": 0.9436522722244263,
+      "learning_rate": 0.0001,
+      "loss": 0.0538,
+      "step": 23270
+    },
+    {
+      "epoch": 66.13636363636364,
+      "grad_norm": 0.6383907794952393,
+      "learning_rate": 0.0001,
+      "loss": 0.0526,
+      "step": 23280
+    },
+    {
+      "epoch": 66.16477272727273,
+      "grad_norm": 0.8061172962188721,
+      "learning_rate": 0.0001,
+      "loss": 0.0569,
+      "step": 23290
+    },
+    {
+      "epoch": 66.19318181818181,
+      "grad_norm": 0.6677034497261047,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 23300
+    },
+    {
+      "epoch": 66.2215909090909,
+      "grad_norm": 0.8233653903007507,
+      "learning_rate": 0.0001,
+      "loss": 0.0542,
+      "step": 23310
+    },
+    {
+      "epoch": 66.25,
+      "grad_norm": 0.8707202672958374,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 23320
+    },
+    {
+      "epoch": 66.2784090909091,
+      "grad_norm": 0.7756959199905396,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "step": 23330
+    },
+    {
+      "epoch": 66.30681818181819,
+      "grad_norm": 0.8138574361801147,
+      "learning_rate": 0.0001,
+      "loss": 0.055,
+      "step": 23340
+    },
+    {
+      "epoch": 66.33522727272727,
+      "grad_norm": 0.8435407876968384,
+      "learning_rate": 0.0001,
+      "loss": 0.0572,
+      "step": 23350
+    },
+    {
+      "epoch": 66.36363636363636,
+      "grad_norm": 0.8531373143196106,
+      "learning_rate": 0.0001,
+      "loss": 0.0558,
+      "step": 23360
+    },
+    {
+      "epoch": 66.39204545454545,
+      "grad_norm": 0.9886962175369263,
+      "learning_rate": 0.0001,
+      "loss": 0.0562,
+      "step": 23370
+    },
+    {
+      "epoch": 66.42045454545455,
+      "grad_norm": 0.9955214262008667,
+      "learning_rate": 0.0001,
+      "loss": 0.0561,
+      "step": 23380
+    },
+    {
+      "epoch": 66.44886363636364,
+      "grad_norm": 1.1781306266784668,
+      "learning_rate": 0.0001,
+      "loss": 0.0532,
+      "step": 23390
+    },
+    {
+      "epoch": 66.47727272727273,
+      "grad_norm": 1.1715068817138672,
+      "learning_rate": 0.0001,
+      "loss": 0.0557,
+      "step": 23400
+    },
+    {
+      "epoch": 66.50568181818181,
+      "grad_norm": 1.1786881685256958,
+      "learning_rate": 0.0001,
+      "loss": 0.0519,
+      "step": 23410
+    },
+    {
+      "epoch": 66.5340909090909,
+      "grad_norm": 1.2845433950424194,
+      "learning_rate": 0.0001,
+      "loss": 0.056,
+      "step": 23420
+    },
+    {
+      "epoch": 66.5625,
+      "grad_norm": 1.0063714981079102,
+      "learning_rate": 0.0001,
+      "loss": 0.0524,
+      "step": 23430
+    },
+    {
+      "epoch": 66.5909090909091,
+      "grad_norm": 1.013217806816101,
+      "learning_rate": 0.0001,
+      "loss": 0.0532,
+      "step": 23440
+    },
+    {
+      "epoch": 66.61931818181819,
+      "grad_norm": 1.0957231521606445,
+      "learning_rate": 0.0001,
+      "loss": 0.0509,
+      "step": 23450
+    },
+    {
+      "epoch": 66.64772727272727,
+      "grad_norm": 0.9889658689498901,
+      "learning_rate": 0.0001,
+      "loss": 0.0519,
+      "step": 23460
+    },
+    {
+      "epoch": 66.67613636363636,
+      "grad_norm": 0.9741299748420715,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 23470
+    },
+    {
+      "epoch": 66.70454545454545,
+      "grad_norm": 1.234862208366394,
+      "learning_rate": 0.0001,
+      "loss": 0.0529,
+      "step": 23480
+    },
+    {
+      "epoch": 66.73295454545455,
+      "grad_norm": 0.8468987345695496,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 23490
+    },
+    {
+      "epoch": 66.76136363636364,
+      "grad_norm": 0.765661358833313,
+      "learning_rate": 0.0001,
+      "loss": 0.0508,
+      "step": 23500
+    },
+    {
+      "epoch": 66.78977272727273,
+      "grad_norm": 1.1269277334213257,
+      "learning_rate": 0.0001,
+      "loss": 0.0516,
+      "step": 23510
+    },
+    {
+      "epoch": 66.81818181818181,
+      "grad_norm": 0.9379168748855591,
+      "learning_rate": 0.0001,
+      "loss": 0.0516,
+      "step": 23520
+    },
+    {
+      "epoch": 66.8465909090909,
+      "grad_norm": 1.3234306573867798,
+      "learning_rate": 0.0001,
+      "loss": 0.0518,
+      "step": 23530
+    },
+    {
+      "epoch": 66.875,
+      "grad_norm": 0.9817354083061218,
+      "learning_rate": 0.0001,
+      "loss": 0.0539,
+      "step": 23540
+    },
+    {
+      "epoch": 66.9034090909091,
+      "grad_norm": 1.1395480632781982,
+      "learning_rate": 0.0001,
+      "loss": 0.0524,
+      "step": 23550
+    },
+    {
+      "epoch": 66.93181818181819,
+      "grad_norm": 0.9638949036598206,
+      "learning_rate": 0.0001,
+      "loss": 0.0511,
+      "step": 23560
+    },
+    {
+      "epoch": 66.96022727272727,
+      "grad_norm": 0.8169605135917664,
+      "learning_rate": 0.0001,
+      "loss": 0.0517,
+      "step": 23570
+    },
+    {
+      "epoch": 66.98863636363636,
+      "grad_norm": 0.861230731010437,
+      "learning_rate": 0.0001,
+      "loss": 0.0537,
+      "step": 23580
+    },
+    {
+      "epoch": 67.01704545454545,
+      "grad_norm": 0.656604528427124,
+      "learning_rate": 0.0001,
+      "loss": 0.0541,
+      "step": 23590
+    },
+    {
+      "epoch": 67.04545454545455,
+      "grad_norm": 0.7038812041282654,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 23600
+    },
+    {
+      "epoch": 67.07386363636364,
+      "grad_norm": 0.627716064453125,
+      "learning_rate": 0.0001,
+      "loss": 0.0517,
+      "step": 23610
+    },
+    {
+      "epoch": 67.10227272727273,
+      "grad_norm": 0.857126772403717,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 23620
+    },
+    {
+      "epoch": 67.13068181818181,
+      "grad_norm": 0.8336479067802429,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 23630
+    },
+    {
+      "epoch": 67.1590909090909,
+      "grad_norm": 0.6836590766906738,
+      "learning_rate": 0.0001,
+      "loss": 0.0527,
+      "step": 23640
+    },
+    {
+      "epoch": 67.1875,
+      "grad_norm": 0.8017722964286804,
+      "learning_rate": 0.0001,
+      "loss": 0.0535,
+      "step": 23650
+    },
+    {
+      "epoch": 67.2159090909091,
+      "grad_norm": 0.7780610918998718,
+      "learning_rate": 0.0001,
+      "loss": 0.0535,
+      "step": 23660
+    },
+    {
+      "epoch": 67.24431818181819,
+      "grad_norm": 0.7484387755393982,
+      "learning_rate": 0.0001,
+      "loss": 0.0526,
+      "step": 23670
+    },
+    {
+      "epoch": 67.27272727272727,
+      "grad_norm": 0.6655145287513733,
+      "learning_rate": 0.0001,
+      "loss": 0.0513,
+      "step": 23680
+    },
+    {
+      "epoch": 67.30113636363636,
+      "grad_norm": 0.6643372774124146,
+      "learning_rate": 0.0001,
+      "loss": 0.0535,
+      "step": 23690
+    },
+    {
+      "epoch": 67.32954545454545,
+      "grad_norm": 0.7950687408447266,
+      "learning_rate": 0.0001,
+      "loss": 0.0517,
+      "step": 23700
+    },
+    {
+      "epoch": 67.35795454545455,
+      "grad_norm": 0.8306936621665955,
+      "learning_rate": 0.0001,
+      "loss": 0.0534,
+      "step": 23710
+    },
+    {
+      "epoch": 67.38636363636364,
+      "grad_norm": 0.8002460598945618,
+      "learning_rate": 0.0001,
+      "loss": 0.0513,
+      "step": 23720
+    },
+    {
+      "epoch": 67.41477272727273,
+      "grad_norm": 0.7444122433662415,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 23730
+    },
+    {
+      "epoch": 67.44318181818181,
+      "grad_norm": 0.6906100511550903,
+      "learning_rate": 0.0001,
+      "loss": 0.0509,
+      "step": 23740
+    },
+    {
+      "epoch": 67.4715909090909,
+      "grad_norm": 0.7418623566627502,
+      "learning_rate": 0.0001,
+      "loss": 0.0546,
+      "step": 23750
+    },
+    {
+      "epoch": 67.5,
+      "grad_norm": 0.7137703895568848,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 23760
+    },
+    {
+      "epoch": 67.5284090909091,
+      "grad_norm": 0.793405294418335,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 23770
+    },
+    {
+      "epoch": 67.55681818181819,
+      "grad_norm": 0.7513002157211304,
+      "learning_rate": 0.0001,
+      "loss": 0.0568,
+      "step": 23780
+    },
+    {
+      "epoch": 67.58522727272727,
+      "grad_norm": 0.7225285172462463,
+      "learning_rate": 0.0001,
+      "loss": 0.0534,
+      "step": 23790
+    },
+    {
+      "epoch": 67.61363636363636,
+      "grad_norm": 0.690243661403656,
+      "learning_rate": 0.0001,
+      "loss": 0.0517,
+      "step": 23800
+    },
+    {
+      "epoch": 67.64204545454545,
+      "grad_norm": 0.6749371290206909,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 23810
+    },
+    {
+      "epoch": 67.67045454545455,
+      "grad_norm": 0.7889542579650879,
+      "learning_rate": 0.0001,
+      "loss": 0.0535,
+      "step": 23820
+    },
+    {
+      "epoch": 67.69886363636364,
+      "grad_norm": 1.2622177600860596,
+      "learning_rate": 0.0001,
+      "loss": 0.052,
+      "step": 23830
+    },
+    {
+      "epoch": 67.72727272727273,
+      "grad_norm": 1.8038215637207031,
+      "learning_rate": 0.0001,
+      "loss": 0.0534,
+      "step": 23840
+    },
+    {
+      "epoch": 67.75568181818181,
+      "grad_norm": 1.4471378326416016,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 23850
+    },
+    {
+      "epoch": 67.7840909090909,
+      "grad_norm": 1.2452491521835327,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "step": 23860
+    },
+    {
+      "epoch": 67.8125,
+      "grad_norm": 1.2193328142166138,
+      "learning_rate": 0.0001,
+      "loss": 0.0498,
+      "step": 23870
+    },
+    {
+      "epoch": 67.8409090909091,
+      "grad_norm": 1.1352566480636597,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 23880
+    },
+    {
+      "epoch": 67.86931818181819,
+      "grad_norm": 1.0166912078857422,
+      "learning_rate": 0.0001,
+      "loss": 0.0527,
+      "step": 23890
+    },
+    {
+      "epoch": 67.89772727272727,
+      "grad_norm": 0.9358308911323547,
+      "learning_rate": 0.0001,
+      "loss": 0.054,
+      "step": 23900
+    },
+    {
+      "epoch": 67.92613636363636,
+      "grad_norm": 1.0616742372512817,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 23910
+    },
+    {
+      "epoch": 67.95454545454545,
+      "grad_norm": 0.9217783808708191,
+      "learning_rate": 0.0001,
+      "loss": 0.0502,
+      "step": 23920
+    },
+    {
+      "epoch": 67.98295454545455,
+      "grad_norm": 1.0423084497451782,
+      "learning_rate": 0.0001,
+      "loss": 0.0518,
+      "step": 23930
+    },
+    {
+      "epoch": 68.01136363636364,
+      "grad_norm": 1.1823982000350952,
+      "learning_rate": 0.0001,
+      "loss": 0.0534,
+      "step": 23940
+    },
+    {
+      "epoch": 68.03977272727273,
+      "grad_norm": 0.9482648968696594,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 23950
+    },
+    {
+      "epoch": 68.06818181818181,
+      "grad_norm": 0.7669751644134521,
+      "learning_rate": 0.0001,
+      "loss": 0.0516,
+      "step": 23960
+    },
+    {
+      "epoch": 68.0965909090909,
+      "grad_norm": 1.1928632259368896,
+      "learning_rate": 0.0001,
+      "loss": 0.0535,
+      "step": 23970
+    },
+    {
+      "epoch": 68.125,
+      "grad_norm": 0.9698597192764282,
+      "learning_rate": 0.0001,
+      "loss": 0.0541,
+      "step": 23980
+    },
+    {
+      "epoch": 68.1534090909091,
+      "grad_norm": 1.0423868894577026,
+      "learning_rate": 0.0001,
+      "loss": 0.0541,
+      "step": 23990
+    },
+    {
+      "epoch": 68.18181818181819,
+      "grad_norm": 1.2554688453674316,
+      "learning_rate": 0.0001,
+      "loss": 0.0532,
+      "step": 24000
+    },
+    {
+      "epoch": 68.21022727272727,
+      "grad_norm": 1.3134796619415283,
+      "learning_rate": 0.0001,
+      "loss": 0.0545,
+      "step": 24010
+    },
+    {
+      "epoch": 68.23863636363636,
+      "grad_norm": 1.2554820775985718,
+      "learning_rate": 0.0001,
+      "loss": 0.0509,
+      "step": 24020
+    },
+    {
+      "epoch": 68.26704545454545,
+      "grad_norm": 0.9832156896591187,
+      "learning_rate": 0.0001,
+      "loss": 0.0497,
+      "step": 24030
+    },
+    {
+      "epoch": 68.29545454545455,
+      "grad_norm": 1.0172799825668335,
+      "learning_rate": 0.0001,
+      "loss": 0.0482,
+      "step": 24040
+    },
+    {
+      "epoch": 68.32386363636364,
+      "grad_norm": 1.321234107017517,
+      "learning_rate": 0.0001,
+      "loss": 0.0529,
+      "step": 24050
+    },
+    {
+      "epoch": 68.35227272727273,
+      "grad_norm": 1.5265092849731445,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "step": 24060
+    },
+    {
+      "epoch": 68.38068181818181,
+      "grad_norm": 1.1719361543655396,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 24070
+    },
+    {
+      "epoch": 68.4090909090909,
+      "grad_norm": 1.2583420276641846,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 24080
+    },
+    {
+      "epoch": 68.4375,
+      "grad_norm": 1.0527803897857666,
+      "learning_rate": 0.0001,
+      "loss": 0.0516,
+      "step": 24090
+    },
+    {
+      "epoch": 68.4659090909091,
+      "grad_norm": 1.011395812034607,
+      "learning_rate": 0.0001,
+      "loss": 0.05,
+      "step": 24100
+    },
+    {
+      "epoch": 68.49431818181819,
+      "grad_norm": 1.037021517753601,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 24110
+    },
+    {
+      "epoch": 68.52272727272727,
+      "grad_norm": 0.9442154765129089,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "step": 24120
+    },
+    {
+      "epoch": 68.55113636363636,
+      "grad_norm": 0.8959128260612488,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "step": 24130
+    },
+    {
+      "epoch": 68.57954545454545,
+      "grad_norm": 0.9579172730445862,
+      "learning_rate": 0.0001,
+      "loss": 0.0504,
+      "step": 24140
+    },
+    {
+      "epoch": 68.60795454545455,
+      "grad_norm": 1.0895936489105225,
+      "learning_rate": 0.0001,
+      "loss": 0.0488,
+      "step": 24150
+    },
+    {
+      "epoch": 68.63636363636364,
+      "grad_norm": 0.7573409676551819,
+      "learning_rate": 0.0001,
+      "loss": 0.0491,
+      "step": 24160
+    },
+    {
+      "epoch": 68.66477272727273,
+      "grad_norm": 0.8774531483650208,
+      "learning_rate": 0.0001,
+      "loss": 0.0481,
+      "step": 24170
+    },
+    {
+      "epoch": 68.69318181818181,
+      "grad_norm": 0.8967164158821106,
+      "learning_rate": 0.0001,
+      "loss": 0.0509,
+      "step": 24180
+    },
+    {
+      "epoch": 68.7215909090909,
+      "grad_norm": 1.0587197542190552,
+      "learning_rate": 0.0001,
+      "loss": 0.0541,
+      "step": 24190
+    },
+    {
+      "epoch": 68.75,
+      "grad_norm": 1.1898106336593628,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 24200
+    },
+    {
+      "epoch": 68.7784090909091,
+      "grad_norm": 1.0361658334732056,
+      "learning_rate": 0.0001,
+      "loss": 0.0513,
+      "step": 24210
+    },
+    {
+      "epoch": 68.80681818181819,
+      "grad_norm": 0.935555100440979,
+      "learning_rate": 0.0001,
+      "loss": 0.0518,
+      "step": 24220
+    },
+    {
+      "epoch": 68.83522727272727,
+      "grad_norm": 1.032720923423767,
+      "learning_rate": 0.0001,
+      "loss": 0.0511,
+      "step": 24230
+    },
+    {
+      "epoch": 68.86363636363636,
+      "grad_norm": 0.8149229884147644,
+      "learning_rate": 0.0001,
+      "loss": 0.052,
+      "step": 24240
+    },
+    {
+      "epoch": 68.89204545454545,
+      "grad_norm": 0.8590128421783447,
+      "learning_rate": 0.0001,
+      "loss": 0.0515,
+      "step": 24250
+    },
+    {
+      "epoch": 68.92045454545455,
+      "grad_norm": 0.9247688055038452,
+      "learning_rate": 0.0001,
+      "loss": 0.0518,
+      "step": 24260
+    },
+    {
+      "epoch": 68.94886363636364,
+      "grad_norm": 0.9311433434486389,
+      "learning_rate": 0.0001,
+      "loss": 0.0522,
+      "step": 24270
+    },
+    {
+      "epoch": 68.97727272727273,
+      "grad_norm": 0.7276850938796997,
+      "learning_rate": 0.0001,
+      "loss": 0.0533,
+      "step": 24280
+    },
+    {
+      "epoch": 69.00568181818181,
+      "grad_norm": 0.6716181039810181,
+      "learning_rate": 0.0001,
+      "loss": 0.0537,
+      "step": 24290
+    },
+    {
+      "epoch": 69.0340909090909,
+      "grad_norm": 0.9270053505897522,
+      "learning_rate": 0.0001,
+      "loss": 0.0527,
+      "step": 24300
+    },
+    {
+      "epoch": 69.0625,
+      "grad_norm": 0.666446328163147,
+      "learning_rate": 0.0001,
+      "loss": 0.0511,
+      "step": 24310
+    },
+    {
+      "epoch": 69.0909090909091,
+      "grad_norm": 0.8492375016212463,
+      "learning_rate": 0.0001,
+      "loss": 0.0514,
+      "step": 24320
+    },
+    {
+      "epoch": 69.11931818181819,
+      "grad_norm": 0.8447439074516296,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 24330
+    },
+    {
+      "epoch": 69.14772727272727,
+      "grad_norm": 0.7112112045288086,
+      "learning_rate": 0.0001,
+      "loss": 0.0521,
+      "step": 24340
+    },
+    {
+      "epoch": 69.17613636363636,
+      "grad_norm": 0.8900835514068604,
+      "learning_rate": 0.0001,
+      "loss": 0.0554,
+      "step": 24350
+    },
+    {
+      "epoch": 69.20454545454545,
+      "grad_norm": 0.7511789798736572,
+      "learning_rate": 0.0001,
+      "loss": 0.0511,
+      "step": 24360
+    },
+    {
+      "epoch": 69.23295454545455,
+      "grad_norm": 0.6234313249588013,
+      "learning_rate": 0.0001,
+      "loss": 0.0516,
+      "step": 24370
+    },
+    {
+      "epoch": 69.26136363636364,
+      "grad_norm": 0.8581838011741638,
+      "learning_rate": 0.0001,
+      "loss": 0.0503,
+      "step": 24380
+    },
+    {
+      "epoch": 69.28977272727273,
+      "grad_norm": 0.6439953446388245,
+      "learning_rate": 0.0001,
+      "loss": 0.0522,
+      "step": 24390
+    },
+    {
+      "epoch": 69.31818181818181,
+      "grad_norm": 0.805645763874054,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 24400
+    },
+    {
+      "epoch": 69.3465909090909,
+      "grad_norm": 0.7699912786483765,
+      "learning_rate": 0.0001,
+      "loss": 0.0555,
+      "step": 24410
+    },
+    {
+      "epoch": 69.375,
+      "grad_norm": 0.7186166644096375,
+      "learning_rate": 0.0001,
+      "loss": 0.0552,
+      "step": 24420
+    },
+    {
+      "epoch": 69.4034090909091,
+      "grad_norm": 0.8284119963645935,
+      "learning_rate": 0.0001,
+      "loss": 0.0514,
+      "step": 24430
+    },
+    {
+      "epoch": 69.43181818181819,
+      "grad_norm": 0.8688386082649231,
+      "learning_rate": 0.0001,
+      "loss": 0.0529,
+      "step": 24440
+    },
+    {
+      "epoch": 69.46022727272727,
+      "grad_norm": 1.1181520223617554,
+      "learning_rate": 0.0001,
+      "loss": 0.0544,
+      "step": 24450
+    },
+    {
+      "epoch": 69.48863636363636,
+      "grad_norm": 1.0079569816589355,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 24460
+    },
+    {
+      "epoch": 69.51704545454545,
+      "grad_norm": 1.6781340837478638,
+      "learning_rate": 0.0001,
+      "loss": 0.0526,
+      "step": 24470
+    },
+    {
+      "epoch": 69.54545454545455,
+      "grad_norm": 1.3601493835449219,
+      "learning_rate": 0.0001,
+      "loss": 0.0515,
+      "step": 24480
+    },
+    {
+      "epoch": 69.57386363636364,
+      "grad_norm": 0.9561741948127747,
+      "learning_rate": 0.0001,
+      "loss": 0.0514,
+      "step": 24490
+    },
+    {
+      "epoch": 69.60227272727273,
+      "grad_norm": 0.8902744650840759,
+      "learning_rate": 0.0001,
+      "loss": 0.0496,
+      "step": 24500
+    },
+    {
+      "epoch": 69.63068181818181,
+      "grad_norm": 0.9449006915092468,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 24510
+    },
+    {
+      "epoch": 69.6590909090909,
+      "grad_norm": 1.1786518096923828,
+      "learning_rate": 0.0001,
+      "loss": 0.0525,
+      "step": 24520
+    },
+    {
+      "epoch": 69.6875,
+      "grad_norm": 1.233751893043518,
+      "learning_rate": 0.0001,
+      "loss": 0.0507,
+      "step": 24530
+    },
+    {
+      "epoch": 69.7159090909091,
+      "grad_norm": 0.9357305765151978,
+      "learning_rate": 0.0001,
+      "loss": 0.0528,
+      "step": 24540
+    },
+    {
+      "epoch": 69.74431818181819,
+      "grad_norm": 1.226325511932373,
+      "learning_rate": 0.0001,
+      "loss": 0.0515,
+      "step": 24550
+    },
+    {
+      "epoch": 69.77272727272727,
+      "grad_norm": 0.8865256309509277,
+      "learning_rate": 0.0001,
+      "loss": 0.0497,
+      "step": 24560
+    },
+    {
+      "epoch": 69.80113636363636,
+      "grad_norm": 1.6879335641860962,
+      "learning_rate": 0.0001,
+      "loss": 0.0522,
+      "step": 24570
+    },
+    {
+      "epoch": 69.82954545454545,
+      "grad_norm": 1.2305909395217896,
+      "learning_rate": 0.0001,
+      "loss": 0.0502,
+      "step": 24580
+    },
+    {
+      "epoch": 69.85795454545455,
+      "grad_norm": 1.1654038429260254,
+      "learning_rate": 0.0001,
+      "loss": 0.0514,
+      "step": 24590
+    },
+    {
+      "epoch": 69.88636363636364,
+      "grad_norm": 0.8632926940917969,
+      "learning_rate": 0.0001,
+      "loss": 0.0507,
+      "step": 24600
+    },
+    {
+      "epoch": 69.91477272727273,
+      "grad_norm": 0.7934690713882446,
+      "learning_rate": 0.0001,
+      "loss": 0.0497,
+      "step": 24610
+    },
+    {
+      "epoch": 69.94318181818181,
+      "grad_norm": 0.7492729425430298,
+      "learning_rate": 0.0001,
+      "loss": 0.0496,
+      "step": 24620
+    },
+    {
+      "epoch": 69.9715909090909,
+      "grad_norm": 0.7984905242919922,
+      "learning_rate": 0.0001,
+      "loss": 0.0503,
+      "step": 24630
+    },
+    {
+      "epoch": 70.0,
+      "grad_norm": 0.8478935956954956,
+      "learning_rate": 0.0001,
+      "loss": 0.0514,
+      "step": 24640
+    },
+    {
+      "epoch": 70.0284090909091,
+      "grad_norm": 0.7653668522834778,
+      "learning_rate": 0.0001,
+      "loss": 0.0517,
+      "step": 24650
+    },
+    {
+      "epoch": 70.05681818181819,
+      "grad_norm": 0.7579995393753052,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 24660
+    },
+    {
+      "epoch": 70.08522727272727,
+      "grad_norm": 0.9072360992431641,
+      "learning_rate": 0.0001,
+      "loss": 0.0515,
+      "step": 24670
+    },
+    {
+      "epoch": 70.11363636363636,
+      "grad_norm": 0.7853196859359741,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 24680
+    },
+    {
+      "epoch": 70.14204545454545,
+      "grad_norm": 0.7733336091041565,
+      "learning_rate": 0.0001,
+      "loss": 0.0566,
+      "step": 24690
+    },
+    {
+      "epoch": 70.17045454545455,
+      "grad_norm": 0.8603296279907227,
+      "learning_rate": 0.0001,
+      "loss": 0.0522,
+      "step": 24700
+    },
+    {
+      "epoch": 70.19886363636364,
+      "grad_norm": 1.4242461919784546,
+      "learning_rate": 0.0001,
+      "loss": 0.0529,
+      "step": 24710
+    },
+    {
+      "epoch": 70.22727272727273,
+      "grad_norm": 1.4059160947799683,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 24720
+    },
+    {
+      "epoch": 70.25568181818181,
+      "grad_norm": 1.3890278339385986,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "step": 24730
+    },
+    {
+      "epoch": 70.2840909090909,
+      "grad_norm": 1.2617861032485962,
+      "learning_rate": 0.0001,
+      "loss": 0.0472,
+      "step": 24740
+    },
+    {
+      "epoch": 70.3125,
+      "grad_norm": 1.1536449193954468,
+      "learning_rate": 0.0001,
+      "loss": 0.0507,
+      "step": 24750
+    },
+    {
+      "epoch": 70.3409090909091,
+      "grad_norm": 1.032045602798462,
+      "learning_rate": 0.0001,
+      "loss": 0.0502,
+      "step": 24760
+    },
+    {
+      "epoch": 70.36931818181819,
+      "grad_norm": 0.9999845623970032,
+      "learning_rate": 0.0001,
+      "loss": 0.0511,
+      "step": 24770
+    },
+    {
+      "epoch": 70.39772727272727,
+      "grad_norm": 0.9327858090400696,
+      "learning_rate": 0.0001,
+      "loss": 0.0497,
+      "step": 24780
+    },
+    {
+      "epoch": 70.42613636363636,
+      "grad_norm": 1.1740491390228271,
+      "learning_rate": 0.0001,
+      "loss": 0.0509,
+      "step": 24790
+    },
+    {
+      "epoch": 70.45454545454545,
+      "grad_norm": 0.8893155455589294,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 24800
+    },
+    {
+      "epoch": 70.48295454545455,
+      "grad_norm": 0.8836989998817444,
+      "learning_rate": 0.0001,
+      "loss": 0.0522,
+      "step": 24810
+    },
+    {
+      "epoch": 70.51136363636364,
+      "grad_norm": 1.0119452476501465,
+      "learning_rate": 0.0001,
+      "loss": 0.0497,
+      "step": 24820
+    },
+    {
+      "epoch": 70.53977272727273,
+      "grad_norm": 0.9030247330665588,
+      "learning_rate": 0.0001,
+      "loss": 0.0522,
+      "step": 24830
+    },
+    {
+      "epoch": 70.56818181818181,
+      "grad_norm": 0.7199386358261108,
+      "learning_rate": 0.0001,
+      "loss": 0.0514,
+      "step": 24840
+    },
+    {
+      "epoch": 70.5965909090909,
+      "grad_norm": 0.8884567022323608,
+      "learning_rate": 0.0001,
+      "loss": 0.0504,
+      "step": 24850
+    },
+    {
+      "epoch": 70.625,
+      "grad_norm": 0.6433593034744263,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 24860
+    },
+    {
+      "epoch": 70.6534090909091,
+      "grad_norm": 0.6983967423439026,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 24870
+    },
+    {
+      "epoch": 70.68181818181819,
+      "grad_norm": 0.647629976272583,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 24880
+    },
+    {
+      "epoch": 70.71022727272727,
+      "grad_norm": 0.71266108751297,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 24890
+    },
+    {
+      "epoch": 70.73863636363636,
+      "grad_norm": 0.6877172589302063,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 24900
+    },
+    {
+      "epoch": 70.76704545454545,
+      "grad_norm": 0.6937993168830872,
+      "learning_rate": 0.0001,
+      "loss": 0.0496,
+      "step": 24910
+    },
+    {
+      "epoch": 70.79545454545455,
+      "grad_norm": 0.5959415435791016,
+      "learning_rate": 0.0001,
+      "loss": 0.0489,
+      "step": 24920
+    },
+    {
+      "epoch": 70.82386363636364,
+      "grad_norm": 0.6399363279342651,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 24930
+    },
+    {
+      "epoch": 70.85227272727273,
+      "grad_norm": 0.7871550917625427,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 24940
+    },
+    {
+      "epoch": 70.88068181818181,
+      "grad_norm": 0.7523185610771179,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 24950
+    },
+    {
+      "epoch": 70.9090909090909,
+      "grad_norm": 0.7533581852912903,
+      "learning_rate": 0.0001,
+      "loss": 0.0525,
+      "step": 24960
+    },
+    {
+      "epoch": 70.9375,
+      "grad_norm": 0.7682768106460571,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 24970
+    },
+    {
+      "epoch": 70.9659090909091,
+      "grad_norm": 0.8463433980941772,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "step": 24980
+    },
+    {
+      "epoch": 70.99431818181819,
+      "grad_norm": 1.0878268480300903,
+      "learning_rate": 0.0001,
+      "loss": 0.049,
+      "step": 24990
+    },
+    {
+      "epoch": 71.02272727272727,
+      "grad_norm": 0.7139332294464111,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "step": 25000
+    },
+    {
+      "epoch": 71.05113636363636,
+      "grad_norm": 0.6780238747596741,
+      "learning_rate": 0.0001,
+      "loss": 0.05,
+      "step": 25010
+    },
+    {
+      "epoch": 71.07954545454545,
+      "grad_norm": 0.6342650055885315,
+      "learning_rate": 0.0001,
+      "loss": 0.0518,
+      "step": 25020
+    },
+    {
+      "epoch": 71.10795454545455,
+      "grad_norm": 0.6704277992248535,
+      "learning_rate": 0.0001,
+      "loss": 0.0506,
+      "step": 25030
+    },
+    {
+      "epoch": 71.13636363636364,
+      "grad_norm": 0.7333451509475708,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 25040
+    },
+    {
+      "epoch": 71.16477272727273,
+      "grad_norm": 0.8710368275642395,
+      "learning_rate": 0.0001,
+      "loss": 0.0508,
+      "step": 25050
+    },
+    {
+      "epoch": 71.19318181818181,
+      "grad_norm": 0.9135860204696655,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 25060
+    },
+    {
+      "epoch": 71.2215909090909,
+      "grad_norm": 0.7403706908226013,
+      "learning_rate": 0.0001,
+      "loss": 0.0505,
+      "step": 25070
+    },
+    {
+      "epoch": 71.25,
+      "grad_norm": 0.6618191003799438,
+      "learning_rate": 0.0001,
+      "loss": 0.048,
+      "step": 25080
+    },
+    {
+      "epoch": 71.2784090909091,
+      "grad_norm": 0.7856776714324951,
+      "learning_rate": 0.0001,
+      "loss": 0.0476,
+      "step": 25090
+    },
+    {
+      "epoch": 71.30681818181819,
+      "grad_norm": 0.7596649527549744,
+      "learning_rate": 0.0001,
+      "loss": 0.0481,
+      "step": 25100
+    },
+    {
+      "epoch": 71.33522727272727,
+      "grad_norm": 0.8146116733551025,
+      "learning_rate": 0.0001,
+      "loss": 0.0474,
+      "step": 25110
+    },
+    {
+      "epoch": 71.36363636363636,
+      "grad_norm": 0.6791525483131409,
+      "learning_rate": 0.0001,
+      "loss": 0.0484,
+      "step": 25120
+    },
+    {
+      "epoch": 71.39204545454545,
+      "grad_norm": 0.7217307090759277,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 25130
+    },
+    {
+      "epoch": 71.42045454545455,
+      "grad_norm": 0.6544477939605713,
+      "learning_rate": 0.0001,
+      "loss": 0.0531,
+      "step": 25140
+    },
+    {
+      "epoch": 71.44886363636364,
+      "grad_norm": 0.6746852397918701,
+      "learning_rate": 0.0001,
+      "loss": 0.0521,
+      "step": 25150
+    },
+    {
+      "epoch": 71.47727272727273,
+      "grad_norm": 0.5204148888587952,
+      "learning_rate": 0.0001,
+      "loss": 0.0506,
+      "step": 25160
+    },
+    {
+      "epoch": 71.50568181818181,
+      "grad_norm": 0.5109100937843323,
+      "learning_rate": 0.0001,
+      "loss": 0.0493,
+      "step": 25170
+    },
+    {
+      "epoch": 71.5340909090909,
+      "grad_norm": 0.6211031675338745,
+      "learning_rate": 0.0001,
+      "loss": 0.0543,
+      "step": 25180
+    },
+    {
+      "epoch": 71.5625,
+      "grad_norm": 0.677085280418396,
+      "learning_rate": 0.0001,
+      "loss": 0.0527,
+      "step": 25190
+    },
+    {
+      "epoch": 71.5909090909091,
+      "grad_norm": 0.6960747838020325,
+      "learning_rate": 0.0001,
+      "loss": 0.0502,
+      "step": 25200
+    },
+    {
+      "epoch": 71.61931818181819,
+      "grad_norm": 0.6580451130867004,
+      "learning_rate": 0.0001,
+      "loss": 0.0505,
+      "step": 25210
+    },
+    {
+      "epoch": 71.64772727272727,
+      "grad_norm": 0.6775358319282532,
+      "learning_rate": 0.0001,
+      "loss": 0.0491,
+      "step": 25220
+    },
+    {
+      "epoch": 71.67613636363636,
+      "grad_norm": 0.6583216786384583,
+      "learning_rate": 0.0001,
+      "loss": 0.0505,
+      "step": 25230
+    },
+    {
+      "epoch": 71.70454545454545,
+      "grad_norm": 0.7078356146812439,
+      "learning_rate": 0.0001,
+      "loss": 0.0483,
+      "step": 25240
+    },
+    {
+      "epoch": 71.73295454545455,
+      "grad_norm": 0.7176387906074524,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 25250
+    },
+    {
+      "epoch": 71.76136363636364,
+      "grad_norm": 0.749264657497406,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "step": 25260
+    },
+    {
+      "epoch": 71.78977272727273,
+      "grad_norm": 0.6820817589759827,
+      "learning_rate": 0.0001,
+      "loss": 0.0498,
+      "step": 25270
+    },
+    {
+      "epoch": 71.81818181818181,
+      "grad_norm": 0.6245760917663574,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 25280
+    },
+    {
+      "epoch": 71.8465909090909,
+      "grad_norm": 0.5692148804664612,
+      "learning_rate": 0.0001,
+      "loss": 0.0491,
+      "step": 25290
+    },
+    {
+      "epoch": 71.875,
+      "grad_norm": 0.6304931640625,
+      "learning_rate": 0.0001,
+      "loss": 0.0504,
+      "step": 25300
+    },
+    {
+      "epoch": 71.9034090909091,
+      "grad_norm": 0.546541690826416,
+      "learning_rate": 0.0001,
+      "loss": 0.0492,
+      "step": 25310
+    },
+    {
+      "epoch": 71.93181818181819,
+      "grad_norm": 0.5972326993942261,
+      "learning_rate": 0.0001,
+      "loss": 0.0505,
+      "step": 25320
+    },
+    {
+      "epoch": 71.96022727272727,
+      "grad_norm": 0.6020660996437073,
+      "learning_rate": 0.0001,
+      "loss": 0.0505,
+      "step": 25330
+    },
+    {
+      "epoch": 71.98863636363636,
+      "grad_norm": 0.48787808418273926,
+      "learning_rate": 0.0001,
+      "loss": 0.0523,
+      "step": 25340
+    },
+    {
+      "epoch": 72.01704545454545,
+      "grad_norm": 0.7013693451881409,
+      "learning_rate": 0.0001,
+      "loss": 0.0508,
+      "step": 25350
+    },
+    {
+      "epoch": 72.04545454545455,
+      "grad_norm": 0.5541148781776428,
+      "learning_rate": 0.0001,
+      "loss": 0.049,
+      "step": 25360
+    },
+    {
+      "epoch": 72.07386363636364,
+      "grad_norm": 0.6003844738006592,
+      "learning_rate": 0.0001,
+      "loss": 0.0509,
+      "step": 25370
+    },
+    {
+      "epoch": 72.10227272727273,
+      "grad_norm": 0.8124470114707947,
+      "learning_rate": 0.0001,
+      "loss": 0.0532,
+      "step": 25380
+    },
+    {
+      "epoch": 72.13068181818181,
+      "grad_norm": 0.6087120771408081,
+      "learning_rate": 0.0001,
+      "loss": 0.0513,
+      "step": 25390
+    },
+    {
+      "epoch": 72.1590909090909,
+      "grad_norm": 0.6835238337516785,
+      "learning_rate": 0.0001,
+      "loss": 0.0514,
+      "step": 25400
+    },
+    {
+      "epoch": 72.1875,
+      "grad_norm": 0.758734405040741,
+      "learning_rate": 0.0001,
+      "loss": 0.0502,
+      "step": 25410
+    },
+    {
+      "epoch": 72.2159090909091,
+      "grad_norm": 0.745496392250061,
+      "learning_rate": 0.0001,
+      "loss": 0.0496,
+      "step": 25420
+    },
+    {
+      "epoch": 72.24431818181819,
+      "grad_norm": 0.7484995722770691,
+      "learning_rate": 0.0001,
+      "loss": 0.0518,
+      "step": 25430
+    },
+    {
+      "epoch": 72.27272727272727,
+      "grad_norm": 0.7207466959953308,
+      "learning_rate": 0.0001,
+      "loss": 0.0506,
+      "step": 25440
+    },
+    {
+      "epoch": 72.30113636363636,
+      "grad_norm": 0.866812527179718,
+      "learning_rate": 0.0001,
+      "loss": 0.0514,
+      "step": 25450
+    },
+    {
+      "epoch": 72.32954545454545,
+      "grad_norm": 0.7610346674919128,
+      "learning_rate": 0.0001,
+      "loss": 0.0505,
+      "step": 25460
+    },
+    {
+      "epoch": 72.35795454545455,
+      "grad_norm": 0.6176382899284363,
+      "learning_rate": 0.0001,
+      "loss": 0.0493,
+      "step": 25470
+    },
+    {
+      "epoch": 72.38636363636364,
+      "grad_norm": 0.7282941937446594,
+      "learning_rate": 0.0001,
+      "loss": 0.0492,
+      "step": 25480
+    },
+    {
+      "epoch": 72.41477272727273,
+      "grad_norm": 0.6433279514312744,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "step": 25490
+    },
+    {
+      "epoch": 72.44318181818181,
+      "grad_norm": 0.6624048352241516,
+      "learning_rate": 0.0001,
+      "loss": 0.0507,
+      "step": 25500
+    },
+    {
+      "epoch": 72.4715909090909,
+      "grad_norm": 0.6259250044822693,
+      "learning_rate": 0.0001,
+      "loss": 0.0508,
+      "step": 25510
+    },
+    {
+      "epoch": 72.5,
+      "grad_norm": 0.5815131068229675,
+      "learning_rate": 0.0001,
+      "loss": 0.0509,
+      "step": 25520
+    },
+    {
+      "epoch": 72.5284090909091,
+      "grad_norm": 0.4949661195278168,
+      "learning_rate": 0.0001,
+      "loss": 0.0502,
+      "step": 25530
+    },
+    {
+      "epoch": 72.55681818181819,
+      "grad_norm": 0.6070393323898315,
+      "learning_rate": 0.0001,
+      "loss": 0.0536,
+      "step": 25540
+    },
+    {
+      "epoch": 72.58522727272727,
+      "grad_norm": 0.9325839281082153,
+      "learning_rate": 0.0001,
+      "loss": 0.0512,
+      "step": 25550
+    },
+    {
+      "epoch": 72.61363636363636,
+      "grad_norm": 0.6207942962646484,
+      "learning_rate": 0.0001,
+      "loss": 0.051,
+      "step": 25560
+    },
+    {
+      "epoch": 72.64204545454545,
+      "grad_norm": 0.7251754403114319,
+      "learning_rate": 0.0001,
+      "loss": 0.0511,
+      "step": 25570
+    },
+    {
+      "epoch": 72.67045454545455,
+      "grad_norm": 0.7657225131988525,
+      "learning_rate": 0.0001,
+      "loss": 0.053,
+      "step": 25580
+    },
+    {
+      "epoch": 72.69886363636364,
+      "grad_norm": 0.6369885802268982,
+      "learning_rate": 0.0001,
+      "loss": 0.0537,
+      "step": 25590
+    },
+    {
+      "epoch": 72.72727272727273,
+      "grad_norm": 0.82183837890625,
+      "learning_rate": 0.0001,
+      "loss": 0.0503,
+      "step": 25600
+    },
+    {
+      "epoch": 72.75568181818181,
+      "grad_norm": 0.670074999332428,
+      "learning_rate": 0.0001,
+      "loss": 0.0508,
+      "step": 25610
+    },
+    {
+      "epoch": 72.7840909090909,
+      "grad_norm": 0.7039807438850403,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 25620
+    },
+    {
+      "epoch": 72.8125,
+      "grad_norm": 0.6067477464675903,
+      "learning_rate": 0.0001,
+      "loss": 0.0516,
+      "step": 25630
+    },
+    {
+      "epoch": 72.8409090909091,
+      "grad_norm": 0.6139563918113708,
+      "learning_rate": 0.0001,
+      "loss": 0.0485,
+      "step": 25640
+    },
+    {
+      "epoch": 72.86931818181819,
+      "grad_norm": 0.8261796832084656,
+      "learning_rate": 0.0001,
+      "loss": 0.0477,
+      "step": 25650
+    },
+    {
+      "epoch": 72.89772727272727,
+      "grad_norm": 0.7676610350608826,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 25660
+    },
+    {
+      "epoch": 72.92613636363636,
+      "grad_norm": 0.6294236779212952,
+      "learning_rate": 0.0001,
+      "loss": 0.0501,
+      "step": 25670
+    },
+    {
+      "epoch": 72.95454545454545,
+      "grad_norm": 0.5884844660758972,
+      "learning_rate": 0.0001,
+      "loss": 0.047,
+      "step": 25680
+    },
+    {
+      "epoch": 72.98295454545455,
+      "grad_norm": 0.6261518001556396,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 25690
+    },
+    {
+      "epoch": 73.01136363636364,
+      "grad_norm": 0.5874741673469543,
+      "learning_rate": 0.0001,
+      "loss": 0.0492,
+      "step": 25700
+    },
+    {
+      "epoch": 73.03977272727273,
+      "grad_norm": 0.558462917804718,
+      "learning_rate": 0.0001,
+      "loss": 0.0485,
+      "step": 25710
+    },
+    {
+      "epoch": 73.06818181818181,
+      "grad_norm": 1.5096638202667236,
+      "learning_rate": 0.0001,
+      "loss": 0.0482,
+      "step": 25720
+    },
+    {
+      "epoch": 73.0965909090909,
+      "grad_norm": 0.8970916867256165,
+      "learning_rate": 0.0001,
+      "loss": 0.0485,
+      "step": 25730
+    },
+    {
+      "epoch": 73.125,
+      "grad_norm": 1.2470207214355469,
+      "learning_rate": 0.0001,
+      "loss": 0.0474,
+      "step": 25740
+    },
+    {
+      "epoch": 73.1534090909091,
+      "grad_norm": 1.1245193481445312,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 25750
+    },
+    {
+      "epoch": 73.18181818181819,
+      "grad_norm": 1.018660068511963,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 25760
+    },
+    {
+      "epoch": 73.21022727272727,
+      "grad_norm": 1.0547358989715576,
+      "learning_rate": 0.0001,
+      "loss": 0.049,
+      "step": 25770
+    },
+    {
+      "epoch": 73.23863636363636,
+      "grad_norm": 1.0074411630630493,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 25780
+    },
+    {
+      "epoch": 73.26704545454545,
+      "grad_norm": 1.107343077659607,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 25790
+    },
+    {
+      "epoch": 73.29545454545455,
+      "grad_norm": 1.0003204345703125,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 25800
+    },
+    {
+      "epoch": 73.32386363636364,
+      "grad_norm": 1.152951955795288,
+      "learning_rate": 0.0001,
+      "loss": 0.0485,
+      "step": 25810
+    },
+    {
+      "epoch": 73.35227272727273,
+      "grad_norm": 1.0957775115966797,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 25820
+    },
+    {
+      "epoch": 73.38068181818181,
+      "grad_norm": 0.9931585192680359,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 25830
+    },
+    {
+      "epoch": 73.4090909090909,
+      "grad_norm": 0.9358384609222412,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 25840
+    },
+    {
+      "epoch": 73.4375,
+      "grad_norm": 1.6903265714645386,
+      "learning_rate": 0.0001,
+      "loss": 0.0561,
+      "step": 25850
+    },
+    {
+      "epoch": 73.4659090909091,
+      "grad_norm": 1.554337501525879,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 25860
+    },
+    {
+      "epoch": 73.49431818181819,
+      "grad_norm": 1.531087875366211,
+      "learning_rate": 0.0001,
+      "loss": 0.0486,
+      "step": 25870
+    },
+    {
+      "epoch": 73.52272727272727,
+      "grad_norm": 1.144465684890747,
+      "learning_rate": 0.0001,
+      "loss": 0.0472,
+      "step": 25880
+    },
+    {
+      "epoch": 73.55113636363636,
+      "grad_norm": 1.1745659112930298,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 25890
+    },
+    {
+      "epoch": 73.57954545454545,
+      "grad_norm": 1.1764057874679565,
+      "learning_rate": 0.0001,
+      "loss": 0.0474,
+      "step": 25900
+    },
+    {
+      "epoch": 73.60795454545455,
+      "grad_norm": 1.2074421644210815,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 25910
+    },
+    {
+      "epoch": 73.63636363636364,
+      "grad_norm": 1.036546230316162,
+      "learning_rate": 0.0001,
+      "loss": 0.0474,
+      "step": 25920
+    },
+    {
+      "epoch": 73.66477272727273,
+      "grad_norm": 1.5946331024169922,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 25930
+    },
+    {
+      "epoch": 73.69318181818181,
+      "grad_norm": 1.7375640869140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0482,
+      "step": 25940
+    },
+    {
+      "epoch": 73.7215909090909,
+      "grad_norm": 1.494059443473816,
+      "learning_rate": 0.0001,
+      "loss": 0.0498,
+      "step": 25950
+    },
+    {
+      "epoch": 73.75,
+      "grad_norm": 1.3823585510253906,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 25960
+    },
+    {
+      "epoch": 73.7784090909091,
+      "grad_norm": 1.4736156463623047,
+      "learning_rate": 0.0001,
+      "loss": 0.0476,
+      "step": 25970
+    },
+    {
+      "epoch": 73.80681818181819,
+      "grad_norm": 1.3302404880523682,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 25980
+    },
+    {
+      "epoch": 73.83522727272727,
+      "grad_norm": 1.0495837926864624,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 25990
+    },
+    {
+      "epoch": 73.86363636363636,
+      "grad_norm": 1.1226849555969238,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 26000
+    },
+    {
+      "epoch": 73.89204545454545,
+      "grad_norm": 1.1718180179595947,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 26010
+    },
+    {
+      "epoch": 73.92045454545455,
+      "grad_norm": 1.1437042951583862,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 26020
+    },
+    {
+      "epoch": 73.94886363636364,
+      "grad_norm": 0.9460147619247437,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 26030
+    },
+    {
+      "epoch": 73.97727272727273,
+      "grad_norm": 0.7734537720680237,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 26040
+    },
+    {
+      "epoch": 74.00568181818181,
+      "grad_norm": 0.7331590056419373,
+      "learning_rate": 0.0001,
+      "loss": 0.0479,
+      "step": 26050
+    },
+    {
+      "epoch": 74.0340909090909,
+      "grad_norm": 0.8983361721038818,
+      "learning_rate": 0.0001,
+      "loss": 0.0481,
+      "step": 26060
+    },
+    {
+      "epoch": 74.0625,
+      "grad_norm": 0.7752969861030579,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 26070
+    },
+    {
+      "epoch": 74.0909090909091,
+      "grad_norm": 1.147444725036621,
+      "learning_rate": 0.0001,
+      "loss": 0.0488,
+      "step": 26080
+    },
+    {
+      "epoch": 74.11931818181819,
+      "grad_norm": 1.5672545433044434,
+      "learning_rate": 0.0001,
+      "loss": 0.0646,
+      "step": 26090
+    },
+    {
+      "epoch": 74.14772727272727,
+      "grad_norm": 1.861153483390808,
+      "learning_rate": 0.0001,
+      "loss": 0.0496,
+      "step": 26100
+    },
+    {
+      "epoch": 74.17613636363636,
+      "grad_norm": 2.55692720413208,
+      "learning_rate": 0.0001,
+      "loss": 0.0485,
+      "step": 26110
+    },
+    {
+      "epoch": 74.20454545454545,
+      "grad_norm": 1.756406545639038,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 26120
+    },
+    {
+      "epoch": 74.23295454545455,
+      "grad_norm": 1.4010676145553589,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "step": 26130
+    },
+    {
+      "epoch": 74.26136363636364,
+      "grad_norm": 1.0524970293045044,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 26140
+    },
+    {
+      "epoch": 74.28977272727273,
+      "grad_norm": 1.089568853378296,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 26150
+    },
+    {
+      "epoch": 74.31818181818181,
+      "grad_norm": 1.5746029615402222,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 26160
+    },
+    {
+      "epoch": 74.3465909090909,
+      "grad_norm": 1.353350281715393,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 26170
+    },
+    {
+      "epoch": 74.375,
+      "grad_norm": 0.9193561673164368,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 26180
+    },
+    {
+      "epoch": 74.4034090909091,
+      "grad_norm": 0.794005811214447,
+      "learning_rate": 0.0001,
+      "loss": 0.0444,
+      "step": 26190
+    },
+    {
+      "epoch": 74.43181818181819,
+      "grad_norm": 0.7287346124649048,
+      "learning_rate": 0.0001,
+      "loss": 0.0478,
+      "step": 26200
+    },
+    {
+      "epoch": 74.46022727272727,
+      "grad_norm": 0.9359661340713501,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 26210
+    },
+    {
+      "epoch": 74.48863636363636,
+      "grad_norm": 0.8077235221862793,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 26220
+    },
+    {
+      "epoch": 74.51704545454545,
+      "grad_norm": 0.7505087852478027,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 26230
+    },
+    {
+      "epoch": 74.54545454545455,
+      "grad_norm": 0.9545241594314575,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 26240
+    },
+    {
+      "epoch": 74.57386363636364,
+      "grad_norm": 0.8987447619438171,
+      "learning_rate": 0.0001,
+      "loss": 0.0479,
+      "step": 26250
+    },
+    {
+      "epoch": 74.60227272727273,
+      "grad_norm": 0.7390574812889099,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 26260
+    },
+    {
+      "epoch": 74.63068181818181,
+      "grad_norm": 0.7332533001899719,
+      "learning_rate": 0.0001,
+      "loss": 0.0501,
+      "step": 26270
+    },
+    {
+      "epoch": 74.6590909090909,
+      "grad_norm": 0.9167255163192749,
+      "learning_rate": 0.0001,
+      "loss": 0.0473,
+      "step": 26280
+    },
+    {
+      "epoch": 74.6875,
+      "grad_norm": 0.6819099187850952,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 26290
+    },
+    {
+      "epoch": 74.7159090909091,
+      "grad_norm": 0.8224645853042603,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 26300
+    },
+    {
+      "epoch": 74.74431818181819,
+      "grad_norm": 0.8777086138725281,
+      "learning_rate": 0.0001,
+      "loss": 0.0493,
+      "step": 26310
+    },
+    {
+      "epoch": 74.77272727272727,
+      "grad_norm": 0.8701086044311523,
+      "learning_rate": 0.0001,
+      "loss": 0.0496,
+      "step": 26320
+    },
+    {
+      "epoch": 74.80113636363636,
+      "grad_norm": 0.8876749873161316,
+      "learning_rate": 0.0001,
+      "loss": 0.0486,
+      "step": 26330
+    },
+    {
+      "epoch": 74.82954545454545,
+      "grad_norm": 1.0940346717834473,
+      "learning_rate": 0.0001,
+      "loss": 0.05,
+      "step": 26340
+    },
+    {
+      "epoch": 74.85795454545455,
+      "grad_norm": 0.8126282095909119,
+      "learning_rate": 0.0001,
+      "loss": 0.05,
+      "step": 26350
+    },
+    {
+      "epoch": 74.88636363636364,
+      "grad_norm": 0.6576694250106812,
+      "learning_rate": 0.0001,
+      "loss": 0.0477,
+      "step": 26360
+    },
+    {
+      "epoch": 74.91477272727273,
+      "grad_norm": 0.8096992373466492,
+      "learning_rate": 0.0001,
+      "loss": 0.0492,
+      "step": 26370
+    },
+    {
+      "epoch": 74.94318181818181,
+      "grad_norm": 0.7710022926330566,
+      "learning_rate": 0.0001,
+      "loss": 0.0469,
+      "step": 26380
+    },
+    {
+      "epoch": 74.9715909090909,
+      "grad_norm": 0.6302091479301453,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 26390
+    },
+    {
+      "epoch": 75.0,
+      "grad_norm": 0.5126988291740417,
+      "learning_rate": 0.0001,
+      "loss": 0.0476,
+      "step": 26400
+    },
+    {
+      "epoch": 75.0284090909091,
+      "grad_norm": 0.6366093754768372,
+      "learning_rate": 0.0001,
+      "loss": 0.0481,
+      "step": 26410
+    },
+    {
+      "epoch": 75.05681818181819,
+      "grad_norm": 0.6395828127861023,
+      "learning_rate": 0.0001,
+      "loss": 0.0488,
+      "step": 26420
+    },
+    {
+      "epoch": 75.08522727272727,
+      "grad_norm": 0.6386353969573975,
+      "learning_rate": 0.0001,
+      "loss": 0.0515,
+      "step": 26430
