{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.457236842105264,
  "eval_steps": 500,
  "global_step": 23000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004111842105263158,
      "grad_norm": 82.21414947509766,
      "learning_rate": 9.997532894736843e-06,
      "loss": 1.0101,
      "step": 10
    },
    {
      "epoch": 0.008223684210526315,
      "grad_norm": 79.48388671875,
      "learning_rate": 9.993832236842106e-06,
      "loss": 1.3237,
      "step": 20
    },
    {
      "epoch": 0.012335526315789474,
      "grad_norm": 33.75522232055664,
      "learning_rate": 9.989720394736842e-06,
      "loss": 0.596,
      "step": 30
    },
    {
      "epoch": 0.01644736842105263,
      "grad_norm": 0.831979513168335,
      "learning_rate": 9.98560855263158e-06,
      "loss": 0.8206,
      "step": 40
    },
    {
      "epoch": 0.02055921052631579,
      "grad_norm": 58.79596710205078,
      "learning_rate": 9.981496710526317e-06,
      "loss": 1.0928,
      "step": 50
    },
    {
      "epoch": 0.024671052631578948,
      "grad_norm": 51.91939926147461,
      "learning_rate": 9.977384868421054e-06,
      "loss": 0.7411,
      "step": 60
    },
    {
      "epoch": 0.028782894736842105,
      "grad_norm": 69.77727508544922,
      "learning_rate": 9.97327302631579e-06,
      "loss": 0.5594,
      "step": 70
    },
    {
      "epoch": 0.03289473684210526,
      "grad_norm": 69.18291473388672,
      "learning_rate": 9.969161184210527e-06,
      "loss": 1.1593,
      "step": 80
    },
    {
      "epoch": 0.03700657894736842,
      "grad_norm": 25.222484588623047,
      "learning_rate": 9.965049342105263e-06,
      "loss": 0.4795,
      "step": 90
    },
    {
      "epoch": 0.04111842105263158,
      "grad_norm": 65.4477767944336,
      "learning_rate": 9.9609375e-06,
      "loss": 0.7011,
      "step": 100
    },
    {
      "epoch": 0.04523026315789474,
      "grad_norm": 23.820554733276367,
      "learning_rate": 9.956825657894738e-06,
      "loss": 0.3759,
      "step": 110
    },
    {
      "epoch": 0.049342105263157895,
      "grad_norm": 126.01136016845703,
      "learning_rate": 9.952713815789474e-06,
      "loss": 1.2958,
      "step": 120
    },
    {
      "epoch": 0.05345394736842105,
      "grad_norm": 9.699002265930176,
      "learning_rate": 9.948601973684211e-06,
      "loss": 0.7123,
      "step": 130
    },
    {
      "epoch": 0.05756578947368421,
      "grad_norm": 13.736701965332031,
      "learning_rate": 9.944490131578947e-06,
      "loss": 1.2954,
      "step": 140
    },
    {
      "epoch": 0.061677631578947366,
      "grad_norm": 7.952981472015381,
      "learning_rate": 9.940378289473686e-06,
      "loss": 0.5001,
      "step": 150
    },
    {
      "epoch": 0.06578947368421052,
      "grad_norm": 70.3082504272461,
      "learning_rate": 9.936266447368422e-06,
      "loss": 0.909,
      "step": 160
    },
    {
      "epoch": 0.06990131578947369,
      "grad_norm": 51.05005645751953,
      "learning_rate": 9.93215460526316e-06,
      "loss": 0.9518,
      "step": 170
    },
    {
      "epoch": 0.07401315789473684,
      "grad_norm": 41.0015983581543,
      "learning_rate": 9.928042763157895e-06,
      "loss": 0.7172,
      "step": 180
    },
    {
      "epoch": 0.078125,
      "grad_norm": 51.35481262207031,
      "learning_rate": 9.923930921052633e-06,
      "loss": 1.106,
      "step": 190
    },
    {
      "epoch": 0.08223684210526316,
      "grad_norm": 0.127091646194458,
      "learning_rate": 9.919819078947368e-06,
      "loss": 0.3531,
      "step": 200
    },
    {
      "epoch": 0.08634868421052631,
      "grad_norm": 5.155969142913818,
      "learning_rate": 9.915707236842106e-06,
      "loss": 0.6066,
      "step": 210
    },
    {
      "epoch": 0.09046052631578948,
      "grad_norm": 23.830692291259766,
      "learning_rate": 9.911595394736843e-06,
      "loss": 0.6448,
      "step": 220
    },
    {
      "epoch": 0.09457236842105263,
      "grad_norm": 0.5406038761138916,
      "learning_rate": 9.90748355263158e-06,
      "loss": 0.7219,
      "step": 230
    },
    {
      "epoch": 0.09868421052631579,
      "grad_norm": 29.625507354736328,
      "learning_rate": 9.903371710526317e-06,
      "loss": 0.522,
      "step": 240
    },
    {
      "epoch": 0.10279605263157894,
      "grad_norm": 26.483999252319336,
      "learning_rate": 9.89967105263158e-06,
      "loss": 0.4394,
      "step": 250
    },
    {
      "epoch": 0.1069078947368421,
      "grad_norm": 56.8378791809082,
      "learning_rate": 9.895559210526317e-06,
      "loss": 0.8402,
      "step": 260
    },
    {
      "epoch": 0.11101973684210527,
      "grad_norm": 30.779541015625,
      "learning_rate": 9.891447368421053e-06,
      "loss": 0.639,
      "step": 270
    },
    {
      "epoch": 0.11513157894736842,
      "grad_norm": 0.07414458692073822,
      "learning_rate": 9.88733552631579e-06,
      "loss": 0.6098,
      "step": 280
    },
    {
      "epoch": 0.11924342105263158,
      "grad_norm": 0.05630555748939514,
      "learning_rate": 9.883223684210526e-06,
      "loss": 0.2677,
      "step": 290
    },
    {
      "epoch": 0.12335526315789473,
      "grad_norm": 12.017321586608887,
      "learning_rate": 9.879111842105264e-06,
      "loss": 0.1545,
      "step": 300
    },
    {
      "epoch": 0.12746710526315788,
      "grad_norm": 114.26238250732422,
      "learning_rate": 9.875000000000001e-06,
      "loss": 0.8309,
      "step": 310
    },
    {
      "epoch": 0.13157894736842105,
      "grad_norm": 58.30079650878906,
      "learning_rate": 9.870888157894739e-06,
      "loss": 0.5668,
      "step": 320
    },
    {
      "epoch": 0.1356907894736842,
      "grad_norm": 0.042588502168655396,
      "learning_rate": 9.866776315789474e-06,
      "loss": 0.6739,
      "step": 330
    },
    {
      "epoch": 0.13980263157894737,
      "grad_norm": 53.37122344970703,
      "learning_rate": 9.862664473684212e-06,
      "loss": 1.2924,
      "step": 340
    },
    {
      "epoch": 0.14391447368421054,
      "grad_norm": 66.1600341796875,
      "learning_rate": 9.858552631578948e-06,
      "loss": 0.9956,
      "step": 350
    },
    {
      "epoch": 0.14802631578947367,
      "grad_norm": 26.073034286499023,
      "learning_rate": 9.854440789473685e-06,
      "loss": 0.1696,
      "step": 360
    },
    {
      "epoch": 0.15213815789473684,
      "grad_norm": 86.73162078857422,
      "learning_rate": 9.85032894736842e-06,
      "loss": 0.7324,
      "step": 370
    },
    {
      "epoch": 0.15625,
      "grad_norm": 5.856919288635254,
      "learning_rate": 9.846217105263158e-06,
      "loss": 0.6496,
      "step": 380
    },
    {
      "epoch": 0.16036184210526316,
      "grad_norm": 0.010420837439596653,
      "learning_rate": 9.842105263157896e-06,
      "loss": 1.0035,
      "step": 390
    },
    {
      "epoch": 0.16447368421052633,
      "grad_norm": 4.881348609924316,
      "learning_rate": 9.837993421052633e-06,
      "loss": 0.1168,
      "step": 400
    },
    {
      "epoch": 0.16858552631578946,
      "grad_norm": 1.67402184009552,
      "learning_rate": 9.833881578947369e-06,
      "loss": 0.7795,
      "step": 410
    },
    {
      "epoch": 0.17269736842105263,
      "grad_norm": 28.87067222595215,
      "learning_rate": 9.829769736842106e-06,
      "loss": 0.661,
      "step": 420
    },
    {
      "epoch": 0.1768092105263158,
      "grad_norm": 59.759552001953125,
      "learning_rate": 9.825657894736844e-06,
      "loss": 0.7666,
      "step": 430
    },
    {
      "epoch": 0.18092105263157895,
      "grad_norm": 33.859188079833984,
      "learning_rate": 9.82154605263158e-06,
      "loss": 0.3961,
      "step": 440
    },
    {
      "epoch": 0.18503289473684212,
      "grad_norm": 56.692955017089844,
      "learning_rate": 9.817434210526317e-06,
      "loss": 0.7028,
      "step": 450
    },
    {
      "epoch": 0.18914473684210525,
      "grad_norm": 74.9236068725586,
      "learning_rate": 9.813322368421053e-06,
      "loss": 0.6201,
      "step": 460
    },
    {
      "epoch": 0.19325657894736842,
      "grad_norm": 11.421184539794922,
      "learning_rate": 9.80921052631579e-06,
      "loss": 0.6496,
      "step": 470
    },
    {
      "epoch": 0.19736842105263158,
      "grad_norm": 64.66961669921875,
      "learning_rate": 9.805098684210526e-06,
      "loss": 0.5065,
      "step": 480
    },
    {
      "epoch": 0.20148026315789475,
      "grad_norm": 48.3864860534668,
      "learning_rate": 9.800986842105264e-06,
      "loss": 0.5425,
      "step": 490
    },
    {
      "epoch": 0.20559210526315788,
      "grad_norm": 21.042997360229492,
      "learning_rate": 9.796875000000001e-06,
      "loss": 0.6503,
      "step": 500
    },
    {
      "epoch": 0.20970394736842105,
      "grad_norm": 13.386051177978516,
      "learning_rate": 9.792763157894738e-06,
      "loss": 0.411,
      "step": 510
    },
    {
      "epoch": 0.2138157894736842,
      "grad_norm": 0.02692234143614769,
      "learning_rate": 9.788651315789474e-06,
      "loss": 0.8217,
      "step": 520
    },
    {
      "epoch": 0.21792763157894737,
      "grad_norm": 0.34673550724983215,
      "learning_rate": 9.784539473684212e-06,
      "loss": 0.6008,
      "step": 530
    },
    {
      "epoch": 0.22203947368421054,
      "grad_norm": 67.46952056884766,
      "learning_rate": 9.780427631578947e-06,
      "loss": 0.7634,
      "step": 540
    },
    {
      "epoch": 0.22615131578947367,
      "grad_norm": 1.0119798183441162,
      "learning_rate": 9.776315789473685e-06,
      "loss": 0.3242,
      "step": 550
    },
    {
      "epoch": 0.23026315789473684,
      "grad_norm": 31.881614685058594,
      "learning_rate": 9.772203947368422e-06,
      "loss": 0.8713,
      "step": 560
    },
    {
      "epoch": 0.234375,
      "grad_norm": 0.031223086640238762,
      "learning_rate": 9.768092105263158e-06,
      "loss": 0.4535,
      "step": 570
    },
    {
      "epoch": 0.23848684210526316,
      "grad_norm": 3.292686700820923,
      "learning_rate": 9.763980263157896e-06,
      "loss": 0.3269,
      "step": 580
    },
    {
      "epoch": 0.24259868421052633,
      "grad_norm": 44.72856521606445,
      "learning_rate": 9.759868421052633e-06,
      "loss": 0.6943,
      "step": 590
    },
    {
      "epoch": 0.24671052631578946,
      "grad_norm": 2.3683433532714844,
      "learning_rate": 9.75575657894737e-06,
      "loss": 0.4183,
      "step": 600
    },
    {
      "epoch": 0.25082236842105265,
      "grad_norm": 63.95539093017578,
      "learning_rate": 9.751644736842106e-06,
      "loss": 0.5693,
      "step": 610
    },
    {
      "epoch": 0.25493421052631576,
      "grad_norm": 60.73301315307617,
      "learning_rate": 9.747532894736844e-06,
      "loss": 1.1358,
      "step": 620
    },
    {
      "epoch": 0.2590460526315789,
      "grad_norm": 56.232845306396484,
      "learning_rate": 9.74342105263158e-06,
      "loss": 0.4148,
      "step": 630
    },
    {
      "epoch": 0.2631578947368421,
      "grad_norm": 39.33661651611328,
      "learning_rate": 9.739309210526317e-06,
      "loss": 0.6761,
      "step": 640
    },
    {
      "epoch": 0.26726973684210525,
      "grad_norm": 16.530174255371094,
      "learning_rate": 9.735197368421053e-06,
      "loss": 1.1621,
      "step": 650
    },
    {
      "epoch": 0.2713815789473684,
      "grad_norm": 8.869096755981445,
      "learning_rate": 9.73108552631579e-06,
      "loss": 0.6395,
      "step": 660
    },
    {
      "epoch": 0.2754934210526316,
      "grad_norm": 48.607025146484375,
      "learning_rate": 9.726973684210526e-06,
      "loss": 0.628,
      "step": 670
    },
    {
      "epoch": 0.27960526315789475,
      "grad_norm": 2.1920928955078125,
      "learning_rate": 9.722861842105263e-06,
      "loss": 0.572,
      "step": 680
    },
    {
      "epoch": 0.2837171052631579,
      "grad_norm": 1.4220335483551025,
      "learning_rate": 9.71875e-06,
      "loss": 0.2606,
      "step": 690
    },
    {
      "epoch": 0.2878289473684211,
      "grad_norm": 8.437749862670898,
      "learning_rate": 9.714638157894738e-06,
      "loss": 0.3371,
      "step": 700
    },
    {
      "epoch": 0.29194078947368424,
      "grad_norm": 1.182974100112915,
      "learning_rate": 9.710526315789474e-06,
      "loss": 0.9404,
      "step": 710
    },
    {
      "epoch": 0.29605263157894735,
      "grad_norm": 17.592485427856445,
      "learning_rate": 9.706414473684211e-06,
      "loss": 0.891,
      "step": 720
    },
    {
      "epoch": 0.3001644736842105,
      "grad_norm": 0.006371844094246626,
      "learning_rate": 9.702302631578949e-06,
      "loss": 0.5336,
      "step": 730
    },
    {
      "epoch": 0.3042763157894737,
      "grad_norm": 2.269035577774048,
      "learning_rate": 9.698190789473685e-06,
      "loss": 0.4996,
      "step": 740
    },
    {
      "epoch": 0.30838815789473684,
      "grad_norm": 3.4349544048309326,
      "learning_rate": 9.694078947368422e-06,
      "loss": 0.7136,
      "step": 750
    },
    {
      "epoch": 0.3125,
      "grad_norm": 31.934001922607422,
      "learning_rate": 9.689967105263158e-06,
      "loss": 0.6703,
      "step": 760
    },
    {
      "epoch": 0.31661184210526316,
      "grad_norm": 31.40448760986328,
      "learning_rate": 9.685855263157895e-06,
      "loss": 0.7493,
      "step": 770
    },
    {
      "epoch": 0.3207236842105263,
      "grad_norm": 0.03109549917280674,
      "learning_rate": 9.681743421052633e-06,
      "loss": 0.6885,
      "step": 780
    },
    {
      "epoch": 0.3248355263157895,
      "grad_norm": 62.230751037597656,
      "learning_rate": 9.67763157894737e-06,
      "loss": 0.4497,
      "step": 790
    },
    {
      "epoch": 0.32894736842105265,
      "grad_norm": 83.58551025390625,
      "learning_rate": 9.673519736842106e-06,
      "loss": 0.5014,
      "step": 800
    },
    {
      "epoch": 0.33305921052631576,
      "grad_norm": 1.6478629112243652,
      "learning_rate": 9.669407894736843e-06,
      "loss": 0.3641,
      "step": 810
    },
    {
      "epoch": 0.3371710526315789,
      "grad_norm": 0.02929232269525528,
      "learning_rate": 9.66529605263158e-06,
      "loss": 0.5607,
      "step": 820
    },
    {
      "epoch": 0.3412828947368421,
      "grad_norm": 32.008182525634766,
      "learning_rate": 9.661184210526317e-06,
      "loss": 0.4397,
      "step": 830
    },
    {
      "epoch": 0.34539473684210525,
      "grad_norm": 25.76505470275879,
      "learning_rate": 9.657072368421052e-06,
      "loss": 0.9786,
      "step": 840
    },
    {
      "epoch": 0.3495065789473684,
      "grad_norm": 0.04005062207579613,
      "learning_rate": 9.65296052631579e-06,
      "loss": 0.284,
      "step": 850
    },
    {
      "epoch": 0.3536184210526316,
      "grad_norm": 10.596315383911133,
      "learning_rate": 9.648848684210527e-06,
      "loss": 0.9688,
      "step": 860
    },
    {
      "epoch": 0.35773026315789475,
      "grad_norm": 46.72496795654297,
      "learning_rate": 9.644736842105263e-06,
      "loss": 0.594,
      "step": 870
    },
    {
      "epoch": 0.3618421052631579,
      "grad_norm": 43.54978942871094,
      "learning_rate": 9.640625e-06,
      "loss": 0.6063,
      "step": 880
    },
    {
      "epoch": 0.3659539473684211,
      "grad_norm": 2.5766873359680176,
      "learning_rate": 9.636513157894738e-06,
      "loss": 0.3333,
      "step": 890
    },
    {
      "epoch": 0.37006578947368424,
      "grad_norm": 58.31666564941406,
      "learning_rate": 9.632812500000001e-06,
      "loss": 0.8034,
      "step": 900
    },
    {
      "epoch": 0.37417763157894735,
      "grad_norm": 8.071229934692383,
      "learning_rate": 9.628700657894737e-06,
      "loss": 0.6695,
      "step": 910
    },
    {
      "epoch": 0.3782894736842105,
      "grad_norm": 24.1544132232666,
      "learning_rate": 9.624588815789474e-06,
      "loss": 0.6279,
      "step": 920
    },
    {
      "epoch": 0.3824013157894737,
      "grad_norm": 44.78126525878906,
      "learning_rate": 9.620476973684212e-06,
      "loss": 0.5958,
      "step": 930
    },
    {
      "epoch": 0.38651315789473684,
      "grad_norm": 0.43892815709114075,
      "learning_rate": 9.61636513157895e-06,
      "loss": 1.0484,
      "step": 940
    },
    {
      "epoch": 0.390625,
      "grad_norm": 59.37925720214844,
      "learning_rate": 9.612253289473685e-06,
      "loss": 0.9802,
      "step": 950
    },
    {
      "epoch": 0.39473684210526316,
      "grad_norm": 31.142057418823242,
      "learning_rate": 9.608141447368423e-06,
      "loss": 0.339,
      "step": 960
    },
    {
      "epoch": 0.3988486842105263,
      "grad_norm": 40.292205810546875,
      "learning_rate": 9.604029605263158e-06,
      "loss": 0.6531,
      "step": 970
    },
    {
      "epoch": 0.4029605263157895,
      "grad_norm": 38.27806854248047,
      "learning_rate": 9.599917763157896e-06,
      "loss": 0.7292,
      "step": 980
    },
    {
      "epoch": 0.40707236842105265,
      "grad_norm": 8.70391845703125,
      "learning_rate": 9.595805921052632e-06,
      "loss": 0.5728,
      "step": 990
    },
    {
      "epoch": 0.41118421052631576,
      "grad_norm": 26.568729400634766,
      "learning_rate": 9.591694078947369e-06,
      "loss": 0.6767,
      "step": 1000
    },
    {
      "epoch": 0.4152960526315789,
      "grad_norm": 54.55010986328125,
      "learning_rate": 9.587582236842106e-06,
      "loss": 0.7275,
      "step": 1010
    },
    {
      "epoch": 0.4194078947368421,
      "grad_norm": 55.86238098144531,
      "learning_rate": 9.583470394736842e-06,
      "loss": 0.6087,
      "step": 1020
    },
    {
      "epoch": 0.42351973684210525,
      "grad_norm": 67.50110626220703,
      "learning_rate": 9.57935855263158e-06,
      "loss": 1.4051,
      "step": 1030
    },
    {
      "epoch": 0.4276315789473684,
      "grad_norm": 30.782623291015625,
      "learning_rate": 9.575246710526317e-06,
      "loss": 0.75,
      "step": 1040
    },
    {
      "epoch": 0.4317434210526316,
      "grad_norm": 53.02436828613281,
      "learning_rate": 9.571134868421053e-06,
      "loss": 0.4739,
      "step": 1050
    },
    {
      "epoch": 0.43585526315789475,
      "grad_norm": 9.712882995605469,
      "learning_rate": 9.56702302631579e-06,
      "loss": 0.2168,
      "step": 1060
    },
    {
      "epoch": 0.4399671052631579,
      "grad_norm": 18.32390594482422,
      "learning_rate": 9.562911184210528e-06,
      "loss": 0.4673,
      "step": 1070
    },
    {
      "epoch": 0.4440789473684211,
      "grad_norm": 50.921531677246094,
      "learning_rate": 9.558799342105264e-06,
      "loss": 0.4137,
      "step": 1080
    },
    {
      "epoch": 0.44819078947368424,
      "grad_norm": 66.30414581298828,
      "learning_rate": 9.554687500000001e-06,
      "loss": 0.549,
      "step": 1090
    },
    {
      "epoch": 0.45230263157894735,
      "grad_norm": 0.11518318206071854,
      "learning_rate": 9.550575657894737e-06,
      "loss": 0.6499,
      "step": 1100
    },
    {
      "epoch": 0.4564144736842105,
      "grad_norm": 4.3409528732299805,
      "learning_rate": 9.546463815789474e-06,
      "loss": 0.9463,
      "step": 1110
    },
    {
      "epoch": 0.4605263157894737,
      "grad_norm": 30.97076416015625,
      "learning_rate": 9.542351973684212e-06,
      "loss": 0.6719,
      "step": 1120
    },
    {
      "epoch": 0.46463815789473684,
      "grad_norm": 13.25989818572998,
      "learning_rate": 9.538240131578949e-06,
      "loss": 0.4149,
      "step": 1130
    },
    {
      "epoch": 0.46875,
      "grad_norm": 1.624327540397644,
      "learning_rate": 9.534128289473685e-06,
      "loss": 0.7139,
      "step": 1140
    },
    {
      "epoch": 0.47286184210526316,
      "grad_norm": 1.2032309770584106,
      "learning_rate": 9.530016447368422e-06,
      "loss": 0.589,
      "step": 1150
    },
    {