+    },
+    {
+      "epoch": 75.11363636363636,
+      "grad_norm": 0.7400466799736023,
+      "learning_rate": 0.0001,
+      "loss": 0.0502,
+      "step": 26440
+    },
+    {
+      "epoch": 75.14204545454545,
+      "grad_norm": 0.6518636345863342,
+      "learning_rate": 0.0001,
+      "loss": 0.0503,
+      "step": 26450
+    },
+    {
+      "epoch": 75.17045454545455,
+      "grad_norm": 1.10001540184021,
+      "learning_rate": 0.0001,
+      "loss": 0.0503,
+      "step": 26460
+    },
+    {
+      "epoch": 75.19886363636364,
+      "grad_norm": 1.4311150312423706,
+      "learning_rate": 0.0001,
+      "loss": 0.0517,
+      "step": 26470
+    },
+    {
+      "epoch": 75.22727272727273,
+      "grad_norm": 1.1874371767044067,
+      "learning_rate": 0.0001,
+      "loss": 0.0502,
+      "step": 26480
+    },
+    {
+      "epoch": 75.25568181818181,
+      "grad_norm": 1.4475220441818237,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 26490
+    },
+    {
+      "epoch": 75.2840909090909,
+      "grad_norm": 1.1500955820083618,
+      "learning_rate": 0.0001,
+      "loss": 0.049,
+      "step": 26500
+    },
+    {
+      "epoch": 75.3125,
+      "grad_norm": 1.1389282941818237,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 26510
+    },
+    {
+      "epoch": 75.3409090909091,
+      "grad_norm": 0.9480587840080261,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 26520
+    },
+    {
+      "epoch": 75.36931818181819,
+      "grad_norm": 0.8605413436889648,
+      "learning_rate": 0.0001,
+      "loss": 0.0445,
+      "step": 26530
+    },
+    {
+      "epoch": 75.39772727272727,
+      "grad_norm": 0.8746979832649231,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "step": 26540
+    },
+    {
+      "epoch": 75.42613636363636,
+      "grad_norm": 0.9024845957756042,
+      "learning_rate": 0.0001,
+      "loss": 0.0472,
+      "step": 26550
+    },
+    {
+      "epoch": 75.45454545454545,
+      "grad_norm": 1.183405876159668,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "step": 26560
+    },
+    {
+      "epoch": 75.48295454545455,
+      "grad_norm": 0.9437256455421448,
+      "learning_rate": 0.0001,
+      "loss": 0.0477,
+      "step": 26570
+    },
+    {
+      "epoch": 75.51136363636364,
+      "grad_norm": 0.8547454476356506,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 26580
+    },
+    {
+      "epoch": 75.53977272727273,
+      "grad_norm": 0.8393595218658447,
+      "learning_rate": 0.0001,
+      "loss": 0.0486,
+      "step": 26590
+    },
+    {
+      "epoch": 75.56818181818181,
+      "grad_norm": 0.9341287016868591,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 26600
+    },
+    {
+      "epoch": 75.5965909090909,
+      "grad_norm": 0.9698324203491211,
+      "learning_rate": 0.0001,
+      "loss": 0.0483,
+      "step": 26610
+    },
+    {
+      "epoch": 75.625,
+      "grad_norm": 0.8105787038803101,
+      "learning_rate": 0.0001,
+      "loss": 0.0484,
+      "step": 26620
+    },
+    {
+      "epoch": 75.6534090909091,
+      "grad_norm": 1.2497771978378296,
+      "learning_rate": 0.0001,
+      "loss": 0.049,
+      "step": 26630
+    },
+    {
+      "epoch": 75.68181818181819,
+      "grad_norm": 1.0008628368377686,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 26640
+    },
+    {
+      "epoch": 75.71022727272727,
+      "grad_norm": 1.203723669052124,
+      "learning_rate": 0.0001,
+      "loss": 0.0483,
+      "step": 26650
+    },
+    {
+      "epoch": 75.73863636363636,
+      "grad_norm": 1.1215006113052368,
+      "learning_rate": 0.0001,
+      "loss": 0.0483,
+      "step": 26660
+    },
+    {
+      "epoch": 75.76704545454545,
+      "grad_norm": 0.9968708753585815,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 26670
+    },
+    {
+      "epoch": 75.79545454545455,
+      "grad_norm": 0.8633506298065186,
+      "learning_rate": 0.0001,
+      "loss": 0.0472,
+      "step": 26680
+    },
+    {
+      "epoch": 75.82386363636364,
+      "grad_norm": 1.0756628513336182,
+      "learning_rate": 0.0001,
+      "loss": 0.0486,
+      "step": 26690
+    },
+    {
+      "epoch": 75.85227272727273,
+      "grad_norm": 0.7618953585624695,
+      "learning_rate": 0.0001,
+      "loss": 0.0493,
+      "step": 26700
+    },
+    {
+      "epoch": 75.88068181818181,
+      "grad_norm": 0.9463710188865662,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 26710
+    },
+    {
+      "epoch": 75.9090909090909,
+      "grad_norm": 0.7390015125274658,
+      "learning_rate": 0.0001,
+      "loss": 0.0489,
+      "step": 26720
+    },
+    {
+      "epoch": 75.9375,
+      "grad_norm": 0.5806778073310852,
+      "learning_rate": 0.0001,
+      "loss": 0.0506,
+      "step": 26730
+    },
+    {
+      "epoch": 75.9659090909091,
+      "grad_norm": 0.6981925368309021,
+      "learning_rate": 0.0001,
+      "loss": 0.0498,
+      "step": 26740
+    },
+    {
+      "epoch": 75.99431818181819,
+      "grad_norm": 0.7363477349281311,
+      "learning_rate": 0.0001,
+      "loss": 0.0493,
+      "step": 26750
+    },
+    {
+      "epoch": 76.02272727272727,
+      "grad_norm": 0.7735875248908997,
+      "learning_rate": 0.0001,
+      "loss": 0.0491,
+      "step": 26760
+    },
+    {
+      "epoch": 76.05113636363636,
+      "grad_norm": 0.6567436456680298,
+      "learning_rate": 0.0001,
+      "loss": 0.0489,
+      "step": 26770
+    },
+    {
+      "epoch": 76.07954545454545,
+      "grad_norm": 0.6639755368232727,
+      "learning_rate": 0.0001,
+      "loss": 0.0481,
+      "step": 26780
+    },
+    {
+      "epoch": 76.10795454545455,
+      "grad_norm": 0.5334902405738831,
+      "learning_rate": 0.0001,
+      "loss": 0.0481,
+      "step": 26790
+    },
+    {
+      "epoch": 76.13636363636364,
+      "grad_norm": 0.6336926221847534,
+      "learning_rate": 0.0001,
+      "loss": 0.0481,
+      "step": 26800
+    },
+    {
+      "epoch": 76.16477272727273,
+      "grad_norm": 0.5552213191986084,
+      "learning_rate": 0.0001,
+      "loss": 0.0492,
+      "step": 26810
+    },
+    {
+      "epoch": 76.19318181818181,
+      "grad_norm": 0.5877450108528137,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 26820
+    },
+    {
+      "epoch": 76.2215909090909,
+      "grad_norm": 0.6252912878990173,
+      "learning_rate": 0.0001,
+      "loss": 0.0491,
+      "step": 26830
+    },
+    {
+      "epoch": 76.25,
+      "grad_norm": 0.7641172409057617,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 26840
+    },
+    {
+      "epoch": 76.2784090909091,
+      "grad_norm": 0.5870921611785889,
+      "learning_rate": 0.0001,
+      "loss": 0.0491,
+      "step": 26850
+    },
+    {
+      "epoch": 76.30681818181819,
+      "grad_norm": 0.6940385699272156,
+      "learning_rate": 0.0001,
+      "loss": 0.0473,
+      "step": 26860
+    },
+    {
+      "epoch": 76.33522727272727,
+      "grad_norm": 0.6808137893676758,
+      "learning_rate": 0.0001,
+      "loss": 0.0497,
+      "step": 26870
+    },
+    {
+      "epoch": 76.36363636363636,
+      "grad_norm": 0.7470855712890625,
+      "learning_rate": 0.0001,
+      "loss": 0.0478,
+      "step": 26880
+    },
+    {
+      "epoch": 76.39204545454545,
+      "grad_norm": 0.7515180110931396,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 26890
+    },
+    {
+      "epoch": 76.42045454545455,
+      "grad_norm": 0.5889479517936707,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 26900
+    },
+    {
+      "epoch": 76.44886363636364,
+      "grad_norm": 0.600273609161377,
+      "learning_rate": 0.0001,
+      "loss": 0.0479,
+      "step": 26910
+    },
+    {
+      "epoch": 76.47727272727273,
+      "grad_norm": 0.7066619396209717,
+      "learning_rate": 0.0001,
+      "loss": 0.05,
+      "step": 26920
+    },
+    {
+      "epoch": 76.50568181818181,
+      "grad_norm": 0.794434666633606,
+      "learning_rate": 0.0001,
+      "loss": 0.0493,
+      "step": 26930
+    },
+    {
+      "epoch": 76.5340909090909,
+      "grad_norm": 0.517598569393158,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 26940
+    },
+    {
+      "epoch": 76.5625,
+      "grad_norm": 0.7150055170059204,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 26950
+    },
+    {
+      "epoch": 76.5909090909091,
+      "grad_norm": 0.6902920603752136,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 26960
+    },
+    {
+      "epoch": 76.61931818181819,
+      "grad_norm": 0.8069965839385986,
+      "learning_rate": 0.0001,
+      "loss": 0.0476,
+      "step": 26970
+    },
+    {
+      "epoch": 76.64772727272727,
+      "grad_norm": 0.7762559056282043,
+      "learning_rate": 0.0001,
+      "loss": 0.0477,
+      "step": 26980
+    },
+    {
+      "epoch": 76.67613636363636,
+      "grad_norm": 0.7302852272987366,
+      "learning_rate": 0.0001,
+      "loss": 0.0484,
+      "step": 26990
+    },
+    {
+      "epoch": 76.70454545454545,
+      "grad_norm": 0.7422618865966797,
+      "learning_rate": 0.0001,
+      "loss": 0.0469,
+      "step": 27000
+    },
+    {
+      "epoch": 76.73295454545455,
+      "grad_norm": 0.719369649887085,
+      "learning_rate": 0.0001,
+      "loss": 0.0477,
+      "step": 27010
+    },
+    {
+      "epoch": 76.76136363636364,
+      "grad_norm": 0.7470158338546753,
+      "learning_rate": 0.0001,
+      "loss": 0.0466,
+      "step": 27020
+    },
+    {
+      "epoch": 76.78977272727273,
+      "grad_norm": 0.8050602078437805,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 27030
+    },
+    {
+      "epoch": 76.81818181818181,
+      "grad_norm": 0.617423415184021,
+      "learning_rate": 0.0001,
+      "loss": 0.0486,
+      "step": 27040
+    },
+    {
+      "epoch": 76.8465909090909,
+      "grad_norm": 0.505489706993103,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 27050
+    },
+    {
+      "epoch": 76.875,
+      "grad_norm": 0.5494895577430725,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 27060
+    },
+    {
+      "epoch": 76.9034090909091,
+      "grad_norm": 0.5236529111862183,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 27070
+    },
+    {
+      "epoch": 76.93181818181819,
+      "grad_norm": 0.6138285994529724,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 27080
+    },
+    {
+      "epoch": 76.96022727272727,
+      "grad_norm": 0.6185349225997925,
+      "learning_rate": 0.0001,
+      "loss": 0.0499,
+      "step": 27090
+    },
+    {
+      "epoch": 76.98863636363636,
+      "grad_norm": 0.6920495629310608,
+      "learning_rate": 0.0001,
+      "loss": 0.0496,
+      "step": 27100
+    },
+    {
+      "epoch": 77.01704545454545,
+      "grad_norm": 0.7093076109886169,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 27110
+    },
+    {
+      "epoch": 77.04545454545455,
+      "grad_norm": 0.9286143779754639,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 27120
+    },
+    {
+      "epoch": 77.07386363636364,
+      "grad_norm": 1.0601509809494019,
+      "learning_rate": 0.0001,
+      "loss": 0.0479,
+      "step": 27130
+    },
+    {
+      "epoch": 77.10227272727273,
+      "grad_norm": 0.9617673754692078,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 27140
+    },
+    {
+      "epoch": 77.13068181818181,
+      "grad_norm": 1.2405085563659668,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 27150
+    },
+    {
+      "epoch": 77.1590909090909,
+      "grad_norm": 0.8260979056358337,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 27160
+    },
+    {
+      "epoch": 77.1875,
+      "grad_norm": 0.6605196595191956,
+      "learning_rate": 0.0001,
+      "loss": 0.047,
+      "step": 27170
+    },
+    {
+      "epoch": 77.2159090909091,
+      "grad_norm": 0.7948494553565979,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 27180
+    },
+    {
+      "epoch": 77.24431818181819,
+      "grad_norm": 0.8191278576850891,
+      "learning_rate": 0.0001,
+      "loss": 0.0479,
+      "step": 27190
+    },
+    {
+      "epoch": 77.27272727272727,
+      "grad_norm": 0.7254536151885986,
+      "learning_rate": 0.0001,
+      "loss": 0.0496,
+      "step": 27200
+    },
+    {
+      "epoch": 77.30113636363636,
+      "grad_norm": 0.6482180953025818,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "step": 27210
+    },
+    {
+      "epoch": 77.32954545454545,
+      "grad_norm": 0.8210635781288147,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 27220
+    },
+    {
+      "epoch": 77.35795454545455,
+      "grad_norm": 0.6660655736923218,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 27230
+    },
+    {
+      "epoch": 77.38636363636364,
+      "grad_norm": 0.6348584294319153,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 27240
+    },
+    {
+      "epoch": 77.41477272727273,
+      "grad_norm": 0.8718886375427246,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 27250
+    },
+    {
+      "epoch": 77.44318181818181,
+      "grad_norm": 0.919781506061554,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 27260
+    },
+    {
+      "epoch": 77.4715909090909,
+      "grad_norm": 0.9934787154197693,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 27270
+    },
+    {
+      "epoch": 77.5,
+      "grad_norm": 0.9608179330825806,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 27280
+    },
+    {
+      "epoch": 77.5284090909091,
+      "grad_norm": 0.7589172720909119,
+      "learning_rate": 0.0001,
+      "loss": 0.0491,
+      "step": 27290
+    },
+    {
+      "epoch": 77.55681818181819,
+      "grad_norm": 0.9856165647506714,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "step": 27300
+    },
+    {
+      "epoch": 77.58522727272727,
+      "grad_norm": 0.8956001996994019,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 27310
+    },
+    {
+      "epoch": 77.61363636363636,
+      "grad_norm": 1.1567500829696655,
+      "learning_rate": 0.0001,
+      "loss": 0.048,
+      "step": 27320
+    },
+    {
+      "epoch": 77.64204545454545,
+      "grad_norm": 0.9312670826911926,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 27330
+    },
+    {
+      "epoch": 77.67045454545455,
+      "grad_norm": 0.8068075776100159,
+      "learning_rate": 0.0001,
+      "loss": 0.0451,
+      "step": 27340
+    },
+    {
+      "epoch": 77.69886363636364,
+      "grad_norm": 0.5929557085037231,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 27350
+    },
+    {
+      "epoch": 77.72727272727273,
+      "grad_norm": 0.8782048225402832,
+      "learning_rate": 0.0001,
+      "loss": 0.0467,
+      "step": 27360
+    },
+    {
+      "epoch": 77.75568181818181,
+      "grad_norm": 1.0722332000732422,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 27370
+    },
+    {
+      "epoch": 77.7840909090909,
+      "grad_norm": 0.7801492810249329,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 27380
+    },
+    {
+      "epoch": 77.8125,
+      "grad_norm": 0.605384886264801,
+      "learning_rate": 0.0001,
+      "loss": 0.0472,
+      "step": 27390
+    },
+    {
+      "epoch": 77.8409090909091,
+      "grad_norm": 0.9070475101470947,
+      "learning_rate": 0.0001,
+      "loss": 0.0495,
+      "step": 27400
+    },
+    {
+      "epoch": 77.86931818181819,
+      "grad_norm": 1.0343130826950073,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 27410
+    },
+    {
+      "epoch": 77.89772727272727,
+      "grad_norm": 0.7611730098724365,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 27420
+    },
+    {
+      "epoch": 77.92613636363636,
+      "grad_norm": 0.8008614182472229,
+      "learning_rate": 0.0001,
+      "loss": 0.0469,
+      "step": 27430
+    },
+    {
+      "epoch": 77.95454545454545,
+      "grad_norm": 0.7293544411659241,
+      "learning_rate": 0.0001,
+      "loss": 0.0478,
+      "step": 27440
+    },
+    {
+      "epoch": 77.98295454545455,
+      "grad_norm": 0.832565188407898,
+      "learning_rate": 0.0001,
+      "loss": 0.0474,
+      "step": 27450
+    },
+    {
+      "epoch": 78.01136363636364,
+      "grad_norm": 0.7416606545448303,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 27460
+    },
+    {
+      "epoch": 78.03977272727273,
+      "grad_norm": 0.8914027214050293,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 27470
+    },
+    {
+      "epoch": 78.06818181818181,
+      "grad_norm": 0.6268876194953918,
+      "learning_rate": 0.0001,
+      "loss": 0.0446,
+      "step": 27480
+    },
+    {
+      "epoch": 78.0965909090909,
+      "grad_norm": 0.7498577237129211,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 27490
+    },
+    {
+      "epoch": 78.125,
+      "grad_norm": 0.7658631801605225,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "step": 27500
+    },
+    {
+      "epoch": 78.1534090909091,
+      "grad_norm": 0.9924762845039368,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 27510
+    },
+    {
+      "epoch": 78.18181818181819,
+      "grad_norm": 0.8507946133613586,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 27520
+    },
+    {
+      "epoch": 78.21022727272727,
+      "grad_norm": 0.8294076323509216,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 27530
+    },
+    {
+      "epoch": 78.23863636363636,
+      "grad_norm": 0.8344864249229431,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 27540
+    },
+    {
+      "epoch": 78.26704545454545,
+      "grad_norm": 0.6620252132415771,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 27550
+    },
+    {
+      "epoch": 78.29545454545455,
+      "grad_norm": 0.7037463784217834,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 27560
+    },
+    {
+      "epoch": 78.32386363636364,
+      "grad_norm": 0.7051752209663391,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 27570
+    },
+    {
+      "epoch": 78.35227272727273,
+      "grad_norm": 0.7858211398124695,
+      "learning_rate": 0.0001,
+      "loss": 0.0467,
+      "step": 27580
+    },
+    {
+      "epoch": 78.38068181818181,
+      "grad_norm": 0.7993125915527344,
+      "learning_rate": 0.0001,
+      "loss": 0.0469,
+      "step": 27590
+    },
+    {
+      "epoch": 78.4090909090909,
+      "grad_norm": 0.7271113395690918,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 27600
+    },
+    {
+      "epoch": 78.4375,
+      "grad_norm": 0.7963870763778687,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 27610
+    },
+    {
+      "epoch": 78.4659090909091,
+      "grad_norm": 0.9144273400306702,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 27620
+    },
+    {
+      "epoch": 78.49431818181819,
+      "grad_norm": 0.8622909784317017,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 27630
+    },
+    {
+      "epoch": 78.52272727272727,
+      "grad_norm": 0.7054391503334045,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 27640
+    },
+    {
+      "epoch": 78.55113636363636,
+      "grad_norm": 0.7337654232978821,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 27650
+    },
+    {
+      "epoch": 78.57954545454545,
+      "grad_norm": 0.6751934885978699,
+      "learning_rate": 0.0001,
+      "loss": 0.0476,
+      "step": 27660
+    },
+    {
+      "epoch": 78.60795454545455,
+      "grad_norm": 0.7194545269012451,
+      "learning_rate": 0.0001,
+      "loss": 0.047,
+      "step": 27670
+    },
+    {
+      "epoch": 78.63636363636364,
+      "grad_norm": 0.7210686802864075,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 27680
+    },
+    {
+      "epoch": 78.66477272727273,
+      "grad_norm": 0.9098225235939026,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 27690
+    },
+    {
+      "epoch": 78.69318181818181,
+      "grad_norm": 0.9643121361732483,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 27700
+    },
+    {
+      "epoch": 78.7215909090909,
+      "grad_norm": 1.057265043258667,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 27710
+    },
+    {
+      "epoch": 78.75,
+      "grad_norm": 0.8858153820037842,
+      "learning_rate": 0.0001,
+      "loss": 0.0477,
+      "step": 27720
+    },
+    {
+      "epoch": 78.7784090909091,
+      "grad_norm": 0.7570127248764038,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 27730
+    },
+    {
+      "epoch": 78.80681818181819,
+      "grad_norm": 0.8737295269966125,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "step": 27740
+    },
+    {
+      "epoch": 78.83522727272727,
+      "grad_norm": 0.8919824957847595,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 27750
+    },
+    {
+      "epoch": 78.86363636363636,
+      "grad_norm": 0.834999144077301,
+      "learning_rate": 0.0001,
+      "loss": 0.0444,
+      "step": 27760
+    },
+    {
+      "epoch": 78.89204545454545,
+      "grad_norm": 0.7257569432258606,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 27770
+    },
+    {
+      "epoch": 78.92045454545455,
+      "grad_norm": 0.7738620042800903,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 27780
+    },
+    {
+      "epoch": 78.94886363636364,
+      "grad_norm": 0.6493619084358215,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 27790
+    },
+    {
+      "epoch": 78.97727272727273,
+      "grad_norm": 0.7581443190574646,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 27800
+    },
+    {
+      "epoch": 79.00568181818181,
+      "grad_norm": 0.8084586262702942,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 27810
+    },
+    {
+      "epoch": 79.0340909090909,
+      "grad_norm": 0.6828577518463135,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "step": 27820
+    },
+    {
+      "epoch": 79.0625,
+      "grad_norm": 0.6780984401702881,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 27830
+    },
+    {
+      "epoch": 79.0909090909091,
+      "grad_norm": 0.7520745396614075,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 27840
+    },
+    {
+      "epoch": 79.11931818181819,
+      "grad_norm": 0.6034306883811951,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 27850
+    },
+    {
+      "epoch": 79.14772727272727,
+      "grad_norm": 0.8241128921508789,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 27860
+    },
+    {
+      "epoch": 79.17613636363636,
+      "grad_norm": 0.7252616286277771,
+      "learning_rate": 0.0001,
+      "loss": 0.0445,
+      "step": 27870
+    },
+    {
+      "epoch": 79.20454545454545,
+      "grad_norm": 0.8703776597976685,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "step": 27880
+    },
+    {
+      "epoch": 79.23295454545455,
+      "grad_norm": 0.6853988766670227,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 27890
+    },
+    {
+      "epoch": 79.26136363636364,
+      "grad_norm": 0.5899876356124878,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 27900
+    },
+    {
+      "epoch": 79.28977272727273,
+      "grad_norm": 0.5329005122184753,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 27910
+    },
+    {
+      "epoch": 79.31818181818181,
+      "grad_norm": 0.5844208002090454,
+      "learning_rate": 0.0001,
+      "loss": 0.0449,
+      "step": 27920
+    },
+    {
+      "epoch": 79.3465909090909,
+      "grad_norm": 0.5217543840408325,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 27930
+    },
+    {
+      "epoch": 79.375,
+      "grad_norm": 0.631199300289154,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 27940
+    },
+    {
+      "epoch": 79.4034090909091,
+      "grad_norm": 0.5435271859169006,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 27950
+    },
+    {
+      "epoch": 79.43181818181819,
+      "grad_norm": 0.5514788627624512,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 27960
+    },
+    {
+      "epoch": 79.46022727272727,
+      "grad_norm": 0.6063737273216248,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 27970
+    },
+    {
+      "epoch": 79.48863636363636,
+      "grad_norm": 0.6440352201461792,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "step": 27980
+    },
+    {
+      "epoch": 79.51704545454545,
+      "grad_norm": 0.6347674131393433,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 27990
+    },
+    {
+      "epoch": 79.54545454545455,
+      "grad_norm": 0.5119302272796631,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 28000
+    },
+    {
+      "epoch": 79.57386363636364,
+      "grad_norm": 0.665009617805481,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 28010
+    },
+    {
+      "epoch": 79.60227272727273,
+      "grad_norm": 1.1492528915405273,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 28020
+    },
+    {
+      "epoch": 79.63068181818181,
+      "grad_norm": 0.6289621591567993,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 28030
+    },
+    {
+      "epoch": 79.6590909090909,
+      "grad_norm": 0.6949747204780579,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "step": 28040
+    },
+    {
+      "epoch": 79.6875,
+      "grad_norm": 0.7128562331199646,
+      "learning_rate": 0.0001,
+      "loss": 0.0487,
+      "step": 28050
+    },
+    {
+      "epoch": 79.7159090909091,
+      "grad_norm": 0.7679532766342163,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 28060
+    },
+    {
+      "epoch": 79.74431818181819,
+      "grad_norm": 1.129748821258545,
+      "learning_rate": 0.0001,
+      "loss": 0.0485,
+      "step": 28070
+    },
+    {
+      "epoch": 79.77272727272727,
+      "grad_norm": 1.1302276849746704,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 28080
+    },
+    {
+      "epoch": 79.80113636363636,
+      "grad_norm": 1.242452621459961,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 28090
+    },
+    {
+      "epoch": 79.82954545454545,
+      "grad_norm": 1.3404399156570435,
+      "learning_rate": 0.0001,
+      "loss": 0.0474,
+      "step": 28100
+    },
+    {
+      "epoch": 79.85795454545455,
+      "grad_norm": 1.3003270626068115,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 28110
+    },
+    {
+      "epoch": 79.88636363636364,
+      "grad_norm": 1.1596304178237915,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 28120
+    },
+    {
+      "epoch": 79.91477272727273,
+      "grad_norm": 1.0483144521713257,
+      "learning_rate": 0.0001,
+      "loss": 0.0467,
+      "step": 28130
+    },
+    {
+      "epoch": 79.94318181818181,
+      "grad_norm": 0.9983393549919128,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 28140
+    },
+    {
+      "epoch": 79.9715909090909,
+      "grad_norm": 0.8313050270080566,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 28150
+    },
+    {
+      "epoch": 80.0,
+      "grad_norm": 0.9205082654953003,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 28160
+    },
+    {
+      "epoch": 80.0284090909091,
+      "grad_norm": 0.8589319586753845,
+      "learning_rate": 0.0001,
+      "loss": 0.0464,
+      "step": 28170
+    },
+    {
+      "epoch": 80.05681818181819,
+      "grad_norm": 0.9702844023704529,
+      "learning_rate": 0.0001,
+      "loss": 0.0424,
+      "step": 28180
+    },
+    {
+      "epoch": 80.08522727272727,
+      "grad_norm": 0.8930377960205078,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 28190
+    },
+    {
+      "epoch": 80.11363636363636,
+      "grad_norm": 0.721045970916748,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 28200
+    },
+    {
+      "epoch": 80.14204545454545,
+      "grad_norm": 0.7323723435401917,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 28210
+    },
+    {
+      "epoch": 80.17045454545455,
+      "grad_norm": 1.3309814929962158,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 28220
+    },
+    {
+      "epoch": 80.19886363636364,
+      "grad_norm": 1.14047110080719,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 28230
+    },
+    {
+      "epoch": 80.22727272727273,
+      "grad_norm": 1.6657135486602783,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 28240
+    },
+    {
+      "epoch": 80.25568181818181,
+      "grad_norm": 1.1823036670684814,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 28250
+    },
+    {
+      "epoch": 80.2840909090909,
+      "grad_norm": 0.9169923663139343,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 28260
+    },
+    {
+      "epoch": 80.3125,
+      "grad_norm": 0.8880297541618347,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 28270
+    },
+    {
+      "epoch": 80.3409090909091,
+      "grad_norm": 0.778057336807251,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 28280
+    },
+    {
+      "epoch": 80.36931818181819,
+      "grad_norm": 0.837870180606842,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 28290
+    },
+    {
+      "epoch": 80.39772727272727,
+      "grad_norm": 0.8132756352424622,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 28300
+    },
+    {
+      "epoch": 80.42613636363636,
+      "grad_norm": 0.9789218306541443,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 28310
+    },
+    {
+      "epoch": 80.45454545454545,
+      "grad_norm": 1.0025867223739624,
+      "learning_rate": 0.0001,
+      "loss": 0.0424,
+      "step": 28320
+    },
+    {
+      "epoch": 80.48295454545455,
+      "grad_norm": 0.8962274789810181,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 28330
+    },
+    {
+      "epoch": 80.51136363636364,
+      "grad_norm": 0.9207645058631897,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 28340
+    },
+    {
+      "epoch": 80.53977272727273,
+      "grad_norm": 0.7955727577209473,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 28350
+    },
+    {
+      "epoch": 80.56818181818181,
+      "grad_norm": 0.7597567439079285,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 28360
+    },
+    {
+      "epoch": 80.5965909090909,
+      "grad_norm": 0.9111728072166443,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 28370
+    },
+    {
+      "epoch": 80.625,
+      "grad_norm": 0.7924389243125916,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 28380
+    },
+    {
+      "epoch": 80.6534090909091,
+      "grad_norm": 0.7888645529747009,
+      "learning_rate": 0.0001,
+      "loss": 0.047,
+      "step": 28390
+    },
+    {
+      "epoch": 80.68181818181819,
+      "grad_norm": 0.7492277026176453,
+      "learning_rate": 0.0001,
+      "loss": 0.0451,
+      "step": 28400
+    },
+    {
+      "epoch": 80.71022727272727,
+      "grad_norm": 0.5666723251342773,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 28410
+    },
+    {
+      "epoch": 80.73863636363636,
+      "grad_norm": 0.5718184113502502,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 28420
+    },
+    {
+      "epoch": 80.76704545454545,
+      "grad_norm": 0.5507611632347107,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 28430
+    },
+    {
+      "epoch": 80.79545454545455,
+      "grad_norm": 0.6398160457611084,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 28440
+    },
+    {
+      "epoch": 80.82386363636364,
+      "grad_norm": 0.6520999073982239,
+      "learning_rate": 0.0001,
+      "loss": 0.0446,
+      "step": 28450
+    },
+    {
+      "epoch": 80.85227272727273,
+      "grad_norm": 0.6655693054199219,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 28460
+    },
+    {
+      "epoch": 80.88068181818181,
+      "grad_norm": 0.6784167885780334,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 28470
+    },
+    {
+      "epoch": 80.9090909090909,
+      "grad_norm": 0.6713101267814636,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 28480
+    },
+    {
+      "epoch": 80.9375,
+      "grad_norm": 0.6968391537666321,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 28490
+    },
+    {
+      "epoch": 80.9659090909091,
+      "grad_norm": 0.5912553668022156,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 28500
+    },
+    {
+      "epoch": 80.99431818181819,
+      "grad_norm": 0.6505694389343262,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 28510
+    },
+    {
+      "epoch": 81.02272727272727,
+      "grad_norm": 0.6993499994277954,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 28520
+    },
+    {
+      "epoch": 81.05113636363636,
+      "grad_norm": 0.8444068431854248,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 28530
+    },
+    {
+      "epoch": 81.07954545454545,
+      "grad_norm": 0.642077624797821,
+      "learning_rate": 0.0001,
+      "loss": 0.0446,
+      "step": 28540
+    },
+    {
+      "epoch": 81.10795454545455,
+      "grad_norm": 0.6671980023384094,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 28550
+    },
+    {
+      "epoch": 81.13636363636364,
+      "grad_norm": 0.7422840595245361,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 28560
+    },
+    {
+      "epoch": 81.16477272727273,
+      "grad_norm": 0.7244125604629517,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 28570
+    },
+    {
+      "epoch": 81.19318181818181,
+      "grad_norm": 0.5740301609039307,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 28580
+    },
+    {
+      "epoch": 81.2215909090909,
+      "grad_norm": 0.5553819537162781,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 28590
+    },
+    {
+      "epoch": 81.25,
+      "grad_norm": 0.4433813989162445,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 28600
+    },
+    {
+      "epoch": 81.2784090909091,
+      "grad_norm": 0.5169538855552673,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 28610
+    },
+    {
+      "epoch": 81.30681818181819,
+      "grad_norm": 0.5083233118057251,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 28620
+    },
+    {
+      "epoch": 81.33522727272727,
+      "grad_norm": 0.5519469976425171,
+      "learning_rate": 0.0001,
+      "loss": 0.0425,
+      "step": 28630
+    },
+    {
+      "epoch": 81.36363636363636,
+      "grad_norm": 0.731152355670929,
+      "learning_rate": 0.0001,
+      "loss": 0.0425,
+      "step": 28640
+    },
+    {
+      "epoch": 81.39204545454545,
+      "grad_norm": 0.4356805086135864,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 28650
+    },
+    {
+      "epoch": 81.42045454545455,
+      "grad_norm": 0.812091052532196,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 28660
+    },
+    {
+      "epoch": 81.44886363636364,
+      "grad_norm": 0.5545047521591187,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 28670
+    },
+    {
+      "epoch": 81.47727272727273,
+      "grad_norm": 0.8585087656974792,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 28680
+    },
+    {
+      "epoch": 81.50568181818181,
+      "grad_norm": 0.7256617546081543,
+      "learning_rate": 0.0001,
+      "loss": 0.0445,
+      "step": 28690
+    },
+    {
+      "epoch": 81.5340909090909,
+      "grad_norm": 0.7761886119842529,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 28700
+    },
+    {
+      "epoch": 81.5625,
+      "grad_norm": 0.7050015330314636,
+      "learning_rate": 0.0001,
+      "loss": 0.0449,
+      "step": 28710
+    },
+    {
+      "epoch": 81.5909090909091,
+      "grad_norm": 1.3305480480194092,
+      "learning_rate": 0.0001,
+      "loss": 0.047,
+      "step": 28720
+    },
+    {
+      "epoch": 81.61931818181819,
+      "grad_norm": 0.9800511598587036,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 28730
+    },
+    {
+      "epoch": 81.64772727272727,
+      "grad_norm": 1.166397213935852,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 28740
+    },
+    {
+      "epoch": 81.67613636363636,
+      "grad_norm": 1.0785977840423584,
+      "learning_rate": 0.0001,
+      "loss": 0.0446,
+      "step": 28750
+    },
+    {
+      "epoch": 81.70454545454545,
+      "grad_norm": 0.912084698677063,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 28760
+    },
+    {
+      "epoch": 81.73295454545455,
+      "grad_norm": 0.9032609462738037,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 28770
+    },
+    {
+      "epoch": 81.76136363636364,
+      "grad_norm": 1.6895674467086792,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 28780
+    },
+    {
+      "epoch": 81.78977272727273,
+      "grad_norm": 1.4334778785705566,
+      "learning_rate": 0.0001,
+      "loss": 0.0475,
+      "step": 28790
+    },
+    {
+      "epoch": 81.81818181818181,
+      "grad_norm": 1.5415606498718262,
+      "learning_rate": 0.0001,
+      "loss": 0.0449,
+      "step": 28800
+    },
+    {
+      "epoch": 81.8465909090909,
+      "grad_norm": 1.1889894008636475,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 28810
+    },
+    {
+      "epoch": 81.875,
+      "grad_norm": 1.0240460634231567,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 28820
+    },
+    {
+      "epoch": 81.9034090909091,
+      "grad_norm": 0.9590867757797241,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 28830
+    },
+    {
+      "epoch": 81.93181818181819,
+      "grad_norm": 0.776984453201294,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 28840
+    },
+    {
+      "epoch": 81.96022727272727,
+      "grad_norm": 1.2868221998214722,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 28850
+    },
+    {
+      "epoch": 81.98863636363636,
+      "grad_norm": 1.035592794418335,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 28860
+    },
+    {
+      "epoch": 82.01704545454545,
+      "grad_norm": 1.3734735250473022,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 28870
+    },
+    {
+      "epoch": 82.04545454545455,
+      "grad_norm": 1.5599828958511353,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 28880
+    },
+    {
+      "epoch": 82.07386363636364,
+      "grad_norm": 1.185050368309021,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 28890
+    },
+    {
+      "epoch": 82.10227272727273,
+      "grad_norm": 1.5994446277618408,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 28900
+    },
+    {
+      "epoch": 82.13068181818181,
+      "grad_norm": 1.3795400857925415,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 28910
+    },
+    {
+      "epoch": 82.1590909090909,
+      "grad_norm": 1.2823818922042847,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 28920
+    },
+    {
+      "epoch": 82.1875,
+      "grad_norm": 1.39549720287323,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 28930
+    },
+    {
+      "epoch": 82.2159090909091,
+      "grad_norm": 1.1723809242248535,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 28940
+    },
+    {
+      "epoch": 82.24431818181819,
+      "grad_norm": 0.7764673829078674,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 28950
+    },
+    {
+      "epoch": 82.27272727272727,
+      "grad_norm": 0.7350073456764221,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 28960
+    },
+    {
+      "epoch": 82.30113636363636,
+      "grad_norm": 0.8376882672309875,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 28970
+    },
+    {
+      "epoch": 82.32954545454545,
+      "grad_norm": 0.7426922917366028,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 28980
+    },
+    {
+      "epoch": 82.35795454545455,
+      "grad_norm": 1.0266258716583252,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 28990
+    },
+    {
+      "epoch": 82.38636363636364,
+      "grad_norm": 0.6878785490989685,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 29000
+    },
+    {
+      "epoch": 82.41477272727273,
+      "grad_norm": 0.8112396597862244,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 29010
+    },
+    {
+      "epoch": 82.44318181818181,
+      "grad_norm": 0.7438217997550964,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 29020
+    },
+    {
+      "epoch": 82.4715909090909,
+      "grad_norm": 0.5908733010292053,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 29030
+    },
+    {
+      "epoch": 82.5,
+      "grad_norm": 0.7258604168891907,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 29040
+    },
+    {
+      "epoch": 82.5284090909091,
+      "grad_norm": 0.5995661616325378,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 29050
+    },
+    {
+      "epoch": 82.55681818181819,
+      "grad_norm": 0.7221328616142273,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 29060
+    },
+    {
+      "epoch": 82.58522727272727,
+      "grad_norm": 0.596004068851471,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 29070
+    },
+    {
+      "epoch": 82.61363636363636,
+      "grad_norm": 0.5565772652626038,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 29080
+    },
+    {
+      "epoch": 82.64204545454545,
+      "grad_norm": 0.697248637676239,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 29090
+    },
+    {
+      "epoch": 82.67045454545455,
+      "grad_norm": 0.6714693903923035,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 29100
+    },
+    {
+      "epoch": 82.69886363636364,
+      "grad_norm": 0.7975269556045532,
+      "learning_rate": 0.0001,
+      "loss": 0.0467,
+      "step": 29110
+    },
+    {
+      "epoch": 82.72727272727273,
+      "grad_norm": 0.870853841304779,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 29120
+    },
+    {
+      "epoch": 82.75568181818181,
+      "grad_norm": 0.7218012809753418,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 29130
+    },
+    {
+      "epoch": 82.7840909090909,
+      "grad_norm": 0.7033442258834839,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 29140
+    },
+    {
+      "epoch": 82.8125,
+      "grad_norm": 0.7944255471229553,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "step": 29150
+    },
+    {
+      "epoch": 82.8409090909091,
+      "grad_norm": 0.712996244430542,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 29160
+    },
+    {
+      "epoch": 82.86931818181819,
+      "grad_norm": 0.7248801589012146,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 29170
+    },
+    {
+      "epoch": 82.89772727272727,
+      "grad_norm": 0.8134949803352356,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 29180
+    },
+    {
+      "epoch": 82.92613636363636,
+      "grad_norm": 0.5927881002426147,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 29190
+    },
+    {
+      "epoch": 82.95454545454545,
+      "grad_norm": 0.5970407128334045,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 29200
+    },
+    {
+      "epoch": 82.98295454545455,
+      "grad_norm": 0.5957374572753906,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 29210
+    },
+    {
+      "epoch": 83.01136363636364,
+      "grad_norm": 1.196097493171692,
+      "learning_rate": 0.0001,
+      "loss": 0.0513,
+      "step": 29220
+    },
+    {
+      "epoch": 83.03977272727273,
+      "grad_norm": 1.275525450706482,
+      "learning_rate": 0.0001,
+      "loss": 0.0465,
+      "step": 29230
+    },
+    {
+      "epoch": 83.06818181818181,
+      "grad_norm": 1.131506085395813,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 29240
+    },
+    {
+      "epoch": 83.0965909090909,
+      "grad_norm": 1.2795711755752563,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 29250
+    },
+    {
+      "epoch": 83.125,
+      "grad_norm": 1.0409256219863892,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 29260
+    },
+    {
+      "epoch": 83.1534090909091,
+      "grad_norm": 1.1085119247436523,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 29270
+    },
+    {
+      "epoch": 83.18181818181819,
+      "grad_norm": 1.0388301610946655,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 29280
+    },
+    {
+      "epoch": 83.21022727272727,
+      "grad_norm": 0.8974637985229492,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "step": 29290
+    },
+    {
+      "epoch": 83.23863636363636,
+      "grad_norm": 0.8485453724861145,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "step": 29300
+    },
+    {
+      "epoch": 83.26704545454545,
+      "grad_norm": 0.7687119841575623,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 29310
+    },
+    {
+      "epoch": 83.29545454545455,
+      "grad_norm": 0.844149649143219,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 29320
+    },
+    {
+      "epoch": 83.32386363636364,
+      "grad_norm": 0.8516110181808472,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 29330
+    },
+    {
+      "epoch": 83.35227272727273,
+      "grad_norm": 0.6241595149040222,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 29340
+    },
+    {
+      "epoch": 83.38068181818181,
+      "grad_norm": 0.7771625518798828,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 29350
+    },
+    {
+      "epoch": 83.4090909090909,
+      "grad_norm": 0.6839377284049988,
+      "learning_rate": 0.0001,
+      "loss": 0.0459,
+      "step": 29360
+    },
+    {
+      "epoch": 83.4375,
+      "grad_norm": 0.7105209231376648,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 29370
+    },
+    {
+      "epoch": 83.4659090909091,
+      "grad_norm": 1.2164829969406128,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 29380
+    },
+    {
+      "epoch": 83.49431818181819,
+      "grad_norm": 0.969587504863739,
+      "learning_rate": 0.0001,
+      "loss": 0.046,
+      "step": 29390
+    },
+    {
+      "epoch": 83.52272727272727,
+      "grad_norm": 0.8211323618888855,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 29400
+    },
+    {
+      "epoch": 83.55113636363636,
+      "grad_norm": 0.6561827659606934,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 29410
+    },
+    {
+      "epoch": 83.57954545454545,
+      "grad_norm": 0.694988489151001,
+      "learning_rate": 0.0001,
+      "loss": 0.0446,
+      "step": 29420
+    },
+    {
+      "epoch": 83.60795454545455,
+      "grad_norm": 0.8020559549331665,
+      "learning_rate": 0.0001,
+      "loss": 0.0431,
+      "step": 29430
+    },
+    {
+      "epoch": 83.63636363636364,
+      "grad_norm": 0.9569689631462097,
+      "learning_rate": 0.0001,
+      "loss": 0.0449,
+      "step": 29440
+    },
+    {
+      "epoch": 83.66477272727273,
+      "grad_norm": 0.874990701675415,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 29450
+    },
+    {
+      "epoch": 83.69318181818181,
+      "grad_norm": 0.6901691555976868,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 29460
+    },
+    {
+      "epoch": 83.7215909090909,
+      "grad_norm": 0.7540920376777649,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 29470
+    },
+    {
+      "epoch": 83.75,
+      "grad_norm": 0.8184428811073303,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 29480
+    },
+    {
+      "epoch": 83.7784090909091,
+      "grad_norm": 0.9438989162445068,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 29490
+    },
+    {
+      "epoch": 83.80681818181819,
+      "grad_norm": 0.8632564544677734,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 29500
+    },
+    {
+      "epoch": 83.83522727272727,
+      "grad_norm": 0.7080706357955933,
+      "learning_rate": 0.0001,
+      "loss": 0.0444,
+      "step": 29510
+    },
+    {
+      "epoch": 83.86363636363636,
+      "grad_norm": 1.5293716192245483,
+      "learning_rate": 0.0001,
+      "loss": 0.0451,
+      "step": 29520
+    },
+    {
+      "epoch": 83.89204545454545,
+      "grad_norm": 1.1970537900924683,
+      "learning_rate": 0.0001,
+      "loss": 0.0435,
+      "step": 29530
+    },
+    {
+      "epoch": 83.92045454545455,
+      "grad_norm": 1.4087823629379272,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 29540
+    },
+    {
+      "epoch": 83.94886363636364,
+      "grad_norm": 0.9186455011367798,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 29550
+    },
+    {
+      "epoch": 83.97727272727273,
+      "grad_norm": 0.89197838306427,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 29560
+    },
+    {
+      "epoch": 84.00568181818181,
+      "grad_norm": 0.8925532698631287,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 29570
+    },
+    {
+      "epoch": 84.0340909090909,
+      "grad_norm": 0.9491750597953796,
+      "learning_rate": 0.0001,
+      "loss": 0.0458,
+      "step": 29580
+    },
+    {
+      "epoch": 84.0625,
+      "grad_norm": 0.7977159023284912,
+      "learning_rate": 0.0001,
+      "loss": 0.0455,
+      "step": 29590
+    },
+    {
+      "epoch": 84.0909090909091,
+      "grad_norm": 0.7931260466575623,
+      "learning_rate": 0.0001,
+      "loss": 0.0451,
+      "step": 29600
+    },
+    {
+      "epoch": 84.11931818181819,
+      "grad_norm": 0.691261887550354,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 29610
+    },
+    {
+      "epoch": 84.14772727272727,
+      "grad_norm": 0.9392285346984863,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 29620
+    },
+    {
+      "epoch": 84.17613636363636,
+      "grad_norm": 1.9056662321090698,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 29630
+    },
+    {
+      "epoch": 84.20454545454545,
+      "grad_norm": 1.3401696681976318,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 29640
+    },
+    {
+      "epoch": 84.23295454545455,
+      "grad_norm": 1.1489585638046265,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 29650
+    },
+    {
+      "epoch": 84.26136363636364,
+      "grad_norm": 1.2388312816619873,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 29660
+    },
+    {
+      "epoch": 84.28977272727273,
+      "grad_norm": 1.1380677223205566,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 29670
+    },
+    {
+      "epoch": 84.31818181818181,
+      "grad_norm": 0.9601713418960571,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 29680
+    },
+    {
+      "epoch": 84.3465909090909,
+      "grad_norm": 1.2920958995819092,
+      "learning_rate": 0.0001,
+      "loss": 0.0425,
+      "step": 29690
+    },
+    {
+      "epoch": 84.375,
+      "grad_norm": 1.1546963453292847,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 29700
+    },
+    {
+      "epoch": 84.4034090909091,
+      "grad_norm": 1.6051952838897705,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 29710
+    },
+    {
+      "epoch": 84.43181818181819,
+      "grad_norm": 1.979384422302246,
+      "learning_rate": 0.0001,
+      "loss": 0.042,
+      "step": 29720
+    },
+    {
+      "epoch": 84.46022727272727,
+      "grad_norm": 2.053717613220215,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 29730
+    },
+    {
+      "epoch": 84.48863636363636,
+      "grad_norm": 1.1979448795318604,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 29740
+    },
+    {
+      "epoch": 84.51704545454545,
+      "grad_norm": 1.6801135540008545,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 29750
+    },
+    {
+      "epoch": 84.54545454545455,
+      "grad_norm": 0.7504470348358154,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 29760
+    },
+    {
+      "epoch": 84.57386363636364,
+      "grad_norm": 1.1721948385238647,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 29770
+    },
+    {
+      "epoch": 84.60227272727273,
+      "grad_norm": 0.9734560251235962,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 29780
+    },
+    {
+      "epoch": 84.63068181818181,
+      "grad_norm": 1.0226346254348755,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 29790
+    },
+    {
+      "epoch": 84.6590909090909,
+      "grad_norm": 0.7675327658653259,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 29800
+    },
+    {
+      "epoch": 84.6875,
+      "grad_norm": 0.9482449889183044,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 29810
+    },
+    {
+      "epoch": 84.7159090909091,
+      "grad_norm": 0.7545673251152039,
+      "learning_rate": 0.0001,
+      "loss": 0.0422,
+      "step": 29820
+    },
+    {
+      "epoch": 84.74431818181819,
+      "grad_norm": 0.7988566160202026,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 29830
+    },
+    {
+      "epoch": 84.77272727272727,
+      "grad_norm": 0.9122010469436646,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 29840
+    },
+    {
+      "epoch": 84.80113636363636,
+      "grad_norm": 0.8561978936195374,
+      "learning_rate": 0.0001,
+      "loss": 0.0443,
+      "step": 29850
+    },
+    {
+      "epoch": 84.82954545454545,
+      "grad_norm": 0.6838624477386475,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 29860
+    },
+    {
+      "epoch": 84.85795454545455,
+      "grad_norm": 1.1277817487716675,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 29870
+    },
+    {
+      "epoch": 84.88636363636364,
+      "grad_norm": 1.5995237827301025,
+      "learning_rate": 0.0001,
+      "loss": 0.0462,
+      "step": 29880
+    },
+    {
+      "epoch": 84.91477272727273,
+      "grad_norm": 1.4446282386779785,
+      "learning_rate": 0.0001,
+      "loss": 0.0451,
+      "step": 29890
+    },
+    {
+      "epoch": 84.94318181818181,
+      "grad_norm": 1.1589558124542236,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 29900
+    },
+    {
+      "epoch": 84.9715909090909,
+      "grad_norm": 1.063513159751892,
+      "learning_rate": 0.0001,
+      "loss": 0.0427,
+      "step": 29910
+    },
+    {
+      "epoch": 85.0,
+      "grad_norm": 0.8769459128379822,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 29920
+    },
+    {
+      "epoch": 85.0284090909091,
+      "grad_norm": 0.9075056910514832,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 29930
+    },
+    {
+      "epoch": 85.05681818181819,
+      "grad_norm": 0.6981393098831177,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 29940
+    },
+    {
+      "epoch": 85.08522727272727,
+      "grad_norm": 0.757666826248169,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 29950
+    },
+    {
+      "epoch": 85.11363636363636,
+      "grad_norm": 0.7865346074104309,
+      "learning_rate": 0.0001,
+      "loss": 0.0431,
+      "step": 29960
+    },
+    {
+      "epoch": 85.14204545454545,
+      "grad_norm": 0.8140910267829895,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 29970
+    },
+    {
+      "epoch": 85.17045454545455,
+      "grad_norm": 0.6491031646728516,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 29980
+    },
+    {
+      "epoch": 85.19886363636364,
+      "grad_norm": 0.8176889419555664,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 29990
+    },
+    {
+      "epoch": 85.22727272727273,
+      "grad_norm": 0.8975611925125122,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 30000
+    },
+    {
+      "epoch": 85.25568181818181,
+      "grad_norm": 0.6582433581352234,
+      "learning_rate": 0.0001,
+      "loss": 0.0431,
+      "step": 30010
+    },
+    {
+      "epoch": 85.2840909090909,
+      "grad_norm": 0.7108902931213379,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 30020
+    },
+    {
+      "epoch": 85.3125,
+      "grad_norm": 0.8365264534950256,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 30030
+    },
+    {
+      "epoch": 85.3409090909091,
+      "grad_norm": 1.003644347190857,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 30040
+    },
+    {
+      "epoch": 85.36931818181819,
+      "grad_norm": 0.8627144694328308,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 30050
+    },
+    {
+      "epoch": 85.39772727272727,
+      "grad_norm": 0.8255655765533447,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 30060
+    },
+    {
+      "epoch": 85.42613636363636,
+      "grad_norm": 0.7838973999023438,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 30070
+    },
+    {
+      "epoch": 85.45454545454545,
+      "grad_norm": 0.7875524163246155,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 30080
+    },
+    {
+      "epoch": 85.48295454545455,
+      "grad_norm": 0.9446814656257629,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 30090
+    },
+    {
+      "epoch": 85.51136363636364,
+      "grad_norm": 0.6380667686462402,
+      "learning_rate": 0.0001,
+      "loss": 0.0431,
+      "step": 30100
+    },
+    {
+      "epoch": 85.53977272727273,
+      "grad_norm": 0.8755031228065491,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 30110
+    },
+    {
+      "epoch": 85.56818181818181,
+      "grad_norm": 0.6563933491706848,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 30120
+    },
+    {
+      "epoch": 85.5965909090909,
+      "grad_norm": 0.7544918060302734,
+      "learning_rate": 0.0001,
+      "loss": 0.0435,
+      "step": 30130
+    },
+    {
+      "epoch": 85.625,
+      "grad_norm": 1.112111210823059,
+      "learning_rate": 0.0001,
+      "loss": 0.0463,
+      "step": 30140
+    },
+    {
+      "epoch": 85.6534090909091,
+      "grad_norm": 0.9833115935325623,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 30150
+    },
+    {
+      "epoch": 85.68181818181819,
+      "grad_norm": 0.914084255695343,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 30160
+    },
+    {
+      "epoch": 85.71022727272727,
+      "grad_norm": 1.0748567581176758,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 30170
+    },
+    {
+      "epoch": 85.73863636363636,
+      "grad_norm": 0.6668802499771118,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 30180
+    },
+    {
+      "epoch": 85.76704545454545,
+      "grad_norm": 0.8170040845870972,
+      "learning_rate": 0.0001,
+      "loss": 0.0435,
+      "step": 30190
+    },
+    {
+      "epoch": 85.79545454545455,
+      "grad_norm": 0.7252139449119568,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 30200
+    },
+    {
+      "epoch": 85.82386363636364,
+      "grad_norm": 0.621457576751709,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 30210
+    },
+    {
+      "epoch": 85.85227272727273,
+      "grad_norm": 0.7499610185623169,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 30220
+    },
+    {
+      "epoch": 85.88068181818181,
+      "grad_norm": 0.6016923785209656,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 30230
+    },
+    {
+      "epoch": 85.9090909090909,
+      "grad_norm": 0.9071959853172302,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 30240
+    },
+    {
+      "epoch": 85.9375,
+      "grad_norm": 0.6232897043228149,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 30250
+    },
+    {
+      "epoch": 85.9659090909091,
+      "grad_norm": 0.7610146999359131,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 30260
+    },
+    {
+      "epoch": 85.99431818181819,
+      "grad_norm": 0.6071455478668213,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "step": 30270
+    },
+    {
+      "epoch": 86.02272727272727,
+      "grad_norm": 0.6175063848495483,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 30280
+    },
+    {
+      "epoch": 86.05113636363636,
+      "grad_norm": 0.7343127727508545,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 30290
+    },
+    {
+      "epoch": 86.07954545454545,
+      "grad_norm": 0.7600955367088318,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 30300
+    },
+    {
+      "epoch": 86.10795454545455,
+      "grad_norm": 0.5361849665641785,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 30310
+    },
+    {
+      "epoch": 86.13636363636364,
+      "grad_norm": 0.6238926649093628,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 30320
+    },
+    {
+      "epoch": 86.16477272727273,
+      "grad_norm": 0.530532717704773,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "step": 30330
+    },
+    {
+      "epoch": 86.19318181818181,
+      "grad_norm": 0.6140005588531494,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 30340
+    },
+    {
+      "epoch": 86.2215909090909,
+      "grad_norm": 0.5914357900619507,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "step": 30350
+    },
+    {
+      "epoch": 86.25,
+      "grad_norm": 0.5596780776977539,
+      "learning_rate": 0.0001,
+      "loss": 0.0427,
+      "step": 30360
+    },
+    {
+      "epoch": 86.2784090909091,
+      "grad_norm": 0.5929566025733948,
+      "learning_rate": 0.0001,
+      "loss": 0.0425,
+      "step": 30370
+    },
+    {
+      "epoch": 86.30681818181819,
+      "grad_norm": 0.6372137069702148,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 30380
+    },
+    {
+      "epoch": 86.33522727272727,
+      "grad_norm": 0.6501078009605408,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "step": 30390
+    },
+    {
+      "epoch": 86.36363636363636,
+      "grad_norm": 0.57587730884552,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 30400
+    },
+    {
+      "epoch": 86.39204545454545,
+      "grad_norm": 0.6445661187171936,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 30410
+    },
+    {
+      "epoch": 86.42045454545455,
+      "grad_norm": 0.7396770715713501,
+      "learning_rate": 0.0001,
+      "loss": 0.0457,
+      "step": 30420
+    },
+    {
+      "epoch": 86.44886363636364,
+      "grad_norm": 0.673376202583313,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 30430
+    },
+    {
+      "epoch": 86.47727272727273,
+      "grad_norm": 0.6946069598197937,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 30440
+    },
+    {
+      "epoch": 86.50568181818181,
+      "grad_norm": 0.6759048104286194,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 30450
+    },
+    {
+      "epoch": 86.5340909090909,
+      "grad_norm": 0.5420788526535034,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 30460
+    },
+    {
+      "epoch": 86.5625,
+      "grad_norm": 0.6719872355461121,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 30470
+    },
+    {
+      "epoch": 86.5909090909091,
+      "grad_norm": 0.6998466849327087,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 30480
+    },
+    {
+      "epoch": 86.61931818181819,
+      "grad_norm": 0.6579269766807556,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 30490
+    },
+    {
+      "epoch": 86.64772727272727,
+      "grad_norm": 0.7772161364555359,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 30500
+    },
+    {
+      "epoch": 86.67613636363636,
+      "grad_norm": 0.5968173742294312,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 30510
+    },
+    {
+      "epoch": 86.70454545454545,
+      "grad_norm": 0.6808553338050842,
+      "learning_rate": 0.0001,
+      "loss": 0.0435,
+      "step": 30520
+    },
+    {
+      "epoch": 86.73295454545455,
+      "grad_norm": 0.7286439538002014,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 30530
+    },
+    {
+      "epoch": 86.76136363636364,
+      "grad_norm": 0.6226254105567932,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "step": 30540
+    },
+    {
+      "epoch": 86.78977272727273,
+      "grad_norm": 0.61110919713974,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 30550
+    },
+    {
+      "epoch": 86.81818181818181,
+      "grad_norm": 0.6020199656486511,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 30560
+    },
+    {
+      "epoch": 86.8465909090909,
+      "grad_norm": 0.630531907081604,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 30570
+    },
+    {
+      "epoch": 86.875,
+      "grad_norm": 0.5699210166931152,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 30580
+    },
+    {
+      "epoch": 86.9034090909091,
+      "grad_norm": 0.5879133343696594,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 30590
+    },
+    {
+      "epoch": 86.93181818181819,
+      "grad_norm": 0.6827641129493713,
+      "learning_rate": 0.0001,
+      "loss": 0.0446,
+      "step": 30600
+    },
+    {
+      "epoch": 86.96022727272727,
+      "grad_norm": 0.581780731678009,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 30610
+    },
+    {
+      "epoch": 86.98863636363636,
+      "grad_norm": 0.5517546534538269,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 30620
+    },
+    {
+      "epoch": 87.01704545454545,
+      "grad_norm": 0.6034563779830933,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 30630
+    },
+    {
+      "epoch": 87.04545454545455,
+      "grad_norm": 0.5161349177360535,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 30640
+    },
+    {
+      "epoch": 87.07386363636364,
+      "grad_norm": 0.6517345309257507,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 30650
+    },
+    {
+      "epoch": 87.10227272727273,
+      "grad_norm": 0.7236988544464111,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 30660
+    },
+    {
+      "epoch": 87.13068181818181,
+      "grad_norm": 0.5659297704696655,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 30670
+    },
+    {
+      "epoch": 87.1590909090909,
+      "grad_norm": 0.5743705630302429,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 30680
+    },
+    {
+      "epoch": 87.1875,
+      "grad_norm": 0.4794277846813202,
+      "learning_rate": 0.0001,
+      "loss": 0.0442,
+      "step": 30690
+    },
+    {
+      "epoch": 87.2159090909091,
+      "grad_norm": 0.6366981267929077,
+      "learning_rate": 0.0001,
+      "loss": 0.0425,
+      "step": 30700
+    },
+    {
+      "epoch": 87.24431818181819,
+      "grad_norm": 0.6959528923034668,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 30710
+    },
+    {
+      "epoch": 87.27272727272727,
+      "grad_norm": 0.5670670866966248,
+      "learning_rate": 0.0001,
+      "loss": 0.0451,
+      "step": 30720
+    },
+    {
+      "epoch": 87.30113636363636,
+      "grad_norm": 0.5087947845458984,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 30730
+    },
+    {
+      "epoch": 87.32954545454545,
+      "grad_norm": 0.638629138469696,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "step": 30740
+    },
+    {
+      "epoch": 87.35795454545455,
+      "grad_norm": 0.5405479073524475,
+      "learning_rate": 0.0001,
+      "loss": 0.0424,
+      "step": 30750
+    },
+    {
+      "epoch": 87.38636363636364,
+      "grad_norm": 0.5827491283416748,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 30760
+    },
+    {
+      "epoch": 87.41477272727273,
+      "grad_norm": 0.8291541337966919,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 30770
+    },
+    {
+      "epoch": 87.44318181818181,
+      "grad_norm": 0.5378076434135437,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 30780
+    },
+    {
+      "epoch": 87.4715909090909,
+      "grad_norm": 0.5673062801361084,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 30790
+    },
+    {
+      "epoch": 87.5,
+      "grad_norm": 0.6628111004829407,
+      "learning_rate": 0.0001,
+      "loss": 0.0454,
+      "step": 30800
+    },
+    {
+      "epoch": 87.5284090909091,
+      "grad_norm": 0.5737188458442688,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 30810
+    },
+    {
+      "epoch": 87.55681818181819,
+      "grad_norm": 0.49959149956703186,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 30820
+    },
+    {
+      "epoch": 87.58522727272727,
+      "grad_norm": 0.579260528087616,
+      "learning_rate": 0.0001,
+      "loss": 0.0461,
+      "step": 30830
+    },
+    {
+      "epoch": 87.61363636363636,
+      "grad_norm": 0.6071043014526367,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 30840
+    },
+    {
+      "epoch": 87.64204545454545,
+      "grad_norm": 0.8908697366714478,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 30850
+    },
+    {
+      "epoch": 87.67045454545455,
+      "grad_norm": 0.7816733717918396,
+      "learning_rate": 0.0001,
+      "loss": 0.0456,
+      "step": 30860
+    },
+    {
+      "epoch": 87.69886363636364,
+      "grad_norm": 0.837706446647644,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 30870
+    },
+    {
+      "epoch": 87.72727272727273,
+      "grad_norm": 0.6439931988716125,
+      "learning_rate": 0.0001,
+      "loss": 0.0448,
+      "step": 30880
+    },
+    {
+      "epoch": 87.75568181818181,
+      "grad_norm": 0.5524504780769348,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 30890
+    },
+    {
+      "epoch": 87.7840909090909,
+      "grad_norm": 0.5548602342605591,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 30900
+    },
+    {
+      "epoch": 87.8125,
+      "grad_norm": 0.5937913656234741,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 30910
+    },
+    {
+      "epoch": 87.8409090909091,
+      "grad_norm": 0.49119383096694946,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 30920
+    },
+    {
+      "epoch": 87.86931818181819,
+      "grad_norm": 0.5523015260696411,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 30930
+    },
+    {
+      "epoch": 87.89772727272727,
+      "grad_norm": 0.6836235523223877,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 30940
+    },
+    {
+      "epoch": 87.92613636363636,
+      "grad_norm": 0.728946328163147,
+      "learning_rate": 0.0001,
+      "loss": 0.0435,
+      "step": 30950
+    },
+    {
+      "epoch": 87.95454545454545,
+      "grad_norm": 0.7053698301315308,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 30960
+    },
+    {
+      "epoch": 87.98295454545455,
+      "grad_norm": 0.7172105312347412,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 30970
+    },
+    {
+      "epoch": 88.01136363636364,
+      "grad_norm": 0.8821631073951721,
+      "learning_rate": 0.0001,
+      "loss": 0.0425,
+      "step": 30980
+    },
+    {
+      "epoch": 88.03977272727273,
+      "grad_norm": 0.837773323059082,
+      "learning_rate": 0.0001,
+      "loss": 0.045,
+      "step": 30990
+    },
+    {
+      "epoch": 88.06818181818181,
+      "grad_norm": 1.007065773010254,
+      "learning_rate": 0.0001,
+      "loss": 0.0422,
+      "step": 31000
+    },
+    {
+      "epoch": 88.0965909090909,
+      "grad_norm": 0.8076823353767395,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 31010
+    },
+    {
+      "epoch": 88.125,
+      "grad_norm": 1.02508544921875,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 31020
+    },
+    {
+      "epoch": 88.1534090909091,
+      "grad_norm": 0.726391077041626,
+      "learning_rate": 0.0001,
+      "loss": 0.0424,
+      "step": 31030
+    },
+    {
+      "epoch": 88.18181818181819,
+      "grad_norm": 0.78676837682724,
+      "learning_rate": 0.0001,
+      "loss": 0.0427,
+      "step": 31040
+    },
+    {
+      "epoch": 88.21022727272727,
+      "grad_norm": 0.7329301834106445,
+      "learning_rate": 0.0001,
+      "loss": 0.0427,
+      "step": 31050
+    },
+    {
+      "epoch": 88.23863636363636,
+      "grad_norm": 0.6445389986038208,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 31060
+    },
+    {
+      "epoch": 88.26704545454545,
+      "grad_norm": 0.6451675295829773,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 31070
+    },
+    {
+      "epoch": 88.29545454545455,
+      "grad_norm": 0.7502676248550415,
+      "learning_rate": 0.0001,
+      "loss": 0.0437,
+      "step": 31080
+    },
+    {
+      "epoch": 88.32386363636364,
+      "grad_norm": 0.8322815299034119,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 31090
+    },
+    {
+      "epoch": 88.35227272727273,
+      "grad_norm": 0.8641359210014343,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 31100
+    },
+    {
+      "epoch": 88.38068181818181,
+      "grad_norm": 0.8692095279693604,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 31110
+    },
+    {
+      "epoch": 88.4090909090909,
+      "grad_norm": 0.6443942785263062,
+      "learning_rate": 0.0001,
+      "loss": 0.0427,
+      "step": 31120
+    },
+    {
+      "epoch": 88.4375,
+      "grad_norm": 0.7591652274131775,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 31130
+    },
+    {
+      "epoch": 88.4659090909091,
+      "grad_norm": 0.8515008687973022,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 31140
+    },
+    {
+      "epoch": 88.49431818181819,
+      "grad_norm": 0.7373746633529663,
+      "learning_rate": 0.0001,
+      "loss": 0.0452,
+      "step": 31150
+    },
+    {
+      "epoch": 88.52272727272727,
+      "grad_norm": 2.620330810546875,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 31160
+    },
+    {
+      "epoch": 88.55113636363636,
+      "grad_norm": 0.7088080048561096,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 31170
+    },
+    {
+      "epoch": 88.57954545454545,
+      "grad_norm": 0.8215885758399963,
+      "learning_rate": 0.0001,
+      "loss": 0.0429,
+      "step": 31180
+    },
+    {
+      "epoch": 88.60795454545455,
+      "grad_norm": 0.8965012431144714,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 31190
+    },
+    {
+      "epoch": 88.63636363636364,
+      "grad_norm": 1.090306282043457,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 31200
+    },
+    {
+      "epoch": 88.66477272727273,
+      "grad_norm": 0.9115955829620361,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 31210
+    },
+    {
+      "epoch": 88.69318181818181,
+      "grad_norm": 1.4948323965072632,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "step": 31220
+    },
+    {
+      "epoch": 88.7215909090909,
+      "grad_norm": 1.177667498588562,
+      "learning_rate": 0.0001,
+      "loss": 0.0427,
+      "step": 31230
+    },
+    {
+      "epoch": 88.75,
+      "grad_norm": 1.1721715927124023,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 31240
+    },
+    {
+      "epoch": 88.7784090909091,
+      "grad_norm": 1.04111647605896,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "step": 31250
+    },
+    {
+      "epoch": 88.80681818181819,
+      "grad_norm": 1.1286450624465942,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 31260
+    },
+    {
+      "epoch": 88.83522727272727,
+      "grad_norm": 0.8720760941505432,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 31270
+    },
+    {
+      "epoch": 88.86363636363636,
+      "grad_norm": 0.7654427289962769,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 31280
+    },
+    {
+      "epoch": 88.89204545454545,
+      "grad_norm": 0.6938006281852722,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 31290
+    },
+    {
+      "epoch": 88.92045454545455,
+      "grad_norm": 0.6255007386207581,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 31300
+    },
+    {
+      "epoch": 88.94886363636364,
+      "grad_norm": 0.881568193435669,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 31310
+    },
+    {
+      "epoch": 88.97727272727273,
+      "grad_norm": 0.7803657650947571,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 31320
+    },
+    {
+      "epoch": 89.00568181818181,
+      "grad_norm": 0.5337231159210205,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 31330
+    },
+    {
+      "epoch": 89.0340909090909,
+      "grad_norm": 0.6003879904747009,
+      "learning_rate": 0.0001,
+      "loss": 0.0422,
+      "step": 31340
+    },
+    {
+      "epoch": 89.0625,
+      "grad_norm": 0.694319486618042,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 31350
+    },
+    {
+      "epoch": 89.0909090909091,
+      "grad_norm": 0.6574826836585999,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 31360
+    },
+    {
+      "epoch": 89.11931818181819,
+      "grad_norm": 0.6312698125839233,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 31370
+    },
+    {
+      "epoch": 89.14772727272727,
+      "grad_norm": 0.5977025032043457,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 31380
+    },
+    {
+      "epoch": 89.17613636363636,
+      "grad_norm": 0.6093351244926453,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 31390
+    },
+    {
+      "epoch": 89.20454545454545,
+      "grad_norm": 0.563823938369751,
+      "learning_rate": 0.0001,
+      "loss": 0.042,
+      "step": 31400
+    },
+    {
+      "epoch": 89.23295454545455,
+      "grad_norm": 0.6710460782051086,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 31410
+    },
+    {
+      "epoch": 89.26136363636364,
+      "grad_norm": 0.6708618998527527,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 31420
+    },
+    {
+      "epoch": 89.28977272727273,
+      "grad_norm": 0.7974550127983093,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 31430
+    },
+    {
+      "epoch": 89.31818181818181,
+      "grad_norm": 0.7905569672584534,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 31440
+    },
+    {
+      "epoch": 89.3465909090909,
+      "grad_norm": 0.6526768207550049,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 31450
+    },
+    {
+      "epoch": 89.375,
+      "grad_norm": 0.6570956707000732,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 31460
+    },
+    {
+      "epoch": 89.4034090909091,
+      "grad_norm": 0.5645592212677002,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 31470
+    },
+    {
+      "epoch": 89.43181818181819,
+      "grad_norm": 0.7741744518280029,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 31480
+    },
+    {
+      "epoch": 89.46022727272727,
+      "grad_norm": 0.8912363052368164,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 31490
+    },
+    {
+      "epoch": 89.48863636363636,
+      "grad_norm": 1.4925516843795776,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 31500
+    },
+    {
+      "epoch": 89.51704545454545,
+      "grad_norm": 1.1449397802352905,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 31510
+    },
+    {
+      "epoch": 89.54545454545455,
+      "grad_norm": 0.7447580099105835,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 31520
+    },
+    {
+      "epoch": 89.57386363636364,
+      "grad_norm": 0.8966802358627319,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 31530
+    },
+    {
+      "epoch": 89.60227272727273,
+      "grad_norm": 0.8590907454490662,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 31540
+    },
+    {
+      "epoch": 89.63068181818181,
+      "grad_norm": 1.0364528894424438,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 31550
+    },
+    {
+      "epoch": 89.6590909090909,
+      "grad_norm": 1.0565521717071533,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 31560
+    },
+    {
+      "epoch": 89.6875,
+      "grad_norm": 0.8635243773460388,
+      "learning_rate": 0.0001,
+      "loss": 0.0435,
+      "step": 31570
+    },
+    {
+      "epoch": 89.7159090909091,
+      "grad_norm": 0.8231905698776245,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 31580
+    },
+    {
+      "epoch": 89.74431818181819,
+      "grad_norm": 0.683319628238678,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 31590
+    },
+    {
+      "epoch": 89.77272727272727,
+      "grad_norm": 0.696625292301178,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 31600
+    },
+    {
+      "epoch": 89.80113636363636,
+      "grad_norm": 0.6028037667274475,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 31610
+    },
+    {
+      "epoch": 89.82954545454545,
+      "grad_norm": 1.206365704536438,
+      "learning_rate": 0.0001,
+      "loss": 0.044,
+      "step": 31620
+    },
+    {
+      "epoch": 89.85795454545455,
+      "grad_norm": 1.4085060358047485,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 31630
+    },
+    {
+      "epoch": 89.88636363636364,
+      "grad_norm": 1.3718105554580688,
+      "learning_rate": 0.0001,
+      "loss": 0.0439,
+      "step": 31640
+    },
+    {
+      "epoch": 89.91477272727273,
+      "grad_norm": 0.872251570224762,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 31650
+    },
+    {
+      "epoch": 89.94318181818181,
+      "grad_norm": 0.8787030577659607,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 31660
+    },
+    {
+      "epoch": 89.9715909090909,
+      "grad_norm": 0.9750413298606873,
+      "learning_rate": 0.0001,
+      "loss": 0.0412,
+      "step": 31670
+    },
+    {
+      "epoch": 90.0,
+      "grad_norm": 1.8003432750701904,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 31680
+    },
+    {
+      "epoch": 90.0284090909091,
+      "grad_norm": 1.7299727201461792,
+      "learning_rate": 0.0001,
+      "loss": 0.0432,
+      "step": 31690
+    },
+    {
+      "epoch": 90.05681818181819,
+      "grad_norm": 2.0806267261505127,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 31700
+    },
+    {
+      "epoch": 90.08522727272727,
+      "grad_norm": 1.1748923063278198,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 31710
+    },
+    {
+      "epoch": 90.11363636363636,
+      "grad_norm": 1.2577301263809204,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 31720
+    },
+    {
+      "epoch": 90.14204545454545,
+      "grad_norm": 1.1263160705566406,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 31730
+    },
+    {
+      "epoch": 90.17045454545455,
+      "grad_norm": 0.8332096338272095,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "step": 31740
+    },
+    {
+      "epoch": 90.19886363636364,
+      "grad_norm": 0.9236270189285278,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 31750
+    },
+    {
+      "epoch": 90.22727272727273,
+      "grad_norm": 0.826349675655365,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 31760
+    },
+    {
+      "epoch": 90.25568181818181,
+      "grad_norm": 0.7999365329742432,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 31770
+    },
+    {
+      "epoch": 90.2840909090909,
+      "grad_norm": 0.8490392565727234,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 31780
+    },
+    {
+      "epoch": 90.3125,
+      "grad_norm": 0.8082181811332703,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 31790
+    },
+    {
+      "epoch": 90.3409090909091,
+      "grad_norm": 0.9047965407371521,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 31800
+    },
+    {
+      "epoch": 90.36931818181819,
+      "grad_norm": 0.7220473885536194,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 31810
+    },
+    {
+      "epoch": 90.39772727272727,
+      "grad_norm": 1.0218350887298584,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 31820
+    },
+    {
+      "epoch": 90.42613636363636,
+      "grad_norm": 0.6703020930290222,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 31830
+    },
+    {
+      "epoch": 90.45454545454545,
+      "grad_norm": 0.6317295432090759,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "step": 31840
+    },
+    {
+      "epoch": 90.48295454545455,
+      "grad_norm": 0.5803297758102417,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 31850
+    },
+    {
+      "epoch": 90.51136363636364,
+      "grad_norm": 0.7607895135879517,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 31860
+    },
+    {
+      "epoch": 90.53977272727273,
+      "grad_norm": 0.6137414574623108,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 31870
+    },
+    {
+      "epoch": 90.56818181818181,
+      "grad_norm": 0.5825350284576416,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 31880
+    },
+    {
+      "epoch": 90.5965909090909,
+      "grad_norm": 0.564761757850647,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 31890
+    },
+    {
+      "epoch": 90.625,
+      "grad_norm": 1.0057430267333984,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 31900
+    },
+    {
+      "epoch": 90.6534090909091,
+      "grad_norm": 0.7206169962882996,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 31910
+    },
+    {
+      "epoch": 90.68181818181819,
+      "grad_norm": 0.8694084286689758,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 31920
+    },
+    {
+      "epoch": 90.71022727272727,
+      "grad_norm": 0.9109015464782715,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 31930
+    },
+    {
+      "epoch": 90.73863636363636,
+      "grad_norm": 1.422331690788269,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 31940
+    },
+    {
+      "epoch": 90.76704545454545,
+      "grad_norm": 0.6990547180175781,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "step": 31950
+    },
+    {
+      "epoch": 90.79545454545455,
+      "grad_norm": 0.8138213753700256,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 31960
+    },
+    {
+      "epoch": 90.82386363636364,
+      "grad_norm": 0.9061129093170166,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 31970
+    },
+    {
+      "epoch": 90.85227272727273,
+      "grad_norm": 0.5697906613349915,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 31980
+    },
+    {
+      "epoch": 90.88068181818181,
+      "grad_norm": 0.6935226917266846,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 31990
+    },
+    {
+      "epoch": 90.9090909090909,
+      "grad_norm": 0.6459117531776428,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 32000
+    },
+    {
+      "epoch": 90.9375,
+      "grad_norm": 0.8231947422027588,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 32010
+    },
+    {
+      "epoch": 90.9659090909091,
+      "grad_norm": 1.047978401184082,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 32020
+    },
+    {
+      "epoch": 90.99431818181819,
+      "grad_norm": 0.5649544596672058,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "step": 32030
+    },
+    {
+      "epoch": 91.02272727272727,
+      "grad_norm": 0.6621559262275696,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 32040
+    },
+    {
+      "epoch": 91.05113636363636,
+      "grad_norm": 0.8017844557762146,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 32050
+    },
+    {
+      "epoch": 91.07954545454545,
+      "grad_norm": 0.7462131977081299,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 32060
+    },
+    {
+      "epoch": 91.10795454545455,
+      "grad_norm": 0.7743386030197144,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 32070
+    },
+    {
+      "epoch": 91.13636363636364,
+      "grad_norm": 0.6021215915679932,
+      "learning_rate": 0.0001,
+      "loss": 0.0424,
+      "step": 32080
+    },
+    {
+      "epoch": 91.16477272727273,
+      "grad_norm": 0.6302787065505981,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 32090
+    },
+    {
+      "epoch": 91.19318181818181,
+      "grad_norm": 0.5769550800323486,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 32100
+    },
+    {
+      "epoch": 91.2215909090909,
+      "grad_norm": 0.6698492169380188,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 32110
+    },
+    {
+      "epoch": 91.25,
+      "grad_norm": 0.6460458040237427,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 32120
+    },
+    {
+      "epoch": 91.2784090909091,
+      "grad_norm": 0.693789541721344,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 32130
+    },
+    {
+      "epoch": 91.30681818181819,
+      "grad_norm": 0.7158985733985901,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 32140
+    },
+    {
+      "epoch": 91.33522727272727,
+      "grad_norm": 0.618126630783081,
+      "learning_rate": 0.0001,
+      "loss": 0.0433,
+      "step": 32150
+    },
+    {
+      "epoch": 91.36363636363636,
+      "grad_norm": 0.6397842168807983,
+      "learning_rate": 0.0001,
+      "loss": 0.0434,
+      "step": 32160
+    },
+    {
+      "epoch": 91.39204545454545,
+      "grad_norm": 0.7809488773345947,
+      "learning_rate": 0.0001,
+      "loss": 0.042,
+      "step": 32170
+    },
+    {
+      "epoch": 91.42045454545455,
+      "grad_norm": 0.5916877388954163,
+      "learning_rate": 0.0001,
+      "loss": 0.0438,
+      "step": 32180
+    },
+    {
+      "epoch": 91.44886363636364,
+      "grad_norm": 0.6435518860816956,
+      "learning_rate": 0.0001,
+      "loss": 0.0428,
+      "step": 32190
+    },
+    {
+      "epoch": 91.47727272727273,
+      "grad_norm": 0.7912642955780029,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 32200
+    },
+    {
+      "epoch": 91.50568181818181,
+      "grad_norm": 0.692492663860321,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 32210
+    },
+    {
+      "epoch": 91.5340909090909,
+      "grad_norm": 0.6788442730903625,
+      "learning_rate": 0.0001,
+      "loss": 0.0416,
+      "step": 32220
+    },
+    {
+      "epoch": 91.5625,
+      "grad_norm": 0.7223365902900696,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 32230
+    },
+    {
+      "epoch": 91.5909090909091,
+      "grad_norm": 0.7962009310722351,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 32240
+    },
+    {
+      "epoch": 91.61931818181819,
+      "grad_norm": 0.8391971588134766,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 32250
+    },
+    {
+      "epoch": 91.64772727272727,
+      "grad_norm": 0.963829517364502,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 32260
+    },
+    {
+      "epoch": 91.67613636363636,
+      "grad_norm": 0.7977566719055176,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 32270
+    },
+    {
+      "epoch": 91.70454545454545,
+      "grad_norm": 0.9379525780677795,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 32280
+    },
+    {
+      "epoch": 91.73295454545455,
+      "grad_norm": 0.8582242727279663,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 32290
+    },
+    {
+      "epoch": 91.76136363636364,
+      "grad_norm": 0.8872998952865601,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 32300
+    },
+    {
+      "epoch": 91.78977272727273,
+      "grad_norm": 0.6711391806602478,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 32310
+    },
+    {
+      "epoch": 91.81818181818181,
+      "grad_norm": 0.890733540058136,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 32320
+    },
+    {
+      "epoch": 91.8465909090909,
+      "grad_norm": 0.9312843680381775,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 32330
+    },
+    {
+      "epoch": 91.875,
+      "grad_norm": 0.6852923035621643,
+      "learning_rate": 0.0001,
+      "loss": 0.0425,
+      "step": 32340
+    },
+    {
+      "epoch": 91.9034090909091,
+      "grad_norm": 0.7489289045333862,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 32350
+    },
+    {
+      "epoch": 91.93181818181819,
+      "grad_norm": 0.7574262022972107,
+      "learning_rate": 0.0001,
+      "loss": 0.0431,
+      "step": 32360
+    },
+    {
+      "epoch": 91.96022727272727,
+      "grad_norm": 0.7518380284309387,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 32370
+    },
+    {
+      "epoch": 91.98863636363636,
+      "grad_norm": 0.7089730501174927,
+      "learning_rate": 0.0001,
+      "loss": 0.0427,
+      "step": 32380
+    },
+    {
+      "epoch": 92.01704545454545,
+      "grad_norm": 0.7275684475898743,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 32390
+    },
+    {
+      "epoch": 92.04545454545455,
+      "grad_norm": 0.6037976741790771,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 32400
+    },
+    {
+      "epoch": 92.07386363636364,
+      "grad_norm": 0.5874007940292358,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 32410
+    },
+    {
+      "epoch": 92.10227272727273,
+      "grad_norm": 0.5332598686218262,
+      "learning_rate": 0.0001,
+      "loss": 0.0412,
+      "step": 32420
+    },
+    {
+      "epoch": 92.13068181818181,
+      "grad_norm": 0.7063932418823242,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 32430
+    },
+    {
+      "epoch": 92.1590909090909,
+      "grad_norm": 0.7319120168685913,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 32440
+    },
+    {
+      "epoch": 92.1875,
+      "grad_norm": 0.6438412070274353,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 32450
+    },
+    {
+      "epoch": 92.2159090909091,
+      "grad_norm": 0.5752390027046204,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 32460
+    },
+    {
+      "epoch": 92.24431818181819,
+      "grad_norm": 0.7674922943115234,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 32470
+    },
+    {
+      "epoch": 92.27272727272727,
+      "grad_norm": 0.5381972789764404,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 32480
+    },
+    {
+      "epoch": 92.30113636363636,
+      "grad_norm": 0.766947329044342,
+      "learning_rate": 0.0001,
+      "loss": 0.0431,
+      "step": 32490
+    },
+    {
+      "epoch": 92.32954545454545,
+      "grad_norm": 0.5970397591590881,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 32500
+    },
+    {
+      "epoch": 92.35795454545455,
+      "grad_norm": 0.5418734550476074,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 32510
+    },
+    {
+      "epoch": 92.38636363636364,
+      "grad_norm": 0.5782895684242249,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 32520
+    },
+    {
+      "epoch": 92.41477272727273,
+      "grad_norm": 0.5378401875495911,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 32530
+    },
+    {
+      "epoch": 92.44318181818181,
+      "grad_norm": 0.5437403321266174,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 32540
+    },
+    {
+      "epoch": 92.4715909090909,
+      "grad_norm": 0.574937641620636,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 32550
+    },
+    {
+      "epoch": 92.5,
+      "grad_norm": 0.4711826741695404,
+      "learning_rate": 0.0001,
+      "loss": 0.042,
+      "step": 32560
+    },
+    {
+      "epoch": 92.5284090909091,
+      "grad_norm": 0.5091038346290588,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "step": 32570
+    },
+    {
+      "epoch": 92.55681818181819,
+      "grad_norm": 0.5985186696052551,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 32580
+    },
+    {
+      "epoch": 92.58522727272727,
+      "grad_norm": 0.6167530417442322,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 32590
+    },
+    {
+      "epoch": 92.61363636363636,
+      "grad_norm": 0.7481162548065186,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 32600
+    },
+    {
+      "epoch": 92.64204545454545,
+      "grad_norm": 0.6328353881835938,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 32610
+    },
+    {
+      "epoch": 92.67045454545455,
+      "grad_norm": 0.4634016752243042,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 32620
+    },
+    {
+      "epoch": 92.69886363636364,
+      "grad_norm": 0.5572225451469421,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 32630
+    },
+    {
+      "epoch": 92.72727272727273,
+      "grad_norm": 0.5547319650650024,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 32640
+    },
+    {
+      "epoch": 92.75568181818181,
+      "grad_norm": 0.5432265996932983,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 32650
+    },
+    {
+      "epoch": 92.7840909090909,
+      "grad_norm": 0.7217846512794495,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 32660
+    },
+    {
+      "epoch": 92.8125,
+      "grad_norm": 0.7317110896110535,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 32670
+    },
+    {
+      "epoch": 92.8409090909091,
+      "grad_norm": 0.76151442527771,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 32680
+    },
+    {
+      "epoch": 92.86931818181819,
+      "grad_norm": 0.6238445043563843,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 32690
+    },
+    {
+      "epoch": 92.89772727272727,
+      "grad_norm": 0.5886904001235962,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 32700
+    },
+    {
+      "epoch": 92.92613636363636,
+      "grad_norm": 0.46394994854927063,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 32710
+    },
+    {
+      "epoch": 92.95454545454545,
+      "grad_norm": 0.7471411824226379,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 32720
+    },
+    {
+      "epoch": 92.98295454545455,
+      "grad_norm": 0.6481496095657349,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 32730
+    },
+    {
+      "epoch": 93.01136363636364,
+      "grad_norm": 0.6740915775299072,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 32740
+    },
+    {
+      "epoch": 93.03977272727273,
+      "grad_norm": 0.4988252520561218,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 32750
+    },
+    {
+      "epoch": 93.06818181818181,
+      "grad_norm": 0.5601629614830017,
+      "learning_rate": 0.0001,
+      "loss": 0.0417,
+      "step": 32760
+    },
+    {
+      "epoch": 93.0965909090909,
+      "grad_norm": 0.5621158480644226,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 32770
+    },
+    {
+      "epoch": 93.125,
+      "grad_norm": 0.5953294038772583,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 32780
+    },
+    {
+      "epoch": 93.1534090909091,
+      "grad_norm": 0.6276609897613525,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 32790
+    },
+    {
+      "epoch": 93.18181818181819,
+      "grad_norm": 0.6082143783569336,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 32800
+    },
+    {
+      "epoch": 93.21022727272727,
+      "grad_norm": 0.7014224529266357,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "step": 32810
+    },
+    {
+      "epoch": 93.23863636363636,
+      "grad_norm": 0.6491138339042664,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 32820
+    },
+    {
+      "epoch": 93.26704545454545,
+      "grad_norm": 0.7243189215660095,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 32830
+    },
+    {
+      "epoch": 93.29545454545455,
+      "grad_norm": 0.5766690969467163,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 32840
+    },
+    {
+      "epoch": 93.32386363636364,
+      "grad_norm": 0.7540706992149353,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 32850
+    },
+    {
+      "epoch": 93.35227272727273,
+      "grad_norm": 0.7168294787406921,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 32860
+    },
+    {
+      "epoch": 93.38068181818181,
+      "grad_norm": 0.8406569361686707,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 32870
+    },
+    {
+      "epoch": 93.4090909090909,
+      "grad_norm": 0.8490883708000183,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 32880
+    },
+    {
+      "epoch": 93.4375,
+      "grad_norm": 0.5935730338096619,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "step": 32890
+    },
+    {
+      "epoch": 93.4659090909091,
+      "grad_norm": 0.5843430161476135,
+      "learning_rate": 0.0001,
+      "loss": 0.0418,
+      "step": 32900
+    },
+    {
+      "epoch": 93.49431818181819,
+      "grad_norm": 0.8435496091842651,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 32910
+    },
+    {
+      "epoch": 93.52272727272727,
+      "grad_norm": 0.599806547164917,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 32920
+    },
+    {
+      "epoch": 93.55113636363636,
+      "grad_norm": 0.6157388687133789,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 32930
+    },
+    {
+      "epoch": 93.57954545454545,
+      "grad_norm": 0.6935839056968689,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 32940
+    },
+    {
+      "epoch": 93.60795454545455,
+      "grad_norm": 0.6349316835403442,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 32950
+    },
+    {
+      "epoch": 93.63636363636364,
+      "grad_norm": 0.5609252452850342,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 32960
+    },
+    {
+      "epoch": 93.66477272727273,
+      "grad_norm": 0.6227506399154663,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 32970
+    },
+    {
+      "epoch": 93.69318181818181,
+      "grad_norm": 0.671882688999176,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 32980
+    },
+    {
+      "epoch": 93.7215909090909,
+      "grad_norm": 0.8065037727355957,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 32990
+    },
+    {
+      "epoch": 93.75,
+      "grad_norm": 0.7018570303916931,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 33000
+    },
+    {
+      "epoch": 93.7784090909091,
+      "grad_norm": 0.7340584993362427,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "step": 33010
+    },
+    {
+      "epoch": 93.80681818181819,
+      "grad_norm": 0.6570980548858643,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 33020
+    },
+    {
+      "epoch": 93.83522727272727,
+      "grad_norm": 0.6870690584182739,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 33030
+    },
+    {
+      "epoch": 93.86363636363636,
+      "grad_norm": 0.6674038171768188,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 33040
+    },
+    {
+      "epoch": 93.89204545454545,
+      "grad_norm": 0.5651020407676697,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 33050
+    },
+    {
+      "epoch": 93.92045454545455,
+      "grad_norm": 0.7089354991912842,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 33060
+    },
+    {
+      "epoch": 93.94886363636364,
+      "grad_norm": 0.5789372324943542,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 33070
+    },
+    {
+      "epoch": 93.97727272727273,
+      "grad_norm": 0.8415607213973999,
+      "learning_rate": 0.0001,
+      "loss": 0.043,
+      "step": 33080
+    },
+    {
+      "epoch": 94.00568181818181,
+      "grad_norm": 0.7195010185241699,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 33090
+    },
+    {