      "epoch": 0.4769736842105263,
      "grad_norm": 125.88443756103516,
      "learning_rate": 9.525904605263158e-06,
      "loss": 0.5495,
      "step": 1160
    },
    {
      "epoch": 0.4810855263157895,
      "grad_norm": 31.080154418945312,
      "learning_rate": 9.521792763157896e-06,
      "loss": 0.7089,
      "step": 1170
    },
    {
      "epoch": 0.48519736842105265,
      "grad_norm": 39.169124603271484,
      "learning_rate": 9.517680921052633e-06,
      "loss": 0.8776,
      "step": 1180
    },
    {
      "epoch": 0.48930921052631576,
      "grad_norm": 157.91415405273438,
      "learning_rate": 9.513569078947369e-06,
      "loss": 0.713,
      "step": 1190
    },
    {
      "epoch": 0.4934210526315789,
      "grad_norm": 21.97187614440918,
      "learning_rate": 9.509457236842106e-06,
      "loss": 0.4497,
      "step": 1200
    },
    {
      "epoch": 0.4975328947368421,
      "grad_norm": 48.10848617553711,
      "learning_rate": 9.50575657894737e-06,
      "loss": 0.8274,
      "step": 1210
    },
    {
      "epoch": 0.5016447368421053,
      "grad_norm": 3.884164571762085,
      "learning_rate": 9.501644736842107e-06,
      "loss": 0.9336,
      "step": 1220
    },
    {
      "epoch": 0.5057565789473685,
      "grad_norm": 6.460564136505127,
      "learning_rate": 9.497532894736843e-06,
      "loss": 0.8036,
      "step": 1230
    },
    {
      "epoch": 0.5098684210526315,
      "grad_norm": 32.25432205200195,
      "learning_rate": 9.49342105263158e-06,
      "loss": 0.5282,
      "step": 1240
    },
    {
      "epoch": 0.5139802631578947,
      "grad_norm": 1.8110394477844238,
      "learning_rate": 9.489309210526316e-06,
      "loss": 0.1622,
      "step": 1250
    },
    {
      "epoch": 0.5180921052631579,
      "grad_norm": 66.42652893066406,
      "learning_rate": 9.485197368421053e-06,
      "loss": 1.3015,
      "step": 1260
    },
    {
      "epoch": 0.522203947368421,
      "grad_norm": 0.5288456678390503,
      "learning_rate": 9.481085526315789e-06,
      "loss": 1.0417,
      "step": 1270
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 43.16275405883789,
      "learning_rate": 9.476973684210528e-06,
      "loss": 0.2968,
      "step": 1280
    },
    {
      "epoch": 0.5304276315789473,
      "grad_norm": 2.046732187271118,
      "learning_rate": 9.472861842105264e-06,
      "loss": 0.9133,
      "step": 1290
    },
    {
      "epoch": 0.5345394736842105,
      "grad_norm": 0.09054580330848694,
      "learning_rate": 9.468750000000001e-06,
      "loss": 0.2698,
      "step": 1300
    },
    {
      "epoch": 0.5386513157894737,
      "grad_norm": 0.03111335076391697,
      "learning_rate": 9.464638157894737e-06,
      "loss": 0.2832,
      "step": 1310
    },
    {
      "epoch": 0.5427631578947368,
      "grad_norm": 78.9267578125,
      "learning_rate": 9.460526315789475e-06,
      "loss": 0.8709,
      "step": 1320
    },
    {
      "epoch": 0.546875,
      "grad_norm": 55.27962875366211,
      "learning_rate": 9.45641447368421e-06,
      "loss": 0.7564,
      "step": 1330
    },
    {
      "epoch": 0.5509868421052632,
      "grad_norm": 53.56753921508789,
      "learning_rate": 9.452302631578948e-06,
      "loss": 0.6463,
      "step": 1340
    },
    {
      "epoch": 0.5550986842105263,
      "grad_norm": 61.4496955871582,
      "learning_rate": 9.448190789473685e-06,
      "loss": 0.8631,
      "step": 1350
    },
    {
      "epoch": 0.5592105263157895,
      "grad_norm": 34.20602035522461,
      "learning_rate": 9.444078947368421e-06,
      "loss": 0.3161,
      "step": 1360
    },
    {
      "epoch": 0.5633223684210527,
      "grad_norm": 35.03656768798828,
      "learning_rate": 9.439967105263159e-06,
      "loss": 0.2618,
      "step": 1370
    },
    {
      "epoch": 0.5674342105263158,
      "grad_norm": 0.7599917054176331,
      "learning_rate": 9.435855263157896e-06,
      "loss": 0.9035,
      "step": 1380
    },
    {
      "epoch": 0.571546052631579,
      "grad_norm": 29.473997116088867,
      "learning_rate": 9.431743421052633e-06,
      "loss": 0.8102,
      "step": 1390
    },
    {
      "epoch": 0.5756578947368421,
      "grad_norm": 13.44682502746582,
      "learning_rate": 9.42763157894737e-06,
      "loss": 0.5657,
      "step": 1400
    },
    {
      "epoch": 0.5797697368421053,
      "grad_norm": 40.96477127075195,
      "learning_rate": 9.423519736842107e-06,
      "loss": 0.3194,
      "step": 1410
    },
    {
      "epoch": 0.5838815789473685,
      "grad_norm": 53.700321197509766,
      "learning_rate": 9.419407894736842e-06,
      "loss": 0.6133,
      "step": 1420
    },
    {
      "epoch": 0.5879934210526315,
      "grad_norm": 57.40962219238281,
      "learning_rate": 9.41529605263158e-06,
      "loss": 1.0728,
      "step": 1430
    },
    {
      "epoch": 0.5921052631578947,
      "grad_norm": 81.72367095947266,
      "learning_rate": 9.411184210526316e-06,
      "loss": 0.9684,
      "step": 1440
    },
    {
      "epoch": 0.5962171052631579,
      "grad_norm": 7.812744617462158,
      "learning_rate": 9.407072368421053e-06,
      "loss": 0.496,
      "step": 1450
    },
    {
      "epoch": 0.600328947368421,
      "grad_norm": 12.160445213317871,
      "learning_rate": 9.402960526315789e-06,
      "loss": 0.6885,
      "step": 1460
    },
    {
      "epoch": 0.6044407894736842,
      "grad_norm": 51.21662902832031,
      "learning_rate": 9.398848684210528e-06,
      "loss": 0.6898,
      "step": 1470
    },
    {
      "epoch": 0.6085526315789473,
      "grad_norm": 5.085622310638428,
      "learning_rate": 9.394736842105264e-06,
      "loss": 0.3369,
      "step": 1480
    },
    {
      "epoch": 0.6126644736842105,
      "grad_norm": 0.29845261573791504,
      "learning_rate": 9.390625000000001e-06,
      "loss": 1.106,
      "step": 1490
    },
    {
      "epoch": 0.6167763157894737,
      "grad_norm": 13.685811042785645,
      "learning_rate": 9.386513157894737e-06,
      "loss": 0.5248,
      "step": 1500
    },
    {
      "epoch": 0.6208881578947368,
      "grad_norm": 29.665889739990234,
      "learning_rate": 9.382401315789474e-06,
      "loss": 0.568,
      "step": 1510
    },
    {
      "epoch": 0.625,
      "grad_norm": 18.520368576049805,
      "learning_rate": 9.378289473684212e-06,
      "loss": 0.6348,
      "step": 1520
    },
    {
      "epoch": 0.6291118421052632,
      "grad_norm": 3.3405580520629883,
      "learning_rate": 9.374177631578948e-06,
      "loss": 0.595,
      "step": 1530
    },
    {
      "epoch": 0.6332236842105263,
      "grad_norm": 29.315523147583008,
      "learning_rate": 9.370065789473685e-06,
      "loss": 0.3298,
      "step": 1540
    },
    {
      "epoch": 0.6373355263157895,
      "grad_norm": 46.490848541259766,
      "learning_rate": 9.365953947368421e-06,
      "loss": 0.8335,
      "step": 1550
    },
    {
      "epoch": 0.6414473684210527,
      "grad_norm": 12.853484153747559,
      "learning_rate": 9.361842105263158e-06,
      "loss": 0.5452,
      "step": 1560
    },
    {
      "epoch": 0.6455592105263158,
      "grad_norm": 11.238713264465332,
      "learning_rate": 9.357730263157896e-06,
      "loss": 0.8813,
      "step": 1570
    },
    {
      "epoch": 0.649671052631579,
      "grad_norm": 1.7914977073669434,
      "learning_rate": 9.353618421052633e-06,
      "loss": 0.5766,
      "step": 1580
    },
    {
      "epoch": 0.6537828947368421,
      "grad_norm": 9.993241310119629,
      "learning_rate": 9.349506578947369e-06,
      "loss": 1.0686,
      "step": 1590
    },
    {
      "epoch": 0.6578947368421053,
      "grad_norm": 5.760276794433594,
      "learning_rate": 9.345394736842106e-06,
      "loss": 0.9441,
      "step": 1600
    },
    {
      "epoch": 0.6620065789473685,
      "grad_norm": 58.067134857177734,
      "learning_rate": 9.341282894736842e-06,
      "loss": 0.5564,
      "step": 1610
    },
    {
      "epoch": 0.6661184210526315,
      "grad_norm": 44.08545684814453,
      "learning_rate": 9.33717105263158e-06,
      "loss": 0.4461,
      "step": 1620
    },
    {
      "epoch": 0.6702302631578947,
      "grad_norm": 28.851512908935547,
      "learning_rate": 9.333059210526315e-06,
      "loss": 1.1042,
      "step": 1630
    },
    {
      "epoch": 0.6743421052631579,
      "grad_norm": 0.11420683562755585,
      "learning_rate": 9.328947368421053e-06,
      "loss": 0.7105,
      "step": 1640
    },
    {
      "epoch": 0.678453947368421,
      "grad_norm": 44.5568962097168,
      "learning_rate": 9.32483552631579e-06,
      "loss": 0.5754,
      "step": 1650
    },
    {
      "epoch": 0.6825657894736842,
      "grad_norm": 16.143844604492188,
      "learning_rate": 9.320723684210528e-06,
      "loss": 0.5564,
      "step": 1660
    },
    {
      "epoch": 0.6866776315789473,
      "grad_norm": 87.19486236572266,
      "learning_rate": 9.316611842105264e-06,
      "loss": 0.7184,
      "step": 1670
    },
    {
      "epoch": 0.6907894736842105,
      "grad_norm": 26.366031646728516,
      "learning_rate": 9.312500000000001e-06,
      "loss": 0.4315,
      "step": 1680
    },
    {
      "epoch": 0.6949013157894737,
      "grad_norm": 30.502588272094727,
      "learning_rate": 9.308388157894739e-06,
      "loss": 1.1223,
      "step": 1690
    },
    {
      "epoch": 0.6990131578947368,
      "grad_norm": 59.447601318359375,
      "learning_rate": 9.304276315789474e-06,
      "loss": 0.5216,
      "step": 1700
    },
    {
      "epoch": 0.703125,
      "grad_norm": 25.073314666748047,
      "learning_rate": 9.300164473684212e-06,
      "loss": 0.7732,
      "step": 1710
    },
    {
      "epoch": 0.7072368421052632,
      "grad_norm": 3.718965530395508,
      "learning_rate": 9.296052631578947e-06,
      "loss": 0.7764,
      "step": 1720
    },
    {
      "epoch": 0.7113486842105263,
      "grad_norm": 48.748783111572266,
      "learning_rate": 9.291940789473685e-06,
      "loss": 0.4773,
      "step": 1730
    },
    {
      "epoch": 0.7154605263157895,
      "grad_norm": 17.26929473876953,
      "learning_rate": 9.28782894736842e-06,
      "loss": 0.3774,
      "step": 1740
    },
    {
      "epoch": 0.7195723684210527,
      "grad_norm": 0.05121591314673424,
      "learning_rate": 9.28371710526316e-06,
      "loss": 0.7225,
      "step": 1750
    },
    {
      "epoch": 0.7236842105263158,
      "grad_norm": 1.7986981868743896,
      "learning_rate": 9.279605263157896e-06,
      "loss": 0.6779,
      "step": 1760
    },
    {
      "epoch": 0.727796052631579,
      "grad_norm": 77.67304229736328,
      "learning_rate": 9.275493421052633e-06,
      "loss": 0.3377,
      "step": 1770
    },
    {
      "epoch": 0.7319078947368421,
      "grad_norm": 12.0983304977417,
      "learning_rate": 9.271381578947369e-06,
      "loss": 0.7999,
      "step": 1780
    },
    {
      "epoch": 0.7360197368421053,
      "grad_norm": 0.6760322451591492,
      "learning_rate": 9.267269736842106e-06,
      "loss": 0.6168,
      "step": 1790
    },
    {
      "epoch": 0.7401315789473685,
      "grad_norm": 11.3161039352417,
      "learning_rate": 9.263157894736842e-06,
      "loss": 0.5886,
      "step": 1800
    },
    {
      "epoch": 0.7442434210526315,
      "grad_norm": 5.615332126617432,
      "learning_rate": 9.25904605263158e-06,
      "loss": 0.1935,
      "step": 1810
    },
    {
      "epoch": 0.7483552631578947,
      "grad_norm": 40.399017333984375,
      "learning_rate": 9.254934210526317e-06,
      "loss": 0.4364,
      "step": 1820
    },
    {
      "epoch": 0.7524671052631579,
      "grad_norm": 20.905866622924805,
      "learning_rate": 9.250822368421053e-06,
      "loss": 0.9716,
      "step": 1830
    },
    {
      "epoch": 0.756578947368421,
      "grad_norm": 50.7322883605957,
      "learning_rate": 9.24671052631579e-06,
      "loss": 0.4309,
      "step": 1840
    },
    {
      "epoch": 0.7606907894736842,
      "grad_norm": 36.55461883544922,
      "learning_rate": 9.242598684210528e-06,
      "loss": 1.0015,
      "step": 1850
    },
    {
      "epoch": 0.7648026315789473,
      "grad_norm": 55.39717483520508,
      "learning_rate": 9.238486842105265e-06,
      "loss": 1.1342,
      "step": 1860
    },
    {
      "epoch": 0.7689144736842105,
      "grad_norm": 47.303123474121094,
      "learning_rate": 9.234375e-06,
      "loss": 0.4452,
      "step": 1870
    },
    {
      "epoch": 0.7730263157894737,
      "grad_norm": 0.07194405794143677,
      "learning_rate": 9.230263157894738e-06,
      "loss": 0.6433,
      "step": 1880
    },
    {
      "epoch": 0.7771381578947368,
      "grad_norm": 80.85247039794922,
      "learning_rate": 9.226151315789474e-06,
      "loss": 0.5254,
      "step": 1890
    },
    {
      "epoch": 0.78125,
      "grad_norm": 27.962688446044922,
      "learning_rate": 9.222039473684212e-06,
      "loss": 0.92,
      "step": 1900
    },
    {
      "epoch": 0.7853618421052632,
      "grad_norm": 43.64583969116211,
      "learning_rate": 9.217927631578947e-06,
      "loss": 0.6327,
      "step": 1910
    },
    {
      "epoch": 0.7894736842105263,
      "grad_norm": 25.193130493164062,
      "learning_rate": 9.213815789473685e-06,
      "loss": 0.4634,
      "step": 1920
    },
    {
      "epoch": 0.7935855263157895,
      "grad_norm": 0.6646166443824768,
      "learning_rate": 9.20970394736842e-06,
      "loss": 0.2824,
      "step": 1930
    },
    {
      "epoch": 0.7976973684210527,
      "grad_norm": 66.69795989990234,
      "learning_rate": 9.20559210526316e-06,
      "loss": 0.6365,
      "step": 1940
    },
    {
      "epoch": 0.8018092105263158,
      "grad_norm": 26.328622817993164,
      "learning_rate": 9.201480263157895e-06,
      "loss": 0.6187,
      "step": 1950
    },
    {
      "epoch": 0.805921052631579,
      "grad_norm": 43.69908905029297,
      "learning_rate": 9.197368421052633e-06,
      "loss": 0.7507,
      "step": 1960
    },
    {
      "epoch": 0.8100328947368421,
      "grad_norm": 55.16326904296875,
      "learning_rate": 9.193256578947369e-06,
      "loss": 0.5324,
      "step": 1970
    },
    {
      "epoch": 0.8141447368421053,
      "grad_norm": 0.5004386901855469,
      "learning_rate": 9.189144736842106e-06,
      "loss": 0.3479,
      "step": 1980
    },
    {
      "epoch": 0.8182565789473685,
      "grad_norm": 0.6016833782196045,
      "learning_rate": 9.185032894736844e-06,
      "loss": 0.8248,
      "step": 1990
    },
    {
      "epoch": 0.8223684210526315,
      "grad_norm": 49.71751403808594,
      "learning_rate": 9.18092105263158e-06,
      "loss": 0.7017,
      "step": 2000
    },
    {
      "epoch": 0.8264802631578947,
      "grad_norm": 60.05244064331055,
      "learning_rate": 9.176809210526317e-06,
      "loss": 0.642,
      "step": 2010
    },
    {
      "epoch": 0.8305921052631579,
      "grad_norm": 70.1126708984375,
      "learning_rate": 9.172697368421053e-06,
      "loss": 0.9515,
      "step": 2020
    },
    {
      "epoch": 0.834703947368421,
      "grad_norm": 70.0270004272461,
      "learning_rate": 9.16858552631579e-06,
      "loss": 0.5299,
      "step": 2030
    },
    {
      "epoch": 0.8388157894736842,
      "grad_norm": 0.9353859424591064,
      "learning_rate": 9.164473684210527e-06,
      "loss": 0.8934,
      "step": 2040
    },
    {
      "epoch": 0.8429276315789473,
      "grad_norm": 0.3574899435043335,
      "learning_rate": 9.160361842105265e-06,
      "loss": 0.5327,
      "step": 2050
    },
    {
      "epoch": 0.8470394736842105,
      "grad_norm": 77.09919738769531,
      "learning_rate": 9.15625e-06,
      "loss": 1.1277,
      "step": 2060
    },
    {
      "epoch": 0.8511513157894737,
      "grad_norm": 94.8553695678711,
      "learning_rate": 9.152138157894738e-06,
      "loss": 1.0451,
      "step": 2070
    },
    {
      "epoch": 0.8552631578947368,
      "grad_norm": 6.773098945617676,
      "learning_rate": 9.148026315789474e-06,
      "loss": 0.5617,
      "step": 2080
    },
    {
      "epoch": 0.859375,
      "grad_norm": 3.3853366374969482,
      "learning_rate": 9.143914473684211e-06,
      "loss": 0.5929,
      "step": 2090
    },
    {
      "epoch": 0.8634868421052632,
      "grad_norm": 24.653453826904297,
      "learning_rate": 9.139802631578947e-06,
      "loss": 0.6193,
      "step": 2100
    },
    {
      "epoch": 0.8675986842105263,
      "grad_norm": 51.23694610595703,
      "learning_rate": 9.135690789473685e-06,
      "loss": 1.0407,
      "step": 2110
    },
    {
      "epoch": 0.8717105263157895,
      "grad_norm": 39.51556396484375,
      "learning_rate": 9.131578947368422e-06,
      "loss": 0.7252,
      "step": 2120
    },
    {
      "epoch": 0.8758223684210527,
      "grad_norm": 0.20115692913532257,
      "learning_rate": 9.12746710526316e-06,
      "loss": 0.6013,
      "step": 2130
    },
    {
      "epoch": 0.8799342105263158,
      "grad_norm": 58.84267044067383,
      "learning_rate": 9.123355263157895e-06,
      "loss": 0.686,
      "step": 2140
    },
    {
      "epoch": 0.884046052631579,
      "grad_norm": 0.0019536695908755064,
      "learning_rate": 9.119243421052633e-06,
      "loss": 0.2745,
      "step": 2150
    },
    {
      "epoch": 0.8881578947368421,
      "grad_norm": 66.56216430664062,
      "learning_rate": 9.11513157894737e-06,
      "loss": 0.7467,
      "step": 2160
    },
    {
      "epoch": 0.8922697368421053,
      "grad_norm": 4.7335100173950195,
      "learning_rate": 9.111019736842106e-06,
      "loss": 0.1942,
      "step": 2170
    },
    {
      "epoch": 0.8963815789473685,
      "grad_norm": 39.29569625854492,
      "learning_rate": 9.106907894736843e-06,
      "loss": 0.4847,
      "step": 2180
    },
    {
      "epoch": 0.9004934210526315,
      "grad_norm": 5.849125862121582,
      "learning_rate": 9.102796052631579e-06,
      "loss": 0.4814,
      "step": 2190
    },
    {
      "epoch": 0.9046052631578947,
      "grad_norm": 15.409236907958984,
      "learning_rate": 9.098684210526317e-06,
      "loss": 0.2912,
      "step": 2200
    },
    {
      "epoch": 0.9087171052631579,
      "grad_norm": 0.6259574890136719,
      "learning_rate": 9.094572368421052e-06,
      "loss": 0.6347,
      "step": 2210
    },
    {
      "epoch": 0.912828947368421,
      "grad_norm": 65.06391143798828,
      "learning_rate": 9.09046052631579e-06,
      "loss": 0.3179,
      "step": 2220
    },
    {
      "epoch": 0.9169407894736842,
      "grad_norm": 0.6020347476005554,
      "learning_rate": 9.086348684210527e-06,
      "loss": 0.6884,
      "step": 2230
    },
    {
      "epoch": 0.9210526315789473,
      "grad_norm": 2.7630629539489746,
      "learning_rate": 9.082236842105265e-06,
      "loss": 0.7149,
      "step": 2240
    },
    {
      "epoch": 0.9251644736842105,
      "grad_norm": 12.887619972229004,
      "learning_rate": 9.078125e-06,
      "loss": 0.59,
      "step": 2250
    },
    {
      "epoch": 0.9292763157894737,
      "grad_norm": 49.26618194580078,
      "learning_rate": 9.074013157894738e-06,
      "loss": 0.4737,
      "step": 2260
    },
    {
      "epoch": 0.9333881578947368,
      "grad_norm": 1.3609040975570679,
      "learning_rate": 9.069901315789474e-06,
      "loss": 0.4908,
      "step": 2270
    },
    {
      "epoch": 0.9375,
      "grad_norm": 56.066932678222656,
      "learning_rate": 9.065789473684211e-06,
      "loss": 0.7488,
      "step": 2280
    },
    {
      "epoch": 0.9416118421052632,
      "grad_norm": 79.9988784790039,
      "learning_rate": 9.061677631578949e-06,
      "loss": 0.7794,
      "step": 2290
    },
    {
      "epoch": 0.9457236842105263,
      "grad_norm": 0.05063479766249657,
      "learning_rate": 9.057565789473684e-06,
      "loss": 0.2889,
      "step": 2300
    },
    {
      "epoch": 0.9498355263157895,
      "grad_norm": 55.820369720458984,
      "learning_rate": 9.053453947368422e-06,
      "loss": 0.7358,
      "step": 2310