+      "epoch": 94.0340909090909,
+      "grad_norm": 0.6305752992630005,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 33100
+    },
+    {
+      "epoch": 94.0625,
+      "grad_norm": 0.6140927076339722,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 33110
+    },
+    {
+      "epoch": 94.0909090909091,
+      "grad_norm": 0.761303186416626,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 33120
+    },
+    {
+      "epoch": 94.11931818181819,
+      "grad_norm": 0.7136927247047424,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 33130
+    },
+    {
+      "epoch": 94.14772727272727,
+      "grad_norm": 0.5686725378036499,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 33140
+    },
+    {
+      "epoch": 94.17613636363636,
+      "grad_norm": 0.7701740860939026,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 33150
+    },
+    {
+      "epoch": 94.20454545454545,
+      "grad_norm": 1.336498737335205,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 33160
+    },
+    {
+      "epoch": 94.23295454545455,
+      "grad_norm": 1.33478581905365,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 33170
+    },
+    {
+      "epoch": 94.26136363636364,
+      "grad_norm": 1.4353957176208496,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 33180
+    },
+    {
+      "epoch": 94.28977272727273,
+      "grad_norm": 1.2320867776870728,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 33190
+    },
+    {
+      "epoch": 94.31818181818181,
+      "grad_norm": 1.4484091997146606,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 33200
+    },
+    {
+      "epoch": 94.3465909090909,
+      "grad_norm": 1.3309117555618286,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 33210
+    },
+    {
+      "epoch": 94.375,
+      "grad_norm": 1.4127764701843262,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 33220
+    },
+    {
+      "epoch": 94.4034090909091,
+      "grad_norm": 1.5341440439224243,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 33230
+    },
+    {
+      "epoch": 94.43181818181819,
+      "grad_norm": 1.1439406871795654,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 33240
+    },
+    {
+      "epoch": 94.46022727272727,
+      "grad_norm": 1.584952473640442,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 33250
+    },
+    {
+      "epoch": 94.48863636363636,
+      "grad_norm": 1.6550278663635254,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 33260
+    },
+    {
+      "epoch": 94.51704545454545,
+      "grad_norm": 1.1844305992126465,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 33270
+    },
+    {
+      "epoch": 94.54545454545455,
+      "grad_norm": 1.3612699508666992,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 33280
+    },
+    {
+      "epoch": 94.57386363636364,
+      "grad_norm": 1.3791062831878662,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 33290
+    },
+    {
+      "epoch": 94.60227272727273,
+      "grad_norm": 0.9146853089332581,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 33300
+    },
+    {
+      "epoch": 94.63068181818181,
+      "grad_norm": 0.9118925929069519,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 33310
+    },
+    {
+      "epoch": 94.6590909090909,
+      "grad_norm": 0.9594295024871826,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 33320
+    },
+    {
+      "epoch": 94.6875,
+      "grad_norm": 0.9613752365112305,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 33330
+    },
+    {
+      "epoch": 94.7159090909091,
+      "grad_norm": 0.8596879839897156,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 33340
+    },
+    {
+      "epoch": 94.74431818181819,
+      "grad_norm": 0.8098438382148743,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 33350
+    },
+    {
+      "epoch": 94.77272727272727,
+      "grad_norm": 1.0036510229110718,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 33360
+    },
+    {
+      "epoch": 94.80113636363636,
+      "grad_norm": 0.8176660537719727,
+      "learning_rate": 0.0001,
+      "loss": 0.041,
+      "step": 33370
+    },
+    {
+      "epoch": 94.82954545454545,
+      "grad_norm": 1.1036738157272339,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 33380
+    },
+    {
+      "epoch": 94.85795454545455,
+      "grad_norm": 0.7849661111831665,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 33390
+    },
+    {
+      "epoch": 94.88636363636364,
+      "grad_norm": 0.7073894739151001,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 33400
+    },
+    {
+      "epoch": 94.91477272727273,
+      "grad_norm": 0.8646548986434937,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 33410
+    },
+    {
+      "epoch": 94.94318181818181,
+      "grad_norm": 0.7283008694648743,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 33420
+    },
+    {
+      "epoch": 94.9715909090909,
+      "grad_norm": 0.8491483926773071,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 33430
+    },
+    {
+      "epoch": 95.0,
+      "grad_norm": 0.985192060470581,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 33440
+    },
+    {
+      "epoch": 95.0284090909091,
+      "grad_norm": 0.7137681841850281,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 33450
+    },
+    {
+      "epoch": 95.05681818181819,
+      "grad_norm": 0.6383925676345825,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 33460
+    },
+    {
+      "epoch": 95.08522727272727,
+      "grad_norm": 0.6039495468139648,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 33470
+    },
+    {
+      "epoch": 95.11363636363636,
+      "grad_norm": 0.7503993511199951,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 33480
+    },
+    {
+      "epoch": 95.14204545454545,
+      "grad_norm": 0.8905356526374817,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 33490
+    },
+    {
+      "epoch": 95.17045454545455,
+      "grad_norm": 0.8237714171409607,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 33500
+    },
+    {
+      "epoch": 95.19886363636364,
+      "grad_norm": 0.9115204811096191,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 33510
+    },
+    {
+      "epoch": 95.22727272727273,
+      "grad_norm": 0.7420920729637146,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 33520
+    },
+    {
+      "epoch": 95.25568181818181,
+      "grad_norm": 0.6619880199432373,
+      "learning_rate": 0.0001,
+      "loss": 0.0426,
+      "step": 33530
+    },
+    {
+      "epoch": 95.2840909090909,
+      "grad_norm": 0.7541500926017761,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 33540
+    },
+    {
+      "epoch": 95.3125,
+      "grad_norm": 0.6301935911178589,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 33550
+    },
+    {
+      "epoch": 95.3409090909091,
+      "grad_norm": 0.677110493183136,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 33560
+    },
+    {
+      "epoch": 95.36931818181819,
+      "grad_norm": 0.6058023571968079,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 33570
+    },
+    {
+      "epoch": 95.39772727272727,
+      "grad_norm": 0.6188281774520874,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 33580
+    },
+    {
+      "epoch": 95.42613636363636,
+      "grad_norm": 0.6876928210258484,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 33590
+    },
+    {
+      "epoch": 95.45454545454545,
+      "grad_norm": 0.7519726753234863,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "step": 33600
+    },
+    {
+      "epoch": 95.48295454545455,
+      "grad_norm": 0.9113184809684753,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 33610
+    },
+    {
+      "epoch": 95.51136363636364,
+      "grad_norm": 0.6155601739883423,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 33620
+    },
+    {
+      "epoch": 95.53977272727273,
+      "grad_norm": 0.5960917472839355,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 33630
+    },
+    {
+      "epoch": 95.56818181818181,
+      "grad_norm": 0.849075198173523,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 33640
+    },
+    {
+      "epoch": 95.5965909090909,
+      "grad_norm": 0.6028590798377991,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 33650
+    },
+    {
+      "epoch": 95.625,
+      "grad_norm": 0.7840140461921692,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 33660
+    },
+    {
+      "epoch": 95.6534090909091,
+      "grad_norm": 1.1838630437850952,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 33670
+    },
+    {
+      "epoch": 95.68181818181819,
+      "grad_norm": 1.1603631973266602,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 33680
+    },
+    {
+      "epoch": 95.71022727272727,
+      "grad_norm": 1.3737505674362183,
+      "learning_rate": 0.0001,
+      "loss": 0.0423,
+      "step": 33690
+    },
+    {
+      "epoch": 95.73863636363636,
+      "grad_norm": 1.0241883993148804,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "step": 33700
+    },
+    {
+      "epoch": 95.76704545454545,
+      "grad_norm": 1.056270718574524,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 33710
+    },
+    {
+      "epoch": 95.79545454545455,
+      "grad_norm": 0.6747000217437744,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 33720
+    },
+    {
+      "epoch": 95.82386363636364,
+      "grad_norm": 0.7106748819351196,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 33730
+    },
+    {
+      "epoch": 95.85227272727273,
+      "grad_norm": 0.7365654706954956,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 33740
+    },
+    {
+      "epoch": 95.88068181818181,
+      "grad_norm": 0.7549445033073425,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 33750
+    },
+    {
+      "epoch": 95.9090909090909,
+      "grad_norm": 0.8147190809249878,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 33760
+    },
+    {
+      "epoch": 95.9375,
+      "grad_norm": 0.7287954092025757,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 33770
+    },
+    {
+      "epoch": 95.9659090909091,
+      "grad_norm": 0.771390974521637,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 33780
+    },
+    {
+      "epoch": 95.99431818181819,
+      "grad_norm": 1.2401736974716187,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 33790
+    },
+    {
+      "epoch": 96.02272727272727,
+      "grad_norm": 0.8751268982887268,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 33800
+    },
+    {
+      "epoch": 96.05113636363636,
+      "grad_norm": 0.7138180732727051,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 33810
+    },
+    {
+      "epoch": 96.07954545454545,
+      "grad_norm": 0.7193799614906311,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 33820
+    },
+    {
+      "epoch": 96.10795454545455,
+      "grad_norm": 0.8099432587623596,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 33830
+    },
+    {
+      "epoch": 96.13636363636364,
+      "grad_norm": 0.76226407289505,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 33840
+    },
+    {
+      "epoch": 96.16477272727273,
+      "grad_norm": 0.6789332628250122,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 33850
+    },
+    {
+      "epoch": 96.19318181818181,
+      "grad_norm": 0.6385184526443481,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 33860
+    },
+    {
+      "epoch": 96.2215909090909,
+      "grad_norm": 0.6390976309776306,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 33870
+    },
+    {
+      "epoch": 96.25,
+      "grad_norm": 0.597517192363739,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 33880
+    },
+    {
+      "epoch": 96.2784090909091,
+      "grad_norm": 0.7059200406074524,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 33890
+    },
+    {
+      "epoch": 96.30681818181819,
+      "grad_norm": 0.7767623066902161,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 33900
+    },
+    {
+      "epoch": 96.33522727272727,
+      "grad_norm": 0.5218889117240906,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 33910
+    },
+    {
+      "epoch": 96.36363636363636,
+      "grad_norm": 0.6701022386550903,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 33920
+    },
+    {
+      "epoch": 96.39204545454545,
+      "grad_norm": 0.7549053430557251,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 33930
+    },
+    {
+      "epoch": 96.42045454545455,
+      "grad_norm": 0.7081325650215149,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 33940
+    },
+    {
+      "epoch": 96.44886363636364,
+      "grad_norm": 0.7790707349777222,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 33950
+    },
+    {
+      "epoch": 96.47727272727273,
+      "grad_norm": 0.6598635315895081,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 33960
+    },
+    {
+      "epoch": 96.50568181818181,
+      "grad_norm": 0.6724303364753723,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 33970
+    },
+    {
+      "epoch": 96.5340909090909,
+      "grad_norm": 0.5733104944229126,
+      "learning_rate": 0.0001,
+      "loss": 0.0419,
+      "step": 33980
+    },
+    {
+      "epoch": 96.5625,
+      "grad_norm": 0.5401538610458374,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 33990
+    },
+    {
+      "epoch": 96.5909090909091,
+      "grad_norm": 0.614717423915863,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 34000
+    },
+    {
+      "epoch": 96.61931818181819,
+      "grad_norm": 0.5657342672348022,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 34010
+    },
+    {
+      "epoch": 96.64772727272727,
+      "grad_norm": 1.142454981803894,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 34020
+    },
+    {
+      "epoch": 96.67613636363636,
+      "grad_norm": 1.0728636980056763,
+      "learning_rate": 0.0001,
+      "loss": 0.0411,
+      "step": 34030
+    },
+    {
+      "epoch": 96.70454545454545,
+      "grad_norm": 1.2795010805130005,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 34040
+    },
+    {
+      "epoch": 96.73295454545455,
+      "grad_norm": 1.0543229579925537,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 34050
+    },
+    {
+      "epoch": 96.76136363636364,
+      "grad_norm": 1.0960166454315186,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 34060
+    },
+    {
+      "epoch": 96.78977272727273,
+      "grad_norm": 1.0084160566329956,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 34070
+    },
+    {
+      "epoch": 96.81818181818181,
+      "grad_norm": 0.9923943281173706,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 34080
+    },
+    {
+      "epoch": 96.8465909090909,
+      "grad_norm": 0.8087642788887024,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 34090
+    },
+    {
+      "epoch": 96.875,
+      "grad_norm": 1.208733081817627,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 34100
+    },
+    {
+      "epoch": 96.9034090909091,
+      "grad_norm": 0.9292431473731995,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 34110
+    },
+    {
+      "epoch": 96.93181818181819,
+      "grad_norm": 0.8942785263061523,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 34120
+    },
+    {
+      "epoch": 96.96022727272727,
+      "grad_norm": 0.863674521446228,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 34130
+    },
+    {
+      "epoch": 96.98863636363636,
+      "grad_norm": 1.0691416263580322,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 34140
+    },
+    {
+      "epoch": 97.01704545454545,
+      "grad_norm": 0.5712941884994507,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 34150
+    },
+    {
+      "epoch": 97.04545454545455,
+      "grad_norm": 0.6319429278373718,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 34160
+    },
+    {
+      "epoch": 97.07386363636364,
+      "grad_norm": 0.9614266157150269,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 34170
+    },
+    {
+      "epoch": 97.10227272727273,
+      "grad_norm": 0.7226883769035339,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 34180
+    },
+    {
+      "epoch": 97.13068181818181,
+      "grad_norm": 0.7708411812782288,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "step": 34190
+    },
+    {
+      "epoch": 97.1590909090909,
+      "grad_norm": 0.7161945104598999,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 34200
+    },
+    {
+      "epoch": 97.1875,
+      "grad_norm": 0.650200366973877,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 34210
+    },
+    {
+      "epoch": 97.2159090909091,
+      "grad_norm": 0.6397100687026978,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 34220
+    },
+    {
+      "epoch": 97.24431818181819,
+      "grad_norm": 0.7142146825790405,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 34230
+    },
+    {
+      "epoch": 97.27272727272727,
+      "grad_norm": 0.6408595442771912,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 34240
+    },
+    {
+      "epoch": 97.30113636363636,
+      "grad_norm": 0.6152955889701843,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 34250
+    },
+    {
+      "epoch": 97.32954545454545,
+      "grad_norm": 0.7230101227760315,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 34260
+    },
+    {
+      "epoch": 97.35795454545455,
+      "grad_norm": 0.945869505405426,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 34270
+    },
+    {
+      "epoch": 97.38636363636364,
+      "grad_norm": 0.9501050114631653,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 34280
+    },
+    {
+      "epoch": 97.41477272727273,
+      "grad_norm": 0.9380021691322327,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 34290
+    },
+    {
+      "epoch": 97.44318181818181,
+      "grad_norm": 1.3023756742477417,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "step": 34300
+    },
+    {
+      "epoch": 97.4715909090909,
+      "grad_norm": 1.0887079238891602,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 34310
+    },
+    {
+      "epoch": 97.5,
+      "grad_norm": 1.2671388387680054,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 34320
+    },
+    {
+      "epoch": 97.5284090909091,
+      "grad_norm": 0.7356063723564148,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 34330
+    },
+    {
+      "epoch": 97.55681818181819,
+      "grad_norm": 1.0338929891586304,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 34340
+    },
+    {
+      "epoch": 97.58522727272727,
+      "grad_norm": 1.080224633216858,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 34350
+    },
+    {
+      "epoch": 97.61363636363636,
+      "grad_norm": 1.2610585689544678,
+      "learning_rate": 0.0001,
+      "loss": 0.0413,
+      "step": 34360
+    },
+    {
+      "epoch": 97.64204545454545,
+      "grad_norm": 1.3167310953140259,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 34370
+    },
+    {
+      "epoch": 97.67045454545455,
+      "grad_norm": 1.0706456899642944,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 34380
+    },
+    {
+      "epoch": 97.69886363636364,
+      "grad_norm": 0.8218298554420471,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 34390
+    },
+    {
+      "epoch": 97.72727272727273,
+      "grad_norm": 1.1640470027923584,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 34400
+    },
+    {
+      "epoch": 97.75568181818181,
+      "grad_norm": 0.8588812947273254,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 34410
+    },
+    {
+      "epoch": 97.7840909090909,
+      "grad_norm": 0.892999529838562,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 34420
+    },
+    {
+      "epoch": 97.8125,
+      "grad_norm": 0.7717360258102417,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 34430
+    },
+    {
+      "epoch": 97.8409090909091,
+      "grad_norm": 0.8617984652519226,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 34440
+    },
+    {
+      "epoch": 97.86931818181819,
+      "grad_norm": 0.8200704455375671,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 34450
+    },
+    {
+      "epoch": 97.89772727272727,
+      "grad_norm": 0.6376478672027588,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 34460
+    },
+    {
+      "epoch": 97.92613636363636,
+      "grad_norm": 0.9599566459655762,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 34470
+    },
+    {
+      "epoch": 97.95454545454545,
+      "grad_norm": 0.7544838786125183,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 34480
+    },
+    {
+      "epoch": 97.98295454545455,
+      "grad_norm": 0.6836613416671753,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 34490
+    },
+    {
+      "epoch": 98.01136363636364,
+      "grad_norm": 0.806623101234436,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 34500
+    },
+    {
+      "epoch": 98.03977272727273,
+      "grad_norm": 0.9929761290550232,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 34510
+    },
+    {
+      "epoch": 98.06818181818181,
+      "grad_norm": 0.7506119012832642,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 34520
+    },
+    {
+      "epoch": 98.0965909090909,
+      "grad_norm": 0.7990569472312927,
+      "learning_rate": 0.0001,
+      "loss": 0.0409,
+      "step": 34530
+    },
+    {
+      "epoch": 98.125,
+      "grad_norm": 0.8026459813117981,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 34540
+    },
+    {
+      "epoch": 98.1534090909091,
+      "grad_norm": 0.7655097842216492,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 34550
+    },
+    {
+      "epoch": 98.18181818181819,
+      "grad_norm": 0.626440703868866,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 34560
+    },
+    {
+      "epoch": 98.21022727272727,
+      "grad_norm": 0.5965021252632141,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 34570
+    },
+    {
+      "epoch": 98.23863636363636,
+      "grad_norm": 0.6253120303153992,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 34580
+    },
+    {
+      "epoch": 98.26704545454545,
+      "grad_norm": 0.728787362575531,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "step": 34590
+    },
+    {
+      "epoch": 98.29545454545455,
+      "grad_norm": 0.6482828259468079,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 34600
+    },
+    {
+      "epoch": 98.32386363636364,
+      "grad_norm": 0.6943103671073914,
+      "learning_rate": 0.0001,
+      "loss": 0.0405,
+      "step": 34610
+    },
+    {
+      "epoch": 98.35227272727273,
+      "grad_norm": 0.7795711159706116,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 34620
+    },
+    {
+      "epoch": 98.38068181818181,
+      "grad_norm": 0.7584307193756104,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 34630
+    },
+    {
+      "epoch": 98.4090909090909,
+      "grad_norm": 0.583733320236206,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 34640
+    },
+    {
+      "epoch": 98.4375,
+      "grad_norm": 1.0233768224716187,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 34650
+    },
+    {
+      "epoch": 98.4659090909091,
+      "grad_norm": 0.7381771206855774,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 34660
+    },
+    {
+      "epoch": 98.49431818181819,
+      "grad_norm": 1.0933961868286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 34670
+    },
+    {
+      "epoch": 98.52272727272727,
+      "grad_norm": 0.8754884600639343,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 34680
+    },
+    {
+      "epoch": 98.55113636363636,
+      "grad_norm": 0.658818781375885,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 34690
+    },
+    {
+      "epoch": 98.57954545454545,
+      "grad_norm": 0.8832113146781921,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 34700
+    },
+    {
+      "epoch": 98.60795454545455,
+      "grad_norm": 0.6543091535568237,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 34710
+    },
+    {
+      "epoch": 98.63636363636364,
+      "grad_norm": 0.5386546850204468,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 34720
+    },
+    {
+      "epoch": 98.66477272727273,
+      "grad_norm": 0.4984776973724365,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 34730
+    },
+    {
+      "epoch": 98.69318181818181,
+      "grad_norm": 0.5178702473640442,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 34740
+    },
+    {
+      "epoch": 98.7215909090909,
+      "grad_norm": 0.7301501035690308,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 34750
+    },
+    {
+      "epoch": 98.75,
+      "grad_norm": 0.5040386319160461,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 34760
+    },
+    {
+      "epoch": 98.7784090909091,
+      "grad_norm": 0.4964589774608612,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 34770
+    },
+    {
+      "epoch": 98.80681818181819,
+      "grad_norm": 0.6709886193275452,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 34780
+    },
+    {
+      "epoch": 98.83522727272727,
+      "grad_norm": 0.6795845031738281,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 34790
+    },
+    {
+      "epoch": 98.86363636363636,
+      "grad_norm": 0.6201198101043701,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 34800
+    },
+    {
+      "epoch": 98.89204545454545,
+      "grad_norm": 0.5602060556411743,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 34810
+    },
+    {
+      "epoch": 98.92045454545455,
+      "grad_norm": 0.6460253000259399,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 34820
+    },
+    {
+      "epoch": 98.94886363636364,
+      "grad_norm": 0.6049633026123047,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 34830
+    },
+    {
+      "epoch": 98.97727272727273,
+      "grad_norm": 0.6341941356658936,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 34840
+    },
+    {
+      "epoch": 99.00568181818181,
+      "grad_norm": 0.6305556893348694,
+      "learning_rate": 0.0001,
+      "loss": 0.0414,
+      "step": 34850
+    },
+    {
+      "epoch": 99.0340909090909,
+      "grad_norm": 0.5669991970062256,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 34860
+    },
+    {
+      "epoch": 99.0625,
+      "grad_norm": 0.6804741024971008,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 34870
+    },
+    {
+      "epoch": 99.0909090909091,
+      "grad_norm": 0.6002970933914185,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 34880
+    },
+    {
+      "epoch": 99.11931818181819,
+      "grad_norm": 0.6289021968841553,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 34890
+    },
+    {
+      "epoch": 99.14772727272727,
+      "grad_norm": 0.5393396019935608,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 34900
+    },
+    {
+      "epoch": 99.17613636363636,
+      "grad_norm": 0.6519724130630493,
+      "learning_rate": 0.0001,
+      "loss": 0.0402,
+      "step": 34910
+    },
+    {
+      "epoch": 99.20454545454545,
+      "grad_norm": 0.6385529637336731,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 34920
+    },
+    {
+      "epoch": 99.23295454545455,
+      "grad_norm": 0.7283846139907837,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 34930
+    },
+    {
+      "epoch": 99.26136363636364,
+      "grad_norm": 0.5138035416603088,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 34940
+    },
+    {
+      "epoch": 99.28977272727273,
+      "grad_norm": 0.6235365867614746,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 34950
+    },
+    {
+      "epoch": 99.31818181818181,
+      "grad_norm": 0.6972271800041199,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 34960
+    },
+    {
+      "epoch": 99.3465909090909,
+      "grad_norm": 0.7025635242462158,
+      "learning_rate": 0.0001,
+      "loss": 0.0403,
+      "step": 34970
+    },
+    {
+      "epoch": 99.375,
+      "grad_norm": 0.5961763858795166,
+      "learning_rate": 0.0001,
+      "loss": 0.0396,
+      "step": 34980
+    },
+    {
+      "epoch": 99.4034090909091,
+      "grad_norm": 0.6492330431938171,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 34990
+    },
+    {
+      "epoch": 99.43181818181819,
+      "grad_norm": 0.6340757012367249,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 35000
+    },
+    {
+      "epoch": 99.46022727272727,
+      "grad_norm": 0.6484765410423279,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 35010
+    },
+    {
+      "epoch": 99.48863636363636,
+      "grad_norm": 0.5331886410713196,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 35020
+    },
+    {
+      "epoch": 99.51704545454545,
+      "grad_norm": 0.4786685109138489,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 35030
+    },
+    {
+      "epoch": 99.54545454545455,
+      "grad_norm": 0.5610359311103821,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 35040
+    },
+    {
+      "epoch": 99.57386363636364,
+      "grad_norm": 0.5391741394996643,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 35050
+    },
+    {
+      "epoch": 99.60227272727273,
+      "grad_norm": 0.610522985458374,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 35060
+    },
+    {
+      "epoch": 99.63068181818181,
+      "grad_norm": 0.6739444732666016,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 35070
+    },
+    {
+      "epoch": 99.6590909090909,
+      "grad_norm": 0.5937843918800354,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 35080
+    },
+    {
+      "epoch": 99.6875,
+      "grad_norm": 0.9213070869445801,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 35090
+    },
+    {
+      "epoch": 99.7159090909091,
+      "grad_norm": 1.3140711784362793,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 35100
+    },
+    {
+      "epoch": 99.74431818181819,
+      "grad_norm": 1.2353553771972656,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 35110
+    },
+    {
+      "epoch": 99.77272727272727,
+      "grad_norm": 0.7020501494407654,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 35120
+    },
+    {
+      "epoch": 99.80113636363636,
+      "grad_norm": 0.8266453742980957,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 35130
+    },
+    {
+      "epoch": 99.82954545454545,
+      "grad_norm": 0.7972448468208313,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 35140
+    },
+    {
+      "epoch": 99.85795454545455,
+      "grad_norm": 0.7136993408203125,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 35150
+    },
+    {
+      "epoch": 99.88636363636364,
+      "grad_norm": 0.794268012046814,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 35160
+    },
+    {
+      "epoch": 99.91477272727273,
+      "grad_norm": 0.7449502348899841,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 35170
+    },
+    {
+      "epoch": 99.94318181818181,
+      "grad_norm": 0.5898183584213257,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 35180
+    },
+    {
+      "epoch": 99.9715909090909,
+      "grad_norm": 0.5253759622573853,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 35190
+    },
+    {
+      "epoch": 100.0,
+      "grad_norm": 0.5536308288574219,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 35200
+    },
+    {
+      "epoch": 100.0284090909091,
+      "grad_norm": 0.712748110294342,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 35210
+    },
+    {
+      "epoch": 100.05681818181819,
+      "grad_norm": 0.6205607056617737,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 35220
+    },
+    {
+      "epoch": 100.08522727272727,
+      "grad_norm": 0.8103065490722656,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 35230
+    },
+    {
+      "epoch": 100.11363636363636,
+      "grad_norm": 0.8175324201583862,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 35240
+    },
+    {
+      "epoch": 100.14204545454545,
+      "grad_norm": 0.6265504956245422,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 35250
+    },
+    {
+      "epoch": 100.17045454545455,
+      "grad_norm": 0.7531624436378479,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 35260
+    },
+    {
+      "epoch": 100.19886363636364,
+      "grad_norm": 0.7789162993431091,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 35270
+    },
+    {
+      "epoch": 100.22727272727273,
+      "grad_norm": 0.7431286573410034,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 35280
+    },
+    {
+      "epoch": 100.25568181818181,
+      "grad_norm": 0.6338279247283936,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 35290
+    },
+    {
+      "epoch": 100.2840909090909,
+      "grad_norm": 0.6069151759147644,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 35300
+    },
+    {
+      "epoch": 100.3125,
+      "grad_norm": 0.6290576457977295,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 35310
+    },
+    {
+      "epoch": 100.3409090909091,
+      "grad_norm": 0.5072388052940369,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 35320
+    },
+    {
+      "epoch": 100.36931818181819,
+      "grad_norm": 0.9051946401596069,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 35330
+    },
+    {
+      "epoch": 100.39772727272727,
+      "grad_norm": 0.9270437955856323,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 35340
+    },
+    {
+      "epoch": 100.42613636363636,
+      "grad_norm": 1.1097337007522583,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 35350
+    },
+    {
+      "epoch": 100.45454545454545,
+      "grad_norm": 0.753572404384613,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 35360
+    },
+    {
+      "epoch": 100.48295454545455,
+      "grad_norm": 0.8269745111465454,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 35370
+    },
+    {
+      "epoch": 100.51136363636364,
+      "grad_norm": 0.9077835083007812,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 35380
+    },
+    {
+      "epoch": 100.53977272727273,
+      "grad_norm": 0.863038182258606,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 35390
+    },
+    {
+      "epoch": 100.56818181818181,
+      "grad_norm": 0.7924647927284241,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 35400
+    },
+    {
+      "epoch": 100.5965909090909,
+      "grad_norm": 0.8401536345481873,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 35410
+    },
+    {
+      "epoch": 100.625,
+      "grad_norm": 0.8418139219284058,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 35420
+    },
+    {
+      "epoch": 100.6534090909091,
+      "grad_norm": 0.8648441433906555,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 35430
+    },
+    {
+      "epoch": 100.68181818181819,
+      "grad_norm": 0.7955145239830017,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 35440
+    },
+    {
+      "epoch": 100.71022727272727,
+      "grad_norm": 0.759017825126648,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 35450
+    },
+    {
+      "epoch": 100.73863636363636,
+      "grad_norm": 0.7723873853683472,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 35460
+    },
+    {
+      "epoch": 100.76704545454545,
+      "grad_norm": 0.5847647190093994,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 35470
+    },
+    {
+      "epoch": 100.79545454545455,
+      "grad_norm": 0.7090848684310913,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 35480
+    },
+    {
+      "epoch": 100.82386363636364,
+      "grad_norm": 0.5850080847740173,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 35490
+    },
+    {
+      "epoch": 100.85227272727273,
+      "grad_norm": 0.6634331345558167,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 35500
+    },
+    {
+      "epoch": 100.88068181818181,
+      "grad_norm": 0.5659244060516357,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 35510
+    },
+    {
+      "epoch": 100.9090909090909,
+      "grad_norm": 0.6139445304870605,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 35520
+    },
+    {
+      "epoch": 100.9375,
+      "grad_norm": 0.5992445945739746,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 35530
+    },
+    {
+      "epoch": 100.9659090909091,
+      "grad_norm": 0.705426037311554,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 35540
+    },
+    {
+      "epoch": 100.99431818181819,
+      "grad_norm": 0.8458812832832336,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 35550
+    },
+    {
+      "epoch": 101.02272727272727,
+      "grad_norm": 0.8477011919021606,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 35560
+    },
+    {
+      "epoch": 101.05113636363636,
+      "grad_norm": 0.7048535346984863,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 35570
+    },
+    {
+      "epoch": 101.07954545454545,
+      "grad_norm": 0.9626438617706299,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 35580
+    },
+    {
+      "epoch": 101.10795454545455,
+      "grad_norm": 0.8232958912849426,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 35590
+    },
+    {
+      "epoch": 101.13636363636364,
+      "grad_norm": 1.2208539247512817,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 35600
+    },
+    {
+      "epoch": 101.16477272727273,
+      "grad_norm": 0.8574521541595459,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 35610
+    },
+    {
+      "epoch": 101.19318181818181,
+      "grad_norm": 0.536631166934967,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 35620
+    },
+    {
+      "epoch": 101.2215909090909,
+      "grad_norm": 0.6252682209014893,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 35630
+    },
+    {
+      "epoch": 101.25,
+      "grad_norm": 0.6429247260093689,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 35640
+    },
+    {
+      "epoch": 101.2784090909091,
+      "grad_norm": 0.6166262626647949,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 35650
+    },
+    {
+      "epoch": 101.30681818181819,
+      "grad_norm": 0.6511677503585815,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 35660
+    },
+    {
+      "epoch": 101.33522727272727,
+      "grad_norm": 0.6906519532203674,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 35670
+    },
+    {
+      "epoch": 101.36363636363636,
+      "grad_norm": 0.6717009544372559,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 35680
+    },
+    {
+      "epoch": 101.39204545454545,
+      "grad_norm": 0.6030964255332947,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 35690
+    },
+    {
+      "epoch": 101.42045454545455,
+      "grad_norm": 0.6418792009353638,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 35700
+    },
+    {
+      "epoch": 101.44886363636364,
+      "grad_norm": 0.7006118297576904,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 35710
+    },
+    {
+      "epoch": 101.47727272727273,
+      "grad_norm": 0.7065404653549194,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 35720
+    },
+    {
+      "epoch": 101.50568181818181,
+      "grad_norm": 0.7001347541809082,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 35730
+    },
+    {
+      "epoch": 101.5340909090909,
+      "grad_norm": 0.7850422859191895,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 35740
+    },
+    {
+      "epoch": 101.5625,
+      "grad_norm": 0.6983082294464111,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 35750
+    },
+    {
+      "epoch": 101.5909090909091,
+      "grad_norm": 0.6168388724327087,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 35760
+    },
+    {
+      "epoch": 101.61931818181819,
+      "grad_norm": 0.6485419869422913,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 35770
+    },
+    {
+      "epoch": 101.64772727272727,
+      "grad_norm": 0.694078803062439,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 35780
+    },
+    {
+      "epoch": 101.67613636363636,
+      "grad_norm": 0.6054084300994873,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 35790
+    },
+    {
+      "epoch": 101.70454545454545,
+      "grad_norm": 0.7031105756759644,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 35800
+    },
+    {
+      "epoch": 101.73295454545455,
+      "grad_norm": 0.6772159337997437,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 35810
+    },
+    {
+      "epoch": 101.76136363636364,
+      "grad_norm": 0.742162823677063,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 35820
+    },
+    {
+      "epoch": 101.78977272727273,
+      "grad_norm": 0.6221441626548767,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 35830
+    },
+    {
+      "epoch": 101.81818181818181,
+      "grad_norm": 0.5692674517631531,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 35840
+    },
+    {
+      "epoch": 101.8465909090909,
+      "grad_norm": 0.5200670957565308,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 35850
+    },
+    {
+      "epoch": 101.875,
+      "grad_norm": 0.44014811515808105,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 35860
+    },
+    {
+      "epoch": 101.9034090909091,
+      "grad_norm": 0.5131399035453796,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 35870
+    },
+    {
+      "epoch": 101.93181818181819,
+      "grad_norm": 0.7803551554679871,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 35880
+    },
+    {
+      "epoch": 101.96022727272727,
+      "grad_norm": 1.0714529752731323,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 35890
+    },
+    {
+      "epoch": 101.98863636363636,
+      "grad_norm": 0.8870387077331543,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 35900
+    },
+    {
+      "epoch": 102.01704545454545,
+      "grad_norm": 0.814504861831665,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 35910
+    },
+    {
+      "epoch": 102.04545454545455,
+      "grad_norm": 1.2375514507293701,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 35920
+    },
+    {
+      "epoch": 102.07386363636364,
+      "grad_norm": 1.0883283615112305,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 35930
+    },
+    {
+      "epoch": 102.10227272727273,
+      "grad_norm": 0.7217296361923218,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 35940
+    },
+    {
+      "epoch": 102.13068181818181,
+      "grad_norm": 0.9220253229141235,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 35950
+    },
+    {
+      "epoch": 102.1590909090909,
+      "grad_norm": 0.9065287113189697,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 35960
+    },
+    {
+      "epoch": 102.1875,
+      "grad_norm": 1.2304972410202026,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 35970
+    },
+    {
+      "epoch": 102.2159090909091,
+      "grad_norm": 0.9404845833778381,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 35980
+    },
+    {
+      "epoch": 102.24431818181819,
+      "grad_norm": 0.9212031960487366,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 35990
+    },
+    {
+      "epoch": 102.27272727272727,
+      "grad_norm": 0.7656883597373962,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 36000
+    },
+    {
+      "epoch": 102.30113636363636,
+      "grad_norm": 0.7965297102928162,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 36010
+    },
+    {
+      "epoch": 102.32954545454545,
+      "grad_norm": 0.6216439604759216,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 36020
+    },
+    {
+      "epoch": 102.35795454545455,
+      "grad_norm": 0.6738339066505432,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 36030
+    },
+    {
+      "epoch": 102.38636363636364,
+      "grad_norm": 0.6152580380439758,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 36040
+    },
+    {
+      "epoch": 102.41477272727273,
+      "grad_norm": 0.6139420866966248,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 36050
+    },
+    {
+      "epoch": 102.44318181818181,
+      "grad_norm": 0.5595870614051819,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 36060
+    },
+    {
+      "epoch": 102.4715909090909,
+      "grad_norm": 0.721856951713562,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 36070
+    },
+    {
+      "epoch": 102.5,
+      "grad_norm": 0.7998674511909485,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 36080
+    },
+    {
+      "epoch": 102.5284090909091,
+      "grad_norm": 0.8735951781272888,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 36090
+    },
+    {
+      "epoch": 102.55681818181819,
+      "grad_norm": 0.7308524250984192,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 36100
+    },
+    {
+      "epoch": 102.58522727272727,
+      "grad_norm": 0.7623921036720276,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 36110
+    },
+    {
+      "epoch": 102.61363636363636,
+      "grad_norm": 0.8306724429130554,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 36120
+    },
+    {
+      "epoch": 102.64204545454545,
+      "grad_norm": 0.8680564165115356,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 36130
+    },
+    {
+      "epoch": 102.67045454545455,
+      "grad_norm": 0.8162680864334106,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 36140
+    },
+    {
+      "epoch": 102.69886363636364,
+      "grad_norm": 0.6912002563476562,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 36150
+    },
+    {
+      "epoch": 102.72727272727273,
+      "grad_norm": 0.7910269498825073,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 36160
+    },
+    {
+      "epoch": 102.75568181818181,
+      "grad_norm": 0.6541531682014465,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 36170
+    },
+    {
+      "epoch": 102.7840909090909,
+      "grad_norm": 0.5540030002593994,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 36180
+    },
+    {
+      "epoch": 102.8125,
+      "grad_norm": 0.6944588422775269,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 36190
+    },
+    {
+      "epoch": 102.8409090909091,
+      "grad_norm": 0.6281765699386597,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 36200
+    },
+    {
+      "epoch": 102.86931818181819,
+      "grad_norm": 0.5334039330482483,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 36210
+    },
+    {
+      "epoch": 102.89772727272727,
+      "grad_norm": 0.6771912574768066,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 36220
+    },
+    {
+      "epoch": 102.92613636363636,
+      "grad_norm": 0.5143633484840393,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 36230
+    },
+    {
+      "epoch": 102.95454545454545,
+      "grad_norm": 0.5875553488731384,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 36240
+    },
+    {
+      "epoch": 102.98295454545455,
+      "grad_norm": 0.5696831941604614,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 36250
+    },
+    {
+      "epoch": 103.01136363636364,
+      "grad_norm": 0.6309694647789001,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 36260
+    },
+    {
+      "epoch": 103.03977272727273,
+      "grad_norm": 0.8848923444747925,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 36270
+    },
+    {
+      "epoch": 103.06818181818181,
+      "grad_norm": 0.578700840473175,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 36280
+    },
+    {
+      "epoch": 103.0965909090909,
+      "grad_norm": 0.6599058508872986,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 36290
+    },
+    {
+      "epoch": 103.125,
+      "grad_norm": 0.7933474779129028,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 36300
+    },
+    {
+      "epoch": 103.1534090909091,
+      "grad_norm": 0.8254300355911255,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 36310
+    },
+    {
+      "epoch": 103.18181818181819,
+      "grad_norm": 0.6838820576667786,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 36320
+    },
+    {
+      "epoch": 103.21022727272727,
+      "grad_norm": 0.686337947845459,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 36330
+    },
+    {
+      "epoch": 103.23863636363636,
+      "grad_norm": 0.768295168876648,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 36340
+    },
+    {
+      "epoch": 103.26704545454545,
+      "grad_norm": 0.7727903127670288,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 36350
+    },
+    {
+      "epoch": 103.29545454545455,
+      "grad_norm": 0.4438778758049011,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 36360
+    },
+    {
+      "epoch": 103.32386363636364,
+      "grad_norm": 0.5905418395996094,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 36370
+    },
+    {
+      "epoch": 103.35227272727273,
+      "grad_norm": 0.6414166688919067,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 36380
+    },
+    {
+      "epoch": 103.38068181818181,
+      "grad_norm": 0.6797493100166321,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 36390
+    },
+    {
+      "epoch": 103.4090909090909,
+      "grad_norm": 0.653895914554596,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 36400
+    },
+    {
+      "epoch": 103.4375,
+      "grad_norm": 0.6173151731491089,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 36410
+    },
+    {
+      "epoch": 103.4659090909091,
+      "grad_norm": 0.47727423906326294,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 36420
+    },
+    {
+      "epoch": 103.49431818181819,
+      "grad_norm": 0.6933102011680603,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 36430
+    },
+    {
+      "epoch": 103.52272727272727,
+      "grad_norm": 0.6333000063896179,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 36440
+    },
+    {
+      "epoch": 103.55113636363636,
+      "grad_norm": 0.5522160530090332,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 36450
+    },
+    {
+      "epoch": 103.57954545454545,
+      "grad_norm": 0.6237143278121948,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 36460
+    },
+    {
+      "epoch": 103.60795454545455,
+      "grad_norm": 0.7558906674385071,
+      "learning_rate": 0.0001,
+      "loss": 0.0395,
+      "step": 36470
+    },
+    {
+      "epoch": 103.63636363636364,
+      "grad_norm": 0.5823392868041992,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 36480
+    },
+    {
+      "epoch": 103.66477272727273,
+      "grad_norm": 0.6411099433898926,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 36490
+    },
+    {
+      "epoch": 103.69318181818181,
+      "grad_norm": 0.5531007051467896,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 36500
+    },
+    {
+      "epoch": 103.7215909090909,
+      "grad_norm": 0.5332525968551636,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 36510
+    },
+    {
+      "epoch": 103.75,
+      "grad_norm": 0.6390560269355774,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 36520
+    },
+    {
+      "epoch": 103.7784090909091,
+      "grad_norm": 0.6145923137664795,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 36530
+    },
+    {
+      "epoch": 103.80681818181819,
+      "grad_norm": 0.663539707660675,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 36540
+    },
+    {
+      "epoch": 103.83522727272727,
+      "grad_norm": 0.6167774796485901,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 36550
+    },
+    {
+      "epoch": 103.86363636363636,
+      "grad_norm": 0.7627246975898743,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 36560
+    },
+    {
+      "epoch": 103.89204545454545,
+      "grad_norm": 0.5639641880989075,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 36570
+    },
+    {
+      "epoch": 103.92045454545455,
+      "grad_norm": 0.6381582617759705,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 36580
+    },
+    {
+      "epoch": 103.94886363636364,
+      "grad_norm": 0.6323047876358032,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 36590
+    },
+    {
+      "epoch": 103.97727272727273,
+      "grad_norm": 0.5740509629249573,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 36600
+    },
+    {
+      "epoch": 104.00568181818181,
+      "grad_norm": 0.6786510348320007,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 36610
+    },
+    {
+      "epoch": 104.0340909090909,
+      "grad_norm": 0.6328359246253967,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 36620
+    },
+    {
+      "epoch": 104.0625,
+      "grad_norm": 0.5744404792785645,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 36630
+    },
+    {
+      "epoch": 104.0909090909091,
+      "grad_norm": 0.6046051979064941,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 36640
+    },
+    {
+      "epoch": 104.11931818181819,
+      "grad_norm": 0.6407805681228638,
+      "learning_rate": 0.0001,
+      "loss": 0.0387,
+      "step": 36650
+    },
+    {
+      "epoch": 104.14772727272727,
+      "grad_norm": 0.8871857523918152,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 36660
+    },
+    {
+      "epoch": 104.17613636363636,
+      "grad_norm": 0.8872131705284119,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 36670
+    },
+    {
+      "epoch": 104.20454545454545,
+      "grad_norm": 0.7679946422576904,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 36680
+    },
+    {
+      "epoch": 104.23295454545455,
+      "grad_norm": 0.6552764177322388,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 36690
+    },
+    {
+      "epoch": 104.26136363636364,
+      "grad_norm": 0.8447787761688232,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 36700
+    },
+    {
+      "epoch": 104.28977272727273,
+      "grad_norm": 0.6279290318489075,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 36710
+    },
+    {
+      "epoch": 104.31818181818181,
+      "grad_norm": 0.5877067446708679,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 36720
+    },
+    {
+      "epoch": 104.3465909090909,
+      "grad_norm": 0.8654998540878296,
+      "learning_rate": 0.0001,
+      "loss": 0.0408,
+      "step": 36730
+    },
+    {
+      "epoch": 104.375,
+      "grad_norm": 0.6628789901733398,
+      "learning_rate": 0.0001,
+      "loss": 0.0415,
+      "step": 36740
+    },
+    {
+      "epoch": 104.4034090909091,
+      "grad_norm": 1.0221258401870728,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 36750
+    },
+    {
+      "epoch": 104.43181818181819,
+      "grad_norm": 1.2567273378372192,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 36760
+    },
+    {
+      "epoch": 104.46022727272727,
+      "grad_norm": 1.0356628894805908,
+      "learning_rate": 0.0001,
+      "loss": 0.0404,
+      "step": 36770
+    },
+    {
+      "epoch": 104.48863636363636,
+      "grad_norm": 1.1455295085906982,
+      "learning_rate": 0.0001,
+      "loss": 0.0397,
+      "step": 36780
+    },
+    {
+      "epoch": 104.51704545454545,
+      "grad_norm": 1.0565097332000732,
+      "learning_rate": 0.0001,
+      "loss": 0.0392,
+      "step": 36790
+    },
+    {
+      "epoch": 104.54545454545455,
+      "grad_norm": 0.8333950638771057,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 36800
+    },
+    {
+      "epoch": 104.57386363636364,
+      "grad_norm": 0.8816116452217102,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 36810
+    },
+    {
+      "epoch": 104.60227272727273,
+      "grad_norm": 0.8084827661514282,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 36820
+    },
+    {
+      "epoch": 104.63068181818181,
+      "grad_norm": 0.7814130187034607,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 36830
+    },
+    {
+      "epoch": 104.6590909090909,
+      "grad_norm": 0.6202470064163208,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 36840
+    },
+    {
+      "epoch": 104.6875,
+      "grad_norm": 0.6300608515739441,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 36850
+    },
+    {
+      "epoch": 104.7159090909091,
+      "grad_norm": 0.6419079303741455,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 36860
+    },
+    {
+      "epoch": 104.74431818181819,
+      "grad_norm": 0.6812740564346313,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 36870
+    },
+    {
+      "epoch": 104.77272727272727,
+      "grad_norm": 0.8751927614212036,
+      "learning_rate": 0.0001,
+      "loss": 0.0398,
+      "step": 36880
+    },
+    {
+      "epoch": 104.80113636363636,
+      "grad_norm": 0.694141149520874,
+      "learning_rate": 0.0001,
+      "loss": 0.0394,
+      "step": 36890
+    },
+    {
+      "epoch": 104.82954545454545,
+      "grad_norm": 0.6310260891914368,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 36900
+    },
+    {
+      "epoch": 104.85795454545455,
+      "grad_norm": 0.45405858755111694,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 36910
+    },
+    {
+      "epoch": 104.88636363636364,
+      "grad_norm": 0.7782346606254578,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 36920
+    },
+    {
+      "epoch": 104.91477272727273,
+      "grad_norm": 0.8152375221252441,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 36930
+    },
+    {
+      "epoch": 104.94318181818181,
+      "grad_norm": 0.6959502696990967,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 36940
+    },
+    {
+      "epoch": 104.9715909090909,
+      "grad_norm": 0.805332362651825,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 36950
+    },
+    {
+      "epoch": 105.0,
+      "grad_norm": 0.8097184896469116,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 36960
+    },
+    {
+      "epoch": 105.0284090909091,
+      "grad_norm": 0.611344039440155,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 36970
+    },
+    {
+      "epoch": 105.05681818181819,
+      "grad_norm": 0.8121931552886963,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 36980
+    },
+    {
+      "epoch": 105.08522727272727,
+      "grad_norm": 0.965552568435669,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 36990
+    },
+    {
+      "epoch": 105.11363636363636,
+      "grad_norm": 0.772780179977417,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 37000
+    },
+    {
+      "epoch": 105.14204545454545,
+      "grad_norm": 0.8075849413871765,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 37010
+    },
+    {
+      "epoch": 105.17045454545455,
+      "grad_norm": 0.8520001173019409,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 37020
+    },
+    {
+      "epoch": 105.19886363636364,
+      "grad_norm": 0.9963288903236389,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 37030
+    },
+    {
+      "epoch": 105.22727272727273,
+      "grad_norm": 1.0980024337768555,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 37040
+    },
+    {
+      "epoch": 105.25568181818181,
+      "grad_norm": 0.8528324961662292,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 37050
+    },
+    {
+      "epoch": 105.2840909090909,
+      "grad_norm": 1.0183886289596558,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 37060
+    },
+    {
+      "epoch": 105.3125,
+      "grad_norm": 0.9727984666824341,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 37070
+    },
+    {
+      "epoch": 105.3409090909091,
+      "grad_norm": 1.1144496202468872,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 37080
+    },
+    {
+      "epoch": 105.36931818181819,
+      "grad_norm": 1.2052735090255737,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 37090
+    },
+    {
+      "epoch": 105.39772727272727,
+      "grad_norm": 0.855655312538147,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 37100
+    },
+    {
+      "epoch": 105.42613636363636,
+      "grad_norm": 0.714371383190155,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 37110
+    },
+    {
+      "epoch": 105.45454545454545,
+      "grad_norm": 0.7759353518486023,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 37120
+    },
+    {
+      "epoch": 105.48295454545455,
+      "grad_norm": 0.8793251514434814,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 37130
+    },
+    {
+      "epoch": 105.51136363636364,
+      "grad_norm": 0.8084626793861389,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 37140
+    },
+    {
+      "epoch": 105.53977272727273,
+      "grad_norm": 0.8977124094963074,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 37150
+    },
+    {
+      "epoch": 105.56818181818181,
+      "grad_norm": 0.6372682452201843,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 37160
+    },
+    {
+      "epoch": 105.5965909090909,
+      "grad_norm": 0.6709948182106018,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 37170
+    },
+    {
+      "epoch": 105.625,
+      "grad_norm": 0.8853369951248169,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 37180
+    },
+    {
+      "epoch": 105.6534090909091,
+      "grad_norm": 0.6816346645355225,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 37190
+    },
+    {
+      "epoch": 105.68181818181819,
+      "grad_norm": 0.8214898705482483,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 37200
+    },
+    {
+      "epoch": 105.71022727272727,
+      "grad_norm": 0.7285953164100647,
+      "learning_rate": 0.0001,
+      "loss": 0.0407,
+      "step": 37210
+    },
+    {
+      "epoch": 105.73863636363636,
+      "grad_norm": 0.7041904926300049,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 37220
+    },
+    {
+      "epoch": 105.76704545454545,
+      "grad_norm": 0.7178189754486084,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 37230
+    },
+    {
+      "epoch": 105.79545454545455,
+      "grad_norm": 1.0095086097717285,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 37240
+    },
+    {
+      "epoch": 105.82386363636364,
+      "grad_norm": 0.9176076054573059,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 37250
+    },
+    {
+      "epoch": 105.85227272727273,
+      "grad_norm": 0.8059961795806885,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 37260
+    },
+    {
+      "epoch": 105.88068181818181,
+      "grad_norm": 0.8492456078529358,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 37270
+    },
+    {
+      "epoch": 105.9090909090909,
+      "grad_norm": 0.8832942247390747,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 37280
+    },
+    {
+      "epoch": 105.9375,
+      "grad_norm": 0.7502552270889282,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 37290
+    },
+    {
+      "epoch": 105.9659090909091,
+      "grad_norm": 0.6862600445747375,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 37300
+    },
+    {
+      "epoch": 105.99431818181819,
+      "grad_norm": 0.6588712334632874,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 37310
+    },
+    {
+      "epoch": 106.02272727272727,
+      "grad_norm": 0.5885828137397766,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 37320
+    },
+    {
+      "epoch": 106.05113636363636,
+      "grad_norm": 0.5425651669502258,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 37330
+    },
+    {
+      "epoch": 106.07954545454545,
+      "grad_norm": 0.5327818393707275,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 37340
+    },
+    {
+      "epoch": 106.10795454545455,
+      "grad_norm": 0.6177202463150024,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 37350
+    },
+    {
+      "epoch": 106.13636363636364,
+      "grad_norm": 0.473233699798584,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 37360
+    },
+    {
+      "epoch": 106.16477272727273,
+      "grad_norm": 0.48036202788352966,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 37370
+    },
+    {
+      "epoch": 106.19318181818181,
+      "grad_norm": 0.699809193611145,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 37380
+    },
+    {
+      "epoch": 106.2215909090909,
+      "grad_norm": 0.865032434463501,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 37390
+    },
+    {
+      "epoch": 106.25,
+      "grad_norm": 0.707669198513031,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 37400
+    },
+    {
+      "epoch": 106.2784090909091,
+      "grad_norm": 0.6514762043952942,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 37410
+    },
+    {
+      "epoch": 106.30681818181819,
+      "grad_norm": 0.7002271413803101,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 37420
+    },
+    {
+      "epoch": 106.33522727272727,
+      "grad_norm": 0.6672790050506592,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 37430
+    },
+    {
+      "epoch": 106.36363636363636,
+      "grad_norm": 0.7932654619216919,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 37440
+    },
+    {
+      "epoch": 106.39204545454545,
+      "grad_norm": 0.6026307940483093,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 37450
+    },
+    {
+      "epoch": 106.42045454545455,
+      "grad_norm": 0.6472262144088745,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 37460
+    },
+    {
+      "epoch": 106.44886363636364,
+      "grad_norm": 0.6252856254577637,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 37470
+    },
+    {
+      "epoch": 106.47727272727273,
+      "grad_norm": 0.5449088215827942,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 37480
+    },
+    {
+      "epoch": 106.50568181818181,
+      "grad_norm": 0.7351489663124084,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 37490
+    },
+    {
+      "epoch": 106.5340909090909,
+      "grad_norm": 0.684830367565155,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 37500
+    },
+    {
+      "epoch": 106.5625,
+      "grad_norm": 0.522684633731842,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 37510
+    },
+    {
+      "epoch": 106.5909090909091,
+      "grad_norm": 0.5345920920372009,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 37520
+    },
+    {
+      "epoch": 106.61931818181819,
+      "grad_norm": 0.5353555083274841,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 37530
+    },
+    {
+      "epoch": 106.64772727272727,
+      "grad_norm": 0.631646454334259,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 37540
+    },
+    {
+      "epoch": 106.67613636363636,
+      "grad_norm": 0.6440027952194214,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 37550
+    },
+    {
+      "epoch": 106.70454545454545,
+      "grad_norm": 0.5962269306182861,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 37560
+    },
+    {
+      "epoch": 106.73295454545455,
+      "grad_norm": 0.6566577553749084,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 37570
+    },
+    {
+      "epoch": 106.76136363636364,
+      "grad_norm": 0.6731533408164978,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 37580
+    },
+    {
+      "epoch": 106.78977272727273,
+      "grad_norm": 0.8614497780799866,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 37590
+    },
+    {
+      "epoch": 106.81818181818181,
+      "grad_norm": 0.7013105154037476,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 37600
+    },
+    {
+      "epoch": 106.8465909090909,
+      "grad_norm": 0.7781490087509155,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 37610
+    },
+    {
+      "epoch": 106.875,
+      "grad_norm": 0.6093642711639404,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 37620
+    },
+    {
+      "epoch": 106.9034090909091,
+      "grad_norm": 0.5649243593215942,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 37630
+    },
+    {
+      "epoch": 106.93181818181819,
+      "grad_norm": 0.6890942454338074,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 37640
+    },
+    {
+      "epoch": 106.96022727272727,
+      "grad_norm": 0.7046976089477539,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 37650
+    },
+    {
+      "epoch": 106.98863636363636,
+      "grad_norm": 0.6145200729370117,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 37660
+    },
+    {
+      "epoch": 107.01704545454545,
+      "grad_norm": 0.4534114897251129,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 37670
+    },
+    {
+      "epoch": 107.04545454545455,
+      "grad_norm": 0.519280195236206,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 37680
+    },
+    {
+      "epoch": 107.07386363636364,
+      "grad_norm": 0.6525757312774658,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 37690
+    },
+    {
+      "epoch": 107.10227272727273,
+      "grad_norm": 0.6470338106155396,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 37700
+    },
+    {
+      "epoch": 107.13068181818181,
+      "grad_norm": 0.5215091705322266,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 37710
+    },
+    {
+      "epoch": 107.1590909090909,
+      "grad_norm": 0.48153358697891235,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 37720
+    },
+    {
+      "epoch": 107.1875,
+      "grad_norm": 0.6757826209068298,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 37730
+    },
+    {
+      "epoch": 107.2159090909091,
+      "grad_norm": 0.7484776377677917,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 37740
+    },
+    {
+      "epoch": 107.24431818181819,
+      "grad_norm": 0.6073766946792603,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 37750
+    },
+    {
+      "epoch": 107.27272727272727,
+      "grad_norm": 0.45944106578826904,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 37760
+    },
+    {
+      "epoch": 107.30113636363636,
+      "grad_norm": 0.481180876493454,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 37770
+    },
+    {
+      "epoch": 107.32954545454545,
+      "grad_norm": 0.5007697343826294,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 37780
+    },
+    {
+      "epoch": 107.35795454545455,
+      "grad_norm": 0.6877015233039856,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 37790
+    },
+    {
+      "epoch": 107.38636363636364,
+      "grad_norm": 1.0906400680541992,
+      "learning_rate": 0.0001,
+      "loss": 0.0383,
+      "step": 37800
+    },
+    {
+      "epoch": 107.41477272727273,
+      "grad_norm": 0.9992173910140991,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 37810
+    },
+    {
+      "epoch": 107.44318181818181,
+      "grad_norm": 1.065973162651062,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 37820
+    },
+    {
+      "epoch": 107.4715909090909,
+      "grad_norm": 0.8227842450141907,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 37830
+    },
+    {
+      "epoch": 107.5,
+      "grad_norm": 0.830697238445282,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 37840
+    },
+    {
+      "epoch": 107.5284090909091,
+      "grad_norm": 0.652164101600647,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 37850
+    },
+    {
+      "epoch": 107.55681818181819,
+      "grad_norm": 0.6938498020172119,
+      "learning_rate": 0.0001,
+      "loss": 0.0389,
+      "step": 37860
+    },
+    {
+      "epoch": 107.58522727272727,
+      "grad_norm": 0.680587887763977,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 37870
+    },
+    {
+      "epoch": 107.61363636363636,
+      "grad_norm": 0.6912381649017334,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 37880
+    },
+    {
+      "epoch": 107.64204545454545,
+      "grad_norm": 0.6087629795074463,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 37890
+    },
+    {
+      "epoch": 107.67045454545455,
+      "grad_norm": 0.8172582983970642,
+      "learning_rate": 0.0001,
+      "loss": 0.0385,
+      "step": 37900
+    },
+    {
+      "epoch": 107.69886363636364,
+      "grad_norm": 0.7405523657798767,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 37910
+    },
+    {
+      "epoch": 107.72727272727273,
+      "grad_norm": 0.6305014491081238,
+      "learning_rate": 0.0001,
+      "loss": 0.0384,
+      "step": 37920
+    },
+    {
+      "epoch": 107.75568181818181,
+      "grad_norm": 0.6484267711639404,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 37930
+    },
+    {
+      "epoch": 107.7840909090909,
+      "grad_norm": 0.6926651000976562,
+      "learning_rate": 0.0001,
+      "loss": 0.0382,
+      "step": 37940
+    },
+    {
+      "epoch": 107.8125,
+      "grad_norm": 0.7457519769668579,
+      "learning_rate": 0.0001,
+      "loss": 0.04,
+      "step": 37950
+    },
+    {
+      "epoch": 107.8409090909091,
+      "grad_norm": 0.6257100701332092,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 37960
+    },
+    {
+      "epoch": 107.86931818181819,
+      "grad_norm": 0.6201730966567993,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 37970
+    },
+    {
+      "epoch": 107.89772727272727,
+      "grad_norm": 0.6591887474060059,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 37980
+    },
+    {
+      "epoch": 107.92613636363636,
+      "grad_norm": 0.827599287033081,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 37990
+    },
+    {
+      "epoch": 107.95454545454545,
+      "grad_norm": 1.455817699432373,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 38000
+    },
+    {
+      "epoch": 107.98295454545455,
+      "grad_norm": 0.6563143730163574,
+      "learning_rate": 0.0001,
+      "loss": 0.0386,
+      "step": 38010
+    },
+    {
+      "epoch": 108.01136363636364,
+      "grad_norm": 0.7723172307014465,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 38020
+    },
+    {
+      "epoch": 108.03977272727273,
+      "grad_norm": 1.2289658784866333,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 38030
+    },
+    {
+      "epoch": 108.06818181818181,
+      "grad_norm": 0.977079451084137,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 38040
+    },
+    {
+      "epoch": 108.0965909090909,
+      "grad_norm": 0.7695228457450867,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 38050
+    },
+    {
+      "epoch": 108.125,
+      "grad_norm": 0.8520819544792175,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 38060
+    },
+    {
+      "epoch": 108.1534090909091,
+      "grad_norm": 0.6254966855049133,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 38070
+    },
+    {
+      "epoch": 108.18181818181819,
+      "grad_norm": 0.7900024056434631,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 38080
+    },
+    {
+      "epoch": 108.21022727272727,
+      "grad_norm": 0.7880393266677856,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 38090
+    },
+    {
+      "epoch": 108.23863636363636,
+      "grad_norm": 0.7509173154830933,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 38100
+    },
+    {
+      "epoch": 108.26704545454545,
+      "grad_norm": 0.5946815013885498,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 38110
+    },
+    {
+      "epoch": 108.29545454545455,
+      "grad_norm": 0.703536331653595,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 38120
+    },
+    {
+      "epoch": 108.32386363636364,
+      "grad_norm": 0.6964169144630432,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 38130
+    },
+    {
+      "epoch": 108.35227272727273,
+      "grad_norm": 0.778841495513916,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 38140
+    },
+    {
+      "epoch": 108.38068181818181,
+      "grad_norm": 0.8367446064949036,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 38150
+    },
+    {
+      "epoch": 108.4090909090909,
+      "grad_norm": 0.4970041811466217,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 38160
+    },
+    {
+      "epoch": 108.4375,
+      "grad_norm": 0.584697961807251,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 38170
+    },
+    {
+      "epoch": 108.4659090909091,
+      "grad_norm": 0.5434656143188477,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 38180
+    },
+    {
+      "epoch": 108.49431818181819,
+      "grad_norm": 0.8836389183998108,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 38190
+    },
+    {
+      "epoch": 108.52272727272727,
+      "grad_norm": 0.7299078702926636,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 38200
+    },
+    {
+      "epoch": 108.55113636363636,
+      "grad_norm": 0.6886661052703857,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 38210
+    },
+    {
+      "epoch": 108.57954545454545,
+      "grad_norm": 0.6602329611778259,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 38220
+    },
+    {
+      "epoch": 108.60795454545455,
+      "grad_norm": 0.5685455799102783,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 38230
+    },
+    {
+      "epoch": 108.63636363636364,
+      "grad_norm": 0.5193608403205872,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 38240
+    },
+    {
+      "epoch": 108.66477272727273,
+      "grad_norm": 0.4227427840232849,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 38250
+    },
+    {
+      "epoch": 108.69318181818181,
+      "grad_norm": 0.5614402890205383,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 38260
+    },
+    {
+      "epoch": 108.7215909090909,
+      "grad_norm": 0.4981783330440521,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 38270
+    },
+    {
+      "epoch": 108.75,
+      "grad_norm": 0.5758175849914551,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 38280
+    },
+    {
+      "epoch": 108.7784090909091,
+      "grad_norm": 0.5388387441635132,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 38290
+    },
+    {
+      "epoch": 108.80681818181819,
+      "grad_norm": 0.6330029964447021,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 38300
+    },
+    {
+      "epoch": 108.83522727272727,
+      "grad_norm": 0.6576165556907654,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 38310
+    },
+    {
+      "epoch": 108.86363636363636,
+      "grad_norm": 0.6450666189193726,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 38320
+    },
+    {
+      "epoch": 108.89204545454545,
+      "grad_norm": 0.5270466208457947,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 38330
+    },
+    {
+      "epoch": 108.92045454545455,
+      "grad_norm": 0.6218364834785461,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 38340
+    },
+    {
+      "epoch": 108.94886363636364,
+      "grad_norm": 0.7764090895652771,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 38350
+    },
+    {
+      "epoch": 108.97727272727273,
+      "grad_norm": 0.7030161619186401,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 38360
+    },
+    {
+      "epoch": 109.00568181818181,
+      "grad_norm": 0.9898335933685303,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 38370
+    },
+    {
+      "epoch": 109.0340909090909,
+      "grad_norm": 1.0687246322631836,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 38380
+    },
+    {
+      "epoch": 109.0625,
+      "grad_norm": 0.8941261768341064,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 38390
+    },
+    {
+      "epoch": 109.0909090909091,
+      "grad_norm": 0.9911275506019592,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 38400
+    },
+    {
+      "epoch": 109.11931818181819,
+      "grad_norm": 0.7970879077911377,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 38410
+    },
+    {
+      "epoch": 109.14772727272727,
+      "grad_norm": 0.5866600871086121,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 38420
+    },
+    {
+      "epoch": 109.17613636363636,
+      "grad_norm": 0.7543792128562927,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 38430
+    },
+    {
+      "epoch": 109.20454545454545,
+      "grad_norm": 0.9520198702812195,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 38440
+    },
+    {
+      "epoch": 109.23295454545455,
+      "grad_norm": 0.898628830909729,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 38450
+    },
+    {
+      "epoch": 109.26136363636364,
+      "grad_norm": 0.7924801707267761,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 38460
+    },
+    {
+      "epoch": 109.28977272727273,
+      "grad_norm": 0.6956962943077087,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 38470
+    },
+    {
+      "epoch": 109.31818181818181,
+      "grad_norm": 0.6093046069145203,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 38480
+    },
+    {
+      "epoch": 109.3465909090909,
+      "grad_norm": 0.8115942478179932,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 38490
+    },
+    {
+      "epoch": 109.375,
+      "grad_norm": 0.6574427485466003,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 38500
+    },
+    {
+      "epoch": 109.4034090909091,
+      "grad_norm": 0.48985546827316284,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 38510
+    },
+    {
+      "epoch": 109.43181818181819,
+      "grad_norm": 0.5606568455696106,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 38520
+    },
+    {
+      "epoch": 109.46022727272727,
+      "grad_norm": 0.5550665855407715,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 38530
+    },
+    {
+      "epoch": 109.48863636363636,
+      "grad_norm": 0.6146281361579895,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 38540
+    },
+    {
+      "epoch": 109.51704545454545,
+      "grad_norm": 0.5746130347251892,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 38550
+    },
+    {
+      "epoch": 109.54545454545455,
+      "grad_norm": 0.5612075328826904,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 38560
+    },
+    {
+      "epoch": 109.57386363636364,
+      "grad_norm": 0.623532235622406,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 38570
+    },
+    {
+      "epoch": 109.60227272727273,
+      "grad_norm": 0.5237706899642944,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 38580
+    },
+    {
+      "epoch": 109.63068181818181,
+      "grad_norm": 0.7601568102836609,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 38590
+    },
+    {
+      "epoch": 109.6590909090909,
+      "grad_norm": 0.7682206630706787,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 38600
+    },
+    {
+      "epoch": 109.6875,
+      "grad_norm": 0.6917203664779663,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 38610
+    },
+    {
+      "epoch": 109.7159090909091,
+      "grad_norm": 0.7227908372879028,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 38620
+    },
+    {
+      "epoch": 109.74431818181819,
+      "grad_norm": 0.6287977695465088,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 38630
+    },
+    {
+      "epoch": 109.77272727272727,
+      "grad_norm": 0.41634276509284973,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 38640
+    },
+    {
+      "epoch": 109.80113636363636,
+      "grad_norm": 0.5135223865509033,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 38650
+    },
+    {
+      "epoch": 109.82954545454545,
+      "grad_norm": 0.57159423828125,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 38660
+    },
+    {
+      "epoch": 109.85795454545455,
+      "grad_norm": 0.7393842935562134,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 38670
+    },
+    {
+      "epoch": 109.88636363636364,
+      "grad_norm": 0.8454589247703552,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 38680
+    },
+    {
+      "epoch": 109.91477272727273,
+      "grad_norm": 0.6690040230751038,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 38690
+    },
+    {
+      "epoch": 109.94318181818181,
+      "grad_norm": 1.1583824157714844,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 38700
+    },
+    {
+      "epoch": 109.9715909090909,
+      "grad_norm": 0.675502598285675,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 38710
+    },
+    {
+      "epoch": 110.0,
+      "grad_norm": 0.711791455745697,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 38720
+    },
+    {
+      "epoch": 110.0284090909091,
+      "grad_norm": 1.1118009090423584,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 38730
+    },
+    {
+      "epoch": 110.05681818181819,
+      "grad_norm": 1.0338175296783447,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 38740
+    },
+    {
+      "epoch": 110.08522727272727,
+      "grad_norm": 1.348634123802185,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 38750
+    },
+    {
+      "epoch": 110.11363636363636,
+      "grad_norm": 1.231924295425415,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 38760
+    },
+    {
+      "epoch": 110.14204545454545,
+      "grad_norm": 1.2100741863250732,
+      "learning_rate": 0.0001,
+      "loss": 0.039,
+      "step": 38770
+    },
+    {
+      "epoch": 110.17045454545455,
+      "grad_norm": 0.7521945834159851,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 38780
+    },
+    {
+      "epoch": 110.19886363636364,
+      "grad_norm": 1.066770315170288,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 38790
+    },
+    {
+      "epoch": 110.22727272727273,
+      "grad_norm": 0.9602776765823364,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 38800
+    },
+    {
+      "epoch": 110.25568181818181,
+      "grad_norm": 1.193345069885254,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 38810
+    },
+    {
+      "epoch": 110.2840909090909,
+      "grad_norm": 0.9646350741386414,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 38820
+    },
+    {
+      "epoch": 110.3125,
+      "grad_norm": 1.0762289762496948,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 38830
+    },
+    {
+      "epoch": 110.3409090909091,
+      "grad_norm": 1.0231586694717407,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 38840
+    },
+    {
+      "epoch": 110.36931818181819,
+      "grad_norm": 0.8765637278556824,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 38850
+    },
+    {
+      "epoch": 110.39772727272727,
+      "grad_norm": 0.7653793692588806,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 38860
+    },
+    {
+      "epoch": 110.42613636363636,
+      "grad_norm": 0.7483477592468262,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 38870
+    },
+    {
+      "epoch": 110.45454545454545,
+      "grad_norm": 0.7942159175872803,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 38880
+    },
+    {
+      "epoch": 110.48295454545455,
+      "grad_norm": 0.7775100469589233,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 38890
+    },
+    {
+      "epoch": 110.51136363636364,
+      "grad_norm": 0.6128475666046143,
+      "learning_rate": 0.0001,
+      "loss": 0.0381,
+      "step": 38900
+    },
+    {
+      "epoch": 110.53977272727273,
+      "grad_norm": 0.6405518054962158,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 38910
+    },
+    {
+      "epoch": 110.56818181818181,
+      "grad_norm": 0.6473897695541382,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 38920
+    },
+    {
+      "epoch": 110.5965909090909,
+      "grad_norm": 0.8890331387519836,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 38930
+    },
+    {
+      "epoch": 110.625,
+      "grad_norm": 0.9068458676338196,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 38940
+    },
+    {
+      "epoch": 110.6534090909091,
+      "grad_norm": 0.8818080425262451,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 38950
+    },
+    {
+      "epoch": 110.68181818181819,
+      "grad_norm": 1.1605606079101562,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 38960
+    },
+    {
+      "epoch": 110.71022727272727,
+      "grad_norm": 0.8575087785720825,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 38970
+    },
+    {
+      "epoch": 110.73863636363636,
+      "grad_norm": 0.68806391954422,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 38980
+    },
+    {
+      "epoch": 110.76704545454545,
+      "grad_norm": 0.7244641184806824,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 38990
+    },
+    {
+      "epoch": 110.79545454545455,
+      "grad_norm": 0.7463381886482239,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 39000
+    },
+    {
+      "epoch": 110.82386363636364,
+      "grad_norm": 0.6760275959968567,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 39010
+    },
+    {
+      "epoch": 110.85227272727273,
+      "grad_norm": 0.6048280000686646,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 39020
+    },
+    {
+      "epoch": 110.88068181818181,
+      "grad_norm": 0.6237334609031677,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 39030
+    },
+    {
+      "epoch": 110.9090909090909,
+      "grad_norm": 0.5601615309715271,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 39040
+    },
+    {
+      "epoch": 110.9375,
+      "grad_norm": 0.657729983329773,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 39050
+    },
+    {
+      "epoch": 110.9659090909091,
+      "grad_norm": 0.7096065282821655,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 39060
+    },
+    {
+      "epoch": 110.99431818181819,
+      "grad_norm": 0.6007034182548523,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 39070
+    },
+    {
+      "epoch": 111.02272727272727,
+      "grad_norm": 0.675838053226471,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 39080
+    },
+    {
+      "epoch": 111.05113636363636,
+      "grad_norm": 0.6892589330673218,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 39090
+    },
+    {
+      "epoch": 111.07954545454545,
+      "grad_norm": 0.8822354674339294,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 39100
+    },
+    {
+      "epoch": 111.10795454545455,
+      "grad_norm": 0.7115966081619263,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 39110
+    },
+    {
+      "epoch": 111.13636363636364,
+      "grad_norm": 0.7949591279029846,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 39120
+    },
+    {
+      "epoch": 111.16477272727273,
+      "grad_norm": 0.8462267518043518,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 39130
+    },
+    {
+      "epoch": 111.19318181818181,
+      "grad_norm": 0.7834585905075073,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 39140
+    },
+    {
+      "epoch": 111.2215909090909,
+      "grad_norm": 0.7656166553497314,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 39150
+    },
+    {
+      "epoch": 111.25,
+      "grad_norm": 0.599009096622467,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 39160
+    },
+    {
+      "epoch": 111.2784090909091,
+      "grad_norm": 0.8071345090866089,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 39170
+    },
+    {
+      "epoch": 111.30681818181819,
+      "grad_norm": 0.6010581254959106,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 39180
+    },
+    {
+      "epoch": 111.33522727272727,
+      "grad_norm": 0.6166114807128906,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 39190
+    },
+    {
+      "epoch": 111.36363636363636,
+      "grad_norm": 0.5775583982467651,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 39200
+    },
+    {
+      "epoch": 111.39204545454545,
+      "grad_norm": 0.5263673067092896,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 39210
+    },
+    {
+      "epoch": 111.42045454545455,
+      "grad_norm": 1.2878767251968384,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 39220
+    },
+    {
+      "epoch": 111.44886363636364,
+      "grad_norm": 1.5461721420288086,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 39230
+    },
+    {
+      "epoch": 111.47727272727273,
+      "grad_norm": 1.5255823135375977,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 39240
+    },
+    {
+      "epoch": 111.50568181818181,
+      "grad_norm": 1.0932375192642212,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 39250
+    },
+    {
+      "epoch": 111.5340909090909,
+      "grad_norm": 0.901646614074707,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 39260
+    },
+    {
+      "epoch": 111.5625,
+      "grad_norm": 0.6453569531440735,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 39270
+    },
+    {
+      "epoch": 111.5909090909091,
+      "grad_norm": 0.5468852519989014,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 39280
+    },
+    {
+      "epoch": 111.61931818181819,
+      "grad_norm": 0.7907230257987976,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 39290
+    },
+    {
+      "epoch": 111.64772727272727,
+      "grad_norm": 0.8564003109931946,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 39300
+    },
+    {
+      "epoch": 111.67613636363636,
+      "grad_norm": 0.9764010310173035,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 39310
+    },
+    {
+      "epoch": 111.70454545454545,
+      "grad_norm": 0.8942014575004578,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 39320
+    },
+    {
+      "epoch": 111.73295454545455,
+      "grad_norm": 0.5502811074256897,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 39330
+    },
+    {
+      "epoch": 111.76136363636364,
+      "grad_norm": 0.8766093850135803,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 39340
+    },
+    {
+      "epoch": 111.78977272727273,
+      "grad_norm": 0.98386150598526,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 39350
+    },
+    {
+      "epoch": 111.81818181818181,
+      "grad_norm": 0.9076448678970337,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 39360
+    },
+    {
+      "epoch": 111.8465909090909,
+      "grad_norm": 0.9458385109901428,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 39370
+    },
+    {
+      "epoch": 111.875,
+      "grad_norm": 0.7242578864097595,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 39380
+    },
+    {
+      "epoch": 111.9034090909091,
+      "grad_norm": 0.9158487915992737,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 39390
+    },
+    {
+      "epoch": 111.93181818181819,
+      "grad_norm": 0.7818379402160645,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 39400
+    },
+    {
+      "epoch": 111.96022727272727,
+      "grad_norm": 0.8776116967201233,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 39410
+    },
+    {
+      "epoch": 111.98863636363636,
+      "grad_norm": 0.8241115808486938,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 39420
+    },
+    {
+      "epoch": 112.01704545454545,
+      "grad_norm": 0.5597853660583496,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 39430
+    },
+    {
+      "epoch": 112.04545454545455,
+      "grad_norm": 0.6248016953468323,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 39440
+    },
+    {
+      "epoch": 112.07386363636364,
+      "grad_norm": 0.5273939967155457,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 39450
+    },
+    {
+      "epoch": 112.10227272727273,
+      "grad_norm": 0.4962223172187805,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 39460
+    },
+    {
+      "epoch": 112.13068181818181,
+      "grad_norm": 0.6138328313827515,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 39470
+    },
+    {
+      "epoch": 112.1590909090909,
+      "grad_norm": 0.4526226222515106,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 39480
+    },
+    {
+      "epoch": 112.1875,
+      "grad_norm": 0.5858993530273438,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 39490
+    },
+    {
+      "epoch": 112.2159090909091,
+      "grad_norm": 0.6115566492080688,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 39500
+    },
+    {
+      "epoch": 112.24431818181819,
+      "grad_norm": 0.5232250094413757,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 39510
+    },
+    {
+      "epoch": 112.27272727272727,
+      "grad_norm": 0.5535911321640015,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 39520
+    },
+    {
+      "epoch": 112.30113636363636,
+      "grad_norm": 0.7334476113319397,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 39530
+    },
+    {
+      "epoch": 112.32954545454545,
+      "grad_norm": 0.569206178188324,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 39540
+    },
+    {
+      "epoch": 112.35795454545455,
+      "grad_norm": 0.9461883902549744,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 39550
+    },
+    {
+      "epoch": 112.38636363636364,
+      "grad_norm": 0.8097552061080933,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 39560
+    },
+    {
+      "epoch": 112.41477272727273,
+      "grad_norm": 0.6988733410835266,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 39570
+    },
+    {
+      "epoch": 112.44318181818181,
+      "grad_norm": 0.7010588645935059,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 39580
+    },
+    {
+      "epoch": 112.4715909090909,
+      "grad_norm": 0.6968778967857361,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 39590
+    },
+    {
+      "epoch": 112.5,
+      "grad_norm": 0.7270570397377014,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 39600
+    },
+    {
+      "epoch": 112.5284090909091,
+      "grad_norm": 0.5744696855545044,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 39610
+    },
+    {
+      "epoch": 112.55681818181819,
+      "grad_norm": 0.6304721832275391,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 39620
+    },
+    {
+      "epoch": 112.58522727272727,
+      "grad_norm": 0.6305207014083862,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 39630
+    },
+    {
+      "epoch": 112.61363636363636,
+      "grad_norm": 0.7027773261070251,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 39640
+    },
+    {
+      "epoch": 112.64204545454545,
+      "grad_norm": 0.6099282503128052,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 39650
+    },
+    {
+      "epoch": 112.67045454545455,
+      "grad_norm": 1.234322428703308,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 39660
+    },
+    {
+      "epoch": 112.69886363636364,
+      "grad_norm": 1.157665729522705,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 39670
+    },
+    {
+      "epoch": 112.72727272727273,
+      "grad_norm": 1.371219277381897,
+      "learning_rate": 0.0001,
+      "loss": 0.0377,
+      "step": 39680
+    },
+    {
+      "epoch": 112.75568181818181,
+      "grad_norm": 1.147692322731018,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 39690
+    },
+    {
+      "epoch": 112.7840909090909,
+      "grad_norm": 0.7648851871490479,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 39700
+    },
+    {
+      "epoch": 112.8125,
+      "grad_norm": 0.6350628733634949,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 39710
+    },
+    {
+      "epoch": 112.8409090909091,
+      "grad_norm": 0.6837986707687378,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 39720
+    },
+    {
+      "epoch": 112.86931818181819,
+      "grad_norm": 0.6625135540962219,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 39730
+    },
+    {
+      "epoch": 112.89772727272727,
+      "grad_norm": 0.8770467638969421,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 39740
+    },
+    {
+      "epoch": 112.92613636363636,
+      "grad_norm": 0.8206093311309814,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 39750
+    },
+    {
+      "epoch": 112.95454545454545,
+      "grad_norm": 0.8660224080085754,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 39760
+    },
+    {
+      "epoch": 112.98295454545455,
+      "grad_norm": 0.6569976806640625,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 39770
+    },
+    {
+      "epoch": 113.01136363636364,
+      "grad_norm": 0.9233992099761963,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 39780
+    },
+    {
+      "epoch": 113.03977272727273,
+      "grad_norm": 0.8801051378250122,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 39790
+    },
+    {
+      "epoch": 113.06818181818181,
+      "grad_norm": 0.773917019367218,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 39800
+    },
+    {
+      "epoch": 113.0965909090909,
+      "grad_norm": 0.7063732147216797,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 39810
+    },
+    {
+      "epoch": 113.125,
+      "grad_norm": 0.7631141543388367,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 39820
+    },
+    {
+      "epoch": 113.1534090909091,
+      "grad_norm": 0.7640942931175232,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 39830
+    },
+    {
+      "epoch": 113.18181818181819,
+      "grad_norm": 0.6806530356407166,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 39840
+    },
+    {
+      "epoch": 113.21022727272727,
+      "grad_norm": 0.627130389213562,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 39850
+    },
+    {
+      "epoch": 113.23863636363636,
+      "grad_norm": 0.4664798378944397,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 39860
+    },
+    {
+      "epoch": 113.26704545454545,
+      "grad_norm": 0.6047282814979553,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 39870
+    },
+    {
+      "epoch": 113.29545454545455,
+      "grad_norm": 0.8172615170478821,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 39880
+    },
+    {