    },
    {
      "epoch": 0.9539473684210527,
      "grad_norm": 84.37875366210938,
      "learning_rate": 9.04934210526316e-06,
      "loss": 0.8118,
      "step": 2320
    },
    {
      "epoch": 0.9580592105263158,
      "grad_norm": 69.74197387695312,
      "learning_rate": 9.045230263157897e-06,
      "loss": 1.1243,
      "step": 2330
    },
    {
      "epoch": 0.962171052631579,
      "grad_norm": 0.7944409251213074,
      "learning_rate": 9.041118421052632e-06,
      "loss": 0.1485,
      "step": 2340
    },
    {
      "epoch": 0.9662828947368421,
      "grad_norm": 9.75572681427002,
      "learning_rate": 9.03700657894737e-06,
      "loss": 0.6436,
      "step": 2350
    },
    {
      "epoch": 0.9703947368421053,
      "grad_norm": 0.3801949918270111,
      "learning_rate": 9.032894736842106e-06,
      "loss": 0.6525,
      "step": 2360
    },
    {
      "epoch": 0.9745065789473685,
      "grad_norm": 0.04239730164408684,
      "learning_rate": 9.028782894736843e-06,
      "loss": 0.4073,
      "step": 2370
    },
    {
      "epoch": 0.9786184210526315,
      "grad_norm": 0.5113763809204102,
      "learning_rate": 9.024671052631579e-06,
      "loss": 0.4306,
      "step": 2380
    },
    {
      "epoch": 0.9827302631578947,
      "grad_norm": 67.6680679321289,
      "learning_rate": 9.020559210526316e-06,
      "loss": 0.8971,
      "step": 2390
    },
    {
      "epoch": 0.9868421052631579,
      "grad_norm": 64.32469177246094,
      "learning_rate": 9.016447368421054e-06,
      "loss": 0.4565,
      "step": 2400
    },
    {
      "epoch": 0.990953947368421,
      "grad_norm": 2.2221884727478027,
      "learning_rate": 9.01233552631579e-06,
      "loss": 0.5263,
      "step": 2410
    },
    {
      "epoch": 0.9950657894736842,
      "grad_norm": 7.352302551269531,
      "learning_rate": 9.008223684210527e-06,
      "loss": 0.6523,
      "step": 2420
    },
    {
      "epoch": 0.9991776315789473,
      "grad_norm": 66.60418701171875,
      "learning_rate": 9.004111842105264e-06,
      "loss": 0.401,
      "step": 2430
    },
    {
      "epoch": 1.0032894736842106,
      "grad_norm": 10.393697738647461,
      "learning_rate": 9e-06,
      "loss": 0.3328,
      "step": 2440
    },
    {
      "epoch": 1.0074013157894737,
      "grad_norm": 33.00063705444336,
      "learning_rate": 8.995888157894738e-06,
      "loss": 0.1872,
      "step": 2450
    },
    {
      "epoch": 1.011513157894737,
      "grad_norm": 53.126712799072266,
      "learning_rate": 8.991776315789475e-06,
      "loss": 1.1614,
      "step": 2460
    },
    {
      "epoch": 1.015625,
      "grad_norm": 0.4837985634803772,
      "learning_rate": 8.987664473684211e-06,
      "loss": 0.2477,
      "step": 2470
    },
    {
      "epoch": 1.019736842105263,
      "grad_norm": 5.717953681945801,
      "learning_rate": 8.983552631578948e-06,
      "loss": 0.1087,
      "step": 2480
    },
    {
      "epoch": 1.0238486842105263,
      "grad_norm": 13.705921173095703,
      "learning_rate": 8.979440789473684e-06,
      "loss": 0.1692,
      "step": 2490
    },
    {
      "epoch": 1.0279605263157894,
      "grad_norm": 61.79713439941406,
      "learning_rate": 8.975328947368422e-06,
      "loss": 0.6781,
      "step": 2500
    },
    {
      "epoch": 1.0320723684210527,
      "grad_norm": 5.752686023712158,
      "learning_rate": 8.971217105263159e-06,
      "loss": 0.5315,
      "step": 2510
    },
    {
      "epoch": 1.0361842105263157,
      "grad_norm": 23.530588150024414,
      "learning_rate": 8.967105263157896e-06,
      "loss": 0.585,
      "step": 2520
    },
    {
      "epoch": 1.040296052631579,
      "grad_norm": 0.11146017163991928,
      "learning_rate": 8.962993421052632e-06,
      "loss": 0.4786,
      "step": 2530
    },
    {
      "epoch": 1.044407894736842,
      "grad_norm": 4.48652458190918,
      "learning_rate": 8.95888157894737e-06,
      "loss": 0.659,
      "step": 2540
    },
    {
      "epoch": 1.0485197368421053,
      "grad_norm": 3.888078451156616,
      "learning_rate": 8.954769736842105e-06,
      "loss": 0.1793,
      "step": 2550
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 0.5999749302864075,
      "learning_rate": 8.950657894736843e-06,
      "loss": 0.3992,
      "step": 2560
    },
    {
      "epoch": 1.0567434210526316,
      "grad_norm": 20.677486419677734,
      "learning_rate": 8.946546052631579e-06,
      "loss": 0.4838,
      "step": 2570
    },
    {
      "epoch": 1.0608552631578947,
      "grad_norm": 0.10241834819316864,
      "learning_rate": 8.942434210526316e-06,
      "loss": 0.6778,
      "step": 2580
    },
    {
      "epoch": 1.064967105263158,
      "grad_norm": 87.81678009033203,
      "learning_rate": 8.938322368421054e-06,
      "loss": 0.4954,
      "step": 2590
    },
    {
      "epoch": 1.069078947368421,
      "grad_norm": 109.55130767822266,
      "learning_rate": 8.93421052631579e-06,
      "loss": 0.489,
      "step": 2600
    },
    {
      "epoch": 1.0731907894736843,
      "grad_norm": 52.265106201171875,
      "learning_rate": 8.930098684210527e-06,
      "loss": 0.5496,
      "step": 2610
    },
    {
      "epoch": 1.0773026315789473,
      "grad_norm": 4.881550312042236,
      "learning_rate": 8.925986842105264e-06,
      "loss": 0.3602,
      "step": 2620
    },
    {
      "epoch": 1.0814144736842106,
      "grad_norm": 59.2353401184082,
      "learning_rate": 8.921875000000002e-06,
      "loss": 0.7403,
      "step": 2630
    },
    {
      "epoch": 1.0855263157894737,
      "grad_norm": 64.6301040649414,
      "learning_rate": 8.917763157894737e-06,
      "loss": 0.6835,
      "step": 2640
    },
    {
      "epoch": 1.0896381578947367,
      "grad_norm": 45.62979507446289,
      "learning_rate": 8.913651315789475e-06,
      "loss": 0.4795,
      "step": 2650
    },
    {
      "epoch": 1.09375,
      "grad_norm": 18.216320037841797,
      "learning_rate": 8.90953947368421e-06,
      "loss": 0.557,
      "step": 2660
    },
    {
      "epoch": 1.0978618421052633,
      "grad_norm": 0.05309246852993965,
      "learning_rate": 8.905427631578948e-06,
      "loss": 0.3478,
      "step": 2670
    },
    {
      "epoch": 1.1019736842105263,
      "grad_norm": 0.35603123903274536,
      "learning_rate": 8.901315789473684e-06,
      "loss": 0.1215,
      "step": 2680
    },
    {
      "epoch": 1.1060855263157894,
      "grad_norm": 3.711970567703247,
      "learning_rate": 8.897203947368421e-06,
      "loss": 0.2544,
      "step": 2690
    },
    {
      "epoch": 1.1101973684210527,
      "grad_norm": 5.010875225067139,
      "learning_rate": 8.893092105263159e-06,
      "loss": 0.2661,
      "step": 2700
    },
    {
      "epoch": 1.1143092105263157,
      "grad_norm": 51.760868072509766,
      "learning_rate": 8.888980263157896e-06,
      "loss": 0.3528,
      "step": 2710
    },
    {
      "epoch": 1.118421052631579,
      "grad_norm": 0.11445479840040207,
      "learning_rate": 8.884868421052632e-06,
      "loss": 0.542,
      "step": 2720
    },
    {
      "epoch": 1.122532894736842,
      "grad_norm": 4.898531913757324,
      "learning_rate": 8.88075657894737e-06,
      "loss": 0.7877,
      "step": 2730
    },
    {
      "epoch": 1.1266447368421053,
      "grad_norm": 12.105345726013184,
      "learning_rate": 8.876644736842105e-06,
      "loss": 0.7329,
      "step": 2740
    },
    {
      "epoch": 1.1307565789473684,
      "grad_norm": 6.530974864959717,
      "learning_rate": 8.872532894736843e-06,
      "loss": 0.2338,
      "step": 2750
    },
    {
      "epoch": 1.1348684210526316,
      "grad_norm": 5.893588066101074,
      "learning_rate": 8.86842105263158e-06,
      "loss": 0.7787,
      "step": 2760
    },
    {
      "epoch": 1.1389802631578947,
      "grad_norm": 52.82911682128906,
      "learning_rate": 8.864309210526316e-06,
      "loss": 0.2331,
      "step": 2770
    },
    {
      "epoch": 1.143092105263158,
      "grad_norm": 15.883142471313477,
      "learning_rate": 8.860197368421053e-06,
      "loss": 0.3743,
      "step": 2780
    },
    {
      "epoch": 1.147203947368421,
      "grad_norm": 99.91140747070312,
      "learning_rate": 8.856085526315789e-06,
      "loss": 0.1293,
      "step": 2790
    },
    {
      "epoch": 1.1513157894736843,
      "grad_norm": 0.10743849724531174,
      "learning_rate": 8.851973684210528e-06,
      "loss": 0.1448,
      "step": 2800
    },
    {
      "epoch": 1.1554276315789473,
      "grad_norm": 0.12402098625898361,
      "learning_rate": 8.847861842105264e-06,
      "loss": 0.2003,
      "step": 2810
    },
    {
      "epoch": 1.1595394736842106,
      "grad_norm": 0.9490414261817932,
      "learning_rate": 8.843750000000002e-06,
      "loss": 0.1746,
      "step": 2820
    },
    {
      "epoch": 1.1636513157894737,
      "grad_norm": 8.182841300964355,
      "learning_rate": 8.839638157894737e-06,
      "loss": 0.2,
      "step": 2830
    },
    {
      "epoch": 1.1677631578947367,
      "grad_norm": 29.3283748626709,
      "learning_rate": 8.835526315789475e-06,
      "loss": 1.0814,
      "step": 2840
    },
    {
      "epoch": 1.171875,
      "grad_norm": 0.44098392128944397,
      "learning_rate": 8.83141447368421e-06,
      "loss": 0.1534,
      "step": 2850
    },
    {
      "epoch": 1.1759868421052633,
      "grad_norm": 26.898929595947266,
      "learning_rate": 8.827302631578948e-06,
      "loss": 0.3173,
      "step": 2860
    },
    {
      "epoch": 1.1800986842105263,
      "grad_norm": 49.62389373779297,
      "learning_rate": 8.823190789473685e-06,
      "loss": 0.6717,
      "step": 2870
    },
    {
      "epoch": 1.1842105263157894,
      "grad_norm": 54.11696243286133,
      "learning_rate": 8.819078947368421e-06,
      "loss": 0.4769,
      "step": 2880
    },
    {
      "epoch": 1.1883223684210527,
      "grad_norm": 0.8223657011985779,
      "learning_rate": 8.814967105263159e-06,
      "loss": 0.1133,
      "step": 2890
    },
    {
      "epoch": 1.1924342105263157,
      "grad_norm": 0.30546388030052185,
      "learning_rate": 8.810855263157896e-06,
      "loss": 0.1435,
      "step": 2900
    },
    {
      "epoch": 1.196546052631579,
      "grad_norm": 0.005724355112761259,
      "learning_rate": 8.806743421052632e-06,
      "loss": 0.0993,
      "step": 2910
    },
    {
      "epoch": 1.200657894736842,
      "grad_norm": 2.1313202381134033,
      "learning_rate": 8.80263157894737e-06,
      "loss": 0.301,
      "step": 2920
    },
    {
      "epoch": 1.2047697368421053,
      "grad_norm": 2.2508060932159424,
      "learning_rate": 8.798519736842107e-06,
      "loss": 0.3936,
      "step": 2930
    },
    {
      "epoch": 1.2088815789473684,
      "grad_norm": 17.970064163208008,
      "learning_rate": 8.794407894736843e-06,
      "loss": 0.2133,
      "step": 2940
    },
    {
      "epoch": 1.2129934210526316,
      "grad_norm": 3.051180839538574,
      "learning_rate": 8.79029605263158e-06,
      "loss": 0.3686,
      "step": 2950
    },
    {
      "epoch": 1.2171052631578947,
      "grad_norm": 63.62922286987305,
      "learning_rate": 8.786184210526316e-06,
      "loss": 0.6775,
      "step": 2960
    },
    {
      "epoch": 1.221217105263158,
      "grad_norm": 91.9775390625,
      "learning_rate": 8.782072368421053e-06,
      "loss": 0.1896,
      "step": 2970
    },
    {
      "epoch": 1.225328947368421,
      "grad_norm": 13.494383811950684,
      "learning_rate": 8.777960526315789e-06,
      "loss": 0.446,
      "step": 2980
    },
    {
      "epoch": 1.2294407894736843,
      "grad_norm": 12.346921920776367,
      "learning_rate": 8.773848684210528e-06,
      "loss": 0.2766,
      "step": 2990
    },
    {
      "epoch": 1.2335526315789473,
      "grad_norm": 69.7927017211914,
      "learning_rate": 8.769736842105264e-06,
      "loss": 0.7206,
      "step": 3000
    },
    {
      "epoch": 1.2376644736842106,
      "grad_norm": 0.10051664710044861,
      "learning_rate": 8.765625000000001e-06,
      "loss": 0.8094,
      "step": 3010
    },
    {
      "epoch": 1.2417763157894737,
      "grad_norm": 73.73413848876953,
      "learning_rate": 8.761513157894737e-06,
      "loss": 0.3381,
      "step": 3020
    },
    {
      "epoch": 1.2458881578947367,
      "grad_norm": 5.243391513824463,
      "learning_rate": 8.757401315789475e-06,
      "loss": 0.424,
      "step": 3030
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.6941500902175903,
      "learning_rate": 8.75328947368421e-06,
      "loss": 0.3621,
      "step": 3040
    },
    {
      "epoch": 1.2541118421052633,
      "grad_norm": 1.7824820280075073,
      "learning_rate": 8.749177631578948e-06,
      "loss": 0.5639,
      "step": 3050
    },
    {
      "epoch": 1.2582236842105263,
      "grad_norm": 74.03777313232422,
      "learning_rate": 8.745065789473685e-06,
      "loss": 0.3756,
      "step": 3060
    },
    {
      "epoch": 1.2623355263157894,
      "grad_norm": 63.435646057128906,
      "learning_rate": 8.740953947368421e-06,
      "loss": 0.4663,
      "step": 3070
    },
    {
      "epoch": 1.2664473684210527,
      "grad_norm": 20.51471519470215,
      "learning_rate": 8.736842105263158e-06,
      "loss": 0.5241,
      "step": 3080
    },
    {
      "epoch": 1.2705592105263157,
      "grad_norm": 31.2130069732666,
      "learning_rate": 8.732730263157896e-06,
      "loss": 0.049,
      "step": 3090
    },
    {
      "epoch": 1.274671052631579,
      "grad_norm": 0.4100927412509918,
      "learning_rate": 8.728618421052633e-06,
      "loss": 0.6768,
      "step": 3100
    },
    {
      "epoch": 1.278782894736842,
      "grad_norm": 8.450210571289062,
      "learning_rate": 8.724506578947369e-06,
      "loss": 0.2578,
      "step": 3110
    },
    {
      "epoch": 1.2828947368421053,
      "grad_norm": 47.065731048583984,
      "learning_rate": 8.720394736842107e-06,
      "loss": 0.2957,
      "step": 3120
    },
    {
      "epoch": 1.2870065789473684,
      "grad_norm": 20.626476287841797,
      "learning_rate": 8.716282894736842e-06,
      "loss": 0.0919,
      "step": 3130
    },
    {
      "epoch": 1.2911184210526316,
      "grad_norm": 2.748300313949585,
      "learning_rate": 8.71217105263158e-06,
      "loss": 0.3204,
      "step": 3140
    },
    {
      "epoch": 1.2952302631578947,
      "grad_norm": 18.729766845703125,
      "learning_rate": 8.708059210526316e-06,
      "loss": 0.1864,
      "step": 3150
    },
    {
      "epoch": 1.299342105263158,
      "grad_norm": 2.0139780044555664,
      "learning_rate": 8.703947368421053e-06,
      "loss": 0.1559,
      "step": 3160
    },
    {
      "epoch": 1.303453947368421,
      "grad_norm": 2.845900774002075,
      "learning_rate": 8.69983552631579e-06,
      "loss": 0.6238,
      "step": 3170
    },
    {
      "epoch": 1.3075657894736843,
      "grad_norm": 46.37712860107422,
      "learning_rate": 8.695723684210528e-06,
      "loss": 0.2323,
      "step": 3180
    },
    {
      "epoch": 1.3116776315789473,
      "grad_norm": 271.55926513671875,
      "learning_rate": 8.691611842105264e-06,
      "loss": 0.2997,
      "step": 3190
    },
    {
      "epoch": 1.3157894736842106,
      "grad_norm": 0.07383125275373459,
      "learning_rate": 8.687500000000001e-06,
      "loss": 0.3775,
      "step": 3200
    },
    {
      "epoch": 1.3199013157894737,
      "grad_norm": 0.09837435185909271,
      "learning_rate": 8.683388157894737e-06,
      "loss": 0.7621,
      "step": 3210
    },
    {
      "epoch": 1.3240131578947367,
      "grad_norm": 9.785972595214844,
      "learning_rate": 8.679276315789474e-06,
      "loss": 0.1313,
      "step": 3220
    },
    {
      "epoch": 1.328125,
      "grad_norm": 12.865886688232422,
      "learning_rate": 8.675164473684212e-06,
      "loss": 0.3697,
      "step": 3230
    },
    {
      "epoch": 1.3322368421052633,
      "grad_norm": 65.38900756835938,
      "learning_rate": 8.671052631578948e-06,
      "loss": 0.1807,
      "step": 3240
    },
    {
      "epoch": 1.3363486842105263,
      "grad_norm": 21.760971069335938,
      "learning_rate": 8.666940789473685e-06,
      "loss": 0.427,
      "step": 3250
    },
    {
      "epoch": 1.3404605263157894,
      "grad_norm": 0.22508497536182404,
      "learning_rate": 8.66282894736842e-06,
      "loss": 0.2474,
      "step": 3260
    },
    {
      "epoch": 1.3445723684210527,
      "grad_norm": 0.027158482000231743,
      "learning_rate": 8.65871710526316e-06,
      "loss": 0.0662,
      "step": 3270
    },
    {
      "epoch": 1.3486842105263157,
      "grad_norm": 0.5353153944015503,
      "learning_rate": 8.654605263157896e-06,
      "loss": 0.6432,
      "step": 3280
    },
    {
      "epoch": 1.352796052631579,
      "grad_norm": 34.469608306884766,
      "learning_rate": 8.650493421052633e-06,
      "loss": 0.2062,
      "step": 3290
    },
    {
      "epoch": 1.356907894736842,
      "grad_norm": 1.8599954843521118,
      "learning_rate": 8.646381578947369e-06,
      "loss": 0.4208,
      "step": 3300
    },
    {
      "epoch": 1.3610197368421053,
      "grad_norm": 38.51445770263672,
      "learning_rate": 8.642269736842106e-06,
      "loss": 0.2114,
      "step": 3310
    },
    {
      "epoch": 1.3651315789473684,
      "grad_norm": 69.46410369873047,
      "learning_rate": 8.638157894736842e-06,
      "loss": 0.6396,
      "step": 3320
    },
    {
      "epoch": 1.3692434210526316,
      "grad_norm": 5.565284729003906,
      "learning_rate": 8.63404605263158e-06,
      "loss": 0.6757,
      "step": 3330
    },
    {
      "epoch": 1.3733552631578947,
      "grad_norm": 5.994679927825928,
      "learning_rate": 8.629934210526317e-06,
      "loss": 0.2594,
      "step": 3340
    },
    {
      "epoch": 1.377467105263158,
      "grad_norm": 30.572546005249023,
      "learning_rate": 8.625822368421053e-06,
      "loss": 0.2005,
      "step": 3350
    },
    {
      "epoch": 1.381578947368421,
      "grad_norm": 78.56243896484375,
      "learning_rate": 8.62171052631579e-06,
      "loss": 0.6431,
      "step": 3360
    },
    {
      "epoch": 1.3856907894736843,
      "grad_norm": 54.2548828125,
      "learning_rate": 8.617598684210528e-06,
      "loss": 0.3055,
      "step": 3370
    },
    {
      "epoch": 1.3898026315789473,
      "grad_norm": 0.4007514715194702,
      "learning_rate": 8.613486842105263e-06,
      "loss": 0.3877,
      "step": 3380
    },
    {
      "epoch": 1.3939144736842106,
      "grad_norm": 20.567773818969727,
      "learning_rate": 8.609375000000001e-06,
      "loss": 0.483,
      "step": 3390
    },
    {
      "epoch": 1.3980263157894737,
      "grad_norm": 57.93979263305664,
      "learning_rate": 8.605263157894738e-06,
      "loss": 0.5153,
      "step": 3400
    },
    {
      "epoch": 1.4021381578947367,
      "grad_norm": 0.26199084520339966,
      "learning_rate": 8.601151315789474e-06,
      "loss": 0.3544,
      "step": 3410
    },
    {
      "epoch": 1.40625,
      "grad_norm": 13.385703086853027,
      "learning_rate": 8.597039473684212e-06,
      "loss": 0.5427,
      "step": 3420
    },
    {
      "epoch": 1.4103618421052633,
      "grad_norm": 9.025935173034668,
      "learning_rate": 8.592927631578947e-06,
      "loss": 0.3185,
      "step": 3430
    },
    {
      "epoch": 1.4144736842105263,
      "grad_norm": 53.11691665649414,
      "learning_rate": 8.588815789473685e-06,
      "loss": 0.3702,
      "step": 3440
    },
    {
      "epoch": 1.4185855263157894,
      "grad_norm": 0.0294374767690897,
      "learning_rate": 8.58470394736842e-06,
      "loss": 0.3489,
      "step": 3450
    },
    {
      "epoch": 1.4226973684210527,
      "grad_norm": 73.63797760009766,
      "learning_rate": 8.58059210526316e-06,
      "loss": 0.1992,
      "step": 3460
    },
    {
      "epoch": 1.4268092105263157,
      "grad_norm": 10.585562705993652,
      "learning_rate": 8.576480263157895e-06,
      "loss": 0.2284,