+      "epoch": 113.32386363636364,
+      "grad_norm": 0.7047287821769714,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 39890
+    },
+    {
+      "epoch": 113.35227272727273,
+      "grad_norm": 0.6957646012306213,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 39900
+    },
+    {
+      "epoch": 113.38068181818181,
+      "grad_norm": 0.7700342535972595,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 39910
+    },
+    {
+      "epoch": 113.4090909090909,
+      "grad_norm": 0.6716678738594055,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 39920
+    },
+    {
+      "epoch": 113.4375,
+      "grad_norm": 0.6276105046272278,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 39930
+    },
+    {
+      "epoch": 113.4659090909091,
+      "grad_norm": 0.7621968388557434,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 39940
+    },
+    {
+      "epoch": 113.49431818181819,
+      "grad_norm": 0.5619763135910034,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 39950
+    },
+    {
+      "epoch": 113.52272727272727,
+      "grad_norm": 0.5187546014785767,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 39960
+    },
+    {
+      "epoch": 113.55113636363636,
+      "grad_norm": 0.7270740866661072,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 39970
+    },
+    {
+      "epoch": 113.57954545454545,
+      "grad_norm": 0.49849000573158264,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 39980
+    },
+    {
+      "epoch": 113.60795454545455,
+      "grad_norm": 1.0504893064498901,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 39990
+    },
+    {
+      "epoch": 113.63636363636364,
+      "grad_norm": 0.6417130827903748,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 40000
+    },
+    {
+      "epoch": 113.66477272727273,
+      "grad_norm": 0.6625685691833496,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 40010
+    },
+    {
+      "epoch": 113.69318181818181,
+      "grad_norm": 0.6271093487739563,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 40020
+    },
+    {
+      "epoch": 113.7215909090909,
+      "grad_norm": 0.6028963923454285,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 40030
+    },
+    {
+      "epoch": 113.75,
+      "grad_norm": 0.5805331468582153,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 40040
+    },
+    {
+      "epoch": 113.7784090909091,
+      "grad_norm": 0.594214677810669,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 40050
+    },
+    {
+      "epoch": 113.80681818181819,
+      "grad_norm": 0.6908999681472778,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 40060
+    },
+    {
+      "epoch": 113.83522727272727,
+      "grad_norm": 0.53608238697052,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 40070
+    },
+    {
+      "epoch": 113.86363636363636,
+      "grad_norm": 0.4885925352573395,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 40080
+    },
+    {
+      "epoch": 113.89204545454545,
+      "grad_norm": 0.5205367803573608,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 40090
+    },
+    {
+      "epoch": 113.92045454545455,
+      "grad_norm": 0.5592257976531982,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 40100
+    },
+    {
+      "epoch": 113.94886363636364,
+      "grad_norm": 0.5714760422706604,
+      "learning_rate": 0.0001,
+      "loss": 0.0373,
+      "step": 40110
+    },
+    {
+      "epoch": 113.97727272727273,
+      "grad_norm": 0.6751970648765564,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 40120
+    },
+    {
+      "epoch": 114.00568181818181,
+      "grad_norm": 0.6371946334838867,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 40130
+    },
+    {
+      "epoch": 114.0340909090909,
+      "grad_norm": 0.6978771686553955,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 40140
+    },
+    {
+      "epoch": 114.0625,
+      "grad_norm": 0.598162829875946,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 40150
+    },
+    {
+      "epoch": 114.0909090909091,
+      "grad_norm": 0.6158064007759094,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 40160
+    },
+    {
+      "epoch": 114.11931818181819,
+      "grad_norm": 0.5208731889724731,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 40170
+    },
+    {
+      "epoch": 114.14772727272727,
+      "grad_norm": 0.5302230715751648,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 40180
+    },
+    {
+      "epoch": 114.17613636363636,
+      "grad_norm": 0.5512319803237915,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 40190
+    },
+    {
+      "epoch": 114.20454545454545,
+      "grad_norm": 0.43418312072753906,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 40200
+    },
+    {
+      "epoch": 114.23295454545455,
+      "grad_norm": 0.5973665118217468,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 40210
+    },
+    {
+      "epoch": 114.26136363636364,
+      "grad_norm": 0.5855032801628113,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 40220
+    },
+    {
+      "epoch": 114.28977272727273,
+      "grad_norm": 0.5427087545394897,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 40230
+    },
+    {
+      "epoch": 114.31818181818181,
+      "grad_norm": 0.7128482460975647,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 40240
+    },
+    {
+      "epoch": 114.3465909090909,
+      "grad_norm": 0.4120272696018219,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 40250
+    },
+    {
+      "epoch": 114.375,
+      "grad_norm": 0.4165058135986328,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 40260
+    },
+    {
+      "epoch": 114.4034090909091,
+      "grad_norm": 0.4229552447795868,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 40270
+    },
+    {
+      "epoch": 114.43181818181819,
+      "grad_norm": 0.526804506778717,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 40280
+    },
+    {
+      "epoch": 114.46022727272727,
+      "grad_norm": 0.5471256971359253,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 40290
+    },
+    {
+      "epoch": 114.48863636363636,
+      "grad_norm": 0.7126163840293884,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 40300
+    },
+    {
+      "epoch": 114.51704545454545,
+      "grad_norm": 0.5300715565681458,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 40310
+    },
+    {
+      "epoch": 114.54545454545455,
+      "grad_norm": 0.5716768503189087,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 40320
+    },
+    {
+      "epoch": 114.57386363636364,
+      "grad_norm": 0.5660290122032166,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 40330
+    },
+    {
+      "epoch": 114.60227272727273,
+      "grad_norm": 0.4344553053379059,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 40340
+    },
+    {
+      "epoch": 114.63068181818181,
+      "grad_norm": 0.5412406921386719,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 40350
+    },
+    {
+      "epoch": 114.6590909090909,
+      "grad_norm": 0.6711289286613464,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 40360
+    },
+    {
+      "epoch": 114.6875,
+      "grad_norm": 0.783184289932251,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 40370
+    },
+    {
+      "epoch": 114.7159090909091,
+      "grad_norm": 0.819743812084198,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 40380
+    },
+    {
+      "epoch": 114.74431818181819,
+      "grad_norm": 0.7587737441062927,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 40390
+    },
+    {
+      "epoch": 114.77272727272727,
+      "grad_norm": 0.6675707101821899,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 40400
+    },
+    {
+      "epoch": 114.80113636363636,
+      "grad_norm": 0.5723679065704346,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 40410
+    },
+    {
+      "epoch": 114.82954545454545,
+      "grad_norm": 0.6985281109809875,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 40420
+    },
+    {
+      "epoch": 114.85795454545455,
+      "grad_norm": 1.0145719051361084,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 40430
+    },
+    {
+      "epoch": 114.88636363636364,
+      "grad_norm": 1.0415736436843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 40440
+    },
+    {
+      "epoch": 114.91477272727273,
+      "grad_norm": 0.74834805727005,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 40450
+    },
+    {
+      "epoch": 114.94318181818181,
+      "grad_norm": 1.022322654724121,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 40460
+    },
+    {
+      "epoch": 114.9715909090909,
+      "grad_norm": 0.8849312663078308,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 40470
+    },
+    {
+      "epoch": 115.0,
+      "grad_norm": 0.602628231048584,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 40480
+    },
+    {
+      "epoch": 115.0284090909091,
+      "grad_norm": 0.8007441163063049,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 40490
+    },
+    {
+      "epoch": 115.05681818181819,
+      "grad_norm": 0.8990688323974609,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 40500
+    },
+    {
+      "epoch": 115.08522727272727,
+      "grad_norm": 0.8828384876251221,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 40510
+    },
+    {
+      "epoch": 115.11363636363636,
+      "grad_norm": 1.0378295183181763,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 40520
+    },
+    {
+      "epoch": 115.14204545454545,
+      "grad_norm": 1.0264496803283691,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 40530
+    },
+    {
+      "epoch": 115.17045454545455,
+      "grad_norm": 1.1999590396881104,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 40540
+    },
+    {
+      "epoch": 115.19886363636364,
+      "grad_norm": 0.8743994832038879,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 40550
+    },
+    {
+      "epoch": 115.22727272727273,
+      "grad_norm": 0.5800889730453491,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 40560
+    },
+    {
+      "epoch": 115.25568181818181,
+      "grad_norm": 0.8396671414375305,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 40570
+    },
+    {
+      "epoch": 115.2840909090909,
+      "grad_norm": 0.8173549175262451,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 40580
+    },
+    {
+      "epoch": 115.3125,
+      "grad_norm": 0.7395663857460022,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 40590
+    },
+    {
+      "epoch": 115.3409090909091,
+      "grad_norm": 0.6188920140266418,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 40600
+    },
+    {
+      "epoch": 115.36931818181819,
+      "grad_norm": 0.5487187504768372,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 40610
+    },
+    {
+      "epoch": 115.39772727272727,
+      "grad_norm": 0.5410171151161194,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 40620
+    },
+    {
+      "epoch": 115.42613636363636,
+      "grad_norm": 0.5235713124275208,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 40630
+    },
+    {
+      "epoch": 115.45454545454545,
+      "grad_norm": 0.7061126232147217,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 40640
+    },
+    {
+      "epoch": 115.48295454545455,
+      "grad_norm": 0.7242769598960876,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 40650
+    },
+    {
+      "epoch": 115.51136363636364,
+      "grad_norm": 0.6157847046852112,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 40660
+    },
+    {
+      "epoch": 115.53977272727273,
+      "grad_norm": 0.6777865886688232,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 40670
+    },
+    {
+      "epoch": 115.56818181818181,
+      "grad_norm": 0.5488711595535278,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 40680
+    },
+    {
+      "epoch": 115.5965909090909,
+      "grad_norm": 0.7855844497680664,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 40690
+    },
+    {
+      "epoch": 115.625,
+      "grad_norm": 0.640532910823822,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 40700
+    },
+    {
+      "epoch": 115.6534090909091,
+      "grad_norm": 0.5868560671806335,
+      "learning_rate": 0.0001,
+      "loss": 0.038,
+      "step": 40710
+    },
+    {
+      "epoch": 115.68181818181819,
+      "grad_norm": 0.8778387308120728,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 40720
+    },
+    {
+      "epoch": 115.71022727272727,
+      "grad_norm": 0.6132923364639282,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 40730
+    },
+    {
+      "epoch": 115.73863636363636,
+      "grad_norm": 0.6299883127212524,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 40740
+    },
+    {
+      "epoch": 115.76704545454545,
+      "grad_norm": 0.6556488871574402,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 40750
+    },
+    {
+      "epoch": 115.79545454545455,
+      "grad_norm": 0.7025827169418335,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 40760
+    },
+    {
+      "epoch": 115.82386363636364,
+      "grad_norm": 0.49530184268951416,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 40770
+    },
+    {
+      "epoch": 115.85227272727273,
+      "grad_norm": 0.6061105728149414,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 40780
+    },
+    {
+      "epoch": 115.88068181818181,
+      "grad_norm": 0.5180947184562683,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 40790
+    },
+    {
+      "epoch": 115.9090909090909,
+      "grad_norm": 0.584722101688385,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 40800
+    },
+    {
+      "epoch": 115.9375,
+      "grad_norm": 0.5374404788017273,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 40810
+    },
+    {
+      "epoch": 115.9659090909091,
+      "grad_norm": 0.5954856276512146,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 40820
+    },
+    {
+      "epoch": 115.99431818181819,
+      "grad_norm": 0.5415761470794678,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 40830
+    },
+    {
+      "epoch": 116.02272727272727,
+      "grad_norm": 0.6761354804039001,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 40840
+    },
+    {
+      "epoch": 116.05113636363636,
+      "grad_norm": 0.5434542894363403,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 40850
+    },
+    {
+      "epoch": 116.07954545454545,
+      "grad_norm": 0.67738276720047,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 40860
+    },
+    {
+      "epoch": 116.10795454545455,
+      "grad_norm": 0.5063143968582153,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 40870
+    },
+    {
+      "epoch": 116.13636363636364,
+      "grad_norm": 0.642759382724762,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 40880
+    },
+    {
+      "epoch": 116.16477272727273,
+      "grad_norm": 0.6486865878105164,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 40890
+    },
+    {
+      "epoch": 116.19318181818181,
+      "grad_norm": 0.6188929080963135,
+      "learning_rate": 0.0001,
+      "loss": 0.0401,
+      "step": 40900
+    },
+    {
+      "epoch": 116.2215909090909,
+      "grad_norm": 0.5677132606506348,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 40910
+    },
+    {
+      "epoch": 116.25,
+      "grad_norm": 0.583200991153717,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 40920
+    },
+    {
+      "epoch": 116.2784090909091,
+      "grad_norm": 0.6927564740180969,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 40930
+    },
+    {
+      "epoch": 116.30681818181819,
+      "grad_norm": 0.5694494247436523,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 40940
+    },
+    {
+      "epoch": 116.33522727272727,
+      "grad_norm": 0.6229391098022461,
+      "learning_rate": 0.0001,
+      "loss": 0.0379,
+      "step": 40950
+    },
+    {
+      "epoch": 116.36363636363636,
+      "grad_norm": 0.6101709008216858,
+      "learning_rate": 0.0001,
+      "loss": 0.0375,
+      "step": 40960
+    },
+    {
+      "epoch": 116.39204545454545,
+      "grad_norm": 0.5544459819793701,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 40970
+    },
+    {
+      "epoch": 116.42045454545455,
+      "grad_norm": 0.6003863215446472,
+      "learning_rate": 0.0001,
+      "loss": 0.0368,
+      "step": 40980
+    },
+    {
+      "epoch": 116.44886363636364,
+      "grad_norm": 0.735579788684845,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 40990
+    },
+    {
+      "epoch": 116.47727272727273,
+      "grad_norm": 0.5912768244743347,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 41000
+    },
+    {
+      "epoch": 116.50568181818181,
+      "grad_norm": 0.5649675726890564,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 41010
+    },
+    {
+      "epoch": 116.5340909090909,
+      "grad_norm": 0.9016214609146118,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41020
+    },
+    {
+      "epoch": 116.5625,
+      "grad_norm": 0.6981037259101868,
+      "learning_rate": 0.0001,
+      "loss": 0.0372,
+      "step": 41030
+    },
+    {
+      "epoch": 116.5909090909091,
+      "grad_norm": 0.7360560297966003,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 41040
+    },
+    {
+      "epoch": 116.61931818181819,
+      "grad_norm": 0.9419438242912292,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 41050
+    },
+    {
+      "epoch": 116.64772727272727,
+      "grad_norm": 0.7652466297149658,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41060
+    },
+    {
+      "epoch": 116.67613636363636,
+      "grad_norm": 0.8323368430137634,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 41070
+    },
+    {
+      "epoch": 116.70454545454545,
+      "grad_norm": 0.625653088092804,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 41080
+    },
+    {
+      "epoch": 116.73295454545455,
+      "grad_norm": 0.8039312362670898,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 41090
+    },
+    {
+      "epoch": 116.76136363636364,
+      "grad_norm": 0.7268533706665039,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 41100
+    },
+    {
+      "epoch": 116.78977272727273,
+      "grad_norm": 0.7572269439697266,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41110
+    },
+    {
+      "epoch": 116.81818181818181,
+      "grad_norm": 0.695713460445404,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 41120
+    },
+    {
+      "epoch": 116.8465909090909,
+      "grad_norm": 0.7137623429298401,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 41130
+    },
+    {
+      "epoch": 116.875,
+      "grad_norm": 0.6217263340950012,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 41140
+    },
+    {
+      "epoch": 116.9034090909091,
+      "grad_norm": 0.6812773942947388,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 41150
+    },
+    {
+      "epoch": 116.93181818181819,
+      "grad_norm": 0.6269494891166687,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41160
+    },
+    {
+      "epoch": 116.96022727272727,
+      "grad_norm": 0.5813413262367249,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41170
+    },
+    {
+      "epoch": 116.98863636363636,
+      "grad_norm": 0.47251439094543457,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 41180
+    },
+    {
+      "epoch": 117.01704545454545,
+      "grad_norm": 0.82036954164505,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 41190
+    },
+    {
+      "epoch": 117.04545454545455,
+      "grad_norm": 0.8919968605041504,
+      "learning_rate": 0.0001,
+      "loss": 0.0374,
+      "step": 41200
+    },
+    {
+      "epoch": 117.07386363636364,
+      "grad_norm": 0.5334138870239258,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 41210
+    },
+    {
+      "epoch": 117.10227272727273,
+      "grad_norm": 0.4660698473453522,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 41220
+    },
+    {
+      "epoch": 117.13068181818181,
+      "grad_norm": 0.6906890869140625,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 41230
+    },
+    {
+      "epoch": 117.1590909090909,
+      "grad_norm": 0.8132134675979614,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 41240
+    },
+    {
+      "epoch": 117.1875,
+      "grad_norm": 0.8870444297790527,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 41250
+    },
+    {
+      "epoch": 117.2159090909091,
+      "grad_norm": 0.7599701285362244,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 41260
+    },
+    {
+      "epoch": 117.24431818181819,
+      "grad_norm": 0.8724075555801392,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 41270
+    },
+    {
+      "epoch": 117.27272727272727,
+      "grad_norm": 0.6574482917785645,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 41280
+    },
+    {
+      "epoch": 117.30113636363636,
+      "grad_norm": 0.7404472231864929,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 41290
+    },
+    {
+      "epoch": 117.32954545454545,
+      "grad_norm": 0.8209742903709412,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 41300
+    },
+    {
+      "epoch": 117.35795454545455,
+      "grad_norm": 1.0132182836532593,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 41310
+    },
+    {
+      "epoch": 117.38636363636364,
+      "grad_norm": 1.2387603521347046,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 41320
+    },
+    {
+      "epoch": 117.41477272727273,
+      "grad_norm": 1.3428541421890259,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 41330
+    },
+    {
+      "epoch": 117.44318181818181,
+      "grad_norm": 1.3659002780914307,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 41340
+    },
+    {
+      "epoch": 117.4715909090909,
+      "grad_norm": 1.2876734733581543,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 41350
+    },
+    {
+      "epoch": 117.5,
+      "grad_norm": 1.0242540836334229,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 41360
+    },
+    {
+      "epoch": 117.5284090909091,
+      "grad_norm": 0.6913342475891113,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 41370
+    },
+    {
+      "epoch": 117.55681818181819,
+      "grad_norm": 0.7995432019233704,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 41380
+    },
+    {
+      "epoch": 117.58522727272727,
+      "grad_norm": 0.8866602182388306,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 41390
+    },
+    {
+      "epoch": 117.61363636363636,
+      "grad_norm": 0.6081330180168152,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 41400
+    },
+    {
+      "epoch": 117.64204545454545,
+      "grad_norm": 0.7164961099624634,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 41410
+    },
+    {
+      "epoch": 117.67045454545455,
+      "grad_norm": 0.6436015963554382,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 41420
+    },
+    {
+      "epoch": 117.69886363636364,
+      "grad_norm": 0.7437272667884827,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 41430
+    },
+    {
+      "epoch": 117.72727272727273,
+      "grad_norm": 0.723030686378479,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 41440
+    },
+    {
+      "epoch": 117.75568181818181,
+      "grad_norm": 0.7317346334457397,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 41450
+    },
+    {
+      "epoch": 117.7840909090909,
+      "grad_norm": 0.7966662049293518,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 41460
+    },
+    {
+      "epoch": 117.8125,
+      "grad_norm": 0.6614232659339905,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 41470
+    },
+    {
+      "epoch": 117.8409090909091,
+      "grad_norm": 0.7168434858322144,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 41480
+    },
+    {
+      "epoch": 117.86931818181819,
+      "grad_norm": 0.5962741374969482,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 41490
+    },
+    {
+      "epoch": 117.89772727272727,
+      "grad_norm": 0.6524445414543152,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 41500
+    },
+    {
+      "epoch": 117.92613636363636,
+      "grad_norm": 0.4581736624240875,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 41510
+    },
+    {
+      "epoch": 117.95454545454545,
+      "grad_norm": 0.6047722101211548,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 41520
+    },
+    {
+      "epoch": 117.98295454545455,
+      "grad_norm": 0.9901228547096252,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 41530
+    },
+    {
+      "epoch": 118.01136363636364,
+      "grad_norm": 0.8791798949241638,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 41540
+    },
+    {
+      "epoch": 118.03977272727273,
+      "grad_norm": 0.727450966835022,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41550
+    },
+    {
+      "epoch": 118.06818181818181,
+      "grad_norm": 0.638188898563385,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 41560
+    },
+    {
+      "epoch": 118.0965909090909,
+      "grad_norm": 0.7026476263999939,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 41570
+    },
+    {
+      "epoch": 118.125,
+      "grad_norm": 0.7015557885169983,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 41580
+    },
+    {
+      "epoch": 118.1534090909091,
+      "grad_norm": 0.7547580599784851,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 41590
+    },
+    {
+      "epoch": 118.18181818181819,
+      "grad_norm": 0.6024928092956543,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 41600
+    },
+    {
+      "epoch": 118.21022727272727,
+      "grad_norm": 0.7490041255950928,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 41610
+    },
+    {
+      "epoch": 118.23863636363636,
+      "grad_norm": 0.540759265422821,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 41620
+    },
+    {
+      "epoch": 118.26704545454545,
+      "grad_norm": 0.6325246691703796,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 41630
+    },
+    {
+      "epoch": 118.29545454545455,
+      "grad_norm": 0.8828673362731934,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 41640
+    },
+    {
+      "epoch": 118.32386363636364,
+      "grad_norm": 1.1287953853607178,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 41650
+    },
+    {
+      "epoch": 118.35227272727273,
+      "grad_norm": 1.034847617149353,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 41660
+    },
+    {
+      "epoch": 118.38068181818181,
+      "grad_norm": 0.6673224568367004,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 41670
+    },
+    {
+      "epoch": 118.4090909090909,
+      "grad_norm": 0.5625375509262085,
+      "learning_rate": 0.0001,
+      "loss": 0.0363,
+      "step": 41680
+    },
+    {
+      "epoch": 118.4375,
+      "grad_norm": 0.759629487991333,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 41690
+    },
+    {
+      "epoch": 118.4659090909091,
+      "grad_norm": 0.9632283449172974,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 41700
+    },
+    {
+      "epoch": 118.49431818181819,
+      "grad_norm": 0.9728202819824219,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 41710
+    },
+    {
+      "epoch": 118.52272727272727,
+      "grad_norm": 0.6623541712760925,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41720
+    },
+    {
+      "epoch": 118.55113636363636,
+      "grad_norm": 0.5848730206489563,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 41730
+    },
+    {
+      "epoch": 118.57954545454545,
+      "grad_norm": 0.5939080715179443,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41740
+    },
+    {
+      "epoch": 118.60795454545455,
+      "grad_norm": 0.4702250063419342,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 41750
+    },
+    {
+      "epoch": 118.63636363636364,
+      "grad_norm": 0.6767942905426025,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 41760
+    },
+    {
+      "epoch": 118.66477272727273,
+      "grad_norm": 0.7504387497901917,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 41770
+    },
+    {
+      "epoch": 118.69318181818181,
+      "grad_norm": 0.5079829096794128,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 41780
+    },
+    {
+      "epoch": 118.7215909090909,
+      "grad_norm": 0.5891208648681641,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 41790
+    },
+    {
+      "epoch": 118.75,
+      "grad_norm": 0.8426125049591064,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 41800
+    },
+    {
+      "epoch": 118.7784090909091,
+      "grad_norm": 0.6993535757064819,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 41810
+    },
+    {
+      "epoch": 118.80681818181819,
+      "grad_norm": 0.8501553535461426,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 41820
+    },
+    {
+      "epoch": 118.83522727272727,
+      "grad_norm": 0.7962374091148376,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 41830
+    },
+    {
+      "epoch": 118.86363636363636,
+      "grad_norm": 0.6130684018135071,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 41840
+    },
+    {
+      "epoch": 118.89204545454545,
+      "grad_norm": 0.6343328952789307,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 41850
+    },
+    {
+      "epoch": 118.92045454545455,
+      "grad_norm": 0.5049582123756409,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41860
+    },
+    {
+      "epoch": 118.94886363636364,
+      "grad_norm": 0.6091402769088745,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41870
+    },
+    {
+      "epoch": 118.97727272727273,
+      "grad_norm": 0.5600488781929016,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 41880
+    },
+    {
+      "epoch": 119.00568181818181,
+      "grad_norm": 0.6329433917999268,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 41890
+    },
+    {
+      "epoch": 119.0340909090909,
+      "grad_norm": 0.589053213596344,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 41900
+    },
+    {
+      "epoch": 119.0625,
+      "grad_norm": 0.456606924533844,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 41910
+    },
+    {
+      "epoch": 119.0909090909091,
+      "grad_norm": 0.5378241539001465,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 41920
+    },
+    {
+      "epoch": 119.11931818181819,
+      "grad_norm": 0.6176024675369263,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 41930
+    },
+    {
+      "epoch": 119.14772727272727,
+      "grad_norm": 0.50550776720047,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 41940
+    },
+    {
+      "epoch": 119.17613636363636,
+      "grad_norm": 0.5080638527870178,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41950
+    },
+    {
+      "epoch": 119.20454545454545,
+      "grad_norm": 0.5049692392349243,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 41960
+    },
+    {
+      "epoch": 119.23295454545455,
+      "grad_norm": 0.5414020419120789,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 41970
+    },
+    {
+      "epoch": 119.26136363636364,
+      "grad_norm": 0.592162549495697,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 41980
+    },
+    {
+      "epoch": 119.28977272727273,
+      "grad_norm": 1.7480474710464478,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 41990
+    },
+    {
+      "epoch": 119.31818181818181,
+      "grad_norm": 1.2184784412384033,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 42000
+    },
+    {
+      "epoch": 119.3465909090909,
+      "grad_norm": 1.2475852966308594,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 42010
+    },
+    {
+      "epoch": 119.375,
+      "grad_norm": 1.83725905418396,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 42020
+    },
+    {
+      "epoch": 119.4034090909091,
+      "grad_norm": 1.1545556783676147,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 42030
+    },
+    {
+      "epoch": 119.43181818181819,
+      "grad_norm": 1.2570842504501343,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 42040
+    },
+    {
+      "epoch": 119.46022727272727,
+      "grad_norm": 1.244055151939392,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 42050
+    },
+    {
+      "epoch": 119.48863636363636,
+      "grad_norm": 0.766913652420044,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 42060
+    },
+    {
+      "epoch": 119.51704545454545,
+      "grad_norm": 0.6588661074638367,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 42070
+    },
+    {
+      "epoch": 119.54545454545455,
+      "grad_norm": 0.5511944890022278,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 42080
+    },
+    {
+      "epoch": 119.57386363636364,
+      "grad_norm": 0.7311079502105713,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 42090
+    },
+    {
+      "epoch": 119.60227272727273,
+      "grad_norm": 0.6783902049064636,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 42100
+    },
+    {
+      "epoch": 119.63068181818181,
+      "grad_norm": 0.7213060855865479,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 42110
+    },
+    {
+      "epoch": 119.6590909090909,
+      "grad_norm": 0.6389514207839966,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 42120
+    },
+    {
+      "epoch": 119.6875,
+      "grad_norm": 0.6836166977882385,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 42130
+    },
+    {
+      "epoch": 119.7159090909091,
+      "grad_norm": 0.6873879432678223,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 42140
+    },
+    {
+      "epoch": 119.74431818181819,
+      "grad_norm": 0.6492112278938293,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 42150
+    },
+    {
+      "epoch": 119.77272727272727,
+      "grad_norm": 0.6663733720779419,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 42160
+    },
+    {
+      "epoch": 119.80113636363636,
+      "grad_norm": 1.1575038433074951,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 42170
+    },
+    {
+      "epoch": 119.82954545454545,
+      "grad_norm": 0.878379225730896,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 42180
+    },
+    {
+      "epoch": 119.85795454545455,
+      "grad_norm": 0.6646192073822021,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 42190
+    },
+    {
+      "epoch": 119.88636363636364,
+      "grad_norm": 0.7406749129295349,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 42200
+    },
+    {
+      "epoch": 119.91477272727273,
+      "grad_norm": 0.6425890326499939,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 42210
+    },
+    {
+      "epoch": 119.94318181818181,
+      "grad_norm": 0.8351615071296692,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 42220
+    },
+    {
+      "epoch": 119.9715909090909,
+      "grad_norm": 0.6505674719810486,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 42230
+    },
+    {
+      "epoch": 120.0,
+      "grad_norm": 0.6099982857704163,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 42240
+    },
+    {
+      "epoch": 120.0284090909091,
+      "grad_norm": 0.604333758354187,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 42250
+    },
+    {
+      "epoch": 120.05681818181819,
+      "grad_norm": 0.5962154269218445,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 42260
+    },
+    {
+      "epoch": 120.08522727272727,
+      "grad_norm": 0.5833165645599365,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 42270
+    },
+    {
+      "epoch": 120.11363636363636,
+      "grad_norm": 0.5957491397857666,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 42280
+    },
+    {
+      "epoch": 120.14204545454545,
+      "grad_norm": 0.6122380495071411,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 42290
+    },
+    {
+      "epoch": 120.17045454545455,
+      "grad_norm": 0.5932005643844604,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 42300
+    },
+    {
+      "epoch": 120.19886363636364,
+      "grad_norm": 0.5959727168083191,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 42310
+    },
+    {
+      "epoch": 120.22727272727273,
+      "grad_norm": 0.5512206554412842,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 42320
+    },
+    {
+      "epoch": 120.25568181818181,
+      "grad_norm": 0.535960853099823,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 42330
+    },
+    {
+      "epoch": 120.2840909090909,
+      "grad_norm": 0.5918856263160706,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 42340
+    },
+    {
+      "epoch": 120.3125,
+      "grad_norm": 0.5420950055122375,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 42350
+    },
+    {
+      "epoch": 120.3409090909091,
+      "grad_norm": 0.5215405225753784,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 42360
+    },
+    {
+      "epoch": 120.36931818181819,
+      "grad_norm": 0.48384472727775574,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 42370
+    },
+    {
+      "epoch": 120.39772727272727,
+      "grad_norm": 0.40733209252357483,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 42380
+    },
+    {
+      "epoch": 120.42613636363636,
+      "grad_norm": 0.5102291703224182,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 42390
+    },
+    {
+      "epoch": 120.45454545454545,
+      "grad_norm": 0.7866867184638977,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 42400
+    },
+    {
+      "epoch": 120.48295454545455,
+      "grad_norm": 1.2717626094818115,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 42410
+    },
+    {
+      "epoch": 120.51136363636364,
+      "grad_norm": 1.0195508003234863,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 42420
+    },
+    {
+      "epoch": 120.53977272727273,
+      "grad_norm": 0.6073787212371826,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 42430
+    },
+    {
+      "epoch": 120.56818181818181,
+      "grad_norm": 0.6645169854164124,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 42440
+    },
+    {
+      "epoch": 120.5965909090909,
+      "grad_norm": 0.9627057313919067,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 42450
+    },
+    {
+      "epoch": 120.625,
+      "grad_norm": 0.8808974623680115,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 42460
+    },
+    {
+      "epoch": 120.6534090909091,
+      "grad_norm": 0.8835111856460571,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 42470
+    },
+    {
+      "epoch": 120.68181818181819,
+      "grad_norm": 0.6539061665534973,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 42480
+    },
+    {
+      "epoch": 120.71022727272727,
+      "grad_norm": 0.7671836018562317,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 42490
+    },
+    {
+      "epoch": 120.73863636363636,
+      "grad_norm": 0.6868611574172974,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 42500
+    },
+    {
+      "epoch": 120.76704545454545,
+      "grad_norm": 0.613646388053894,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 42510
+    },
+    {
+      "epoch": 120.79545454545455,
+      "grad_norm": 0.6648271083831787,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 42520
+    },
+    {
+      "epoch": 120.82386363636364,
+      "grad_norm": 0.5934839248657227,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 42530
+    },
+    {
+      "epoch": 120.85227272727273,
+      "grad_norm": 0.6186235547065735,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 42540
+    },
+    {
+      "epoch": 120.88068181818181,
+      "grad_norm": 0.46539390087127686,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 42550
+    },
+    {
+      "epoch": 120.9090909090909,
+      "grad_norm": 0.5254285931587219,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 42560
+    },
+    {
+      "epoch": 120.9375,
+      "grad_norm": 0.6108625531196594,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 42570
+    },
+    {
+      "epoch": 120.9659090909091,
+      "grad_norm": 0.6213340759277344,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 42580
+    },
+    {
+      "epoch": 120.99431818181819,
+      "grad_norm": 0.5728833675384521,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 42590
+    },
+    {
+      "epoch": 121.02272727272727,
+      "grad_norm": 0.47747138142585754,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 42600
+    },
+    {
+      "epoch": 121.05113636363636,
+      "grad_norm": 0.5983991026878357,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 42610
+    },
+    {
+      "epoch": 121.07954545454545,
+      "grad_norm": 1.5157920122146606,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 42620
+    },
+    {
+      "epoch": 121.10795454545455,
+      "grad_norm": 1.188115119934082,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 42630
+    },
+    {
+      "epoch": 121.13636363636364,
+      "grad_norm": 0.8782476782798767,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 42640
+    },
+    {
+      "epoch": 121.16477272727273,
+      "grad_norm": 1.138963222503662,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 42650
+    },
+    {
+      "epoch": 121.19318181818181,
+      "grad_norm": 0.7090156674385071,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 42660
+    },
+    {
+      "epoch": 121.2215909090909,
+      "grad_norm": 1.0206100940704346,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 42670
+    },
+    {
+      "epoch": 121.25,
+      "grad_norm": 0.8041273355484009,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 42680
+    },
+    {
+      "epoch": 121.2784090909091,
+      "grad_norm": 0.7158668637275696,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 42690
+    },
+    {
+      "epoch": 121.30681818181819,
+      "grad_norm": 0.6531086564064026,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 42700
+    },
+    {
+      "epoch": 121.33522727272727,
+      "grad_norm": 0.7232179641723633,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 42710
+    },
+    {
+      "epoch": 121.36363636363636,
+      "grad_norm": 0.6331400275230408,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 42720
+    },
+    {
+      "epoch": 121.39204545454545,
+      "grad_norm": 0.7535783052444458,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 42730
+    },
+    {
+      "epoch": 121.42045454545455,
+      "grad_norm": 0.721626341342926,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 42740
+    },
+    {
+      "epoch": 121.44886363636364,
+      "grad_norm": 0.6356875896453857,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 42750
+    },
+    {
+      "epoch": 121.47727272727273,
+      "grad_norm": 0.6131756901741028,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 42760
+    },
+    {
+      "epoch": 121.50568181818181,
+      "grad_norm": 0.577969491481781,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 42770
+    },
+    {
+      "epoch": 121.5340909090909,
+      "grad_norm": 0.5389184951782227,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 42780
+    },
+    {
+      "epoch": 121.5625,
+      "grad_norm": 0.6019951105117798,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 42790
+    },
+    {
+      "epoch": 121.5909090909091,
+      "grad_norm": 0.5243552923202515,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 42800
+    },
+    {
+      "epoch": 121.61931818181819,
+      "grad_norm": 0.5662066340446472,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 42810
+    },
+    {
+      "epoch": 121.64772727272727,
+      "grad_norm": 0.673316478729248,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 42820
+    },
+    {
+      "epoch": 121.67613636363636,
+      "grad_norm": 0.7163783311843872,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 42830
+    },
+    {
+      "epoch": 121.70454545454545,
+      "grad_norm": 0.8090773820877075,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 42840
+    },
+    {
+      "epoch": 121.73295454545455,
+      "grad_norm": 0.97684246301651,
+      "learning_rate": 0.0001,
+      "loss": 0.0364,
+      "step": 42850
+    },
+    {
+      "epoch": 121.76136363636364,
+      "grad_norm": 0.8781417012214661,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 42860
+    },
+    {
+      "epoch": 121.78977272727273,
+      "grad_norm": 0.6750118136405945,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 42870
+    },
+    {
+      "epoch": 121.81818181818181,
+      "grad_norm": 0.7185530662536621,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 42880
+    },
+    {
+      "epoch": 121.8465909090909,
+      "grad_norm": 0.7951401472091675,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 42890
+    },
+    {
+      "epoch": 121.875,
+      "grad_norm": 0.6739190816879272,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 42900
+    },
+    {
+      "epoch": 121.9034090909091,
+      "grad_norm": 0.5545329451560974,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 42910
+    },
+    {
+      "epoch": 121.93181818181819,
+      "grad_norm": 0.7083759307861328,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 42920
+    },
+    {
+      "epoch": 121.96022727272727,
+      "grad_norm": 0.5985446572303772,
+      "learning_rate": 0.0001,
+      "loss": 0.0362,
+      "step": 42930
+    },
+    {
+      "epoch": 121.98863636363636,
+      "grad_norm": 0.6231947541236877,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 42940
+    },
+    {
+      "epoch": 122.01704545454545,
+      "grad_norm": 0.6297039985656738,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 42950
+    },
+    {
+      "epoch": 122.04545454545455,
+      "grad_norm": 0.5027703046798706,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 42960
+    },
+    {
+      "epoch": 122.07386363636364,
+      "grad_norm": 0.5628147721290588,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 42970
+    },
+    {
+      "epoch": 122.10227272727273,
+      "grad_norm": 0.5758391618728638,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 42980
+    },
+    {
+      "epoch": 122.13068181818181,
+      "grad_norm": 0.5796226263046265,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 42990
+    },
+    {
+      "epoch": 122.1590909090909,
+      "grad_norm": 0.6459672451019287,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 43000
+    },
+    {
+      "epoch": 122.1875,
+      "grad_norm": 0.7128278613090515,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 43010
+    },
+    {
+      "epoch": 122.2159090909091,
+      "grad_norm": 0.522715151309967,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 43020
+    },
+    {
+      "epoch": 122.24431818181819,
+      "grad_norm": 0.6805800795555115,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43030
+    },
+    {
+      "epoch": 122.27272727272727,
+      "grad_norm": 0.8094000220298767,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43040
+    },
+    {
+      "epoch": 122.30113636363636,
+      "grad_norm": 0.617563784122467,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 43050
+    },
+    {
+      "epoch": 122.32954545454545,
+      "grad_norm": 0.6483767628669739,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 43060
+    },
+    {
+      "epoch": 122.35795454545455,
+      "grad_norm": 0.6512372493743896,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 43070
+    },
+    {
+      "epoch": 122.38636363636364,
+      "grad_norm": 0.7526001334190369,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 43080
+    },
+    {
+      "epoch": 122.41477272727273,
+      "grad_norm": 0.7624586820602417,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 43090
+    },
+    {
+      "epoch": 122.44318181818181,
+      "grad_norm": 0.5687345862388611,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 43100
+    },
+    {
+      "epoch": 122.4715909090909,
+      "grad_norm": 0.60966956615448,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43110
+    },
+    {
+      "epoch": 122.5,
+      "grad_norm": 0.8027265667915344,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 43120
+    },
+    {
+      "epoch": 122.5284090909091,
+      "grad_norm": 0.5919138193130493,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 43130
+    },
+    {
+      "epoch": 122.55681818181819,
+      "grad_norm": 0.6522887349128723,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43140
+    },
+    {
+      "epoch": 122.58522727272727,
+      "grad_norm": 0.6283818483352661,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43150
+    },
+    {
+      "epoch": 122.61363636363636,
+      "grad_norm": 0.8638715147972107,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 43160
+    },
+    {
+      "epoch": 122.64204545454545,
+      "grad_norm": 0.9759404063224792,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 43170
+    },
+    {
+      "epoch": 122.67045454545455,
+      "grad_norm": 0.9701955914497375,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 43180
+    },
+    {
+      "epoch": 122.69886363636364,
+      "grad_norm": 0.6288473606109619,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 43190
+    },
+    {
+      "epoch": 122.72727272727273,
+      "grad_norm": 0.798579216003418,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43200
+    },
+    {
+      "epoch": 122.75568181818181,
+      "grad_norm": 0.7400290369987488,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 43210
+    },
+    {
+      "epoch": 122.7840909090909,
+      "grad_norm": 0.6623448133468628,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 43220
+    },
+    {
+      "epoch": 122.8125,
+      "grad_norm": 0.645901083946228,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 43230
+    },
+    {
+      "epoch": 122.8409090909091,
+      "grad_norm": 0.7637975811958313,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 43240
+    },
+    {
+      "epoch": 122.86931818181819,
+      "grad_norm": 0.763724684715271,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 43250
+    },
+    {
+      "epoch": 122.89772727272727,
+      "grad_norm": 0.6974006295204163,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 43260
+    },
+    {
+      "epoch": 122.92613636363636,
+      "grad_norm": 0.749366283416748,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 43270
+    },
+    {
+      "epoch": 122.95454545454545,
+      "grad_norm": 0.6450308561325073,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 43280
+    },
+    {
+      "epoch": 122.98295454545455,
+      "grad_norm": 0.7591984868049622,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 43290
+    },
+    {
+      "epoch": 123.01136363636364,
+      "grad_norm": 0.6023324728012085,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 43300
+    },
+    {
+      "epoch": 123.03977272727273,
+      "grad_norm": 0.6163926720619202,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 43310
+    },
+    {
+      "epoch": 123.06818181818181,
+      "grad_norm": 0.5046906471252441,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 43320
+    },
+    {
+      "epoch": 123.0965909090909,
+      "grad_norm": 0.6651236414909363,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 43330
+    },
+    {
+      "epoch": 123.125,
+      "grad_norm": 0.9277121424674988,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 43340
+    },
+    {
+      "epoch": 123.1534090909091,
+      "grad_norm": 0.847284734249115,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 43350
+    },
+    {
+      "epoch": 123.18181818181819,
+      "grad_norm": 0.7165888547897339,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 43360
+    },
+    {
+      "epoch": 123.21022727272727,
+      "grad_norm": 0.5978952646255493,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 43370
+    },
+    {
+      "epoch": 123.23863636363636,
+      "grad_norm": 0.8708299398422241,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 43380
+    },
+    {
+      "epoch": 123.26704545454545,
+      "grad_norm": 0.6764265298843384,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 43390
+    },
+    {
+      "epoch": 123.29545454545455,
+      "grad_norm": 0.9524985551834106,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 43400
+    },
+    {
+      "epoch": 123.32386363636364,
+      "grad_norm": 0.856083333492279,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43410
+    },
+    {
+      "epoch": 123.35227272727273,
+      "grad_norm": 0.8081461191177368,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 43420
+    },
+    {
+      "epoch": 123.38068181818181,
+      "grad_norm": 0.8981258273124695,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 43430
+    },
+    {
+      "epoch": 123.4090909090909,
+      "grad_norm": 0.7242816090583801,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 43440
+    },
+    {
+      "epoch": 123.4375,
+      "grad_norm": 0.983867347240448,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 43450
+    },
+    {
+      "epoch": 123.4659090909091,
+      "grad_norm": 0.9016323089599609,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 43460
+    },
+    {
+      "epoch": 123.49431818181819,
+      "grad_norm": 0.6101610660552979,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 43470
+    },
+    {
+      "epoch": 123.52272727272727,
+      "grad_norm": 0.6919850707054138,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 43480
+    },
+    {
+      "epoch": 123.55113636363636,
+      "grad_norm": 0.668318510055542,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 43490
+    },
+    {
+      "epoch": 123.57954545454545,
+      "grad_norm": 0.6016658544540405,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 43500
+    },
+    {
+      "epoch": 123.60795454545455,
+      "grad_norm": 0.6011306643486023,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 43510
+    },
+    {
+      "epoch": 123.63636363636364,
+      "grad_norm": 0.5712813138961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 43520
+    },
+    {
+      "epoch": 123.66477272727273,
+      "grad_norm": 0.5258704423904419,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 43530
+    },
+    {
+      "epoch": 123.69318181818181,
+      "grad_norm": 0.6852856278419495,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 43540
+    },
+    {
+      "epoch": 123.7215909090909,
+      "grad_norm": 0.6821662187576294,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 43550
+    },
+    {
+      "epoch": 123.75,
+      "grad_norm": 0.9406258463859558,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 43560
+    },
+    {
+      "epoch": 123.7784090909091,
+      "grad_norm": 0.6541404128074646,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 43570
+    },
+    {
+      "epoch": 123.80681818181819,
+      "grad_norm": 0.6775745153427124,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 43580
+    },
+    {
+      "epoch": 123.83522727272727,
+      "grad_norm": 0.8705558180809021,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 43590
+    },
+    {
+      "epoch": 123.86363636363636,
+      "grad_norm": 0.8164870142936707,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 43600
+    },
+    {
+      "epoch": 123.89204545454545,
+      "grad_norm": 0.6702988147735596,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 43610
+    },
+    {
+      "epoch": 123.92045454545455,
+      "grad_norm": 0.8119181990623474,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 43620
+    },
+    {
+      "epoch": 123.94886363636364,
+      "grad_norm": 0.5951023101806641,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 43630
+    },
+    {
+      "epoch": 123.97727272727273,
+      "grad_norm": 0.7227770090103149,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 43640
+    },
+    {
+      "epoch": 124.00568181818181,
+      "grad_norm": 0.6837309002876282,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 43650
+    },
+    {
+      "epoch": 124.0340909090909,
+      "grad_norm": 0.5477321743965149,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 43660
+    },
+    {
+      "epoch": 124.0625,
+      "grad_norm": 0.5354117751121521,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 43670
+    },
+    {
+      "epoch": 124.0909090909091,
+      "grad_norm": 0.6974219083786011,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 43680
+    },
+    {
+      "epoch": 124.11931818181819,
+      "grad_norm": 0.6769513487815857,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 43690
+    },
+    {
+      "epoch": 124.14772727272727,
+      "grad_norm": 0.7007016539573669,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 43700
+    },
+    {
+      "epoch": 124.17613636363636,
+      "grad_norm": 0.6461037993431091,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 43710
+    },
+    {
+      "epoch": 124.20454545454545,
+      "grad_norm": 0.6010048389434814,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 43720
+    },
+    {
+      "epoch": 124.23295454545455,
+      "grad_norm": 0.5394991040229797,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 43730
+    },
+    {
+      "epoch": 124.26136363636364,
+      "grad_norm": 0.48307615518569946,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 43740
+    },
+    {
+      "epoch": 124.28977272727273,
+      "grad_norm": 0.5091378092765808,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43750
+    },
+    {
+      "epoch": 124.31818181818181,
+      "grad_norm": 0.5722032785415649,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 43760
+    },
+    {
+      "epoch": 124.3465909090909,
+      "grad_norm": 0.5467521548271179,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 43770
+    },
+    {
+      "epoch": 124.375,
+      "grad_norm": 0.49909818172454834,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 43780
+    },
+    {
+      "epoch": 124.4034090909091,
+      "grad_norm": 0.6383428573608398,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 43790
+    },
+    {
+      "epoch": 124.43181818181819,
+      "grad_norm": 0.7279382348060608,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 43800
+    },
+    {
+      "epoch": 124.46022727272727,
+      "grad_norm": 0.4871160089969635,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43810
+    },
+    {
+      "epoch": 124.48863636363636,
+      "grad_norm": 0.5173508524894714,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 43820
+    },
+    {
+      "epoch": 124.51704545454545,
+      "grad_norm": 0.5603179335594177,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 43830
+    },
+    {
+      "epoch": 124.54545454545455,
+      "grad_norm": 0.4561804533004761,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 43840
+    },
+    {
+      "epoch": 124.57386363636364,
+      "grad_norm": 0.6842979788780212,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 43850
+    },
+    {
+      "epoch": 124.60227272727273,
+      "grad_norm": 0.6346740126609802,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 43860
+    },
+    {
+      "epoch": 124.63068181818181,
+      "grad_norm": 0.5946170091629028,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 43870
+    },
+    {
+      "epoch": 124.6590909090909,
+      "grad_norm": 0.4583919942378998,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 43880
+    },
+    {
+      "epoch": 124.6875,
+      "grad_norm": 0.8892697095870972,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43890
+    },
+    {
+      "epoch": 124.7159090909091,
+      "grad_norm": 0.7108684182167053,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 43900
+    },
+    {
+      "epoch": 124.74431818181819,
+      "grad_norm": 0.5772215127944946,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 43910
+    },
+    {
+      "epoch": 124.77272727272727,
+      "grad_norm": 0.6915180683135986,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 43920
+    },
+    {
+      "epoch": 124.80113636363636,
+      "grad_norm": 0.8463635444641113,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 43930
+    },
+    {
+      "epoch": 124.82954545454545,
+      "grad_norm": 0.700923502445221,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 43940
+    },
+    {
+      "epoch": 124.85795454545455,
+      "grad_norm": 0.6794731616973877,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43950
+    },
+    {
+      "epoch": 124.88636363636364,
+      "grad_norm": 0.6753823161125183,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 43960
+    },
+    {
+      "epoch": 124.91477272727273,
+      "grad_norm": 0.7288954257965088,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 43970
+    },
+    {
+      "epoch": 124.94318181818181,
+      "grad_norm": 0.8429861664772034,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 43980
+    },
+    {
+      "epoch": 124.9715909090909,
+      "grad_norm": 0.8071476817131042,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 43990
+    },
+    {
+      "epoch": 125.0,
+      "grad_norm": 0.535753607749939,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 44000
+    },
+    {
+      "epoch": 125.0284090909091,
+      "grad_norm": 0.6149882078170776,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 44010
+    },
+    {
+      "epoch": 125.05681818181819,
+      "grad_norm": 0.4777073860168457,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 44020
+    },
+    {
+      "epoch": 125.08522727272727,
+      "grad_norm": 0.5367774367332458,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 44030
+    },
+    {
+      "epoch": 125.11363636363636,
+      "grad_norm": 0.7416982054710388,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 44040
+    },
+    {
+      "epoch": 125.14204545454545,
+      "grad_norm": 0.5535955429077148,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 44050
+    },
+    {
+      "epoch": 125.17045454545455,
+      "grad_norm": 0.5561586022377014,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 44060
+    },
+    {
+      "epoch": 125.19886363636364,
+      "grad_norm": 0.6382668018341064,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 44070
+    },
+    {
+      "epoch": 125.22727272727273,
+      "grad_norm": 0.7232362031936646,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 44080
+    },
+    {
+      "epoch": 125.25568181818181,
+      "grad_norm": 0.5948888659477234,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 44090
+    },
+    {
+      "epoch": 125.2840909090909,
+      "grad_norm": 0.6899495124816895,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 44100
+    },
+    {
+      "epoch": 125.3125,
+      "grad_norm": 0.612220287322998,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 44110
+    },
+    {
+      "epoch": 125.3409090909091,
+      "grad_norm": 0.731208086013794,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 44120
+    },
+    {
+      "epoch": 125.36931818181819,
+      "grad_norm": 0.6531153917312622,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 44130
+    },
+    {
+      "epoch": 125.39772727272727,
+      "grad_norm": 0.662437379360199,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 44140
+    },
+    {
+      "epoch": 125.42613636363636,
+      "grad_norm": 0.6035165786743164,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 44150
+    },
+    {
+      "epoch": 125.45454545454545,
+      "grad_norm": 0.6566755771636963,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 44160
+    },
+    {
+      "epoch": 125.48295454545455,
+      "grad_norm": 0.80877286195755,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 44170
+    },
+    {
+      "epoch": 125.51136363636364,
+      "grad_norm": 0.7452892065048218,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 44180
+    },
+    {
+      "epoch": 125.53977272727273,
+      "grad_norm": 0.7200114727020264,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 44190
+    },
+    {
+      "epoch": 125.56818181818181,
+      "grad_norm": 0.5483570098876953,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 44200
+    },
+    {
+      "epoch": 125.5965909090909,
+      "grad_norm": 0.7039602398872375,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 44210
+    },
+    {
+      "epoch": 125.625,
+      "grad_norm": 0.5269778966903687,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 44220
+    },
+    {
+      "epoch": 125.6534090909091,
+      "grad_norm": 0.5524904131889343,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 44230
+    },
+    {
+      "epoch": 125.68181818181819,
+      "grad_norm": 0.6056973934173584,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 44240
+    },
+    {
+      "epoch": 125.71022727272727,
+      "grad_norm": 0.7236668467521667,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 44250
+    },
+    {
+      "epoch": 125.73863636363636,
+      "grad_norm": 0.6657916903495789,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 44260
+    },
+    {
+      "epoch": 125.76704545454545,
+      "grad_norm": 0.7973979711532593,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 44270
+    },
+    {
+      "epoch": 125.79545454545455,
+      "grad_norm": 0.7450402975082397,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 44280
+    },
+    {
+      "epoch": 125.82386363636364,
+      "grad_norm": 0.8265330791473389,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 44290
+    },
+    {
+      "epoch": 125.85227272727273,
+      "grad_norm": 0.49795201420783997,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 44300
+    },
+    {
+      "epoch": 125.88068181818181,
+      "grad_norm": 0.5793216824531555,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 44310
+    },
+    {
+      "epoch": 125.9090909090909,
+      "grad_norm": 0.5494784116744995,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 44320
+    },
+    {
+      "epoch": 125.9375,
+      "grad_norm": 0.7741979360580444,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 44330
+    },
+    {
+      "epoch": 125.9659090909091,
+      "grad_norm": 0.5874280333518982,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 44340
+    },
+    {
+      "epoch": 125.99431818181819,
+      "grad_norm": 0.5626118183135986,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 44350
+    },
+    {
+      "epoch": 126.02272727272727,
+      "grad_norm": 0.5672938823699951,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 44360
+    },
+    {
+      "epoch": 126.05113636363636,
+      "grad_norm": 0.5379377603530884,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 44370
+    },
+    {
+      "epoch": 126.07954545454545,
+      "grad_norm": 0.6178792715072632,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 44380
+    },
+    {
+      "epoch": 126.10795454545455,
+      "grad_norm": 0.4595714509487152,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 44390
+    },
+    {
+      "epoch": 126.13636363636364,
+      "grad_norm": 0.5440025925636292,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 44400
+    },
+    {
+      "epoch": 126.16477272727273,
+      "grad_norm": 0.447454035282135,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 44410
+    },
+    {
+      "epoch": 126.19318181818181,
+      "grad_norm": 0.5264921188354492,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 44420
+    },
+    {
+      "epoch": 126.2215909090909,
+      "grad_norm": 0.8086947202682495,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 44430
+    },
+    {
+      "epoch": 126.25,
+      "grad_norm": 0.743067741394043,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 44440
+    },
+    {
+      "epoch": 126.2784090909091,
+      "grad_norm": 0.5720465779304504,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 44450
+    },
+    {
+      "epoch": 126.30681818181819,
+      "grad_norm": 0.6255154609680176,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 44460
+    },
+    {
+      "epoch": 126.33522727272727,
+      "grad_norm": 0.7044152617454529,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 44470
+    },
+    {
+      "epoch": 126.36363636363636,
+      "grad_norm": 0.7242276668548584,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 44480
+    },
+    {
+      "epoch": 126.39204545454545,
+      "grad_norm": 0.5152789354324341,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 44490
+    },
+    {
+      "epoch": 126.42045454545455,
+      "grad_norm": 0.8477872610092163,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 44500
+    },
+    {
+      "epoch": 126.44886363636364,
+      "grad_norm": 1.2072283029556274,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 44510
+    },
+    {
+      "epoch": 126.47727272727273,
+      "grad_norm": 1.0745333433151245,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 44520
+    },
+    {
+      "epoch": 126.50568181818181,
+      "grad_norm": 0.6237938404083252,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 44530
+    },
+    {
+      "epoch": 126.5340909090909,
+      "grad_norm": 1.150746464729309,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 44540
+    },
+    {
+      "epoch": 126.5625,
+      "grad_norm": 0.6744441390037537,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 44550
+    },
+    {
+      "epoch": 126.5909090909091,
+      "grad_norm": 0.8944215178489685,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 44560
+    },
+    {
+      "epoch": 126.61931818181819,
+      "grad_norm": 0.8610298037528992,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 44570
+    },
+    {
+      "epoch": 126.64772727272727,
+      "grad_norm": 1.040744423866272,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 44580
+    },
+    {
+      "epoch": 126.67613636363636,
+      "grad_norm": 0.8564578890800476,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 44590
+    },
+    {
+      "epoch": 126.70454545454545,
+      "grad_norm": 0.9337520599365234,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 44600
+    },
+    {
+      "epoch": 126.73295454545455,
+      "grad_norm": 0.8132588863372803,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 44610
+    },
+    {
+      "epoch": 126.76136363636364,
+      "grad_norm": 0.9135181903839111,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 44620
+    },
+    {
+      "epoch": 126.78977272727273,
+      "grad_norm": 1.0999176502227783,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 44630
+    },
+    {
+      "epoch": 126.81818181818181,
+      "grad_norm": 0.9005665183067322,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 44640
+    },
+    {
+      "epoch": 126.8465909090909,
+      "grad_norm": 0.9482970833778381,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 44650
+    },
+    {
+      "epoch": 126.875,
+      "grad_norm": 0.708218514919281,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 44660
+    },
+    {
+      "epoch": 126.9034090909091,
+      "grad_norm": 0.8633854389190674,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 44670
+    },
+    {
+      "epoch": 126.93181818181819,
+      "grad_norm": 0.7367269992828369,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 44680
+    },
+    {
+      "epoch": 126.96022727272727,
+      "grad_norm": 0.954519510269165,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 44690
+    },
+    {
+      "epoch": 126.98863636363636,
+      "grad_norm": 1.0101993083953857,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 44700
+    },
+    {
+      "epoch": 127.01704545454545,
+      "grad_norm": 1.0522290468215942,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 44710
+    },
+    {
+      "epoch": 127.04545454545455,
+      "grad_norm": 1.0212730169296265,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 44720
+    },
+    {
+      "epoch": 127.07386363636364,
+      "grad_norm": 0.8338181972503662,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 44730
+    },
+    {
+      "epoch": 127.10227272727273,
+      "grad_norm": 0.8435046672821045,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 44740
+    },
+    {
+      "epoch": 127.13068181818181,
+      "grad_norm": 0.6481569409370422,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 44750
+    },
+    {
+      "epoch": 127.1590909090909,
+      "grad_norm": 0.525007426738739,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 44760
+    },
+    {
+      "epoch": 127.1875,
+      "grad_norm": 0.7917028069496155,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 44770
+    },
+    {
+      "epoch": 127.2159090909091,
+      "grad_norm": 0.6355962157249451,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 44780
+    },
+    {
+      "epoch": 127.24431818181819,
+      "grad_norm": 0.4666050970554352,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 44790
+    },
+    {
+      "epoch": 127.27272727272727,
+      "grad_norm": 0.6988716125488281,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 44800
+    },
+    {
+      "epoch": 127.30113636363636,
+      "grad_norm": 0.599835216999054,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 44810
+    },
+    {
+      "epoch": 127.32954545454545,
+      "grad_norm": 0.7451519966125488,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 44820
+    },
+    {
+      "epoch": 127.35795454545455,
+      "grad_norm": 0.7405692934989929,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 44830
+    },
+    {
+      "epoch": 127.38636363636364,
+      "grad_norm": 0.7457634806632996,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 44840
+    },
+    {
+      "epoch": 127.41477272727273,
+      "grad_norm": 0.575497031211853,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 44850
+    },
+    {
+      "epoch": 127.44318181818181,
+      "grad_norm": 0.6736639738082886,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 44860
+    },
+    {
+      "epoch": 127.4715909090909,
+      "grad_norm": 0.7018943428993225,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 44870
+    },
+    {
+      "epoch": 127.5,
+      "grad_norm": 0.6599423289299011,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 44880
+    },
+    {
+      "epoch": 127.5284090909091,
+      "grad_norm": 0.6989256739616394,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 44890
+    },
+    {
+      "epoch": 127.55681818181819,
+      "grad_norm": 0.6219071745872498,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 44900
+    },
+    {
+      "epoch": 127.58522727272727,
+      "grad_norm": 0.6060497164726257,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 44910
+    },
+    {
+      "epoch": 127.61363636363636,
+      "grad_norm": 0.8178271651268005,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 44920
+    },
+    {
+      "epoch": 127.64204545454545,
+      "grad_norm": 0.6442059874534607,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 44930
+    },
+    {
+      "epoch": 127.67045454545455,
+      "grad_norm": 0.8080576062202454,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 44940
+    },
+    {
+      "epoch": 127.69886363636364,
+      "grad_norm": 0.8716291189193726,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 44950
+    },
+    {
+      "epoch": 127.72727272727273,
+      "grad_norm": 0.6577211022377014,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 44960
+    },
+    {
+      "epoch": 127.75568181818181,
+      "grad_norm": 0.6145163774490356,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 44970
+    },
+    {
+      "epoch": 127.7840909090909,
+      "grad_norm": 0.6429737210273743,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 44980
+    },
+    {
+      "epoch": 127.8125,
+      "grad_norm": 1.2113999128341675,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 44990
+    },
+    {
+      "epoch": 127.8409090909091,
+      "grad_norm": 1.2781754732131958,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 45000
+    },
+    {
+      "epoch": 127.86931818181819,
+      "grad_norm": 0.8394537568092346,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 45010
+    },
+    {
+      "epoch": 127.89772727272727,
+      "grad_norm": 1.1930190324783325,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 45020
+    },
+    {
+      "epoch": 127.92613636363636,
+      "grad_norm": 0.6906962990760803,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 45030
+    },
+    {
+      "epoch": 127.95454545454545,
+      "grad_norm": 0.891907811164856,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 45040
+    },
+    {
+      "epoch": 127.98295454545455,
+      "grad_norm": 1.1798697710037231,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 45050
+    },
+    {
+      "epoch": 128.01136363636363,
+      "grad_norm": 1.0021605491638184,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 45060
+    },
+    {
+      "epoch": 128.03977272727272,
+      "grad_norm": 1.0694570541381836,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 45070
+    },
+    {
+      "epoch": 128.0681818181818,
+      "grad_norm": 0.9267218708992004,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 45080
+    },
+    {
+      "epoch": 128.0965909090909,
+      "grad_norm": 0.6276006698608398,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 45090
+    },
+    {
+      "epoch": 128.125,
+      "grad_norm": 0.6347649693489075,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 45100
+    },
+    {
+      "epoch": 128.1534090909091,
+      "grad_norm": 0.6764646768569946,
+      "learning_rate": 0.0001,
+      "loss": 0.0357,
+      "step": 45110
+    },
+    {
+      "epoch": 128.1818181818182,
+      "grad_norm": 0.6223496794700623,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 45120
+    },
+    {
+      "epoch": 128.21022727272728,
+      "grad_norm": 0.7356517910957336,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 45130
+    },
+    {
+      "epoch": 128.23863636363637,
+      "grad_norm": 0.9081064462661743,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 45140
+    },
+    {
+      "epoch": 128.26704545454547,
+      "grad_norm": 0.8737921118736267,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 45150
+    },
+    {
+      "epoch": 128.29545454545453,
+      "grad_norm": 0.9559502005577087,
+      "learning_rate": 0.0001,
+      "loss": 0.0361,
+      "step": 45160
+    },
+    {
+      "epoch": 128.32386363636363,
+      "grad_norm": 0.9400319457054138,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 45170
+    },
+    {
+      "epoch": 128.35227272727272,
+      "grad_norm": 0.9618202447891235,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 45180
+    },
+    {
+      "epoch": 128.3806818181818,
+      "grad_norm": 0.8962883353233337,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 45190
+    },
+    {
+      "epoch": 128.4090909090909,
+      "grad_norm": 0.6514441967010498,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 45200
+    },
+    {
+      "epoch": 128.4375,
+      "grad_norm": 0.7900486588478088,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 45210
+    },
+    {
+      "epoch": 128.4659090909091,
+      "grad_norm": 1.0713646411895752,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 45220
+    },
+    {
+      "epoch": 128.4943181818182,
+      "grad_norm": 0.7977399230003357,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 45230
+    },
+    {
+      "epoch": 128.52272727272728,
+      "grad_norm": 0.7408506870269775,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 45240
+    },
+    {
+      "epoch": 128.55113636363637,
+      "grad_norm": 0.7208636999130249,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 45250
+    },
+    {
+      "epoch": 128.57954545454547,
+      "grad_norm": 0.9026774764060974,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 45260
+    },
+    {
+      "epoch": 128.60795454545453,
+      "grad_norm": 0.7447446584701538,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 45270
+    },
+    {
+      "epoch": 128.63636363636363,
+      "grad_norm": 0.6284778714179993,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 45280
+    },
+    {
+      "epoch": 128.66477272727272,
+      "grad_norm": 0.773349404335022,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 45290
+    },
+    {
+      "epoch": 128.6931818181818,
+      "grad_norm": 0.652649462223053,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 45300
+    },
+    {
+      "epoch": 128.7215909090909,
+      "grad_norm": 0.664982795715332,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 45310
+    },
+    {
+      "epoch": 128.75,
+      "grad_norm": 0.9043540954589844,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 45320
+    },
+    {
+      "epoch": 128.7784090909091,
+      "grad_norm": 0.6659265160560608,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 45330
+    },
+    {
+      "epoch": 128.8068181818182,
+      "grad_norm": 0.6651855111122131,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 45340
+    },
+    {
+      "epoch": 128.83522727272728,
+      "grad_norm": 0.6218985915184021,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 45350
+    },
+    {
+      "epoch": 128.86363636363637,
+      "grad_norm": 0.7133499979972839,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 45360
+    },
+    {
+      "epoch": 128.89204545454547,
+      "grad_norm": 0.6537865996360779,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 45370
+    },
+    {
+      "epoch": 128.92045454545453,
+      "grad_norm": 0.7740278840065002,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 45380
+    },
+    {
+      "epoch": 128.94886363636363,
+      "grad_norm": 0.7691712379455566,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 45390
+    },
+    {
+      "epoch": 128.97727272727272,
+      "grad_norm": 0.7746933698654175,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 45400
+    },
+    {
+      "epoch": 129.0056818181818,
+      "grad_norm": 0.7170177102088928,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 45410
+    },
+    {
+      "epoch": 129.0340909090909,
+      "grad_norm": 0.6429873108863831,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 45420
+    },
+    {
+      "epoch": 129.0625,
+      "grad_norm": 0.8797420263290405,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 45430
+    },
+    {
+      "epoch": 129.0909090909091,
+      "grad_norm": 0.6143895983695984,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 45440
+    },
+    {
+      "epoch": 129.1193181818182,
+      "grad_norm": 0.7619710564613342,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 45450
+    },
+    {
+      "epoch": 129.14772727272728,
+      "grad_norm": 0.5125098824501038,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 45460
+    },
+    {
+      "epoch": 129.17613636363637,
+      "grad_norm": 0.7374431490898132,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 45470
+    },
+    {
+      "epoch": 129.20454545454547,
+      "grad_norm": 0.6298514604568481,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 45480
+    },
+    {
+      "epoch": 129.23295454545453,
+      "grad_norm": 0.570999801158905,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 45490
+    },
+    {
+      "epoch": 129.26136363636363,
+      "grad_norm": 0.5362004041671753,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 45500
+    },
+    {
+      "epoch": 129.28977272727272,
+      "grad_norm": 0.6400303840637207,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 45510
+    },
+    {
+      "epoch": 129.3181818181818,
+      "grad_norm": 0.772942066192627,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 45520
+    },
+    {
+      "epoch": 129.3465909090909,
+      "grad_norm": 0.5503178238868713,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 45530
+    },
+    {
+      "epoch": 129.375,
+      "grad_norm": 0.5466687083244324,
+      "learning_rate": 0.0001,
+      "loss": 0.0353,
+      "step": 45540
+    },
+    {
+      "epoch": 129.4034090909091,
+      "grad_norm": 0.6772168874740601,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 45550
+    },
+    {
+      "epoch": 129.4318181818182,
+      "grad_norm": 0.6691098809242249,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 45560
+    },
+    {
+      "epoch": 129.46022727272728,
+      "grad_norm": 0.8726401925086975,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 45570
+    },
+    {
+      "epoch": 129.48863636363637,
+      "grad_norm": 1.072825312614441,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 45580
+    },
+    {
+      "epoch": 129.51704545454547,
+      "grad_norm": 1.0415953397750854,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 45590
+    },
+    {
+      "epoch": 129.54545454545453,
+      "grad_norm": 0.5183255672454834,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 45600
+    },
+    {
+      "epoch": 129.57386363636363,
+      "grad_norm": 0.8744351267814636,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 45610
+    },
+    {
+      "epoch": 129.60227272727272,
+      "grad_norm": 0.8066695928573608,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 45620
+    },
+    {
+      "epoch": 129.6306818181818,
+      "grad_norm": 0.7792837619781494,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 45630
+    },
+    {
+      "epoch": 129.6590909090909,
+      "grad_norm": 0.8309561610221863,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 45640
+    },
+    {
+      "epoch": 129.6875,
+      "grad_norm": 0.7426981925964355,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 45650
+    },
+    {
+      "epoch": 129.7159090909091,
+      "grad_norm": 0.7349977493286133,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 45660
+    },
+    {
+      "epoch": 129.7443181818182,
+      "grad_norm": 0.6687122583389282,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 45670
+    },
+    {
+      "epoch": 129.77272727272728,
+      "grad_norm": 0.6240840554237366,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 45680
+    },
+    {
+      "epoch": 129.80113636363637,
+      "grad_norm": 0.6070581078529358,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 45690
+    },
+    {
+      "epoch": 129.82954545454547,
+      "grad_norm": 0.6064611673355103,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 45700
+    },
+    {
+      "epoch": 129.85795454545453,
+      "grad_norm": 0.8292164206504822,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 45710
+    },
+    {
+      "epoch": 129.88636363636363,
+      "grad_norm": 0.6600549221038818,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 45720
+    },
+    {
+      "epoch": 129.91477272727272,
+      "grad_norm": 0.8599357604980469,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 45730
+    },
+    {
+      "epoch": 129.9431818181818,
+      "grad_norm": 0.6765527725219727,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 45740
+    },
+    {
+      "epoch": 129.9715909090909,
+      "grad_norm": 0.7478293776512146,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 45750
+    },
+    {
+      "epoch": 130.0,
+      "grad_norm": 0.5045310258865356,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 45760
+    },
+    {
+      "epoch": 130.0284090909091,
+      "grad_norm": 0.5499379634857178,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 45770
+    },
+    {
+      "epoch": 130.0568181818182,
+      "grad_norm": 0.5516977906227112,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 45780
+    },
+    {
+      "epoch": 130.08522727272728,
+      "grad_norm": 0.4980061948299408,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 45790
+    },
+    {
+      "epoch": 130.11363636363637,
+      "grad_norm": 0.47144022583961487,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 45800
+    },
+    {
+      "epoch": 130.14204545454547,
+      "grad_norm": 0.6652446985244751,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 45810
+    },
+    {
+      "epoch": 130.17045454545453,
+      "grad_norm": 0.5668972730636597,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 45820
+    },
+    {
+      "epoch": 130.19886363636363,
+      "grad_norm": 0.5128690004348755,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 45830
+    },
+    {
+      "epoch": 130.22727272727272,
+      "grad_norm": 0.47686073184013367,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 45840
+    },
+    {
+      "epoch": 130.2556818181818,
+      "grad_norm": 0.7872583270072937,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 45850
+    },
+    {
+      "epoch": 130.2840909090909,
+      "grad_norm": 0.6706736087799072,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 45860
+    },
+    {
+      "epoch": 130.3125,
+      "grad_norm": 0.6146107912063599,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 45870
+    },
+    {
+      "epoch": 130.3409090909091,
+      "grad_norm": 0.7997154593467712,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 45880
+    },
+    {
+      "epoch": 130.3693181818182,
+      "grad_norm": 0.8617053031921387,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 45890
+    },
+    {
+      "epoch": 130.39772727272728,
+      "grad_norm": 0.8967577219009399,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 45900
+    },
+    {
+      "epoch": 130.42613636363637,
+      "grad_norm": 0.7063471674919128,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 45910
+    },
+    {
+      "epoch": 130.45454545454547,
+      "grad_norm": 0.6364906430244446,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 45920
+    },
+    {
+      "epoch": 130.48295454545453,
+      "grad_norm": 0.5470952987670898,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 45930
+    },
+    {
+      "epoch": 130.51136363636363,
+      "grad_norm": 0.8126204013824463,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 45940
+    },
+    {
+      "epoch": 130.53977272727272,
+      "grad_norm": 0.9707944989204407,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 45950
+    },
+    {
+      "epoch": 130.5681818181818,
+      "grad_norm": 0.777581512928009,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 45960
+    },
+    {
+      "epoch": 130.5965909090909,
+      "grad_norm": 0.49622127413749695,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 45970
+    },
+    {
+      "epoch": 130.625,
+      "grad_norm": 0.7890464067459106,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 45980
+    },
+    {
+      "epoch": 130.6534090909091,
+      "grad_norm": 0.6376560926437378,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 45990
+    },
+    {
+      "epoch": 130.6818181818182,
+      "grad_norm": 0.5111783742904663,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 46000
+    },
+    {
+      "epoch": 130.71022727272728,
+      "grad_norm": 0.6418190002441406,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 46010
+    },
+    {
+      "epoch": 130.73863636363637,
+      "grad_norm": 0.5361145734786987,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 46020
+    },
+    {
+      "epoch": 130.76704545454547,
+      "grad_norm": 0.5517832040786743,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 46030
+    },
+    {
+      "epoch": 130.79545454545453,
+      "grad_norm": 0.6392123103141785,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 46040
+    },
+    {
+      "epoch": 130.82386363636363,
+      "grad_norm": 0.4958942234516144,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 46050
+    },
+    {
+      "epoch": 130.85227272727272,
+      "grad_norm": 0.5867140293121338,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 46060
+    },
+    {
+      "epoch": 130.8806818181818,
+      "grad_norm": 0.666496217250824,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 46070
+    },
+    {
+      "epoch": 130.9090909090909,
+      "grad_norm": 0.6093063950538635,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 46080
+    },
+    {
+      "epoch": 130.9375,
+      "grad_norm": 0.6793018579483032,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 46090
+    },
+    {
+      "epoch": 130.9659090909091,
+      "grad_norm": 0.8787961602210999,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 46100
+    },
+    {
+      "epoch": 130.9943181818182,
+      "grad_norm": 0.8491075038909912,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 46110
+    },
+    {
+      "epoch": 131.02272727272728,
+      "grad_norm": 0.8769429922103882,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 46120
+    },
+    {
+      "epoch": 131.05113636363637,
+      "grad_norm": 0.6816815137863159,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 46130
+    },
+    {
+      "epoch": 131.07954545454547,
+      "grad_norm": 0.716789722442627,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 46140
+    },
+    {
+      "epoch": 131.10795454545453,
+      "grad_norm": 0.5310009717941284,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 46150
+    },
+    {
+      "epoch": 131.13636363636363,
+      "grad_norm": 0.5277566909790039,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 46160
+    },
+    {
+      "epoch": 131.16477272727272,
+      "grad_norm": 0.48119422793388367,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 46170
+    },
+    {
+      "epoch": 131.1931818181818,
+      "grad_norm": 0.510347843170166,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 46180
+    },
+    {
+      "epoch": 131.2215909090909,
+      "grad_norm": 0.5207124948501587,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 46190
+    },
+    {
+      "epoch": 131.25,
+      "grad_norm": 0.5676156878471375,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 46200
+    },
+    {
+      "epoch": 131.2784090909091,
+      "grad_norm": 0.5713900327682495,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 46210
+    },
+    {
+      "epoch": 131.3068181818182,
+      "grad_norm": 0.6059330701828003,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 46220
+    },
+    {
+      "epoch": 131.33522727272728,
+      "grad_norm": 0.5800050497055054,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 46230
+    },
+    {
+      "epoch": 131.36363636363637,
+      "grad_norm": 0.5849801301956177,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 46240
+    },
+    {
+      "epoch": 131.39204545454547,
+      "grad_norm": 0.46115511655807495,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 46250
+    },
+    {
+      "epoch": 131.42045454545453,
+      "grad_norm": 0.6311548948287964,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 46260
+    },
+    {
+      "epoch": 131.44886363636363,
+      "grad_norm": 0.4590200185775757,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 46270
+    },
+    {
+      "epoch": 131.47727272727272,
+      "grad_norm": 0.4637441635131836,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 46280
+    },
+    {
+      "epoch": 131.5056818181818,
+      "grad_norm": 0.5604123473167419,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 46290
+    },
+    {
+      "epoch": 131.5340909090909,
+      "grad_norm": 0.7339373826980591,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 46300
+    },
+    {
+      "epoch": 131.5625,
+      "grad_norm": 0.49434801936149597,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 46310
+    },
+    {
+      "epoch": 131.5909090909091,
+      "grad_norm": 0.5729549527168274,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 46320
+    },
+    {
+      "epoch": 131.6193181818182,
+      "grad_norm": 0.5489758253097534,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 46330
+    },
+    {
+      "epoch": 131.64772727272728,
+      "grad_norm": 0.5991235375404358,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 46340
+    },
+    {
+      "epoch": 131.67613636363637,
+      "grad_norm": 0.5655492544174194,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 46350
+    },
+    {
+      "epoch": 131.70454545454547,
+      "grad_norm": 0.5596151947975159,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 46360
+    },
+    {
+      "epoch": 131.73295454545453,
+      "grad_norm": 0.6359032988548279,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 46370
+    },
+    {
+      "epoch": 131.76136363636363,
+      "grad_norm": 0.6330766677856445,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 46380
+    },
+    {
+      "epoch": 131.78977272727272,
+      "grad_norm": 0.7287530303001404,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 46390
+    },
+    {
+      "epoch": 131.8181818181818,
+      "grad_norm": 0.4960668087005615,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 46400
+    },
+    {
+      "epoch": 131.8465909090909,
+      "grad_norm": 0.7417987585067749,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 46410
+    },
+    {
+      "epoch": 131.875,
+      "grad_norm": 0.5909314155578613,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 46420
+    },
+    {
+      "epoch": 131.9034090909091,
+      "grad_norm": 0.4722643494606018,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 46430
+    },
+    {
+      "epoch": 131.9318181818182,
+      "grad_norm": 0.5753642320632935,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 46440
+    },
+    {
+      "epoch": 131.96022727272728,
+      "grad_norm": 0.5663985013961792,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 46450
+    },
+    {
+      "epoch": 131.98863636363637,
+      "grad_norm": 0.5400167107582092,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 46460
+    },
+    {
+      "epoch": 132.01704545454547,
+      "grad_norm": 0.46016308665275574,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 46470
+    },
+    {
+      "epoch": 132.04545454545453,
+      "grad_norm": 0.5892532467842102,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 46480
+    },
+    {
+      "epoch": 132.07386363636363,
+      "grad_norm": 0.5685364603996277,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 46490
+    },
+    {
+      "epoch": 132.10227272727272,
+      "grad_norm": 0.5203407406806946,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 46500
+    },
+    {
+      "epoch": 132.1306818181818,
+      "grad_norm": 0.5543718338012695,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 46510
+    },
+    {
+      "epoch": 132.1590909090909,
+      "grad_norm": 0.624247670173645,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 46520
+    },
+    {
+      "epoch": 132.1875,
+      "grad_norm": 0.6007071137428284,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 46530
+    },
+    {
+      "epoch": 132.2159090909091,
+      "grad_norm": 0.68288254737854,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 46540
+    },
+    {
+      "epoch": 132.2443181818182,
+      "grad_norm": 1.2552138566970825,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 46550
+    },
+    {
+      "epoch": 132.27272727272728,
+      "grad_norm": 0.9271278381347656,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 46560
+    },
+    {
+      "epoch": 132.30113636363637,
+      "grad_norm": 0.8253664374351501,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 46570
+    },
+    {
+      "epoch": 132.32954545454547,
+      "grad_norm": 0.5965460538864136,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 46580
+    },
+    {
+      "epoch": 132.35795454545453,
+      "grad_norm": 0.6545840501785278,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 46590
+    },
+    {
+      "epoch": 132.38636363636363,
+      "grad_norm": 0.6397396326065063,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 46600
+    },
+    {
+      "epoch": 132.41477272727272,
+      "grad_norm": 0.6624685525894165,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 46610
+    },
+    {
+      "epoch": 132.4431818181818,
+      "grad_norm": 0.7068834900856018,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 46620
+    },
+    {
+      "epoch": 132.4715909090909,
+      "grad_norm": 0.6984627842903137,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 46630
+    },
+    {
+      "epoch": 132.5,
+      "grad_norm": 0.6401987075805664,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 46640
+    },
+    {
+      "epoch": 132.5284090909091,
+      "grad_norm": 0.5327432155609131,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 46650
+    },
+    {
+      "epoch": 132.5568181818182,
+      "grad_norm": 0.5304933190345764,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 46660
+    },
+    {
+      "epoch": 132.58522727272728,
+      "grad_norm": 0.7458568215370178,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 46670
+    },
+    {
+      "epoch": 132.61363636363637,
+      "grad_norm": 0.6337578892707825,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 46680
+    },
+    {