      "step": 3470
    },
    {
      "epoch": 1.430921052631579,
      "grad_norm": 2.052852153778076,
      "learning_rate": 8.572368421052633e-06,
      "loss": 0.1912,
      "step": 3480
    },
    {
      "epoch": 1.435032894736842,
      "grad_norm": 0.003717209445312619,
      "learning_rate": 8.568256578947369e-06,
      "loss": 0.5507,
      "step": 3490
    },
    {
      "epoch": 1.4391447368421053,
      "grad_norm": 0.7865049242973328,
      "learning_rate": 8.564144736842106e-06,
      "loss": 0.0211,
      "step": 3500
    },
    {
      "epoch": 1.4432565789473684,
      "grad_norm": 15.986753463745117,
      "learning_rate": 8.560032894736842e-06,
      "loss": 0.2161,
      "step": 3510
    },
    {
      "epoch": 1.4473684210526316,
      "grad_norm": 6.693492412567139,
      "learning_rate": 8.55592105263158e-06,
      "loss": 0.3488,
      "step": 3520
    },
    {
      "epoch": 1.4514802631578947,
      "grad_norm": 39.255794525146484,
      "learning_rate": 8.551809210526317e-06,
      "loss": 0.4337,
      "step": 3530
    },
    {
      "epoch": 1.455592105263158,
      "grad_norm": 114.03556823730469,
      "learning_rate": 8.547697368421053e-06,
      "loss": 0.4609,
      "step": 3540
    },
    {
      "epoch": 1.459703947368421,
      "grad_norm": 0.11433131992816925,
      "learning_rate": 8.54358552631579e-06,
      "loss": 0.2146,
      "step": 3550
    },
    {
      "epoch": 1.4638157894736843,
      "grad_norm": 71.40312957763672,
      "learning_rate": 8.539473684210527e-06,
      "loss": 0.3865,
      "step": 3560
    },
    {
      "epoch": 1.4679276315789473,
      "grad_norm": 0.18895581364631653,
      "learning_rate": 8.535361842105265e-06,
      "loss": 0.5562,
      "step": 3570
    },
    {
      "epoch": 1.4720394736842106,
      "grad_norm": 16.85786247253418,
      "learning_rate": 8.53125e-06,
      "loss": 0.2986,
      "step": 3580
    },
    {
      "epoch": 1.4761513157894737,
      "grad_norm": 0.0461287759244442,
      "learning_rate": 8.527138157894738e-06,
      "loss": 0.5559,
      "step": 3590
    },
    {
      "epoch": 1.4802631578947367,
      "grad_norm": 63.416690826416016,
      "learning_rate": 8.523026315789474e-06,
      "loss": 0.4644,
      "step": 3600
    },
    {
      "epoch": 1.484375,
      "grad_norm": 96.18208312988281,
      "learning_rate": 8.518914473684211e-06,
      "loss": 0.4958,
      "step": 3610
    },
    {
      "epoch": 1.4884868421052633,
      "grad_norm": 2.9224765300750732,
      "learning_rate": 8.514802631578947e-06,
      "loss": 0.2215,
      "step": 3620
    },
    {
      "epoch": 1.4925986842105263,
      "grad_norm": 90.86573791503906,
      "learning_rate": 8.510690789473685e-06,
      "loss": 0.4001,
      "step": 3630
    },
    {
      "epoch": 1.4967105263157894,
      "grad_norm": 0.019238781183958054,
      "learning_rate": 8.506578947368422e-06,
      "loss": 0.2813,
      "step": 3640
    },
    {
      "epoch": 1.5008223684210527,
      "grad_norm": 64.77748107910156,
      "learning_rate": 8.50246710526316e-06,
      "loss": 0.4256,
      "step": 3650
    },
    {
      "epoch": 1.504934210526316,
      "grad_norm": 66.73531341552734,
      "learning_rate": 8.498355263157895e-06,
      "loss": 0.5352,
      "step": 3660
    },
    {
      "epoch": 1.509046052631579,
      "grad_norm": 30.41202163696289,
      "learning_rate": 8.494243421052633e-06,
      "loss": 0.4228,
      "step": 3670
    },
    {
      "epoch": 1.513157894736842,
      "grad_norm": 48.70793914794922,
      "learning_rate": 8.490131578947368e-06,
      "loss": 0.4549,
      "step": 3680
    },
    {
      "epoch": 1.5172697368421053,
      "grad_norm": 52.415306091308594,
      "learning_rate": 8.486019736842106e-06,
      "loss": 0.5359,
      "step": 3690
    },
    {
      "epoch": 1.5213815789473686,
      "grad_norm": 0.15045365691184998,
      "learning_rate": 8.481907894736843e-06,
      "loss": 0.2304,
      "step": 3700
    },
    {
      "epoch": 1.5254934210526314,
      "grad_norm": 0.844916582107544,
      "learning_rate": 8.477796052631579e-06,
      "loss": 0.7176,
      "step": 3710
    },
    {
      "epoch": 1.5296052631578947,
      "grad_norm": 1.8648791313171387,
      "learning_rate": 8.473684210526317e-06,
      "loss": 0.2773,
      "step": 3720
    },
    {
      "epoch": 1.533717105263158,
      "grad_norm": 0.052761100232601166,
      "learning_rate": 8.469572368421052e-06,
      "loss": 0.0183,
      "step": 3730
    },
    {
      "epoch": 1.537828947368421,
      "grad_norm": 54.12556838989258,
      "learning_rate": 8.46546052631579e-06,
      "loss": 0.5591,
      "step": 3740
    },
    {
      "epoch": 1.541940789473684,
      "grad_norm": 0.6858662366867065,
      "learning_rate": 8.461348684210527e-06,
      "loss": 0.1663,
      "step": 3750
    },
    {
      "epoch": 1.5460526315789473,
      "grad_norm": 0.3100195527076721,
      "learning_rate": 8.457236842105265e-06,
      "loss": 0.5894,
      "step": 3760
    },
    {
      "epoch": 1.5501644736842106,
      "grad_norm": 29.110191345214844,
      "learning_rate": 8.453125e-06,
      "loss": 0.6844,
      "step": 3770
    },
    {
      "epoch": 1.5542763157894737,
      "grad_norm": 84.61830139160156,
      "learning_rate": 8.449013157894738e-06,
      "loss": 0.8012,
      "step": 3780
    },
    {
      "epoch": 1.5583881578947367,
      "grad_norm": 69.99295043945312,
      "learning_rate": 8.444901315789474e-06,
      "loss": 0.4886,
      "step": 3790
    },
    {
      "epoch": 1.5625,
      "grad_norm": 28.908971786499023,
      "learning_rate": 8.440789473684211e-06,
      "loss": 0.3999,
      "step": 3800
    },
    {
      "epoch": 1.5666118421052633,
      "grad_norm": 0.005273888353258371,
      "learning_rate": 8.436677631578949e-06,
      "loss": 0.5796,
      "step": 3810
    },
    {
      "epoch": 1.5707236842105263,
      "grad_norm": 0.5902597308158875,
      "learning_rate": 8.432565789473684e-06,
      "loss": 0.4326,
      "step": 3820
    },
    {
      "epoch": 1.5748355263157894,
      "grad_norm": 17.558439254760742,
      "learning_rate": 8.428453947368422e-06,
      "loss": 0.1512,
      "step": 3830
    },
    {
      "epoch": 1.5789473684210527,
      "grad_norm": 28.71281623840332,
      "learning_rate": 8.42434210526316e-06,
      "loss": 0.0765,
      "step": 3840
    },
    {
      "epoch": 1.583059210526316,
      "grad_norm": 0.1115703135728836,
      "learning_rate": 8.420230263157895e-06,
      "loss": 0.8147,
      "step": 3850
    },
    {
      "epoch": 1.587171052631579,
      "grad_norm": 105.5567626953125,
      "learning_rate": 8.416118421052633e-06,
      "loss": 0.7391,
      "step": 3860
    },
    {
      "epoch": 1.591282894736842,
      "grad_norm": 15.565473556518555,
      "learning_rate": 8.41200657894737e-06,
      "loss": 0.6355,
      "step": 3870
    },
    {
      "epoch": 1.5953947368421053,
      "grad_norm": 29.326662063598633,
      "learning_rate": 8.407894736842106e-06,
      "loss": 0.3711,
      "step": 3880
    },
    {
      "epoch": 1.5995065789473686,
      "grad_norm": 37.79521560668945,
      "learning_rate": 8.403782894736843e-06,
      "loss": 0.302,
      "step": 3890
    },
    {
      "epoch": 1.6036184210526314,
      "grad_norm": 9.726346015930176,
      "learning_rate": 8.399671052631579e-06,
      "loss": 0.6002,
      "step": 3900
    },
    {
      "epoch": 1.6077302631578947,
      "grad_norm": 47.580997467041016,
      "learning_rate": 8.395559210526316e-06,
      "loss": 0.5547,
      "step": 3910
    },
    {
      "epoch": 1.611842105263158,
      "grad_norm": 35.97808837890625,
      "learning_rate": 8.391447368421052e-06,
      "loss": 0.489,
      "step": 3920
    },
    {
      "epoch": 1.615953947368421,
      "grad_norm": 16.187610626220703,
      "learning_rate": 8.38733552631579e-06,
      "loss": 0.194,
      "step": 3930
    },
    {
      "epoch": 1.620065789473684,
      "grad_norm": 53.040794372558594,
      "learning_rate": 8.383223684210527e-06,
      "loss": 0.2443,
      "step": 3940
    },
    {
      "epoch": 1.6241776315789473,
      "grad_norm": 55.21927261352539,
      "learning_rate": 8.379111842105265e-06,
      "loss": 0.5103,
      "step": 3950
    },
    {
      "epoch": 1.6282894736842106,
      "grad_norm": 41.11189651489258,
      "learning_rate": 8.375e-06,
      "loss": 0.4835,
      "step": 3960
    },
    {
      "epoch": 1.6324013157894737,
      "grad_norm": 3.0493428707122803,
      "learning_rate": 8.370888157894738e-06,
      "loss": 0.2295,
      "step": 3970
    },
    {
      "epoch": 1.6365131578947367,
      "grad_norm": 4.010314464569092,
      "learning_rate": 8.366776315789475e-06,
      "loss": 0.2899,
      "step": 3980
    },
    {
      "epoch": 1.640625,
      "grad_norm": 1.733955979347229,
      "learning_rate": 8.362664473684211e-06,
      "loss": 0.3304,
      "step": 3990
    },
    {
      "epoch": 1.6447368421052633,
      "grad_norm": 0.39239126443862915,
      "learning_rate": 8.358552631578948e-06,
      "loss": 0.7172,
      "step": 4000
    },
    {
      "epoch": 1.6488486842105263,
      "grad_norm": 19.84231185913086,
      "learning_rate": 8.354440789473684e-06,
      "loss": 0.6881,
      "step": 4010
    },
    {
      "epoch": 1.6529605263157894,
      "grad_norm": 3.06638503074646,
      "learning_rate": 8.350328947368422e-06,
      "loss": 0.314,
      "step": 4020
    },
    {
      "epoch": 1.6570723684210527,
      "grad_norm": 0.9069366455078125,
      "learning_rate": 8.346217105263159e-06,
      "loss": 0.3215,
      "step": 4030
    },
    {
      "epoch": 1.661184210526316,
      "grad_norm": 61.60841369628906,
      "learning_rate": 8.342105263157897e-06,
      "loss": 0.5228,
      "step": 4040
    },
    {
      "epoch": 1.665296052631579,
      "grad_norm": 0.013282466679811478,
      "learning_rate": 8.337993421052632e-06,
      "loss": 0.4712,
      "step": 4050
    },
    {
      "epoch": 1.669407894736842,
      "grad_norm": 2.0385208129882812,
      "learning_rate": 8.33388157894737e-06,
      "loss": 0.0717,
      "step": 4060
    },
    {
      "epoch": 1.6735197368421053,
      "grad_norm": 31.112775802612305,
      "learning_rate": 8.329769736842106e-06,
      "loss": 0.2635,
      "step": 4070
    },
    {
      "epoch": 1.6776315789473686,
      "grad_norm": 83.4158935546875,
      "learning_rate": 8.325657894736843e-06,
      "loss": 0.3961,
      "step": 4080
    },
    {
      "epoch": 1.6817434210526314,
      "grad_norm": 0.01563643105328083,
      "learning_rate": 8.321546052631579e-06,
      "loss": 0.142,
      "step": 4090
    },
    {
      "epoch": 1.6858552631578947,
      "grad_norm": 0.2154001146554947,
      "learning_rate": 8.317434210526316e-06,
      "loss": 0.0995,
      "step": 4100
    },
    {
      "epoch": 1.689967105263158,
      "grad_norm": 0.23853525519371033,
      "learning_rate": 8.313322368421054e-06,
      "loss": 0.2759,
      "step": 4110
    },
    {
      "epoch": 1.694078947368421,
      "grad_norm": 35.838157653808594,
      "learning_rate": 8.30921052631579e-06,
      "loss": 0.541,
      "step": 4120
    },
    {
      "epoch": 1.698190789473684,
      "grad_norm": 73.19493865966797,
      "learning_rate": 8.305098684210527e-06,
      "loss": 0.2733,
      "step": 4130
    },
    {
      "epoch": 1.7023026315789473,
      "grad_norm": 0.04348941892385483,
      "learning_rate": 8.300986842105264e-06,
      "loss": 0.4615,
      "step": 4140
    },
    {
      "epoch": 1.7064144736842106,
      "grad_norm": 0.1173626184463501,
      "learning_rate": 8.296875e-06,
      "loss": 0.4318,
      "step": 4150
    },
    {
      "epoch": 1.7105263157894737,
      "grad_norm": 4.591619968414307,
      "learning_rate": 8.292763157894738e-06,
      "loss": 0.203,
      "step": 4160
    },
    {
      "epoch": 1.7146381578947367,
      "grad_norm": 70.83255767822266,
      "learning_rate": 8.288651315789475e-06,
      "loss": 0.3747,
      "step": 4170
    },
    {
      "epoch": 1.71875,
      "grad_norm": 17.581378936767578,
      "learning_rate": 8.28453947368421e-06,
      "loss": 0.1498,
      "step": 4180
    },
    {
      "epoch": 1.7228618421052633,
      "grad_norm": 25.320514678955078,
      "learning_rate": 8.280427631578948e-06,
      "loss": 0.4238,
      "step": 4190
    },
    {
      "epoch": 1.7269736842105263,
      "grad_norm": 60.90825271606445,
      "learning_rate": 8.276315789473684e-06,
      "loss": 0.5959,
      "step": 4200
    },
    {
      "epoch": 1.7310855263157894,
      "grad_norm": 2.9256107807159424,
      "learning_rate": 8.272203947368421e-06,
      "loss": 0.5433,
      "step": 4210
    },
    {
      "epoch": 1.7351973684210527,
      "grad_norm": 0.1645122617483139,
      "learning_rate": 8.268092105263159e-06,
      "loss": 0.3483,
      "step": 4220
    },
    {
      "epoch": 1.739309210526316,
      "grad_norm": 0.030659016221761703,
      "learning_rate": 8.263980263157896e-06,
      "loss": 0.6285,
      "step": 4230
    },
    {
      "epoch": 1.743421052631579,
      "grad_norm": 67.81155395507812,
      "learning_rate": 8.259868421052632e-06,
      "loss": 1.1265,
      "step": 4240
    },
    {
      "epoch": 1.747532894736842,
      "grad_norm": 0.02041751518845558,
      "learning_rate": 8.25575657894737e-06,
      "loss": 0.3916,
      "step": 4250
    },
    {
      "epoch": 1.7516447368421053,
      "grad_norm": 13.793588638305664,
      "learning_rate": 8.251644736842105e-06,
      "loss": 0.2395,
      "step": 4260
    },
    {
      "epoch": 1.7557565789473686,
      "grad_norm": 11.824005126953125,
      "learning_rate": 8.247532894736843e-06,
      "loss": 0.3163,
      "step": 4270
    },
    {
      "epoch": 1.7598684210526314,
      "grad_norm": 6.396011829376221,
      "learning_rate": 8.24342105263158e-06,
      "loss": 0.3761,
      "step": 4280
    },
    {
      "epoch": 1.7639802631578947,
      "grad_norm": 23.77560806274414,
      "learning_rate": 8.239309210526316e-06,
      "loss": 0.632,
      "step": 4290
    },
    {
      "epoch": 1.768092105263158,
      "grad_norm": 10.43442440032959,
      "learning_rate": 8.235197368421053e-06,
      "loss": 0.2882,
      "step": 4300
    },
    {
      "epoch": 1.772203947368421,
      "grad_norm": 42.35690689086914,
      "learning_rate": 8.23108552631579e-06,
      "loss": 0.5975,
      "step": 4310
    },
    {
      "epoch": 1.776315789473684,
      "grad_norm": 1.9145148992538452,
      "learning_rate": 8.226973684210527e-06,
      "loss": 0.1458,
      "step": 4320
    },
    {
      "epoch": 1.7804276315789473,
      "grad_norm": 0.08790173381567001,
      "learning_rate": 8.222861842105264e-06,
      "loss": 0.5101,
      "step": 4330
    },
    {
      "epoch": 1.7845394736842106,
      "grad_norm": 31.585073471069336,
      "learning_rate": 8.218750000000002e-06,
      "loss": 0.7057,
      "step": 4340
    },
    {
      "epoch": 1.7886513157894737,
      "grad_norm": 1.391623616218567,
      "learning_rate": 8.214638157894737e-06,
      "loss": 0.1948,
      "step": 4350
    },
    {
      "epoch": 1.7927631578947367,
      "grad_norm": 50.8994026184082,
      "learning_rate": 8.210526315789475e-06,
      "loss": 0.2784,
      "step": 4360
    },
    {
      "epoch": 1.796875,
      "grad_norm": 61.71848678588867,
      "learning_rate": 8.20641447368421e-06,
      "loss": 0.6412,
      "step": 4370
    },
    {
      "epoch": 1.8009868421052633,
      "grad_norm": 3.166621208190918,
      "learning_rate": 8.202302631578948e-06,
      "loss": 0.484,
      "step": 4380
    },
    {
      "epoch": 1.8050986842105263,
      "grad_norm": 60.430580139160156,
      "learning_rate": 8.198190789473684e-06,
      "loss": 0.5307,
      "step": 4390
    },
    {
      "epoch": 1.8092105263157894,
      "grad_norm": 0.003818509867414832,
      "learning_rate": 8.194078947368421e-06,
      "loss": 0.0689,
      "step": 4400
    },
    {
      "epoch": 1.8133223684210527,
      "grad_norm": 28.11445426940918,
      "learning_rate": 8.189967105263159e-06,
      "loss": 0.5673,
      "step": 4410
    },
    {
      "epoch": 1.817434210526316,
      "grad_norm": 1.9147342443466187,
      "learning_rate": 8.185855263157896e-06,
      "loss": 0.3738,
      "step": 4420
    },
    {
      "epoch": 1.821546052631579,
      "grad_norm": 75.0318832397461,
      "learning_rate": 8.181743421052632e-06,
      "loss": 0.6623,
      "step": 4430
    },
    {
      "epoch": 1.825657894736842,
      "grad_norm": 0.37105000019073486,
      "learning_rate": 8.17763157894737e-06,
      "loss": 0.4921,
      "step": 4440
    },
    {
      "epoch": 1.8297697368421053,
      "grad_norm": 48.0621452331543,
      "learning_rate": 8.173519736842107e-06,
      "loss": 0.1337,
      "step": 4450
    },
    {
      "epoch": 1.8338815789473686,
      "grad_norm": 80.51526641845703,
      "learning_rate": 8.169407894736843e-06,
      "loss": 0.3979,
      "step": 4460
    },
    {
      "epoch": 1.8379934210526314,
      "grad_norm": 43.526065826416016,
      "learning_rate": 8.16529605263158e-06,
      "loss": 0.4044,
      "step": 4470
    },
    {
      "epoch": 1.8421052631578947,
      "grad_norm": 47.46805953979492,
      "learning_rate": 8.161184210526316e-06,
      "loss": 0.215,
      "step": 4480
    },
    {
      "epoch": 1.846217105263158,
      "grad_norm": 48.90501022338867,
      "learning_rate": 8.157072368421053e-06,
      "loss": 0.6912,
      "step": 4490
    },
    {
      "epoch": 1.850328947368421,
      "grad_norm": 9.779630661010742,
      "learning_rate": 8.152960526315789e-06,
      "loss": 0.4418,
      "step": 4500
    },
    {
      "epoch": 1.854440789473684,
      "grad_norm": 0.7416747212409973,
      "learning_rate": 8.148848684210528e-06,
      "loss": 0.678,
      "step": 4510
    },
    {
      "epoch": 1.8585526315789473,
      "grad_norm": 0.008086485788226128,
      "learning_rate": 8.144736842105264e-06,
      "loss": 0.6166,
      "step": 4520
    },
    {
      "epoch": 1.8626644736842106,
      "grad_norm": 4.0084967613220215,
      "learning_rate": 8.140625000000001e-06,
      "loss": 0.1721,
      "step": 4530
    },
    {
      "epoch": 1.8667763157894737,
      "grad_norm": 0.25842565298080444,
      "learning_rate": 8.136513157894737e-06,
      "loss": 0.0703,
      "step": 4540
    },
    {
      "epoch": 1.8708881578947367,
      "grad_norm": 27.43667221069336,
      "learning_rate": 8.132401315789475e-06,
      "loss": 0.3083,
      "step": 4550
    },
    {
      "epoch": 1.875,
      "grad_norm": 48.5318489074707,
      "learning_rate": 8.12828947368421e-06,
      "loss": 0.2443,
      "step": 4560
    },
    {
      "epoch": 1.8791118421052633,
      "grad_norm": 0.08080202341079712,
      "learning_rate": 8.124177631578948e-06,
      "loss": 0.6487,
      "step": 4570
    },
    {
      "epoch": 1.8832236842105263,
      "grad_norm": 0.3895185589790344,
      "learning_rate": 8.120065789473685e-06,
      "loss": 0.1034,
      "step": 4580
    },
    {
      "epoch": 1.8873355263157894,
      "grad_norm": 1.0280920267105103,
      "learning_rate": 8.115953947368421e-06,
      "loss": 0.307,
      "step": 4590
    },
    {
      "epoch": 1.8914473684210527,
      "grad_norm": 65.11549377441406,
      "learning_rate": 8.111842105263158e-06,
      "loss": 0.4791,
      "step": 4600
    },
    {
      "epoch": 1.895559210526316,
      "grad_norm": 74.55525970458984,
      "learning_rate": 8.107730263157896e-06,
      "loss": 0.6433,
      "step": 4610
    },
    {
      "epoch": 1.899671052631579,
      "grad_norm": 22.78346824645996,
      "learning_rate": 8.103618421052632e-06,
      "loss": 0.786,
      "step": 4620
    },
    {
      "epoch": 1.903782894736842,
      "grad_norm": 86.56382751464844,
      "learning_rate": 8.099506578947369e-06,
      "loss": 0.6106,
      "step": 4630
    },
    {
      "epoch": 1.9078947368421053,
      "grad_norm": 0.2657437324523926,
      "learning_rate": 8.095394736842107e-06,
      "loss": 0.6693,
      "step": 4640
    },
    {
      "epoch": 1.9120065789473686,
      "grad_norm": 0.009138388559222221,
      "learning_rate": 8.091282894736842e-06,
      "loss": 0.2074,
      "step": 4650
    },
    {
      "epoch": 1.9161184210526314,
      "grad_norm": 12.050437927246094,
      "learning_rate": 8.08717105263158e-06,
      "loss": 0.3525,
      "step": 4660
    },
    {
      "epoch": 1.9202302631578947,
      "grad_norm": 85.17498016357422,
      "learning_rate": 8.083059210526316e-06,
      "loss": 0.1985,
      "step": 4670
    },
    {
      "epoch": 1.924342105263158,
      "grad_norm": 0.09853484481573105,
      "learning_rate": 8.078947368421053e-06,
      "loss": 0.0842,
      "step": 4680
    },
    {
      "epoch": 1.928453947368421,
      "grad_norm": 58.58940124511719,
      "learning_rate": 8.074835526315789e-06,
      "loss": 0.556,
      "step": 4690
    },
    {
      "epoch": 1.932565789473684,
      "grad_norm": 0.3376722037792206,
      "learning_rate": 8.070723684210528e-06,
      "loss": 0.3301,
      "step": 4700
    },
    {
      "epoch": 1.9366776315789473,
      "grad_norm": 61.81422805786133,
      "learning_rate": 8.066611842105264e-06,
      "loss": 0.5592,
      "step": 4710
    },
    {
      "epoch": 1.9407894736842106,
      "grad_norm": 14.672579765319824,
      "learning_rate": 8.062500000000001e-06,
      "loss": 0.2208,
      "step": 4720
    },
    {
      "epoch": 1.9449013157894737,
      "grad_norm": 11.613296508789062,
      "learning_rate": 8.058388157894737e-06,
      "loss": 0.3177,
      "step": 4730
    },
    {
      "epoch": 1.9490131578947367,
      "grad_norm": 4.859875202178955,
      "learning_rate": 8.054276315789474e-06,
      "loss": 0.1121,
      "step": 4740
    },
    {
      "epoch": 1.953125,
      "grad_norm": 1.6204088926315308,
      "learning_rate": 8.050164473684212e-06,
      "loss": 0.3163,
      "step": 4750
    },
    {
      "epoch": 1.9572368421052633,
      "grad_norm": 22.201229095458984,
      "learning_rate": 8.046052631578948e-06,
      "loss": 0.4466,
      "step": 4760
    },
    {
      "epoch": 1.9613486842105263,
      "grad_norm": 28.351089477539062,
      "learning_rate": 8.041940789473685e-06,
      "loss": 0.1054,
      "step": 4770
    },
    {
      "epoch": 1.9654605263157894,
      "grad_norm": 47.56678771972656,
      "learning_rate": 8.03782894736842e-06,
      "loss": 0.2095,
      "step": 4780
    },
    {
      "epoch": 1.9695723684210527,
      "grad_norm": 3.671384572982788,
      "learning_rate": 8.033717105263158e-06,
      "loss": 0.638,
      "step": 4790
    },
    {
      "epoch": 1.973684210526316,
      "grad_norm": 0.8618096709251404,
      "learning_rate": 8.029605263157896e-06,
      "loss": 0.4759,
      "step": 4800
    },
    {
      "epoch": 1.977796052631579,
      "grad_norm": 0.04880933091044426,
      "learning_rate": 8.025493421052633e-06,
      "loss": 0.6423,
      "step": 4810
    },
    {
      "epoch": 1.981907894736842,
      "grad_norm": 0.03361703082919121,
      "learning_rate": 8.021381578947369e-06,
      "loss": 0.8028,
      "step": 4820
    },
    {
      "epoch": 1.9860197368421053,
      "grad_norm": 0.3052118718624115,
      "learning_rate": 8.017269736842106e-06,
      "loss": 0.0253,
      "step": 4830
    },
    {
      "epoch": 1.9901315789473686,
      "grad_norm": 0.035156674683094025,
      "learning_rate": 8.013157894736842e-06,
      "loss": 0.1967,
      "step": 4840
    },
    {
      "epoch": 1.9942434210526314,
      "grad_norm": 81.99966430664062,
      "learning_rate": 8.00904605263158e-06,
      "loss": 0.5797,
      "step": 4850
    },
    {
      "epoch": 1.9983552631578947,
      "grad_norm": 0.685602068901062,
      "learning_rate": 8.004934210526315e-06,
      "loss": 0.309,
      "step": 4860
    },
    {
      "epoch": 2.002467105263158,
      "grad_norm": 37.46296310424805,
      "learning_rate": 8.000822368421053e-06,
      "loss": 0.2657,
      "step": 4870
    },
    {
      "epoch": 2.0065789473684212,
      "grad_norm": 0.056410618126392365,
      "learning_rate": 7.99671052631579e-06,
      "loss": 0.3954,
      "step": 4880
    },
    {
      "epoch": 2.010690789473684,
      "grad_norm": 112.95884704589844,
      "learning_rate": 7.992598684210528e-06,
      "loss": 0.3842,
      "step": 4890
    },
    {
      "epoch": 2.0148026315789473,
      "grad_norm": 0.022623682394623756,
      "learning_rate": 7.988486842105263e-06,
      "loss": 0.5508,
      "step": 4900
    },
    {
      "epoch": 2.0189144736842106,
      "grad_norm": 58.53015899658203,
      "learning_rate": 7.984375000000001e-06,
      "loss": 0.2577,
      "step": 4910
    },
    {
      "epoch": 2.023026315789474,
      "grad_norm": 7.032954216003418,
      "learning_rate": 7.980263157894738e-06,
      "loss": 0.3766,
      "step": 4920
    },
    {
      "epoch": 2.0271381578947367,
      "grad_norm": 0.21705050766468048,
      "learning_rate": 7.976151315789474e-06,
      "loss": 0.1742,
      "step": 4930
    },
    {
      "epoch": 2.03125,
      "grad_norm": 0.012392381206154823,
      "learning_rate": 7.972039473684212e-06,
      "loss": 0.144,
      "step": 4940
    },
    {
      "epoch": 2.0353618421052633,
      "grad_norm": 46.255126953125,
      "learning_rate": 7.967927631578947e-06,
      "loss": 0.036,
      "step": 4950
    },
    {
      "epoch": 2.039473684210526,
      "grad_norm": 0.034974802285432816,
      "learning_rate": 7.963815789473685e-06,
      "loss": 0.0679,
      "step": 4960
    },
    {
      "epoch": 2.0435855263157894,
      "grad_norm": 0.04458899050951004,
      "learning_rate": 7.95970394736842e-06,
      "loss": 0.0958,
      "step": 4970
    },
    {
      "epoch": 2.0476973684210527,
      "grad_norm": 47.49546813964844,
      "learning_rate": 7.95559210526316e-06,
      "loss": 0.5122,
      "step": 4980
    },
    {
      "epoch": 2.051809210526316,
      "grad_norm": 33.729793548583984,
      "learning_rate": 7.951480263157896e-06,
      "loss": 0.6454,
      "step": 4990
    },
    {
      "epoch": 2.0559210526315788,
      "grad_norm": 12.397784233093262,
      "learning_rate": 7.947368421052633e-06,
      "loss": 0.0289,
      "step": 5000
    },
    {
      "epoch": 2.060032894736842,
      "grad_norm": 45.05794906616211,
      "learning_rate": 7.943256578947369e-06,
      "loss": 0.1298,
      "step": 5010
    },
    {
      "epoch": 2.0641447368421053,
      "grad_norm": 57.769283294677734,
      "learning_rate": 7.939144736842106e-06,
      "loss": 0.4246,
      "step": 5020
    },
    {
      "epoch": 2.0682565789473686,
      "grad_norm": 28.973388671875,
      "learning_rate": 7.935032894736842e-06,
      "loss": 0.2389,
      "step": 5030
    },
    {
      "epoch": 2.0723684210526314,
      "grad_norm": 56.98832321166992,
      "learning_rate": 7.93092105263158e-06,
      "loss": 0.1893,
      "step": 5040
    },
    {
      "epoch": 2.0764802631578947,
      "grad_norm": 56.96243667602539,
      "learning_rate": 7.926809210526317e-06,
      "loss": 0.2976,
      "step": 5050
    },
    {
      "epoch": 2.080592105263158,
      "grad_norm": 55.864070892333984,
      "learning_rate": 7.922697368421053e-06,
      "loss": 0.4408,
      "step": 5060
    },
    {
      "epoch": 2.0847039473684212,
      "grad_norm": 1.9658788442611694,
      "learning_rate": 7.91858552631579e-06,
      "loss": 0.1167,
      "step": 5070
    },
    {
      "epoch": 2.088815789473684,
      "grad_norm": 1.0686661005020142,
      "learning_rate": 7.914473684210528e-06,
      "loss": 0.1516,
      "step": 5080
    },
    {
      "epoch": 2.0929276315789473,
      "grad_norm": 2.143155813217163,
      "learning_rate": 7.910361842105265e-06,
      "loss": 0.1635,
      "step": 5090
    },
    {
      "epoch": 2.0970394736842106,
      "grad_norm": 52.307498931884766,
      "learning_rate": 7.90625e-06,
      "loss": 0.2435,
      "step": 5100
    },
    {
      "epoch": 2.101151315789474,
      "grad_norm": 2.929649829864502,
      "learning_rate": 7.902138157894738e-06,
      "loss": 0.327,
      "step": 5110
    },
    {
      "epoch": 2.1052631578947367,
      "grad_norm": 3.452099561691284,
      "learning_rate": 7.898026315789474e-06,
      "loss": 0.2144,
      "step": 5120
    },
    {
      "epoch": 2.109375,
      "grad_norm": 18.331693649291992,
      "learning_rate": 7.893914473684211e-06,
      "loss": 0.1962,
      "step": 5130
    },
    {
      "epoch": 2.1134868421052633,
      "grad_norm": 6.536662578582764,
      "learning_rate": 7.889802631578947e-06,
      "loss": 0.0614,
      "step": 5140
    },
    {
      "epoch": 2.117598684210526,
      "grad_norm": 0.015528288669884205,
      "learning_rate": 7.885690789473685e-06,
      "loss": 0.218,
      "step": 5150
    },
    {
      "epoch": 2.1217105263157894,
      "grad_norm": 0.06858813762664795,
      "learning_rate": 7.88157894736842e-06,
      "loss": 0.3551,
      "step": 5160
    },
    {
      "epoch": 2.1258223684210527,
      "grad_norm": 2.6634881496429443,
      "learning_rate": 7.87746710526316e-06,
      "loss": 0.2286,
      "step": 5170
    },
    {
      "epoch": 2.129934210526316,
      "grad_norm": 0.09859636425971985,
      "learning_rate": 7.873355263157895e-06,
      "loss": 0.6348,
      "step": 5180
    },
    {
      "epoch": 2.1340460526315788,
      "grad_norm": 0.701995313167572,
      "learning_rate": 7.869243421052633e-06,
      "loss": 0.1765,
      "step": 5190
    },
    {
      "epoch": 2.138157894736842,
      "grad_norm": 0.037273943424224854,
      "learning_rate": 7.865131578947369e-06,
      "loss": 0.1263,
      "step": 5200
    },
    {
      "epoch": 2.1422697368421053,
      "grad_norm": 0.040880296379327774,
      "learning_rate": 7.861019736842106e-06,
      "loss": 0.1073,
      "step": 5210
    },
    {
      "epoch": 2.1463815789473686,
      "grad_norm": 77.20631408691406,
      "learning_rate": 7.856907894736843e-06,
      "loss": 0.4358,
      "step": 5220
    },
    {
      "epoch": 2.1504934210526314,
      "grad_norm": 3.0592525005340576,
      "learning_rate": 7.85279605263158e-06,
      "loss": 0.3378,
      "step": 5230
    },
    {
      "epoch": 2.1546052631578947,
      "grad_norm": 2.437439441680908,
      "learning_rate": 7.848684210526317e-06,
      "loss": 0.3702,
      "step": 5240
    },
    {
      "epoch": 2.158717105263158,
      "grad_norm": 0.14886601269245148,
      "learning_rate": 7.844572368421052e-06,
      "loss": 0.1371,
      "step": 5250
    },
    {
      "epoch": 2.1628289473684212,
      "grad_norm": 45.28163146972656,
      "learning_rate": 7.84046052631579e-06,
      "loss": 0.3802,
      "step": 5260
    },
    {
      "epoch": 2.166940789473684,
      "grad_norm": 72.54158782958984,
      "learning_rate": 7.836348684210527e-06,
      "loss": 0.4916,
      "step": 5270
    },
    {
      "epoch": 2.1710526315789473,
      "grad_norm": 2.2983322143554688,
      "learning_rate": 7.832236842105265e-06,
      "loss": 0.1063,
      "step": 5280
    },
    {
      "epoch": 2.1751644736842106,
      "grad_norm": 41.453407287597656,
      "learning_rate": 7.828125e-06,
      "loss": 0.8019,
      "step": 5290
    },
    {
      "epoch": 2.1792763157894735,
      "grad_norm": 20.001136779785156,
      "learning_rate": 7.824013157894738e-06,
      "loss": 0.0648,
      "step": 5300
    },
    {
      "epoch": 2.1833881578947367,
      "grad_norm": 0.0267824474722147,
      "learning_rate": 7.819901315789474e-06,
      "loss": 0.2172,
      "step": 5310
    },
    {
      "epoch": 2.1875,
      "grad_norm": 79.87635803222656,
      "learning_rate": 7.815789473684211e-06,
      "loss": 0.9097,
      "step": 5320
    },
    {
      "epoch": 2.1916118421052633,
      "grad_norm": 0.06062754988670349,
      "learning_rate": 7.811677631578947e-06,
      "loss": 0.4808,
      "step": 5330
    },
    {
      "epoch": 2.1957236842105265,
      "grad_norm": 0.18611417710781097,
      "learning_rate": 7.807565789473684e-06,
      "loss": 0.3803,
      "step": 5340
    },
    {
      "epoch": 2.1998355263157894,
      "grad_norm": 0.8481597900390625,
      "learning_rate": 7.803453947368422e-06,
      "loss": 0.3331,
      "step": 5350
    },
    {
      "epoch": 2.2039473684210527,
      "grad_norm": 14.719058990478516,
      "learning_rate": 7.79934210526316e-06,
      "loss": 0.4904,
      "step": 5360
    },
    {
      "epoch": 2.208059210526316,
      "grad_norm": 12.377104759216309,
      "learning_rate": 7.795230263157895e-06,
      "loss": 0.1892,
      "step": 5370
    },
    {
      "epoch": 2.2121710526315788,
      "grad_norm": 16.943613052368164,
      "learning_rate": 7.791118421052633e-06,
      "loss": 0.2359,
      "step": 5380
    },
    {
      "epoch": 2.216282894736842,
      "grad_norm": 77.85987091064453,
      "learning_rate": 7.78700657894737e-06,
      "loss": 0.772,
      "step": 5390
    },
    {
      "epoch": 2.2203947368421053,
      "grad_norm": 66.15851593017578,
      "learning_rate": 7.782894736842106e-06,
      "loss": 0.1909,
      "step": 5400
    },
    {
      "epoch": 2.2245065789473686,
      "grad_norm": 67.50369262695312,
      "learning_rate": 7.778782894736843e-06,
      "loss": 0.192,
      "step": 5410
    },
    {
      "epoch": 2.2286184210526314,
      "grad_norm": 0.5703405141830444,
      "learning_rate": 7.774671052631579e-06,
      "loss": 0.0897,
      "step": 5420
    },
    {
      "epoch": 2.2327302631578947,
      "grad_norm": 52.4270133972168,
      "learning_rate": 7.770559210526316e-06,
      "loss": 0.0543,
      "step": 5430
    },
    {
      "epoch": 2.236842105263158,
      "grad_norm": 77.67536163330078,
      "learning_rate": 7.766447368421052e-06,
      "loss": 0.3614,
      "step": 5440
    },
    {
      "epoch": 2.2409539473684212,
      "grad_norm": 61.79861831665039,
      "learning_rate": 7.76233552631579e-06,
      "loss": 0.0899,
      "step": 5450
    },
    {
      "epoch": 2.245065789473684,
      "grad_norm": 0.18140049278736115,
      "learning_rate": 7.758223684210527e-06,
      "loss": 0.411,
      "step": 5460
    },
    {
      "epoch": 2.2491776315789473,
      "grad_norm": 37.416351318359375,
      "learning_rate": 7.754111842105265e-06,
      "loss": 0.1519,
      "step": 5470
    },
    {
      "epoch": 2.2532894736842106,
      "grad_norm": 0.24811792373657227,
      "learning_rate": 7.75e-06,
      "loss": 0.049,
      "step": 5480
    },
    {
      "epoch": 2.2574013157894735,
      "grad_norm": 0.09098932147026062,
      "learning_rate": 7.745888157894738e-06,
      "loss": 0.1216,
      "step": 5490
    },
    {
      "epoch": 2.2615131578947367,
      "grad_norm": 0.39087772369384766,
      "learning_rate": 7.741776315789474e-06,
      "loss": 0.284,
      "step": 5500
    },
    {
      "epoch": 2.265625,
      "grad_norm": 80.84004974365234,
      "learning_rate": 7.737664473684211e-06,
      "loss": 0.2081,
      "step": 5510
    },
    {
      "epoch": 2.2697368421052633,
      "grad_norm": 0.9717281460762024,
      "learning_rate": 7.733552631578948e-06,
      "loss": 0.3674,
      "step": 5520
    },
    {
      "epoch": 2.2738486842105265,
      "grad_norm": 0.6159382462501526,
      "learning_rate": 7.729440789473684e-06,
      "loss": 0.1791,
      "step": 5530
    },
    {
      "epoch": 2.2779605263157894,
      "grad_norm": 0.013826197013258934,
      "learning_rate": 7.725328947368422e-06,
      "loss": 0.0343,
      "step": 5540
    },
    {
      "epoch": 2.2820723684210527,
      "grad_norm": 78.19012451171875,
      "learning_rate": 7.721217105263159e-06,
      "loss": 0.5983,
      "step": 5550
    },
    {
      "epoch": 2.286184210526316,
      "grad_norm": 0.009286273270845413,
      "learning_rate": 7.717105263157897e-06,
      "loss": 0.1152,
      "step": 5560
    },
    {
      "epoch": 2.2902960526315788,
      "grad_norm": 0.7038535475730896,
      "learning_rate": 7.712993421052632e-06,
      "loss": 0.0342,
      "step": 5570
    },
    {
      "epoch": 2.294407894736842,
      "grad_norm": 16.376056671142578,
      "learning_rate": 7.70888157894737e-06,
      "loss": 0.2536,
      "step": 5580
    },
    {
      "epoch": 2.2985197368421053,
      "grad_norm": 75.77314758300781,
      "learning_rate": 7.704769736842106e-06,
      "loss": 0.967,
      "step": 5590
    },
    {
      "epoch": 2.3026315789473686,
      "grad_norm": 0.04217597842216492,
      "learning_rate": 7.700657894736843e-06,
      "loss": 0.0899,
      "step": 5600
    },
    {
      "epoch": 2.3067434210526314,
      "grad_norm": 0.0018036451656371355,
      "learning_rate": 7.696546052631579e-06,
      "loss": 0.0971,
      "step": 5610
    },
    {
      "epoch": 2.3108552631578947,
      "grad_norm": 64.49011993408203,
      "learning_rate": 7.692434210526316e-06,
      "loss": 0.5319,
      "step": 5620
    },
    {
      "epoch": 2.314967105263158,
      "grad_norm": 6.954440593719482,
      "learning_rate": 7.688322368421052e-06,
      "loss": 0.4004,
      "step": 5630
    },
    {
      "epoch": 2.3190789473684212,
      "grad_norm": 83.60736083984375,
      "learning_rate": 7.68421052631579e-06,
      "loss": 0.2637,
      "step": 5640
    },
    {
      "epoch": 2.323190789473684,
      "grad_norm": 0.47764426469802856,
      "learning_rate": 7.680098684210527e-06,
      "loss": 0.2358,
      "step": 5650
    },
    {
      "epoch": 2.3273026315789473,
      "grad_norm": 0.11332187056541443,
      "learning_rate": 7.675986842105264e-06,
      "loss": 0.1717,
      "step": 5660
    },
    {
      "epoch": 2.3314144736842106,
      "grad_norm": 21.913246154785156,
      "learning_rate": 7.671875e-06,
      "loss": 0.3004,
      "step": 5670
    },
    {
      "epoch": 2.3355263157894735,
      "grad_norm": 0.553477942943573,
      "learning_rate": 7.667763157894738e-06,
      "loss": 0.1921,
      "step": 5680
    },
    {
      "epoch": 2.3396381578947367,
      "grad_norm": 94.19501495361328,
      "learning_rate": 7.663651315789475e-06,
      "loss": 0.2609,
      "step": 5690
    },
    {
      "epoch": 2.34375,
      "grad_norm": 21.826440811157227,
      "learning_rate": 7.65953947368421e-06,
      "loss": 0.0104,
      "step": 5700
    },
    {
      "epoch": 2.3478618421052633,
      "grad_norm": 4.665867328643799,
      "learning_rate": 7.655427631578948e-06,
      "loss": 0.3106,
      "step": 5710
    },
    {
      "epoch": 2.3519736842105265,
      "grad_norm": 3.169865846633911,
      "learning_rate": 7.651315789473684e-06,
      "loss": 0.2123,
      "step": 5720
    },
    {
      "epoch": 2.3560855263157894,
      "grad_norm": 64.12752532958984,
      "learning_rate": 7.647203947368421e-06,
      "loss": 0.101,
      "step": 5730
    },
    {
      "epoch": 2.3601973684210527,
      "grad_norm": 0.16324739158153534,
      "learning_rate": 7.643092105263159e-06,
      "loss": 0.2564,
      "step": 5740
    },
    {
      "epoch": 2.364309210526316,
      "grad_norm": 0.007978091016411781,
      "learning_rate": 7.638980263157896e-06,
      "loss": 0.0914,
      "step": 5750
    },
    {
      "epoch": 2.3684210526315788,
      "grad_norm": 6.777188301086426,
      "learning_rate": 7.634868421052632e-06,
      "loss": 0.123,
      "step": 5760
    },
    {
      "epoch": 2.372532894736842,
      "grad_norm": 21.03947639465332,
      "learning_rate": 7.63075657894737e-06,
      "loss": 0.0371,
      "step": 5770
    },
    {
      "epoch": 2.3766447368421053,
      "grad_norm": 0.5910683274269104,
      "learning_rate": 7.626644736842106e-06,
      "loss": 0.0277,
      "step": 5780
    },
    {
      "epoch": 2.3807565789473686,
      "grad_norm": 5.500244140625,