+      "epoch": 132.64204545454547,
+      "grad_norm": 0.712566077709198,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 46690
+    },
+    {
+      "epoch": 132.67045454545453,
+      "grad_norm": 0.6466312408447266,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 46700
+    },
+    {
+      "epoch": 132.69886363636363,
+      "grad_norm": 0.6697583794593811,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 46710
+    },
+    {
+      "epoch": 132.72727272727272,
+      "grad_norm": 0.5255504846572876,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 46720
+    },
+    {
+      "epoch": 132.7556818181818,
+      "grad_norm": 1.0727545022964478,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 46730
+    },
+    {
+      "epoch": 132.7840909090909,
+      "grad_norm": 0.7345282435417175,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 46740
+    },
+    {
+      "epoch": 132.8125,
+      "grad_norm": 0.6004593968391418,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 46750
+    },
+    {
+      "epoch": 132.8409090909091,
+      "grad_norm": 0.4870907664299011,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 46760
+    },
+    {
+      "epoch": 132.8693181818182,
+      "grad_norm": 0.7557304501533508,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 46770
+    },
+    {
+      "epoch": 132.89772727272728,
+      "grad_norm": 0.9188068509101868,
+      "learning_rate": 0.0001,
+      "loss": 0.0347,
+      "step": 46780
+    },
+    {
+      "epoch": 132.92613636363637,
+      "grad_norm": 0.5896967053413391,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 46790
+    },
+    {
+      "epoch": 132.95454545454547,
+      "grad_norm": 0.7609643936157227,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 46800
+    },
+    {
+      "epoch": 132.98295454545453,
+      "grad_norm": 0.8466443419456482,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 46810
+    },
+    {
+      "epoch": 133.01136363636363,
+      "grad_norm": 0.8880951404571533,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 46820
+    },
+    {
+      "epoch": 133.03977272727272,
+      "grad_norm": 0.9683408737182617,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 46830
+    },
+    {
+      "epoch": 133.0681818181818,
+      "grad_norm": 0.871724009513855,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 46840
+    },
+    {
+      "epoch": 133.0965909090909,
+      "grad_norm": 0.5088608264923096,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 46850
+    },
+    {
+      "epoch": 133.125,
+      "grad_norm": 0.7277428507804871,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 46860
+    },
+    {
+      "epoch": 133.1534090909091,
+      "grad_norm": 0.7390486598014832,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 46870
+    },
+    {
+      "epoch": 133.1818181818182,
+      "grad_norm": 0.707007884979248,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 46880
+    },
+    {
+      "epoch": 133.21022727272728,
+      "grad_norm": 0.6008827090263367,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 46890
+    },
+    {
+      "epoch": 133.23863636363637,
+      "grad_norm": 0.6185063123703003,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 46900
+    },
+    {
+      "epoch": 133.26704545454547,
+      "grad_norm": 0.67989182472229,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 46910
+    },
+    {
+      "epoch": 133.29545454545453,
+      "grad_norm": 0.548943817615509,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 46920
+    },
+    {
+      "epoch": 133.32386363636363,
+      "grad_norm": 0.6702325344085693,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 46930
+    },
+    {
+      "epoch": 133.35227272727272,
+      "grad_norm": 0.7883853316307068,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 46940
+    },
+    {
+      "epoch": 133.3806818181818,
+      "grad_norm": 0.7740568518638611,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 46950
+    },
+    {
+      "epoch": 133.4090909090909,
+      "grad_norm": 0.6243300437927246,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 46960
+    },
+    {
+      "epoch": 133.4375,
+      "grad_norm": 0.7169111371040344,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 46970
+    },
+    {
+      "epoch": 133.4659090909091,
+      "grad_norm": 0.6098527312278748,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 46980
+    },
+    {
+      "epoch": 133.4943181818182,
+      "grad_norm": 0.5657358169555664,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 46990
+    },
+    {
+      "epoch": 133.52272727272728,
+      "grad_norm": 0.5001091957092285,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 47000
+    },
+    {
+      "epoch": 133.55113636363637,
+      "grad_norm": 0.5768367052078247,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 47010
+    },
+    {
+      "epoch": 133.57954545454547,
+      "grad_norm": 0.7666851282119751,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 47020
+    },
+    {
+      "epoch": 133.60795454545453,
+      "grad_norm": 0.6639266014099121,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 47030
+    },
+    {
+      "epoch": 133.63636363636363,
+      "grad_norm": 1.1128325462341309,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 47040
+    },
+    {
+      "epoch": 133.66477272727272,
+      "grad_norm": 0.7729830741882324,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 47050
+    },
+    {
+      "epoch": 133.6931818181818,
+      "grad_norm": 0.60756915807724,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 47060
+    },
+    {
+      "epoch": 133.7215909090909,
+      "grad_norm": 0.6942067742347717,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 47070
+    },
+    {
+      "epoch": 133.75,
+      "grad_norm": 0.9247666597366333,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 47080
+    },
+    {
+      "epoch": 133.7784090909091,
+      "grad_norm": 0.9722429513931274,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 47090
+    },
+    {
+      "epoch": 133.8068181818182,
+      "grad_norm": 0.6130657196044922,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 47100
+    },
+    {
+      "epoch": 133.83522727272728,
+      "grad_norm": 0.8849529027938843,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 47110
+    },
+    {
+      "epoch": 133.86363636363637,
+      "grad_norm": 0.865003228187561,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 47120
+    },
+    {
+      "epoch": 133.89204545454547,
+      "grad_norm": 0.6581420302391052,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 47130
+    },
+    {
+      "epoch": 133.92045454545453,
+      "grad_norm": 0.7639325857162476,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 47140
+    },
+    {
+      "epoch": 133.94886363636363,
+      "grad_norm": 0.6666318774223328,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 47150
+    },
+    {
+      "epoch": 133.97727272727272,
+      "grad_norm": 1.2542755603790283,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 47160
+    },
+    {
+      "epoch": 134.0056818181818,
+      "grad_norm": 1.2946697473526,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 47170
+    },
+    {
+      "epoch": 134.0340909090909,
+      "grad_norm": 1.3122329711914062,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 47180
+    },
+    {
+      "epoch": 134.0625,
+      "grad_norm": 1.2413455247879028,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 47190
+    },
+    {
+      "epoch": 134.0909090909091,
+      "grad_norm": 0.801084578037262,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 47200
+    },
+    {
+      "epoch": 134.1193181818182,
+      "grad_norm": 0.6665140986442566,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 47210
+    },
+    {
+      "epoch": 134.14772727272728,
+      "grad_norm": 1.088781714439392,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 47220
+    },
+    {
+      "epoch": 134.17613636363637,
+      "grad_norm": 0.8474469780921936,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 47230
+    },
+    {
+      "epoch": 134.20454545454547,
+      "grad_norm": 0.8532624244689941,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 47240
+    },
+    {
+      "epoch": 134.23295454545453,
+      "grad_norm": 0.6617245078086853,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 47250
+    },
+    {
+      "epoch": 134.26136363636363,
+      "grad_norm": 0.7599644064903259,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 47260
+    },
+    {
+      "epoch": 134.28977272727272,
+      "grad_norm": 0.7604702711105347,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 47270
+    },
+    {
+      "epoch": 134.3181818181818,
+      "grad_norm": 0.6242595314979553,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 47280
+    },
+    {
+      "epoch": 134.3465909090909,
+      "grad_norm": 0.8279832601547241,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 47290
+    },
+    {
+      "epoch": 134.375,
+      "grad_norm": 0.7884923815727234,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 47300
+    },
+    {
+      "epoch": 134.4034090909091,
+      "grad_norm": 0.6375776529312134,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 47310
+    },
+    {
+      "epoch": 134.4318181818182,
+      "grad_norm": 0.5708988308906555,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 47320
+    },
+    {
+      "epoch": 134.46022727272728,
+      "grad_norm": 0.570242702960968,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 47330
+    },
+    {
+      "epoch": 134.48863636363637,
+      "grad_norm": 0.6287363767623901,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 47340
+    },
+    {
+      "epoch": 134.51704545454547,
+      "grad_norm": 0.6968216300010681,
+      "learning_rate": 0.0001,
+      "loss": 0.0371,
+      "step": 47350
+    },
+    {
+      "epoch": 134.54545454545453,
+      "grad_norm": 0.48705530166625977,
+      "learning_rate": 0.0001,
+      "loss": 0.0354,
+      "step": 47360
+    },
+    {
+      "epoch": 134.57386363636363,
+      "grad_norm": 0.7090975046157837,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 47370
+    },
+    {
+      "epoch": 134.60227272727272,
+      "grad_norm": 0.9196219444274902,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 47380
+    },
+    {
+      "epoch": 134.6306818181818,
+      "grad_norm": 0.9817768335342407,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 47390
+    },
+    {
+      "epoch": 134.6590909090909,
+      "grad_norm": 0.8549873232841492,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 47400
+    },
+    {
+      "epoch": 134.6875,
+      "grad_norm": 0.8343051671981812,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 47410
+    },
+    {
+      "epoch": 134.7159090909091,
+      "grad_norm": 0.6651965975761414,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 47420
+    },
+    {
+      "epoch": 134.7443181818182,
+      "grad_norm": 0.8093419075012207,
+      "learning_rate": 0.0001,
+      "loss": 0.0365,
+      "step": 47430
+    },
+    {
+      "epoch": 134.77272727272728,
+      "grad_norm": 0.7498792409896851,
+      "learning_rate": 0.0001,
+      "loss": 0.0349,
+      "step": 47440
+    },
+    {
+      "epoch": 134.80113636363637,
+      "grad_norm": 1.3343470096588135,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 47450
+    },
+    {
+      "epoch": 134.82954545454547,
+      "grad_norm": 1.099355936050415,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 47460
+    },
+    {
+      "epoch": 134.85795454545453,
+      "grad_norm": 1.077439785003662,
+      "learning_rate": 0.0001,
+      "loss": 0.0355,
+      "step": 47470
+    },
+    {
+      "epoch": 134.88636363636363,
+      "grad_norm": 0.9508234262466431,
+      "learning_rate": 0.0001,
+      "loss": 0.0356,
+      "step": 47480
+    },
+    {
+      "epoch": 134.91477272727272,
+      "grad_norm": 0.8098722696304321,
+      "learning_rate": 0.0001,
+      "loss": 0.0346,
+      "step": 47490
+    },
+    {
+      "epoch": 134.9431818181818,
+      "grad_norm": 0.6014454960823059,
+      "learning_rate": 0.0001,
+      "loss": 0.0367,
+      "step": 47500
+    },
+    {
+      "epoch": 134.9715909090909,
+      "grad_norm": 0.7299231886863708,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 47510
+    },
+    {
+      "epoch": 135.0,
+      "grad_norm": 0.6908378005027771,
+      "learning_rate": 0.0001,
+      "loss": 0.0352,
+      "step": 47520
+    },
+    {
+      "epoch": 135.0284090909091,
+      "grad_norm": 0.7811752557754517,
+      "learning_rate": 0.0001,
+      "loss": 0.0348,
+      "step": 47530
+    },
+    {
+      "epoch": 135.0568181818182,
+      "grad_norm": 0.8679901957511902,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 47540
+    },
+    {
+      "epoch": 135.08522727272728,
+      "grad_norm": 0.9140509963035583,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 47550
+    },
+    {
+      "epoch": 135.11363636363637,
+      "grad_norm": 0.7968837022781372,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 47560
+    },
+    {
+      "epoch": 135.14204545454547,
+      "grad_norm": 0.7876217365264893,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 47570
+    },
+    {
+      "epoch": 135.17045454545453,
+      "grad_norm": 0.5965085625648499,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 47580
+    },
+    {
+      "epoch": 135.19886363636363,
+      "grad_norm": 0.5109805464744568,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 47590
+    },
+    {
+      "epoch": 135.22727272727272,
+      "grad_norm": 0.5189054012298584,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 47600
+    },
+    {
+      "epoch": 135.2556818181818,
+      "grad_norm": 0.5483865737915039,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 47610
+    },
+    {
+      "epoch": 135.2840909090909,
+      "grad_norm": 0.8424668312072754,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 47620
+    },
+    {
+      "epoch": 135.3125,
+      "grad_norm": 1.2137010097503662,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 47630
+    },
+    {
+      "epoch": 135.3409090909091,
+      "grad_norm": 0.8658679127693176,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 47640
+    },
+    {
+      "epoch": 135.3693181818182,
+      "grad_norm": 0.8400396704673767,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 47650
+    },
+    {
+      "epoch": 135.39772727272728,
+      "grad_norm": 0.6896227598190308,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 47660
+    },
+    {
+      "epoch": 135.42613636363637,
+      "grad_norm": 0.8099536895751953,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 47670
+    },
+    {
+      "epoch": 135.45454545454547,
+      "grad_norm": 1.0839929580688477,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 47680
+    },
+    {
+      "epoch": 135.48295454545453,
+      "grad_norm": 0.9490888714790344,
+      "learning_rate": 0.0001,
+      "loss": 0.0359,
+      "step": 47690
+    },
+    {
+      "epoch": 135.51136363636363,
+      "grad_norm": 0.9408382177352905,
+      "learning_rate": 0.0001,
+      "loss": 0.0366,
+      "step": 47700
+    },
+    {
+      "epoch": 135.53977272727272,
+      "grad_norm": 0.6724865436553955,
+      "learning_rate": 0.0001,
+      "loss": 0.037,
+      "step": 47710
+    },
+    {
+      "epoch": 135.5681818181818,
+      "grad_norm": 0.889179527759552,
+      "learning_rate": 0.0001,
+      "loss": 0.0378,
+      "step": 47720
+    },
+    {
+      "epoch": 135.5965909090909,
+      "grad_norm": 1.22073495388031,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 47730
+    },
+    {
+      "epoch": 135.625,
+      "grad_norm": 1.0256925821304321,
+      "learning_rate": 0.0001,
+      "loss": 0.035,
+      "step": 47740
+    },
+    {
+      "epoch": 135.6534090909091,
+      "grad_norm": 0.9867517352104187,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 47750
+    },
+    {
+      "epoch": 135.6818181818182,
+      "grad_norm": 0.9394050240516663,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 47760
+    },
+    {
+      "epoch": 135.71022727272728,
+      "grad_norm": 0.7679027318954468,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 47770
+    },
+    {
+      "epoch": 135.73863636363637,
+      "grad_norm": 0.6860771775245667,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 47780
+    },
+    {
+      "epoch": 135.76704545454547,
+      "grad_norm": 0.6097438335418701,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 47790
+    },
+    {
+      "epoch": 135.79545454545453,
+      "grad_norm": 0.8389930129051208,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 47800
+    },
+    {
+      "epoch": 135.82386363636363,
+      "grad_norm": 0.7888689637184143,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 47810
+    },
+    {
+      "epoch": 135.85227272727272,
+      "grad_norm": 0.9129384756088257,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 47820
+    },
+    {
+      "epoch": 135.8806818181818,
+      "grad_norm": 1.0235633850097656,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 47830
+    },
+    {
+      "epoch": 135.9090909090909,
+      "grad_norm": 0.7481328845024109,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 47840
+    },
+    {
+      "epoch": 135.9375,
+      "grad_norm": 0.5055482387542725,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 47850
+    },
+    {
+      "epoch": 135.9659090909091,
+      "grad_norm": 0.5696248412132263,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 47860
+    },
+    {
+      "epoch": 135.9943181818182,
+      "grad_norm": 0.7397261261940002,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 47870
+    },
+    {
+      "epoch": 136.02272727272728,
+      "grad_norm": 0.651668906211853,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 47880
+    },
+    {
+      "epoch": 136.05113636363637,
+      "grad_norm": 0.5386239886283875,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 47890
+    },
+    {
+      "epoch": 136.07954545454547,
+      "grad_norm": 0.4532436430454254,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 47900
+    },
+    {
+      "epoch": 136.10795454545453,
+      "grad_norm": 0.5337704420089722,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 47910
+    },
+    {
+      "epoch": 136.13636363636363,
+      "grad_norm": 0.9169813394546509,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 47920
+    },
+    {
+      "epoch": 136.16477272727272,
+      "grad_norm": 0.7223308682441711,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 47930
+    },
+    {
+      "epoch": 136.1931818181818,
+      "grad_norm": 0.8555115461349487,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 47940
+    },
+    {
+      "epoch": 136.2215909090909,
+      "grad_norm": 0.537327229976654,
+      "learning_rate": 0.0001,
+      "loss": 0.0334,
+      "step": 47950
+    },
+    {
+      "epoch": 136.25,
+      "grad_norm": 0.8894582986831665,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 47960
+    },
+    {
+      "epoch": 136.2784090909091,
+      "grad_norm": 0.555550754070282,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 47970
+    },
+    {
+      "epoch": 136.3068181818182,
+      "grad_norm": 0.6033909916877747,
+      "learning_rate": 0.0001,
+      "loss": 0.0339,
+      "step": 47980
+    },
+    {
+      "epoch": 136.33522727272728,
+      "grad_norm": 0.6103994250297546,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 47990
+    },
+    {
+      "epoch": 136.36363636363637,
+      "grad_norm": 0.6190768480300903,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 48000
+    },
+    {
+      "epoch": 136.39204545454547,
+      "grad_norm": 0.7021390795707703,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 48010
+    },
+    {
+      "epoch": 136.42045454545453,
+      "grad_norm": 0.5402888059616089,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 48020
+    },
+    {
+      "epoch": 136.44886363636363,
+      "grad_norm": 0.7875863909721375,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 48030
+    },
+    {
+      "epoch": 136.47727272727272,
+      "grad_norm": 0.5794070959091187,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 48040
+    },
+    {
+      "epoch": 136.5056818181818,
+      "grad_norm": 0.7281160950660706,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 48050
+    },
+    {
+      "epoch": 136.5340909090909,
+      "grad_norm": 0.6087822318077087,
+      "learning_rate": 0.0001,
+      "loss": 0.0342,
+      "step": 48060
+    },
+    {
+      "epoch": 136.5625,
+      "grad_norm": 0.7562154531478882,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 48070
+    },
+    {
+      "epoch": 136.5909090909091,
+      "grad_norm": 0.7490687370300293,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 48080
+    },
+    {
+      "epoch": 136.6193181818182,
+      "grad_norm": 0.6700184345245361,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 48090
+    },
+    {
+      "epoch": 136.64772727272728,
+      "grad_norm": 0.7165958881378174,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 48100
+    },
+    {
+      "epoch": 136.67613636363637,
+      "grad_norm": 0.5619990229606628,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 48110
+    },
+    {
+      "epoch": 136.70454545454547,
+      "grad_norm": 0.6590405106544495,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 48120
+    },
+    {
+      "epoch": 136.73295454545453,
+      "grad_norm": 0.8965096473693848,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 48130
+    },
+    {
+      "epoch": 136.76136363636363,
+      "grad_norm": 0.932625949382782,
+      "learning_rate": 0.0001,
+      "loss": 0.0332,
+      "step": 48140
+    },
+    {
+      "epoch": 136.78977272727272,
+      "grad_norm": 0.806952714920044,
+      "learning_rate": 0.0001,
+      "loss": 0.0345,
+      "step": 48150
+    },
+    {
+      "epoch": 136.8181818181818,
+      "grad_norm": 0.718571126461029,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 48160
+    },
+    {
+      "epoch": 136.8465909090909,
+      "grad_norm": 0.6784353256225586,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 48170
+    },
+    {
+      "epoch": 136.875,
+      "grad_norm": 0.672789454460144,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 48180
+    },
+    {
+      "epoch": 136.9034090909091,
+      "grad_norm": 0.7343295216560364,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 48190
+    },
+    {
+      "epoch": 136.9318181818182,
+      "grad_norm": 0.6511361598968506,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 48200
+    },
+    {
+      "epoch": 136.96022727272728,
+      "grad_norm": 0.7251712083816528,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 48210
+    },
+    {
+      "epoch": 136.98863636363637,
+      "grad_norm": 0.8382689356803894,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 48220
+    },
+    {
+      "epoch": 137.01704545454547,
+      "grad_norm": 1.2256827354431152,
+      "learning_rate": 0.0001,
+      "loss": 0.0337,
+      "step": 48230
+    },
+    {
+      "epoch": 137.04545454545453,
+      "grad_norm": 0.815129816532135,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 48240
+    },
+    {
+      "epoch": 137.07386363636363,
+      "grad_norm": 1.3422629833221436,
+      "learning_rate": 0.0001,
+      "loss": 0.0351,
+      "step": 48250
+    },
+    {
+      "epoch": 137.10227272727272,
+      "grad_norm": 1.2612097263336182,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 48260
+    },
+    {
+      "epoch": 137.1306818181818,
+      "grad_norm": 1.3112272024154663,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 48270
+    },
+    {
+      "epoch": 137.1590909090909,
+      "grad_norm": 0.8872440457344055,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 48280
+    },
+    {
+      "epoch": 137.1875,
+      "grad_norm": 0.9705610871315002,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 48290
+    },
+    {
+      "epoch": 137.2159090909091,
+      "grad_norm": 1.0388199090957642,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 48300
+    },
+    {
+      "epoch": 137.2443181818182,
+      "grad_norm": 0.8184165358543396,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 48310
+    },
+    {
+      "epoch": 137.27272727272728,
+      "grad_norm": 1.2038040161132812,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 48320
+    },
+    {
+      "epoch": 137.30113636363637,
+      "grad_norm": 0.8519647121429443,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 48330
+    },
+    {
+      "epoch": 137.32954545454547,
+      "grad_norm": 0.7184175252914429,
+      "learning_rate": 0.0001,
+      "loss": 0.0336,
+      "step": 48340
+    },
+    {
+      "epoch": 137.35795454545453,
+      "grad_norm": 0.9237968325614929,
+      "learning_rate": 0.0001,
+      "loss": 0.0338,
+      "step": 48350
+    },
+    {
+      "epoch": 137.38636363636363,
+      "grad_norm": 0.5992299914360046,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 48360
+    },
+    {
+      "epoch": 137.41477272727272,
+      "grad_norm": 0.914725124835968,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 48370
+    },
+    {
+      "epoch": 137.4431818181818,
+      "grad_norm": 0.8184134364128113,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 48380
+    },
+    {
+      "epoch": 137.4715909090909,
+      "grad_norm": 0.5038926601409912,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 48390
+    },
+    {
+      "epoch": 137.5,
+      "grad_norm": 0.6409310102462769,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 48400
+    },
+    {
+      "epoch": 137.5284090909091,
+      "grad_norm": 0.5178235173225403,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 48410
+    },
+    {
+      "epoch": 137.5568181818182,
+      "grad_norm": 0.7749777436256409,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 48420
+    },
+    {
+      "epoch": 137.58522727272728,
+      "grad_norm": 0.5888954401016235,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 48430
+    },
+    {
+      "epoch": 137.61363636363637,
+      "grad_norm": 0.631784200668335,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 48440
+    },
+    {
+      "epoch": 137.64204545454547,
+      "grad_norm": 0.6141875386238098,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 48450
+    },
+    {
+      "epoch": 137.67045454545453,
+      "grad_norm": 0.7147916555404663,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 48460
+    },
+    {
+      "epoch": 137.69886363636363,
+      "grad_norm": 0.6216191649436951,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 48470
+    },
+    {
+      "epoch": 137.72727272727272,
+      "grad_norm": 0.9274272918701172,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 48480
+    },
+    {
+      "epoch": 137.7556818181818,
+      "grad_norm": 1.197992205619812,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 48490
+    },
+    {
+      "epoch": 137.7840909090909,
+      "grad_norm": 1.0352435111999512,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 48500
+    },
+    {
+      "epoch": 137.8125,
+      "grad_norm": 0.9775570631027222,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 48510
+    },
+    {
+      "epoch": 137.8409090909091,
+      "grad_norm": 1.0361013412475586,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 48520
+    },
+    {
+      "epoch": 137.8693181818182,
+      "grad_norm": 0.9069687724113464,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 48530
+    },
+    {
+      "epoch": 137.89772727272728,
+      "grad_norm": 0.6086276173591614,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 48540
+    },
+    {
+      "epoch": 137.92613636363637,
+      "grad_norm": 0.7118426561355591,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 48550
+    },
+    {
+      "epoch": 137.95454545454547,
+      "grad_norm": 0.9204250574111938,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 48560
+    },
+    {
+      "epoch": 137.98295454545453,
+      "grad_norm": 0.7215946316719055,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 48570
+    },
+    {
+      "epoch": 138.01136363636363,
+      "grad_norm": 0.7724602222442627,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 48580
+    },
+    {
+      "epoch": 138.03977272727272,
+      "grad_norm": 0.6827357411384583,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 48590
+    },
+    {
+      "epoch": 138.0681818181818,
+      "grad_norm": 0.46647927165031433,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 48600
+    },
+    {
+      "epoch": 138.0965909090909,
+      "grad_norm": 0.5820374488830566,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 48610
+    },
+    {
+      "epoch": 138.125,
+      "grad_norm": 0.9222242832183838,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 48620
+    },
+    {
+      "epoch": 138.1534090909091,
+      "grad_norm": 1.222086787223816,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 48630
+    },
+    {
+      "epoch": 138.1818181818182,
+      "grad_norm": 0.7875241637229919,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 48640
+    },
+    {
+      "epoch": 138.21022727272728,
+      "grad_norm": 0.7012037634849548,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 48650
+    },
+    {
+      "epoch": 138.23863636363637,
+      "grad_norm": 1.2003134489059448,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 48660
+    },
+    {
+      "epoch": 138.26704545454547,
+      "grad_norm": 0.6346859335899353,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 48670
+    },
+    {
+      "epoch": 138.29545454545453,
+      "grad_norm": 0.5016667246818542,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 48680
+    },
+    {
+      "epoch": 138.32386363636363,
+      "grad_norm": 0.5193179845809937,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 48690
+    },
+    {
+      "epoch": 138.35227272727272,
+      "grad_norm": 0.5191895365715027,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 48700
+    },
+    {
+      "epoch": 138.3806818181818,
+      "grad_norm": 0.4985046088695526,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 48710
+    },
+    {
+      "epoch": 138.4090909090909,
+      "grad_norm": 0.6570383310317993,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 48720
+    },
+    {
+      "epoch": 138.4375,
+      "grad_norm": 0.5687001347541809,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 48730
+    },
+    {
+      "epoch": 138.4659090909091,
+      "grad_norm": 0.6005369424819946,
+      "learning_rate": 0.0001,
+      "loss": 0.0328,
+      "step": 48740
+    },
+    {
+      "epoch": 138.4943181818182,
+      "grad_norm": 0.5225914716720581,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 48750
+    },
+    {
+      "epoch": 138.52272727272728,
+      "grad_norm": 0.6342768669128418,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 48760
+    },
+    {
+      "epoch": 138.55113636363637,
+      "grad_norm": 0.5711255669593811,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 48770
+    },
+    {
+      "epoch": 138.57954545454547,
+      "grad_norm": 0.6120138764381409,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 48780
+    },
+    {
+      "epoch": 138.60795454545453,
+      "grad_norm": 0.7916771769523621,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 48790
+    },
+    {
+      "epoch": 138.63636363636363,
+      "grad_norm": 0.9732442498207092,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 48800
+    },
+    {
+      "epoch": 138.66477272727272,
+      "grad_norm": 1.2003506422042847,
+      "learning_rate": 0.0001,
+      "loss": 0.0341,
+      "step": 48810
+    },
+    {
+      "epoch": 138.6931818181818,
+      "grad_norm": 1.2463781833648682,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 48820
+    },
+    {
+      "epoch": 138.7215909090909,
+      "grad_norm": 0.8527126908302307,
+      "learning_rate": 0.0001,
+      "loss": 0.0329,
+      "step": 48830
+    },
+    {
+      "epoch": 138.75,
+      "grad_norm": 0.9467070698738098,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 48840
+    },
+    {
+      "epoch": 138.7784090909091,
+      "grad_norm": 1.0931123495101929,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 48850
+    },
+    {
+      "epoch": 138.8068181818182,
+      "grad_norm": 1.0193461179733276,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 48860
+    },
+    {
+      "epoch": 138.83522727272728,
+      "grad_norm": 0.9290983080863953,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 48870
+    },
+    {
+      "epoch": 138.86363636363637,
+      "grad_norm": 0.7232368588447571,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 48880
+    },
+    {
+      "epoch": 138.89204545454547,
+      "grad_norm": 0.6270406246185303,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 48890
+    },
+    {
+      "epoch": 138.92045454545453,
+      "grad_norm": 0.8998165726661682,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 48900
+    },
+    {
+      "epoch": 138.94886363636363,
+      "grad_norm": 0.6000069379806519,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 48910
+    },
+    {
+      "epoch": 138.97727272727272,
+      "grad_norm": 0.45096200704574585,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 48920
+    },
+    {
+      "epoch": 139.0056818181818,
+      "grad_norm": 0.6368238925933838,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 48930
+    },
+    {
+      "epoch": 139.0340909090909,
+      "grad_norm": 0.8111124038696289,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 48940
+    },
+    {
+      "epoch": 139.0625,
+      "grad_norm": 0.5319854021072388,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 48950
+    },
+    {
+      "epoch": 139.0909090909091,
+      "grad_norm": 0.6842014193534851,
+      "learning_rate": 0.0001,
+      "loss": 0.0335,
+      "step": 48960
+    },
+    {
+      "epoch": 139.1193181818182,
+      "grad_norm": 0.5523969531059265,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 48970
+    },
+    {
+      "epoch": 139.14772727272728,
+      "grad_norm": 0.6312474608421326,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 48980
+    },
+    {
+      "epoch": 139.17613636363637,
+      "grad_norm": 0.6765289902687073,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 48990
+    },
+    {
+      "epoch": 139.20454545454547,
+      "grad_norm": 0.7405977845191956,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 49000
+    },
+    {
+      "epoch": 139.23295454545453,
+      "grad_norm": 0.9844215512275696,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 49010
+    },
+    {
+      "epoch": 139.26136363636363,
+      "grad_norm": 0.824510395526886,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49020
+    },
+    {
+      "epoch": 139.28977272727272,
+      "grad_norm": 0.6249207258224487,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 49030
+    },
+    {
+      "epoch": 139.3181818181818,
+      "grad_norm": 0.5916060209274292,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 49040
+    },
+    {
+      "epoch": 139.3465909090909,
+      "grad_norm": 0.742417573928833,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 49050
+    },
+    {
+      "epoch": 139.375,
+      "grad_norm": 0.5674957036972046,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 49060
+    },
+    {
+      "epoch": 139.4034090909091,
+      "grad_norm": 0.5021371841430664,
+      "learning_rate": 0.0001,
+      "loss": 0.0309,
+      "step": 49070
+    },
+    {
+      "epoch": 139.4318181818182,
+      "grad_norm": 0.5639374256134033,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 49080
+    },
+    {
+      "epoch": 139.46022727272728,
+      "grad_norm": 0.6540923118591309,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49090
+    },
+    {
+      "epoch": 139.48863636363637,
+      "grad_norm": 0.7910658121109009,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 49100
+    },
+    {
+      "epoch": 139.51704545454547,
+      "grad_norm": 0.9734665155410767,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 49110
+    },
+    {
+      "epoch": 139.54545454545453,
+      "grad_norm": 0.6400315165519714,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 49120
+    },
+    {
+      "epoch": 139.57386363636363,
+      "grad_norm": 0.7147937417030334,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 49130
+    },
+    {
+      "epoch": 139.60227272727272,
+      "grad_norm": 0.7652814984321594,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 49140
+    },
+    {
+      "epoch": 139.6306818181818,
+      "grad_norm": 0.8998138308525085,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 49150
+    },
+    {
+      "epoch": 139.6590909090909,
+      "grad_norm": 0.6835416555404663,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 49160
+    },
+    {
+      "epoch": 139.6875,
+      "grad_norm": 0.6805562376976013,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 49170
+    },
+    {
+      "epoch": 139.7159090909091,
+      "grad_norm": 0.6778475642204285,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49180
+    },
+    {
+      "epoch": 139.7443181818182,
+      "grad_norm": 0.476952463388443,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 49190
+    },
+    {
+      "epoch": 139.77272727272728,
+      "grad_norm": 0.9565654397010803,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 49200
+    },
+    {
+      "epoch": 139.80113636363637,
+      "grad_norm": 0.770005464553833,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 49210
+    },
+    {
+      "epoch": 139.82954545454547,
+      "grad_norm": 0.8685285449028015,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 49220
+    },
+    {
+      "epoch": 139.85795454545453,
+      "grad_norm": 0.6290770769119263,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 49230
+    },
+    {
+      "epoch": 139.88636363636363,
+      "grad_norm": 0.7894107103347778,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 49240
+    },
+    {
+      "epoch": 139.91477272727272,
+      "grad_norm": 0.73750901222229,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 49250
+    },
+    {
+      "epoch": 139.9431818181818,
+      "grad_norm": 0.5105662941932678,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 49260
+    },
+    {
+      "epoch": 139.9715909090909,
+      "grad_norm": 0.5434335470199585,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 49270
+    },
+    {
+      "epoch": 140.0,
+      "grad_norm": 0.5327863097190857,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 49280
+    },
+    {
+      "epoch": 140.0284090909091,
+      "grad_norm": 0.4427546560764313,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 49290
+    },
+    {
+      "epoch": 140.0568181818182,
+      "grad_norm": 0.5727342963218689,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49300
+    },
+    {
+      "epoch": 140.08522727272728,
+      "grad_norm": 0.698470950126648,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49310
+    },
+    {
+      "epoch": 140.11363636363637,
+      "grad_norm": 0.5280812978744507,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 49320
+    },
+    {
+      "epoch": 140.14204545454547,
+      "grad_norm": 0.5219975709915161,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 49330
+    },
+    {
+      "epoch": 140.17045454545453,
+      "grad_norm": 0.603614866733551,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 49340
+    },
+    {
+      "epoch": 140.19886363636363,
+      "grad_norm": 0.48663216829299927,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49350
+    },
+    {
+      "epoch": 140.22727272727272,
+      "grad_norm": 0.6505529284477234,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 49360
+    },
+    {
+      "epoch": 140.2556818181818,
+      "grad_norm": 0.7080401182174683,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 49370
+    },
+    {
+      "epoch": 140.2840909090909,
+      "grad_norm": 0.49827688932418823,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 49380
+    },
+    {
+      "epoch": 140.3125,
+      "grad_norm": 0.6238806843757629,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 49390
+    },
+    {
+      "epoch": 140.3409090909091,
+      "grad_norm": 0.5958791971206665,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 49400
+    },
+    {
+      "epoch": 140.3693181818182,
+      "grad_norm": 0.5299167633056641,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 49410
+    },
+    {
+      "epoch": 140.39772727272728,
+      "grad_norm": 0.534270703792572,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 49420
+    },
+    {
+      "epoch": 140.42613636363637,
+      "grad_norm": 0.4572816789150238,
+      "learning_rate": 0.0001,
+      "loss": 0.0325,
+      "step": 49430
+    },
+    {
+      "epoch": 140.45454545454547,
+      "grad_norm": 0.7355735301971436,
+      "learning_rate": 0.0001,
+      "loss": 0.033,
+      "step": 49440
+    },
+    {
+      "epoch": 140.48295454545453,
+      "grad_norm": 0.8774271607398987,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49450
+    },
+    {
+      "epoch": 140.51136363636363,
+      "grad_norm": 0.5380182266235352,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 49460
+    },
+    {
+      "epoch": 140.53977272727272,
+      "grad_norm": 0.6516682505607605,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 49470
+    },
+    {
+      "epoch": 140.5681818181818,
+      "grad_norm": 0.4943990111351013,
+      "learning_rate": 0.0001,
+      "loss": 0.0301,
+      "step": 49480
+    },
+    {
+      "epoch": 140.5965909090909,
+      "grad_norm": 0.5519253611564636,
+      "learning_rate": 0.0001,
+      "loss": 0.0309,
+      "step": 49490
+    },
+    {
+      "epoch": 140.625,
+      "grad_norm": 0.6039503216743469,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 49500
+    },
+    {
+      "epoch": 140.6534090909091,
+      "grad_norm": 0.754988431930542,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 49510
+    },
+    {
+      "epoch": 140.6818181818182,
+      "grad_norm": 0.6376588344573975,
+      "learning_rate": 0.0001,
+      "loss": 0.0324,
+      "step": 49520
+    },
+    {
+      "epoch": 140.71022727272728,
+      "grad_norm": 0.7755230069160461,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 49530
+    },
+    {
+      "epoch": 140.73863636363637,
+      "grad_norm": 0.6854098439216614,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 49540
+    },
+    {
+      "epoch": 140.76704545454547,
+      "grad_norm": 0.5473119020462036,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 49550
+    },
+    {
+      "epoch": 140.79545454545453,
+      "grad_norm": 0.659775972366333,
+      "learning_rate": 0.0001,
+      "loss": 0.0306,
+      "step": 49560
+    },
+    {
+      "epoch": 140.82386363636363,
+      "grad_norm": 0.7851589918136597,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 49570
+    },
+    {
+      "epoch": 140.85227272727272,
+      "grad_norm": 0.4822283983230591,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 49580
+    },
+    {
+      "epoch": 140.8806818181818,
+      "grad_norm": 0.6676194667816162,
+      "learning_rate": 0.0001,
+      "loss": 0.0311,
+      "step": 49590
+    },
+    {
+      "epoch": 140.9090909090909,
+      "grad_norm": 0.4834653437137604,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 49600
+    },
+    {
+      "epoch": 140.9375,
+      "grad_norm": 0.6345698833465576,
+      "learning_rate": 0.0001,
+      "loss": 0.03,
+      "step": 49610
+    },
+    {
+      "epoch": 140.9659090909091,
+      "grad_norm": 0.45117560029029846,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 49620
+    },
+    {
+      "epoch": 140.9943181818182,
+      "grad_norm": 0.5543921589851379,
+      "learning_rate": 0.0001,
+      "loss": 0.0317,
+      "step": 49630
+    },
+    {
+      "epoch": 141.02272727272728,
+      "grad_norm": 0.5042457580566406,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 49640
+    },
+    {
+      "epoch": 141.05113636363637,
+      "grad_norm": 0.47779983282089233,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 49650
+    },
+    {
+      "epoch": 141.07954545454547,
+      "grad_norm": 0.641806423664093,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 49660
+    },
+    {
+      "epoch": 141.10795454545453,
+      "grad_norm": 0.7229103446006775,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 49670
+    },
+    {
+      "epoch": 141.13636363636363,
+      "grad_norm": 0.5515846014022827,
+      "learning_rate": 0.0001,
+      "loss": 0.0318,
+      "step": 49680
+    },
+    {
+      "epoch": 141.16477272727272,
+      "grad_norm": 0.6664286255836487,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 49690
+    },
+    {
+      "epoch": 141.1931818181818,
+      "grad_norm": 0.5435540080070496,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 49700
+    },
+    {
+      "epoch": 141.2215909090909,
+      "grad_norm": 0.5387634038925171,
+      "learning_rate": 0.0001,
+      "loss": 0.0327,
+      "step": 49710
+    },
+    {
+      "epoch": 141.25,
+      "grad_norm": 0.5133399963378906,
+      "learning_rate": 0.0001,
+      "loss": 0.0331,
+      "step": 49720
+    },
+    {
+      "epoch": 141.2784090909091,
+      "grad_norm": 0.7076446413993835,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 49730
+    },
+    {
+      "epoch": 141.3068181818182,
+      "grad_norm": 0.5965754985809326,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 49740
+    },
+    {
+      "epoch": 141.33522727272728,
+      "grad_norm": 0.5514310598373413,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 49750
+    },
+    {
+      "epoch": 141.36363636363637,
+      "grad_norm": 0.6107453107833862,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49760
+    },
+    {
+      "epoch": 141.39204545454547,
+      "grad_norm": 0.6507061719894409,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 49770
+    },
+    {
+      "epoch": 141.42045454545453,
+      "grad_norm": 0.4469587802886963,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 49780
+    },
+    {
+      "epoch": 141.44886363636363,
+      "grad_norm": 0.5173867344856262,
+      "learning_rate": 0.0001,
+      "loss": 0.0309,
+      "step": 49790
+    },
+    {
+      "epoch": 141.47727272727272,
+      "grad_norm": 0.4976654052734375,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 49800
+    },
+    {
+      "epoch": 141.5056818181818,
+      "grad_norm": 0.7025495171546936,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 49810
+    },
+    {
+      "epoch": 141.5340909090909,
+      "grad_norm": 0.5611472725868225,
+      "learning_rate": 0.0001,
+      "loss": 0.0323,
+      "step": 49820
+    },
+    {
+      "epoch": 141.5625,
+      "grad_norm": 0.5493874549865723,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 49830
+    },
+    {
+      "epoch": 141.5909090909091,
+      "grad_norm": 0.649082601070404,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 49840
+    },
+    {
+      "epoch": 141.6193181818182,
+      "grad_norm": 0.5634473562240601,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49850
+    },
+    {
+      "epoch": 141.64772727272728,
+      "grad_norm": 0.44207829236984253,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 49860
+    },
+    {
+      "epoch": 141.67613636363637,
+      "grad_norm": 1.1749712228775024,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 49870
+    },
+    {
+      "epoch": 141.70454545454547,
+      "grad_norm": 0.8790702223777771,
+      "learning_rate": 0.0001,
+      "loss": 0.0326,
+      "step": 49880
+    },
+    {
+      "epoch": 141.73295454545453,
+      "grad_norm": 1.1796921491622925,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 49890
+    },
+    {
+      "epoch": 141.76136363636363,
+      "grad_norm": 0.7493149638175964,
+      "learning_rate": 0.0001,
+      "loss": 0.0322,
+      "step": 49900
+    },
+    {
+      "epoch": 141.78977272727272,
+      "grad_norm": 0.6880519986152649,
+      "learning_rate": 0.0001,
+      "loss": 0.032,
+      "step": 49910
+    },
+    {
+      "epoch": 141.8181818181818,
+      "grad_norm": 0.7424251437187195,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 49920
+    },
+    {
+      "epoch": 141.8465909090909,
+      "grad_norm": 0.8121177554130554,
+      "learning_rate": 0.0001,
+      "loss": 0.0313,
+      "step": 49930
+    },
+    {
+      "epoch": 141.875,
+      "grad_norm": 0.7063887715339661,
+      "learning_rate": 0.0001,
+      "loss": 0.0307,
+      "step": 49940
+    },
+    {
+      "epoch": 141.9034090909091,
+      "grad_norm": 0.6800211071968079,
+      "learning_rate": 0.0001,
+      "loss": 0.0319,
+      "step": 49950
+    },
+    {
+      "epoch": 141.9318181818182,
+      "grad_norm": 0.800621509552002,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 49960
+    },
+    {
+      "epoch": 141.96022727272728,
+      "grad_norm": 0.7496196627616882,
+      "learning_rate": 0.0001,
+      "loss": 0.0314,
+      "step": 49970
+    },
+    {
+      "epoch": 141.98863636363637,
+      "grad_norm": 0.8301522135734558,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 49980
+    },
+    {
+      "epoch": 142.01704545454547,
+      "grad_norm": 1.3548810482025146,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 49990
+    },
+    {
+      "epoch": 142.04545454545453,
+      "grad_norm": 1.1022731065750122,
+      "learning_rate": 0.0001,
+      "loss": 0.0321,
+      "step": 50000
+    },
+    {
+      "epoch": 142.04545454545453,
+      "step": 50000,
+      "total_flos": 0.0,
+      "train_loss": 0.090032759501338,
+      "train_runtime": 22271.7421,
+      "train_samples_per_second": 287.36,
+      "train_steps_per_second": 2.245
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 50000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 143,
+  "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": 128,
+  "trial_name": null,
+  "trial_params": null
+}