      "learning_rate": 7.622532894736843e-06,
      "loss": 0.3893,
      "step": 5790
    },
    {
      "epoch": 2.3848684210526314,
      "grad_norm": 60.288055419921875,
      "learning_rate": 7.6184210526315794e-06,
      "loss": 0.668,
      "step": 5800
    },
    {
      "epoch": 2.3889802631578947,
      "grad_norm": 0.009432058781385422,
      "learning_rate": 7.614309210526316e-06,
      "loss": 0.0354,
      "step": 5810
    },
    {
      "epoch": 2.393092105263158,
      "grad_norm": 0.014949236996471882,
      "learning_rate": 7.610197368421053e-06,
      "loss": 0.1037,
      "step": 5820
    },
    {
      "epoch": 2.3972039473684212,
      "grad_norm": 2.148726224899292,
      "learning_rate": 7.606085526315789e-06,
      "loss": 0.5846,
      "step": 5830
    },
    {
      "epoch": 2.401315789473684,
      "grad_norm": 2.4273200035095215,
      "learning_rate": 7.6019736842105276e-06,
      "loss": 0.0866,
      "step": 5840
    },
    {
      "epoch": 2.4054276315789473,
      "grad_norm": 70.4505844116211,
      "learning_rate": 7.597861842105264e-06,
      "loss": 0.3915,
      "step": 5850
    },
    {
      "epoch": 2.4095394736842106,
      "grad_norm": 60.22188186645508,
      "learning_rate": 7.593750000000001e-06,
      "loss": 0.5337,
      "step": 5860
    },
    {
      "epoch": 2.4136513157894735,
      "grad_norm": 8.238316535949707,
      "learning_rate": 7.589638157894737e-06,
      "loss": 0.0111,
      "step": 5870
    },
    {
      "epoch": 2.4177631578947367,
      "grad_norm": 14.266987800598145,
      "learning_rate": 7.585526315789474e-06,
      "loss": 0.1612,
      "step": 5880
    },
    {
      "epoch": 2.421875,
      "grad_norm": 0.43882203102111816,
      "learning_rate": 7.581414473684211e-06,
      "loss": 0.3086,
      "step": 5890
    },
    {
      "epoch": 2.4259868421052633,
      "grad_norm": 0.2558574974536896,
      "learning_rate": 7.577302631578948e-06,
      "loss": 0.3342,
      "step": 5900
    },
    {
      "epoch": 2.4300986842105265,
      "grad_norm": 0.09452182054519653,
      "learning_rate": 7.573190789473685e-06,
      "loss": 0.2291,
      "step": 5910
    },
    {
      "epoch": 2.4342105263157894,
      "grad_norm": 0.0014281157637014985,
      "learning_rate": 7.569078947368421e-06,
      "loss": 0.0711,
      "step": 5920
    },
    {
      "epoch": 2.4383223684210527,
      "grad_norm": 37.98161697387695,
      "learning_rate": 7.564967105263159e-06,
      "loss": 0.3051,
      "step": 5930
    },
    {
      "epoch": 2.442434210526316,
      "grad_norm": 3.323979139328003,
      "learning_rate": 7.560855263157895e-06,
      "loss": 0.0245,
      "step": 5940
    },
    {
      "epoch": 2.4465460526315788,
      "grad_norm": 1.3332736492156982,
      "learning_rate": 7.556743421052633e-06,
      "loss": 0.3637,
      "step": 5950
    },
    {
      "epoch": 2.450657894736842,
      "grad_norm": 2.7701385021209717,
      "learning_rate": 7.552631578947369e-06,
      "loss": 0.2454,
      "step": 5960
    },
    {
      "epoch": 2.4547697368421053,
      "grad_norm": 66.12533569335938,
      "learning_rate": 7.548519736842106e-06,
      "loss": 0.4883,
      "step": 5970
    },
    {
      "epoch": 2.4588815789473686,
      "grad_norm": 0.015920128673315048,
      "learning_rate": 7.544407894736843e-06,
      "loss": 0.2759,
      "step": 5980
    },
    {
      "epoch": 2.4629934210526314,
      "grad_norm": 0.8414337635040283,
      "learning_rate": 7.540296052631579e-06,
      "loss": 0.7129,
      "step": 5990
    },
    {
      "epoch": 2.4671052631578947,
      "grad_norm": 14.422956466674805,
      "learning_rate": 7.536184210526316e-06,
      "loss": 0.3284,
      "step": 6000
    },
    {
      "epoch": 2.471217105263158,
      "grad_norm": 0.07774366438388824,
      "learning_rate": 7.5320723684210524e-06,
      "loss": 0.4476,
      "step": 6010
    },
    {
      "epoch": 2.4753289473684212,
      "grad_norm": 0.7951784133911133,
      "learning_rate": 7.52796052631579e-06,
      "loss": 0.3266,
      "step": 6020
    },
    {
      "epoch": 2.479440789473684,
      "grad_norm": 17.519073486328125,
      "learning_rate": 7.523848684210527e-06,
      "loss": 0.2493,
      "step": 6030
    },
    {
      "epoch": 2.4835526315789473,
      "grad_norm": 10.533546447753906,
      "learning_rate": 7.519736842105264e-06,
      "loss": 0.2244,
      "step": 6040
    },
    {
      "epoch": 2.4876644736842106,
      "grad_norm": 0.004280813504010439,
      "learning_rate": 7.5156250000000006e-06,
      "loss": 0.2085,
      "step": 6050
    },
    {
      "epoch": 2.4917763157894735,
      "grad_norm": 5.5901970863342285,
      "learning_rate": 7.511513157894737e-06,
      "loss": 0.0381,
      "step": 6060
    },
    {
      "epoch": 2.4958881578947367,
      "grad_norm": 72.65666961669922,
      "learning_rate": 7.507401315789475e-06,
      "loss": 0.1484,
      "step": 6070
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.09732012450695038,
      "learning_rate": 7.503289473684211e-06,
      "loss": 0.1253,
      "step": 6080
    },
    {
      "epoch": 2.5041118421052633,
      "grad_norm": 1.242153525352478,
      "learning_rate": 7.499177631578948e-06,
      "loss": 0.4389,
      "step": 6090
    },
    {
      "epoch": 2.5082236842105265,
      "grad_norm": 0.5862547159194946,
      "learning_rate": 7.4950657894736845e-06,
      "loss": 0.0854,
      "step": 6100
    },
    {
      "epoch": 2.5123355263157894,
      "grad_norm": 10.629526138305664,
      "learning_rate": 7.490953947368421e-06,
      "loss": 0.0749,
      "step": 6110
    },
    {
      "epoch": 2.5164473684210527,
      "grad_norm": 0.005628309678286314,
      "learning_rate": 7.486842105263159e-06,
      "loss": 0.2196,
      "step": 6120
    },
    {
      "epoch": 2.520559210526316,
      "grad_norm": 0.13058516383171082,
      "learning_rate": 7.482730263157896e-06,
      "loss": 0.1543,
      "step": 6130
    },
    {
      "epoch": 2.5246710526315788,
      "grad_norm": 0.01582922600209713,
      "learning_rate": 7.478618421052633e-06,
      "loss": 0.1814,
      "step": 6140
    },
    {
      "epoch": 2.528782894736842,
      "grad_norm": 0.9425368309020996,
      "learning_rate": 7.474506578947369e-06,
      "loss": 0.4506,
      "step": 6150
    },
    {
      "epoch": 2.5328947368421053,
      "grad_norm": 0.3503407835960388,
      "learning_rate": 7.470394736842106e-06,
      "loss": 0.2199,
      "step": 6160
    },
    {
      "epoch": 2.5370065789473686,
      "grad_norm": 27.474987030029297,
      "learning_rate": 7.466282894736842e-06,
      "loss": 0.2349,
      "step": 6170
    },
    {
      "epoch": 2.5411184210526314,
      "grad_norm": 91.42644500732422,
      "learning_rate": 7.462171052631579e-06,
      "loss": 0.8179,
      "step": 6180
    },
    {
      "epoch": 2.5452302631578947,
      "grad_norm": 66.36519622802734,
      "learning_rate": 7.4580592105263165e-06,
      "loss": 0.547,
      "step": 6190
    },
    {
      "epoch": 2.549342105263158,
      "grad_norm": 92.01309204101562,
      "learning_rate": 7.453947368421053e-06,
      "loss": 0.2581,
      "step": 6200
    },
    {
      "epoch": 2.5534539473684212,
      "grad_norm": 90.81409454345703,
      "learning_rate": 7.44983552631579e-06,
      "loss": 0.7715,
      "step": 6210
    },
    {
      "epoch": 2.557565789473684,
      "grad_norm": 70.20539855957031,
      "learning_rate": 7.445723684210527e-06,
      "loss": 0.1391,
      "step": 6220
    },
    {
      "epoch": 2.5616776315789473,
      "grad_norm": 7.48537015914917,
      "learning_rate": 7.441611842105264e-06,
      "loss": 0.7486,
      "step": 6230
    },
    {
      "epoch": 2.5657894736842106,
      "grad_norm": 0.0066050267778337,
      "learning_rate": 7.437500000000001e-06,
      "loss": 0.6291,
      "step": 6240
    },
    {
      "epoch": 2.5699013157894735,
      "grad_norm": 0.050669994205236435,
      "learning_rate": 7.433388157894738e-06,
      "loss": 0.4743,
      "step": 6250
    },
    {
      "epoch": 2.5740131578947367,
      "grad_norm": 0.9918844699859619,
      "learning_rate": 7.4292763157894744e-06,
      "loss": 0.4746,
      "step": 6260
    },
    {
      "epoch": 2.578125,
      "grad_norm": 69.81021118164062,
      "learning_rate": 7.425164473684211e-06,
      "loss": 0.283,
      "step": 6270
    },
    {
      "epoch": 2.5822368421052633,
      "grad_norm": 0.3285808265209198,
      "learning_rate": 7.421052631578948e-06,
      "loss": 0.575,
      "step": 6280
    },
    {
      "epoch": 2.5863486842105265,
      "grad_norm": 0.7911439538002014,
      "learning_rate": 7.416940789473684e-06,
      "loss": 0.2961,
      "step": 6290
    },
    {
      "epoch": 2.5904605263157894,
      "grad_norm": 8.023531913757324,
      "learning_rate": 7.412828947368421e-06,
      "loss": 0.1326,
      "step": 6300
    },
    {
      "epoch": 2.5945723684210527,
      "grad_norm": 0.36695143580436707,
      "learning_rate": 7.408717105263159e-06,
      "loss": 0.1047,
      "step": 6310
    },
    {
      "epoch": 2.598684210526316,
      "grad_norm": 4.745920658111572,
      "learning_rate": 7.404605263157896e-06,
      "loss": 0.3415,
      "step": 6320
    },
    {
      "epoch": 2.6027960526315788,
      "grad_norm": 14.169811248779297,
      "learning_rate": 7.400493421052632e-06,
      "loss": 0.481,
      "step": 6330
    },
    {
      "epoch": 2.606907894736842,
      "grad_norm": 37.04063034057617,
      "learning_rate": 7.396381578947369e-06,
      "loss": 0.2138,
      "step": 6340
    },
    {
      "epoch": 2.6110197368421053,
      "grad_norm": 72.94132995605469,
      "learning_rate": 7.392269736842106e-06,
      "loss": 0.5681,
      "step": 6350
    },
    {
      "epoch": 2.6151315789473686,
      "grad_norm": 0.2758810520172119,
      "learning_rate": 7.388157894736843e-06,
      "loss": 0.5019,
      "step": 6360
    },
    {
      "epoch": 2.6192434210526314,
      "grad_norm": 22.47516632080078,
      "learning_rate": 7.38404605263158e-06,
      "loss": 0.1939,
      "step": 6370
    },
    {
      "epoch": 2.6233552631578947,
      "grad_norm": 0.08755365759134293,
      "learning_rate": 7.379934210526316e-06,
      "loss": 0.2396,
      "step": 6380
    },
    {
      "epoch": 2.627467105263158,
      "grad_norm": 1.0105468034744263,
      "learning_rate": 7.375822368421053e-06,
      "loss": 0.0383,
      "step": 6390
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 76.56537628173828,
      "learning_rate": 7.3717105263157895e-06,
      "loss": 0.2995,
      "step": 6400
    },
    {
      "epoch": 2.635690789473684,
      "grad_norm": 3.7688522338867188,
      "learning_rate": 7.367598684210528e-06,
      "loss": 0.1928,
      "step": 6410
    },
    {
      "epoch": 2.6398026315789473,
      "grad_norm": 4.840407848358154,
      "learning_rate": 7.363486842105264e-06,
      "loss": 0.5117,
      "step": 6420
    },
    {
      "epoch": 2.6439144736842106,
      "grad_norm": 0.04940143600106239,
      "learning_rate": 7.359375000000001e-06,
      "loss": 0.0273,
      "step": 6430
    },
    {
      "epoch": 2.6480263157894735,
      "grad_norm": 1.0362895727157593,
      "learning_rate": 7.355263157894738e-06,
      "loss": 0.7687,
      "step": 6440
    },
    {
      "epoch": 2.6521381578947367,
      "grad_norm": 0.010492905043065548,
      "learning_rate": 7.351151315789474e-06,
      "loss": 0.2144,
      "step": 6450
    },
    {
      "epoch": 2.65625,
      "grad_norm": 73.12825012207031,
      "learning_rate": 7.347039473684211e-06,
      "loss": 0.2823,
      "step": 6460
    },
    {
      "epoch": 2.6603618421052633,
      "grad_norm": 0.42096540331840515,
      "learning_rate": 7.3429276315789474e-06,
      "loss": 0.3534,
      "step": 6470
    },
    {
      "epoch": 2.6644736842105265,
      "grad_norm": 0.14485445618629456,
      "learning_rate": 7.338815789473684e-06,
      "loss": 0.3453,
      "step": 6480
    },
    {
      "epoch": 2.6685855263157894,
      "grad_norm": 12.482208251953125,
      "learning_rate": 7.3347039473684215e-06,
      "loss": 0.0713,
      "step": 6490
    },
    {
      "epoch": 2.6726973684210527,
      "grad_norm": 0.2503849267959595,
      "learning_rate": 7.330592105263159e-06,
      "loss": 0.2003,
      "step": 6500
    },
    {
      "epoch": 2.676809210526316,
      "grad_norm": 80.90233612060547,
      "learning_rate": 7.3264802631578956e-06,
      "loss": 0.6061,
      "step": 6510
    },
    {
      "epoch": 2.6809210526315788,
      "grad_norm": 0.1269141584634781,
      "learning_rate": 7.322368421052632e-06,
      "loss": 0.1411,
      "step": 6520
    },
    {
      "epoch": 2.685032894736842,
      "grad_norm": 0.40265151858329773,
      "learning_rate": 7.318256578947369e-06,
      "loss": 0.3083,
      "step": 6530
    },
    {
      "epoch": 2.6891447368421053,
      "grad_norm": 0.008912616409361362,
      "learning_rate": 7.314144736842106e-06,
      "loss": 0.2767,
      "step": 6540
    },
    {
      "epoch": 2.6932565789473686,
      "grad_norm": 100.50891876220703,
      "learning_rate": 7.310032894736843e-06,
      "loss": 0.5464,
      "step": 6550
    },
    {
      "epoch": 2.6973684210526314,
      "grad_norm": 1.5742254257202148,
      "learning_rate": 7.3059210526315795e-06,
      "loss": 0.2836,
      "step": 6560
    },
    {
      "epoch": 2.7014802631578947,
      "grad_norm": 40.927833557128906,
      "learning_rate": 7.301809210526316e-06,
      "loss": 0.4598,
      "step": 6570
    },
    {
      "epoch": 2.705592105263158,
      "grad_norm": 31.99708366394043,
      "learning_rate": 7.297697368421053e-06,
      "loss": 0.701,
      "step": 6580
    },
    {
      "epoch": 2.7097039473684212,
      "grad_norm": 154.178955078125,
      "learning_rate": 7.293585526315789e-06,
      "loss": 0.3698,
      "step": 6590
    },
    {
      "epoch": 2.713815789473684,
      "grad_norm": 0.013821637257933617,
      "learning_rate": 7.289473684210528e-06,
      "loss": 0.0203,
      "step": 6600
    },
    {
      "epoch": 2.7179276315789473,
      "grad_norm": 7.3465352058410645,
      "learning_rate": 7.285361842105264e-06,
      "loss": 0.1565,
      "step": 6610
    },
    {
      "epoch": 2.7220394736842106,
      "grad_norm": 22.534740447998047,
      "learning_rate": 7.281250000000001e-06,
      "loss": 0.0484,
      "step": 6620
    },
    {
      "epoch": 2.7261513157894735,
      "grad_norm": 123.59476470947266,
      "learning_rate": 7.277138157894737e-06,
      "loss": 0.3089,
      "step": 6630
    },
    {
      "epoch": 2.7302631578947367,
      "grad_norm": 0.18893316388130188,
      "learning_rate": 7.273026315789474e-06,
      "loss": 0.1738,
      "step": 6640
    },
    {
      "epoch": 2.734375,
      "grad_norm": 0.03182161599397659,
      "learning_rate": 7.268914473684211e-06,
      "loss": 0.1138,
      "step": 6650
    },
    {
      "epoch": 2.7384868421052633,
      "grad_norm": 14.257698059082031,
      "learning_rate": 7.264802631578948e-06,
      "loss": 0.0437,
      "step": 6660
    },
    {
      "epoch": 2.7425986842105265,
      "grad_norm": 0.24907447397708893,
      "learning_rate": 7.260690789473685e-06,
      "loss": 0.1522,
      "step": 6670
    },
    {
      "epoch": 2.7467105263157894,
      "grad_norm": 0.19256246089935303,
      "learning_rate": 7.256578947368421e-06,
      "loss": 0.0405,
      "step": 6680
    },
    {
      "epoch": 2.7508223684210527,
      "grad_norm": 13.120503425598145,
      "learning_rate": 7.252467105263159e-06,
      "loss": 0.3291,
      "step": 6690
    },
    {
      "epoch": 2.754934210526316,
      "grad_norm": 92.6781234741211,
      "learning_rate": 7.248355263157895e-06,
      "loss": 0.3699,
      "step": 6700
    },
    {
      "epoch": 2.7590460526315788,
      "grad_norm": 1.2149583101272583,
      "learning_rate": 7.244243421052633e-06,
      "loss": 0.1675,
      "step": 6710
    },
    {
      "epoch": 2.763157894736842,
      "grad_norm": 45.1708869934082,
      "learning_rate": 7.2401315789473694e-06,
      "loss": 0.7265,
      "step": 6720
    },
    {
      "epoch": 2.7672697368421053,
      "grad_norm": 0.01486230455338955,
      "learning_rate": 7.236019736842106e-06,
      "loss": 0.3647,
      "step": 6730
    },
    {
      "epoch": 2.7713815789473686,
      "grad_norm": 4.854966640472412,
      "learning_rate": 7.231907894736843e-06,
      "loss": 0.0995,
      "step": 6740
    },
    {
      "epoch": 2.7754934210526314,
      "grad_norm": 4.642969608306885,
      "learning_rate": 7.227796052631579e-06,
      "loss": 0.285,
      "step": 6750
    },
    {
      "epoch": 2.7796052631578947,
      "grad_norm": 45.799072265625,
      "learning_rate": 7.223684210526316e-06,
      "loss": 0.4681,
      "step": 6760
    },
    {
      "epoch": 2.783717105263158,
      "grad_norm": 62.760704040527344,
      "learning_rate": 7.2195723684210525e-06,
      "loss": 0.0934,
      "step": 6770
    },
    {
      "epoch": 2.7878289473684212,
      "grad_norm": 1.0289639234542847,
      "learning_rate": 7.21546052631579e-06,
      "loss": 0.1929,
      "step": 6780
    },
    {
      "epoch": 2.791940789473684,
      "grad_norm": 75.18840026855469,
      "learning_rate": 7.211348684210527e-06,
      "loss": 0.218,
      "step": 6790
    },
    {
      "epoch": 2.7960526315789473,
      "grad_norm": 0.4156988561153412,
      "learning_rate": 7.207236842105264e-06,
      "loss": 0.6963,
      "step": 6800
    },
    {
      "epoch": 2.8001644736842106,
      "grad_norm": 34.15106201171875,
      "learning_rate": 7.203125000000001e-06,
      "loss": 0.2267,
      "step": 6810
    },
    {
      "epoch": 2.8042763157894735,
      "grad_norm": 18.53900909423828,
      "learning_rate": 7.199013157894737e-06,
      "loss": 0.1175,
      "step": 6820
    },
    {
      "epoch": 2.8083881578947367,
      "grad_norm": 73.15071105957031,
      "learning_rate": 7.194901315789475e-06,
      "loss": 0.2903,
      "step": 6830
    },
    {
      "epoch": 2.8125,
      "grad_norm": 93.10474395751953,
      "learning_rate": 7.190789473684211e-06,
      "loss": 0.4457,
      "step": 6840
    },
    {
      "epoch": 2.8166118421052633,
      "grad_norm": 0.07544347643852234,
      "learning_rate": 7.186677631578948e-06,
      "loss": 0.5648,
      "step": 6850
    },
    {
      "epoch": 2.8207236842105265,
      "grad_norm": 0.11148311197757721,
      "learning_rate": 7.1825657894736845e-06,
      "loss": 0.31,
      "step": 6860
    },
    {
      "epoch": 2.8248355263157894,
      "grad_norm": 1.118446707725525,
      "learning_rate": 7.178453947368421e-06,
      "loss": 0.7102,
      "step": 6870
    },
    {
      "epoch": 2.8289473684210527,
      "grad_norm": 72.61494445800781,
      "learning_rate": 7.174342105263159e-06,
      "loss": 0.1706,
      "step": 6880
    },
    {
      "epoch": 2.833059210526316,
      "grad_norm": 0.8325988054275513,
      "learning_rate": 7.170230263157896e-06,
      "loss": 0.2599,
      "step": 6890
    },
    {
      "epoch": 2.8371710526315788,
      "grad_norm": 13.365738868713379,
      "learning_rate": 7.166118421052633e-06,
      "loss": 0.1346,
      "step": 6900
    },
    {
      "epoch": 2.841282894736842,
      "grad_norm": 2.8599555492401123,
      "learning_rate": 7.162006578947369e-06,
      "loss": 0.0275,
      "step": 6910
    },
    {
      "epoch": 2.8453947368421053,
      "grad_norm": 0.002780698239803314,
      "learning_rate": 7.157894736842106e-06,
      "loss": 0.1408,
      "step": 6920
    },
    {
      "epoch": 2.8495065789473686,
      "grad_norm": 1.336742877960205,
      "learning_rate": 7.1537828947368424e-06,
      "loss": 0.2551,
      "step": 6930
    },
    {
      "epoch": 2.8536184210526314,
      "grad_norm": 31.96080780029297,
      "learning_rate": 7.149671052631579e-06,
      "loss": 0.3136,
      "step": 6940
    },
    {
      "epoch": 2.8577302631578947,
      "grad_norm": 0.01327523123472929,
      "learning_rate": 7.145559210526316e-06,
      "loss": 0.2752,
      "step": 6950
    },
    {
      "epoch": 2.861842105263158,
      "grad_norm": 3.886087656021118,
      "learning_rate": 7.141447368421053e-06,
      "loss": 0.1399,
      "step": 6960
    },
    {
      "epoch": 2.8659539473684212,
      "grad_norm": 0.03213124722242355,
      "learning_rate": 7.13733552631579e-06,
      "loss": 0.002,
      "step": 6970
    },
    {
      "epoch": 2.870065789473684,
      "grad_norm": 0.41033652424812317,
      "learning_rate": 7.133223684210527e-06,
      "loss": 0.2465,
      "step": 6980
    },
    {
      "epoch": 2.8741776315789473,
      "grad_norm": 0.006523555144667625,
      "learning_rate": 7.129111842105264e-06,
      "loss": 0.2667,
      "step": 6990
    },
    {
      "epoch": 2.8782894736842106,
      "grad_norm": 0.21593809127807617,
      "learning_rate": 7.125e-06,
      "loss": 0.1531,
      "step": 7000
    },
    {
      "epoch": 2.8824013157894735,
      "grad_norm": 0.4313972294330597,
      "learning_rate": 7.120888157894738e-06,
      "loss": 0.2449,
      "step": 7010
    },
    {
      "epoch": 2.8865131578947367,
      "grad_norm": 0.06804917752742767,
      "learning_rate": 7.1167763157894745e-06,
      "loss": 0.4352,
      "step": 7020
    },
    {
      "epoch": 2.890625,
      "grad_norm": 47.29886245727539,
      "learning_rate": 7.112664473684211e-06,
      "loss": 0.1663,
      "step": 7030
    },
    {
      "epoch": 2.8947368421052633,
      "grad_norm": 0.2415972501039505,
      "learning_rate": 7.108552631578948e-06,
      "loss": 0.2693,
      "step": 7040
    },
    {
      "epoch": 2.8988486842105265,
      "grad_norm": 51.08794403076172,
      "learning_rate": 7.104440789473684e-06,
      "loss": 0.0981,
      "step": 7050
    },
    {
      "epoch": 2.9029605263157894,
      "grad_norm": 2.805889129638672,
      "learning_rate": 7.100328947368421e-06,
      "loss": 0.069,
      "step": 7060
    },
    {
      "epoch": 2.9070723684210527,
      "grad_norm": 0.1910848468542099,
      "learning_rate": 7.096217105263159e-06,
      "loss": 0.4011,
      "step": 7070
    },
    {
      "epoch": 2.911184210526316,
      "grad_norm": 18.64550018310547,
      "learning_rate": 7.092105263157896e-06,
      "loss": 0.0614,
      "step": 7080
    },
    {
      "epoch": 2.9152960526315788,
      "grad_norm": 0.04789988696575165,
      "learning_rate": 7.087993421052632e-06,
      "loss": 0.0657,
      "step": 7090
    },
    {
      "epoch": 2.919407894736842,
      "grad_norm": 6.012223243713379,
      "learning_rate": 7.083881578947369e-06,
      "loss": 0.1898,
      "step": 7100
    },
    {
      "epoch": 2.9235197368421053,
      "grad_norm": 14.203520774841309,
      "learning_rate": 7.079769736842106e-06,
      "loss": 0.1308,
      "step": 7110
    },
    {
      "epoch": 2.9276315789473686,
      "grad_norm": 88.35704803466797,
      "learning_rate": 7.075657894736842e-06,
      "loss": 0.4129,
      "step": 7120
    },
    {
      "epoch": 2.9317434210526314,
      "grad_norm": 0.012460517697036266,
      "learning_rate": 7.07154605263158e-06,
      "loss": 0.2988,
      "step": 7130
    },
    {
      "epoch": 2.9358552631578947,
      "grad_norm": 9.11884593963623,
      "learning_rate": 7.067434210526316e-06,
      "loss": 0.1225,
      "step": 7140
    },
    {
      "epoch": 2.939967105263158,
      "grad_norm": 63.315948486328125,
      "learning_rate": 7.063322368421053e-06,
      "loss": 0.6643,
      "step": 7150
    },
    {
      "epoch": 2.9440789473684212,
      "grad_norm": 11.84128475189209,
      "learning_rate": 7.0592105263157895e-06,
      "loss": 0.221,
      "step": 7160
    },
    {
      "epoch": 2.948190789473684,
      "grad_norm": 0.05120432749390602,
      "learning_rate": 7.055098684210527e-06,
      "loss": 0.5354,
      "step": 7170
    },
    {
      "epoch": 2.9523026315789473,
      "grad_norm": 0.002431727247312665,
      "learning_rate": 7.0509868421052644e-06,
      "loss": 0.2658,
      "step": 7180
    },
    {
      "epoch": 2.9564144736842106,
      "grad_norm": 0.0686754584312439,
      "learning_rate": 7.046875000000001e-06,
      "loss": 0.1946,
      "step": 7190
    },
    {
      "epoch": 2.9605263157894735,
      "grad_norm": 0.09617192298173904,
      "learning_rate": 7.042763157894738e-06,
      "loss": 0.1712,
      "step": 7200
    },
    {
      "epoch": 2.9646381578947367,
      "grad_norm": 45.63372039794922,
      "learning_rate": 7.038651315789474e-06,
      "loss": 0.1632,
      "step": 7210
    },
    {
      "epoch": 2.96875,
      "grad_norm": 0.09691905975341797,
      "learning_rate": 7.034539473684211e-06,
      "loss": 0.1844,
      "step": 7220
    },
    {
      "epoch": 2.9728618421052633,
      "grad_norm": 0.13701745867729187,
      "learning_rate": 7.0304276315789475e-06,
      "loss": 0.3205,
      "step": 7230
    },
    {
      "epoch": 2.9769736842105265,
      "grad_norm": 55.56039047241211,
      "learning_rate": 7.026315789473684e-06,
      "loss": 0.0395,
      "step": 7240
    },
    {
      "epoch": 2.9810855263157894,
      "grad_norm": 77.04964447021484,
      "learning_rate": 7.0222039473684215e-06,
      "loss": 0.2569,
      "step": 7250
    },
    {
      "epoch": 2.9851973684210527,
      "grad_norm": 0.3657775819301605,
      "learning_rate": 7.018092105263159e-06,
      "loss": 0.2819,
      "step": 7260
    },
    {
      "epoch": 2.989309210526316,
      "grad_norm": 15.447266578674316,
      "learning_rate": 7.013980263157896e-06,
      "loss": 0.3315,
      "step": 7270
    },
    {
      "epoch": 2.9934210526315788,
      "grad_norm": 3.8167836666107178,
      "learning_rate": 7.009868421052632e-06,
      "loss": 0.3787,
      "step": 7280
    },
    {
      "epoch": 2.997532894736842,
      "grad_norm": 53.00830841064453,
      "learning_rate": 7.005756578947369e-06,
      "loss": 0.1485,
      "step": 7290
    },
    {
      "epoch": 3.0016447368421053,
      "grad_norm": 0.0033909266348928213,
      "learning_rate": 7.001644736842106e-06,
      "loss": 0.4747,
      "step": 7300
    },
    {
      "epoch": 3.0057565789473686,
      "grad_norm": 5.891729831695557,
      "learning_rate": 6.997532894736843e-06,
      "loss": 0.1194,
      "step": 7310
    },
    {
      "epoch": 3.0098684210526314,
      "grad_norm": 16.72593879699707,
      "learning_rate": 6.9934210526315795e-06,
      "loss": 0.3064,
      "step": 7320
    },
    {
      "epoch": 3.0139802631578947,
      "grad_norm": 0.12237309664487839,
      "learning_rate": 6.989309210526316e-06,
      "loss": 0.0441,
      "step": 7330
    },
    {
      "epoch": 3.018092105263158,
      "grad_norm": 0.3185586631298065,
      "learning_rate": 6.985197368421053e-06,
      "loss": 0.01,
      "step": 7340
    },
    {
      "epoch": 3.0222039473684212,
      "grad_norm": 0.26841112971305847,
      "learning_rate": 6.981085526315789e-06,
      "loss": 0.1187,
      "step": 7350
    },
    {
      "epoch": 3.026315789473684,
      "grad_norm": 0.023548761382699013,
      "learning_rate": 6.976973684210528e-06,
      "loss": 0.1564,
      "step": 7360
    },
    {
      "epoch": 3.0304276315789473,
      "grad_norm": 4.014942646026611,
      "learning_rate": 6.972861842105264e-06,
      "loss": 0.0751,
      "step": 7370
    },
    {
      "epoch": 3.0345394736842106,
      "grad_norm": 0.199817955493927,
      "learning_rate": 6.968750000000001e-06,
      "loss": 0.2649,
      "step": 7380
    },
    {
      "epoch": 3.038651315789474,
      "grad_norm": 1.6823475360870361,
      "learning_rate": 6.9646381578947374e-06,
      "loss": 0.1527,
      "step": 7390
    },
    {
      "epoch": 3.0427631578947367,
      "grad_norm": 4.5417375564575195,
      "learning_rate": 6.960526315789474e-06,
      "loss": 0.1669,
      "step": 7400
    },
    {
      "epoch": 3.046875,
      "grad_norm": 0.4437451958656311,
      "learning_rate": 6.956414473684211e-06,
      "loss": 0.0024,
      "step": 7410
    },
    {
      "epoch": 3.0509868421052633,
      "grad_norm": 1.8537052869796753,
      "learning_rate": 6.952302631578948e-06,
      "loss": 0.1486,
      "step": 7420
    },
    {
      "epoch": 3.055098684210526,
      "grad_norm": 0.10921826213598251,
      "learning_rate": 6.948190789473685e-06,
      "loss": 0.1157,
      "step": 7430
    },
    {
      "epoch": 3.0592105263157894,
      "grad_norm": 0.14697521924972534,
      "learning_rate": 6.944078947368421e-06,
      "loss": 0.1615,
      "step": 7440
    },
    {
      "epoch": 3.0633223684210527,
      "grad_norm": 0.3917556703090668,
      "learning_rate": 6.939967105263159e-06,
      "loss": 0.1616,
      "step": 7450
    },
    {
      "epoch": 3.067434210526316,
      "grad_norm": 0.17994216084480286,
      "learning_rate": 6.935855263157895e-06,
      "loss": 0.322,
      "step": 7460
    },
    {
      "epoch": 3.0715460526315788,
      "grad_norm": 0.741560161113739,
      "learning_rate": 6.931743421052633e-06,
      "loss": 0.1822,
      "step": 7470
    },
    {
      "epoch": 3.075657894736842,
      "grad_norm": 2.4355533123016357,
      "learning_rate": 6.9276315789473695e-06,
      "loss": 0.5123,
      "step": 7480
    },
    {
      "epoch": 3.0797697368421053,
      "grad_norm": 5.086973667144775,
      "learning_rate": 6.923519736842106e-06,
      "loss": 0.3002,
      "step": 7490
    },
    {
      "epoch": 3.0838815789473686,
      "grad_norm": 0.2662551999092102,
      "learning_rate": 6.919407894736843e-06,
      "loss": 0.0768,
      "step": 7500
    },
    {
      "epoch": 3.0879934210526314,
      "grad_norm": 0.12516538798809052,
      "learning_rate": 6.915296052631579e-06,
      "loss": 0.2165,
      "step": 7510
    },
    {
      "epoch": 3.0921052631578947,
      "grad_norm": 0.0330631323158741,
      "learning_rate": 6.911184210526316e-06,
      "loss": 0.0027,
      "step": 7520
    },
    {
      "epoch": 3.096217105263158,
      "grad_norm": 91.03437805175781,
      "learning_rate": 6.9070723684210525e-06,
      "loss": 0.1534,
      "step": 7530
    },
    {
      "epoch": 3.1003289473684212,
      "grad_norm": 0.015870722010731697,
      "learning_rate": 6.902960526315789e-06,
      "loss": 0.0099,
      "step": 7540
    },
    {
      "epoch": 3.104440789473684,
      "grad_norm": 1.7429580688476562,
      "learning_rate": 6.898848684210527e-06,
      "loss": 0.1773,
      "step": 7550
    },
    {
      "epoch": 3.1085526315789473,
      "grad_norm": 16.66814422607422,
      "learning_rate": 6.894736842105264e-06,
      "loss": 0.0331,
      "step": 7560
    },
    {
      "epoch": 3.1126644736842106,
      "grad_norm": 0.08903591334819794,
      "learning_rate": 6.890625000000001e-06,
      "loss": 0.2967,
      "step": 7570
    },
    {
      "epoch": 3.116776315789474,
      "grad_norm": 24.660062789916992,
      "learning_rate": 6.886513157894737e-06,
      "loss": 0.63,
      "step": 7580
    },
    {
      "epoch": 3.1208881578947367,
      "grad_norm": 40.910377502441406,
      "learning_rate": 6.882401315789474e-06,
      "loss": 0.0253,
      "step": 7590
    },
    {
      "epoch": 3.125,
      "grad_norm": 0.23835232853889465,
      "learning_rate": 6.878289473684211e-06,
      "loss": 0.1164,
      "step": 7600
    },
    {
      "epoch": 3.1291118421052633,
      "grad_norm": 8.85913372039795,
      "learning_rate": 6.874177631578948e-06,
      "loss": 0.0507,
      "step": 7610
    },
    {
      "epoch": 3.1332236842105265,
      "grad_norm": 0.0227875467389822,
      "learning_rate": 6.8700657894736845e-06,
      "loss": 0.0667,
      "step": 7620
    },
    {
      "epoch": 3.1373355263157894,
      "grad_norm": 0.35952290892601013,
      "learning_rate": 6.865953947368421e-06,
      "loss": 0.035,
      "step": 7630
    },
    {
      "epoch": 3.1414473684210527,
      "grad_norm": 3.871885061264038,
      "learning_rate": 6.861842105263159e-06,
      "loss": 0.4717,
      "step": 7640
    },
    {
      "epoch": 3.145559210526316,
      "grad_norm": 0.0658380314707756,
      "learning_rate": 6.857730263157896e-06,
      "loss": 0.3083,
      "step": 7650
    },
    {
      "epoch": 3.1496710526315788,
      "grad_norm": 0.10795586556196213,
      "learning_rate": 6.853618421052633e-06,
      "loss": 0.0258,
      "step": 7660
    },
    {
      "epoch": 3.153782894736842,
      "grad_norm": 56.6161003112793,
      "learning_rate": 6.849506578947369e-06,
      "loss": 0.223,
      "step": 7670
    },
    {
      "epoch": 3.1578947368421053,
      "grad_norm": 72.3301010131836,
      "learning_rate": 6.845394736842106e-06,
      "loss": 0.1169,
      "step": 7680
    },
    {
      "epoch": 3.1620065789473686,
      "grad_norm": 0.7228354811668396,
      "learning_rate": 6.8412828947368425e-06,
      "loss": 0.0275,
      "step": 7690
    },
    {
      "epoch": 3.1661184210526314,
      "grad_norm": 19.717239379882812,
      "learning_rate": 6.837171052631579e-06,
      "loss": 0.0342,
      "step": 7700
    },
    {
      "epoch": 3.1702302631578947,
      "grad_norm": 2.3942067623138428,
      "learning_rate": 6.833059210526316e-06,
      "loss": 0.3279,
      "step": 7710
    },
    {
      "epoch": 3.174342105263158,
      "grad_norm": 68.46968078613281,
      "learning_rate": 6.828947368421053e-06,
      "loss": 0.0909,
      "step": 7720
    },
    {
      "epoch": 3.1784539473684212,
      "grad_norm": 6.748744010925293,
      "learning_rate": 6.82483552631579e-06,
      "loss": 0.3401,
      "step": 7730
    },
    {
      "epoch": 3.182565789473684,
      "grad_norm": 0.03430832922458649,
      "learning_rate": 6.820723684210527e-06,
      "loss": 0.2739,
      "step": 7740
    },
    {
      "epoch": 3.1866776315789473,
      "grad_norm": 80.0864486694336,
      "learning_rate": 6.816611842105264e-06,
      "loss": 0.2544,
      "step": 7750
    },
    {
      "epoch": 3.1907894736842106,
      "grad_norm": 67.80303192138672,
      "learning_rate": 6.8125e-06,
      "loss": 0.5098,
      "step": 7760
    },
    {
      "epoch": 3.1949013157894735,
      "grad_norm": 0.011421781964600086,
      "learning_rate": 6.808388157894738e-06,
      "loss": 0.0427,
      "step": 7770
    },
    {
      "epoch": 3.1990131578947367,
      "grad_norm": 0.10575248301029205,
      "learning_rate": 6.8042763157894745e-06,
      "loss": 0.1562,
      "step": 7780
    },
    {
      "epoch": 3.203125,
      "grad_norm": 21.22425651550293,
      "learning_rate": 6.800164473684211e-06,
      "loss": 0.13,
      "step": 7790
    },
    {
      "epoch": 3.2072368421052633,
      "grad_norm": 62.44038772583008,
      "learning_rate": 6.796052631578948e-06,
      "loss": 0.1369,
      "step": 7800
    },
    {
      "epoch": 3.2113486842105265,
      "grad_norm": 0.056282613426446915,
      "learning_rate": 6.791940789473684e-06,
      "loss": 0.0594,
      "step": 7810
    },
    {
      "epoch": 3.2154605263157894,
      "grad_norm": 0.05405473709106445,
      "learning_rate": 6.787828947368421e-06,
      "loss": 0.0014,
      "step": 7820
    },
    {
      "epoch": 3.2195723684210527,
      "grad_norm": 0.7672938704490662,
      "learning_rate": 6.783717105263159e-06,
      "loss": 0.2649,
      "step": 7830
    },
    {
      "epoch": 3.223684210526316,
      "grad_norm": 0.012963368557393551,
      "learning_rate": 6.779605263157896e-06,
      "loss": 0.003,
      "step": 7840
    },
    {
      "epoch": 3.2277960526315788,
      "grad_norm": 0.7032648324966431,
      "learning_rate": 6.7754934210526324e-06,
      "loss": 0.0383,
      "step": 7850
    },
    {
      "epoch": 3.231907894736842,
      "grad_norm": 0.00014193853712640703,
      "learning_rate": 6.771381578947369e-06,
      "loss": 0.2203,
      "step": 7860
    },
    {
      "epoch": 3.2360197368421053,
      "grad_norm": 0.16566841304302216,
      "learning_rate": 6.767269736842106e-06,
      "loss": 0.082,
      "step": 7870
    },
    {
      "epoch": 3.2401315789473686,
      "grad_norm": 0.011220881715416908,
      "learning_rate": 6.763157894736842e-06,
      "loss": 0.2433,
      "step": 7880
    },
    {
      "epoch": 3.2442434210526314,
      "grad_norm": 0.031912826001644135,
      "learning_rate": 6.75904605263158e-06,
      "loss": 0.1053,
      "step": 7890
    },
    {
      "epoch": 3.2483552631578947,
      "grad_norm": 0.00015638569311704487,
      "learning_rate": 6.754934210526316e-06,
      "loss": 0.0751,
      "step": 7900
    },
    {
      "epoch": 3.252467105263158,
      "grad_norm": 19.786230087280273,
      "learning_rate": 6.750822368421053e-06,
      "loss": 0.1573,
      "step": 7910
    },
    {
      "epoch": 3.2565789473684212,
      "grad_norm": 25.62039566040039,
      "learning_rate": 6.7467105263157895e-06,
      "loss": 0.0847,
      "step": 7920
    },
    {
      "epoch": 3.260690789473684,
      "grad_norm": 43.73009490966797,
      "learning_rate": 6.742598684210527e-06,
      "loss": 0.2365,
      "step": 7930
    },
    {
      "epoch": 3.2648026315789473,
      "grad_norm": 0.29684680700302124,
      "learning_rate": 6.7384868421052645e-06,
      "loss": 0.5742,
      "step": 7940
    },
    {
      "epoch": 3.2689144736842106,
      "grad_norm": 35.690513610839844,
      "learning_rate": 6.734375000000001e-06,
      "loss": 0.0705,
      "step": 7950
    },
    {
      "epoch": 3.2730263157894735,
      "grad_norm": 0.003651993814855814,
      "learning_rate": 6.730263157894738e-06,
      "loss": 0.0705,
      "step": 7960
    },
    {
      "epoch": 3.2771381578947367,
      "grad_norm": 0.0002773971064016223,
      "learning_rate": 6.726151315789474e-06,
      "loss": 0.2075,
      "step": 7970
    },
    {
      "epoch": 3.28125,
      "grad_norm": 0.35052403807640076,
      "learning_rate": 6.722039473684211e-06,
      "loss": 0.2401,
      "step": 7980
    },
    {
      "epoch": 3.2853618421052633,
      "grad_norm": 0.004375469870865345,
      "learning_rate": 6.7179276315789475e-06,
      "loss": 0.2805,
      "step": 7990
    },
    {
      "epoch": 3.2894736842105265,
      "grad_norm": 51.69245147705078,
      "learning_rate": 6.713815789473684e-06,
      "loss": 0.5704,
      "step": 8000
    },
    {
      "epoch": 3.2935855263157894,
      "grad_norm": 0.09066779911518097,
      "learning_rate": 6.709703947368421e-06,
      "loss": 0.1949,
      "step": 8010
    },
    {
      "epoch": 3.2976973684210527,
      "grad_norm": 20.805805206298828,
      "learning_rate": 6.705592105263159e-06,
      "loss": 0.5101,
      "step": 8020
    },
    {
      "epoch": 3.301809210526316,
      "grad_norm": 3.514925241470337,
      "learning_rate": 6.701480263157896e-06,
      "loss": 0.19,
      "step": 8030
    },
    {
      "epoch": 3.3059210526315788,
      "grad_norm": 1.247069239616394,
      "learning_rate": 6.697368421052632e-06,
      "loss": 0.2137,
      "step": 8040
    },
    {
      "epoch": 3.310032894736842,
      "grad_norm": 88.480712890625,
      "learning_rate": 6.693256578947369e-06,
      "loss": 0.4556,
      "step": 8050
    },
    {
      "epoch": 3.3141447368421053,
      "grad_norm": 0.18627463281154633,
      "learning_rate": 6.6891447368421054e-06,
      "loss": 0.1171,
      "step": 8060
    },
    {
      "epoch": 3.3182565789473686,
      "grad_norm": 0.06986451894044876,
      "learning_rate": 6.685032894736843e-06,
      "loss": 0.4195,
      "step": 8070
    },
    {
      "epoch": 3.3223684210526314,
      "grad_norm": 47.814727783203125,
      "learning_rate": 6.6809210526315795e-06,
      "loss": 0.1274,
      "step": 8080
    },
    {
      "epoch": 3.3264802631578947,
      "grad_norm": 63.75092315673828,
      "learning_rate": 6.676809210526316e-06,
      "loss": 0.4185,
      "step": 8090
    },
    {
      "epoch": 3.330592105263158,
      "grad_norm": 3.7314867973327637,
      "learning_rate": 6.672697368421053e-06,
      "loss": 0.1311,
      "step": 8100
    },
    {
      "epoch": 3.3347039473684212,
      "grad_norm": 16.43247413635254,
      "learning_rate": 6.668585526315789e-06,
      "loss": 0.2694,
      "step": 8110
    },
    {
      "epoch": 3.338815789473684,
      "grad_norm": 0.008259894326329231,
      "learning_rate": 6.664473684210528e-06,
      "loss": 0.2619,
      "step": 8120
    },
    {
      "epoch": 3.3429276315789473,
      "grad_norm": 0.09545180946588516,
      "learning_rate": 6.660361842105264e-06,
      "loss": 0.2593,
      "step": 8130
    },
    {
      "epoch": 3.3470394736842106,
      "grad_norm": 2.6562247276306152,
      "learning_rate": 6.656250000000001e-06,
      "loss": 0.1714,
      "step": 8140
    },
    {
      "epoch": 3.3511513157894735,
      "grad_norm": 1.3511881828308105,
      "learning_rate": 6.6521381578947375e-06,
      "loss": 0.1609,
      "step": 8150
    },
    {
      "epoch": 3.3552631578947367,
      "grad_norm": 0.45312735438346863,
      "learning_rate": 6.648026315789474e-06,
      "loss": 0.402,
      "step": 8160
    },
    {
      "epoch": 3.359375,
      "grad_norm": 0.008526019752025604,
      "learning_rate": 6.643914473684211e-06,
      "loss": 0.1503,
      "step": 8170
    },
    {
      "epoch": 3.3634868421052633,
      "grad_norm": 4.806290149688721,
      "learning_rate": 6.639802631578947e-06,
      "loss": 0.3165,
      "step": 8180
    },
    {
      "epoch": 3.3675986842105265,
      "grad_norm": 0.008733537979424,
      "learning_rate": 6.635690789473685e-06,
      "loss": 0.017,
      "step": 8190
    },
    {
      "epoch": 3.3717105263157894,
      "grad_norm": 11.375794410705566,
      "learning_rate": 6.631578947368421e-06,
      "loss": 0.0327,
      "step": 8200
    },
    {
      "epoch": 3.3758223684210527,
      "grad_norm": 1.7857224941253662,
      "learning_rate": 6.627467105263159e-06,
      "loss": 0.0189,
      "step": 8210
    },
    {
      "epoch": 3.379934210526316,
      "grad_norm": 7.3000102043151855,
      "learning_rate": 6.623355263157895e-06,
      "loss": 0.0515,
      "step": 8220
    },
    {
      "epoch": 3.3840460526315788,
      "grad_norm": 0.0007516624755226076,
      "learning_rate": 6.619243421052632e-06,
      "loss": 0.013,
      "step": 8230
    },
    {
      "epoch": 3.388157894736842,
      "grad_norm": 74.04230499267578,
      "learning_rate": 6.6151315789473695e-06,
      "loss": 0.2848,
      "step": 8240
    },
    {
      "epoch": 3.3922697368421053,
      "grad_norm": 0.06303539872169495,
      "learning_rate": 6.611019736842106e-06,
      "loss": 0.0307,
      "step": 8250
    },
    {
      "epoch": 3.3963815789473686,
      "grad_norm": 24.18715476989746,
      "learning_rate": 6.606907894736843e-06,
      "loss": 0.2593,
      "step": 8260
    },
    {
      "epoch": 3.4004934210526314,
      "grad_norm": 0.29844167828559875,
      "learning_rate": 6.602796052631579e-06,
      "loss": 0.1038,
      "step": 8270
    },
    {
      "epoch": 3.4046052631578947,
      "grad_norm": 0.1648898422718048,
      "learning_rate": 6.598684210526316e-06,
      "loss": 0.1039,
      "step": 8280
    },
    {
      "epoch": 3.408717105263158,
      "grad_norm": 1.139145016670227,
      "learning_rate": 6.5945723684210525e-06,
      "loss": 0.2466,
      "step": 8290
    },
    {
      "epoch": 3.4128289473684212,
      "grad_norm": 7.223357200622559,
      "learning_rate": 6.590460526315789e-06,
      "loss": 0.3805,
      "step": 8300
    },
    {
      "epoch": 3.416940789473684,
      "grad_norm": 0.002951091853901744,
      "learning_rate": 6.5863486842105274e-06,
      "loss": 0.2528,
      "step": 8310
    },
    {
      "epoch": 3.4210526315789473,
      "grad_norm": 0.04810519888997078,
      "learning_rate": 6.582236842105264e-06,
      "loss": 0.0248,
      "step": 8320
    },
    {
      "epoch": 3.4251644736842106,
      "grad_norm": 44.80435562133789,
      "learning_rate": 6.578125000000001e-06,
      "loss": 0.148,
      "step": 8330
    },
    {
      "epoch": 3.4292763157894735,
      "grad_norm": 0.6150760054588318,
      "learning_rate": 6.574013157894737e-06,
      "loss": 0.1437,
      "step": 8340
    },
    {
      "epoch": 3.4333881578947367,
      "grad_norm": 0.06286383420228958,
      "learning_rate": 6.569901315789474e-06,
      "loss": 0.2608,
      "step": 8350
    },
    {
      "epoch": 3.4375,
      "grad_norm": 0.003701314562931657,
      "learning_rate": 6.565789473684211e-06,
      "loss": 0.3989,
      "step": 8360
    },
    {
      "epoch": 3.4416118421052633,
      "grad_norm": 0.033227551728487015,
      "learning_rate": 6.561677631578948e-06,
      "loss": 0.2687,
      "step": 8370
    },
    {
      "epoch": 3.4457236842105265,
      "grad_norm": 74.94076538085938,
      "learning_rate": 6.5575657894736845e-06,
      "loss": 0.2218,
      "step": 8380
    },
    {
      "epoch": 3.4498355263157894,
      "grad_norm": 0.15582340955734253,
      "learning_rate": 6.553453947368421e-06,
      "loss": 0.0771,
      "step": 8390
    },
    {
      "epoch": 3.4539473684210527,
      "grad_norm": 0.005174733232706785,
      "learning_rate": 6.549342105263159e-06,
      "loss": 0.0989,
      "step": 8400
    },
    {
      "epoch": 3.458059210526316,
      "grad_norm": 7.792595863342285,
      "learning_rate": 6.545230263157896e-06,
      "loss": 0.0484,
      "step": 8410
    },
    {
      "epoch": 3.4621710526315788,
      "grad_norm": 0.11872321367263794,
      "learning_rate": 6.541118421052633e-06,
      "loss": 0.0035,
      "step": 8420
    },
    {
      "epoch": 3.466282894736842,
      "grad_norm": 5.740940832765773e-05,
      "learning_rate": 6.537006578947369e-06,
      "loss": 0.2619,
      "step": 8430
    },
    {
      "epoch": 3.4703947368421053,
      "grad_norm": 0.37503787875175476,
      "learning_rate": 6.532894736842106e-06,
      "loss": 0.2062,
      "step": 8440
    },
    {
      "epoch": 3.4745065789473686,
      "grad_norm": 12.798948287963867,
      "learning_rate": 6.5287828947368425e-06,
      "loss": 0.3276,
      "step": 8450
    },
    {
      "epoch": 3.4786184210526314,
      "grad_norm": 4.535250663757324,
      "learning_rate": 6.524671052631579e-06,
      "loss": 0.1852,
      "step": 8460
    },
    {
      "epoch": 3.4827302631578947,
      "grad_norm": 0.47860297560691833,
      "learning_rate": 6.520559210526316e-06,
      "loss": 0.4313,
      "step": 8470
    },
    {
      "epoch": 3.486842105263158,
      "grad_norm": 61.79635238647461,
      "learning_rate": 6.516447368421053e-06,
      "loss": 0.318,
      "step": 8480
    },
    {
      "epoch": 3.4909539473684212,
      "grad_norm": 9.027344703674316,
      "learning_rate": 6.51233552631579e-06,
      "loss": 0.1673,
      "step": 8490
    },
    {
      "epoch": 3.495065789473684,
      "grad_norm": 0.1292915791273117,
      "learning_rate": 6.508223684210527e-06,
      "loss": 0.3923,
      "step": 8500
    },
    {
      "epoch": 3.4991776315789473,
      "grad_norm": 0.0020961903501302004,
      "learning_rate": 6.504111842105264e-06,
      "loss": 0.2699,
      "step": 8510
    },
    {
      "epoch": 3.5032894736842106,
      "grad_norm": 0.14688652753829956,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.2736,
      "step": 8520
    },
    {
      "epoch": 3.5074013157894735,
      "grad_norm": 61.5792121887207,
      "learning_rate": 6.495888157894737e-06,
      "loss": 0.1765,
      "step": 8530
    },
    {
      "epoch": 3.5115131578947367,
      "grad_norm": 0.07907792925834656,
      "learning_rate": 6.4917763157894745e-06,
      "loss": 0.158,
      "step": 8540
    },
    {
      "epoch": 3.515625,
      "grad_norm": 0.29822736978530884,
      "learning_rate": 6.487664473684211e-06,
      "loss": 0.0808,
      "step": 8550
    },
    {
      "epoch": 3.5197368421052633,
      "grad_norm": 5.326796054840088,
      "learning_rate": 6.483552631578948e-06,
      "loss": 0.0521,
      "step": 8560
    },
    {
      "epoch": 3.5238486842105265,
      "grad_norm": 46.79924011230469,
      "learning_rate": 6.479440789473684e-06,
      "loss": 0.5624,
      "step": 8570
    },
    {
      "epoch": 3.5279605263157894,
      "grad_norm": 0.1057400181889534,
      "learning_rate": 6.475328947368421e-06,
      "loss": 0.3374,
      "step": 8580
    },
    {
      "epoch": 3.5320723684210527,
      "grad_norm": 2.302854061126709,
      "learning_rate": 6.471217105263159e-06,
      "loss": 0.1588,
      "step": 8590
    },
    {
      "epoch": 3.536184210526316,
      "grad_norm": 0.31161850690841675,
      "learning_rate": 6.467105263157896e-06,
      "loss": 0.4272,
      "step": 8600
    },
    {
      "epoch": 3.5402960526315788,
      "grad_norm": 1.1857142448425293,
      "learning_rate": 6.4629934210526325e-06,
      "loss": 0.3116,
      "step": 8610
    },
    {
      "epoch": 3.544407894736842,
      "grad_norm": 74.4317398071289,
      "learning_rate": 6.458881578947369e-06,
      "loss": 0.0994,
      "step": 8620
    },
    {
      "epoch": 3.5485197368421053,
      "grad_norm": 1.3740957975387573,
      "learning_rate": 6.454769736842106e-06,
      "loss": 0.5356,
      "step": 8630
    },
    {
      "epoch": 3.5526315789473686,
      "grad_norm": 66.15660858154297,
      "learning_rate": 6.450657894736842e-06,
      "loss": 0.096,
      "step": 8640
    },
    {
      "epoch": 3.5567434210526314,
      "grad_norm": 130.39633178710938,
      "learning_rate": 6.446546052631579e-06,
      "loss": 0.2427,
      "step": 8650
    },
    {
      "epoch": 3.5608552631578947,
      "grad_norm": 0.7252002954483032,
      "learning_rate": 6.442434210526316e-06,
      "loss": 0.0773,
      "step": 8660
    },
    {
      "epoch": 3.564967105263158,
      "grad_norm": 0.32898321747779846,
      "learning_rate": 6.438322368421053e-06,
      "loss": 0.2922,
      "step": 8670
    },
    {
      "epoch": 3.5690789473684212,
      "grad_norm": 0.4259633719921112,
      "learning_rate": 6.4342105263157896e-06,
      "loss": 0.1903,
      "step": 8680
    },
    {
      "epoch": 3.573190789473684,
      "grad_norm": 0.02078865095973015,
      "learning_rate": 6.430098684210527e-06,
      "loss": 0.0631,
      "step": 8690
    },
    {
      "epoch": 3.5773026315789473,
      "grad_norm": 5.149608612060547,
      "learning_rate": 6.425986842105264e-06,
      "loss": 0.0189,
      "step": 8700
    },
    {
      "epoch": 3.5814144736842106,
      "grad_norm": 0.003035628003999591,
      "learning_rate": 6.421875000000001e-06,
      "loss": 0.339,
      "step": 8710
    },
    {
      "epoch": 3.5855263157894735,
      "grad_norm": 81.40657043457031,
      "learning_rate": 6.417763157894738e-06,
      "loss": 0.2999,
      "step": 8720
    },
    {
      "epoch": 3.5896381578947367,
      "grad_norm": 5.7936906814575195,
      "learning_rate": 6.413651315789474e-06,
      "loss": 0.1581,
      "step": 8730
    },
    {
      "epoch": 3.59375,
      "grad_norm": 58.010353088378906,
      "learning_rate": 6.409539473684211e-06,
      "loss": 0.2292,
      "step": 8740
    },
    {
      "epoch": 3.5978618421052633,
      "grad_norm": 10.935487747192383,
      "learning_rate": 6.4054276315789475e-06,
      "loss": 0.0143,
      "step": 8750
    },
    {
      "epoch": 3.6019736842105265,
      "grad_norm": 0.09966960549354553,
      "learning_rate": 6.401315789473684e-06,
      "loss": 0.105,
      "step": 8760
    },
    {
      "epoch": 3.6060855263157894,
      "grad_norm": 0.00877199787646532,
      "learning_rate": 6.397203947368421e-06,
      "loss": 0.2848,
      "step": 8770
    },
    {
      "epoch": 3.6101973684210527,
      "grad_norm": 81.7994613647461,
      "learning_rate": 6.393092105263159e-06,
      "loss": 0.1348,
      "step": 8780
    },
    {
      "epoch": 3.614309210526316,
      "grad_norm": 0.004074628930538893,
      "learning_rate": 6.388980263157896e-06,
      "loss": 0.1723,
      "step": 8790
    },
    {
      "epoch": 3.6184210526315788,
      "grad_norm": 1.1934515237808228,
      "learning_rate": 6.384868421052632e-06,
      "loss": 0.0017,
      "step": 8800
    },
    {
      "epoch": 3.622532894736842,
      "grad_norm": 0.01399979181587696,
      "learning_rate": 6.380756578947369e-06,
      "loss": 0.2224,
      "step": 8810
    },
    {
      "epoch": 3.6266447368421053,
      "grad_norm": 115.63423156738281,
      "learning_rate": 6.3766447368421055e-06,
      "loss": 0.3234,
      "step": 8820
    },
    {
      "epoch": 3.6307565789473686,
      "grad_norm": 0.49743548035621643,
      "learning_rate": 6.372532894736843e-06,
      "loss": 0.5291,
      "step": 8830
    },
    {
      "epoch": 3.6348684210526314,
      "grad_norm": 1.7224241495132446,
      "learning_rate": 6.3684210526315795e-06,
      "loss": 0.2441,
      "step": 8840
    },
    {
      "epoch": 3.6389802631578947,
      "grad_norm": 56.73593521118164,
      "learning_rate": 6.364309210526316e-06,
      "loss": 0.1411,
      "step": 8850
    },
    {
      "epoch": 3.643092105263158,
      "grad_norm": 0.0021920078434050083,
      "learning_rate": 6.360197368421053e-06,
      "loss": 0.2172,
      "step": 8860
    },
    {
      "epoch": 3.6472039473684212,
      "grad_norm": 0.005450670141726732,
      "learning_rate": 6.356085526315789e-06,
      "loss": 0.0721,
      "step": 8870
    },
    {
      "epoch": 3.651315789473684,
      "grad_norm": 0.0028689634054899216,
      "learning_rate": 6.351973684210528e-06,
      "loss": 0.033,
      "step": 8880
    },
    {
      "epoch": 3.6554276315789473,
      "grad_norm": 41.87671661376953,
      "learning_rate": 6.347861842105264e-06,
      "loss": 0.1019,
      "step": 8890
    },
    {
      "epoch": 3.6595394736842106,
      "grad_norm": 0.1014520600438118,
      "learning_rate": 6.343750000000001e-06,
      "loss": 0.0836,
      "step": 8900
    },
    {
      "epoch": 3.6636513157894735,
      "grad_norm": 0.8657021522521973,
      "learning_rate": 6.3396381578947375e-06,
      "loss": 0.0689,
      "step": 8910
    },
    {
      "epoch": 3.6677631578947367,
      "grad_norm": 6.391144275665283,
      "learning_rate": 6.335526315789474e-06,
      "loss": 0.2655,
      "step": 8920
    },
    {
      "epoch": 3.671875,
      "grad_norm": 25.194692611694336,
      "learning_rate": 6.331414473684211e-06,
      "loss": 0.2248,
      "step": 8930
    },
    {
      "epoch": 3.6759868421052633,
      "grad_norm": 87.24070739746094,
      "learning_rate": 6.327302631578947e-06,
      "loss": 0.1769,
      "step": 8940
    },
    {
      "epoch": 3.6800986842105265,
      "grad_norm": 84.9614486694336,
      "learning_rate": 6.323190789473685e-06,
      "loss": 0.1785,
      "step": 8950
    },
    {
      "epoch": 3.6842105263157894,
      "grad_norm": 3.7174837589263916,
      "learning_rate": 6.319078947368421e-06,
      "loss": 0.1043,
      "step": 8960
    },
    {
      "epoch": 3.6883223684210527,
      "grad_norm": 19.11334228515625,
      "learning_rate": 6.314967105263159e-06,
      "loss": 0.1056,
      "step": 8970
    },
    {
      "epoch": 3.692434210526316,
      "grad_norm": 2.9218459129333496,
      "learning_rate": 6.3108552631578954e-06,
      "loss": 0.0715,
      "step": 8980
    },
    {
      "epoch": 3.6965460526315788,
      "grad_norm": 0.7350709438323975,
      "learning_rate": 6.306743421052632e-06,
      "loss": 0.1103,
      "step": 8990
    },
    {
      "epoch": 3.700657894736842,
      "grad_norm": 25.493600845336914,
      "learning_rate": 6.3026315789473695e-06,
      "loss": 0.4917,
      "step": 9000
    },
    {
      "epoch": 3.7047697368421053,
      "grad_norm": 0.3064211905002594,
      "learning_rate": 6.298519736842106e-06,
      "loss": 0.0834,
      "step": 9010
    },
    {
      "epoch": 3.7088815789473686,
      "grad_norm": 0.8027281165122986,
      "learning_rate": 6.294407894736843e-06,
      "loss": 0.2116,
      "step": 9020
    },
    {
      "epoch": 3.7129934210526314,
      "grad_norm": 0.023541197180747986,
      "learning_rate": 6.290296052631579e-06,
      "loss": 0.2796,
      "step": 9030
    },
    {
      "epoch": 3.7171052631578947,
      "grad_norm": 0.007667736150324345,
      "learning_rate": 6.286184210526316e-06,
      "loss": 0.107,
      "step": 9040
    },
    {
      "epoch": 3.721217105263158,
      "grad_norm": 22.70380973815918,
      "learning_rate": 6.2820723684210526e-06,
      "loss": 0.1017,
      "step": 9050
    },
    {
      "epoch": 3.7253289473684212,
      "grad_norm": 0.20407456159591675,
      "learning_rate": 6.277960526315789e-06,
      "loss": 0.0103,
      "step": 9060
    },
    {
      "epoch": 3.729440789473684,
      "grad_norm": 75.71296691894531,
      "learning_rate": 6.2738486842105275e-06,
      "loss": 0.4619,
      "step": 9070
    },
    {
      "epoch": 3.7335526315789473,
      "grad_norm": 58.81220626831055,
      "learning_rate": 6.269736842105264e-06,
      "loss": 0.1691,
      "step": 9080
    },
    {
      "epoch": 3.7376644736842106,
      "grad_norm": 54.02885818481445,
      "learning_rate": 6.265625000000001e-06,
      "loss": 0.2037,
      "step": 9090
    },
    {
      "epoch": 3.7417763157894735,
      "grad_norm": 1.0572305917739868,
      "learning_rate": 6.261513157894737e-06,
      "loss": 0.0768,
      "step": 9100
    },
    {
      "epoch": 3.7458881578947367,
      "grad_norm": 14.10479736328125,
      "learning_rate": 6.257401315789474e-06,
      "loss": 0.3359,
      "step": 9110
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.23570677638053894,
      "learning_rate": 6.2532894736842105e-06,
      "loss": 0.0442,
      "step": 9120
    },
    {
      "epoch": 3.7541118421052633,
      "grad_norm": 1.2006131410598755,
      "learning_rate": 6.249177631578948e-06,
      "loss": 0.1693,
      "step": 9130
    },
    {
      "epoch": 3.7582236842105265,
      "grad_norm": 65.66853332519531,
      "learning_rate": 6.2450657894736846e-06,
      "loss": 0.1355,
      "step": 9140
    },
    {
      "epoch": 3.7623355263157894,
      "grad_norm": 62.854225158691406,
      "learning_rate": 6.240953947368421e-06,
      "loss": 0.1069,
      "step": 9150
    },
    {
      "epoch": 3.7664473684210527,
      "grad_norm": 107.98965454101562,
      "learning_rate": 6.236842105263159e-06,
      "loss": 0.1923,
      "step": 9160
    },
    {
      "epoch": 3.770559210526316,
      "grad_norm": 0.35728099942207336,
      "learning_rate": 6.232730263157895e-06,
      "loss": 0.1733,
      "step": 9170
    },
    {
      "epoch": 3.7746710526315788,
      "grad_norm": 8.636373519897461,
      "learning_rate": 6.228618421052633e-06,
      "loss": 0.2717,
      "step": 9180
    },
    {
      "epoch": 3.778782894736842,
      "grad_norm": 0.19581067562103271,
      "learning_rate": 6.224506578947369e-06,
      "loss": 0.3486,
      "step": 9190
    },
    {
      "epoch": 3.7828947368421053,
      "grad_norm": 0.1582966297864914,
      "learning_rate": 6.220394736842106e-06,
      "loss": 0.0158,
      "step": 9200
    },
    {
      "epoch": 3.7870065789473686,
      "grad_norm": 13.618598937988281,
      "learning_rate": 6.2162828947368425e-06,
      "loss": 0.6965,
      "step": 9210
    },
    {
      "epoch": 3.7911184210526314,
      "grad_norm": 5.897958755493164,
      "learning_rate": 6.212171052631579e-06,
      "loss": 0.2573,
      "step": 9220
    },
    {
      "epoch": 3.7952302631578947,
      "grad_norm": 59.38594055175781,
      "learning_rate": 6.208059210526316e-06,
      "loss": 0.0541,
      "step": 9230
    },
    {
      "epoch": 3.799342105263158,
      "grad_norm": 23.412710189819336,
      "learning_rate": 6.203947368421052e-06,
      "loss": 0.1623,
      "step": 9240
    },
    {
      "epoch": 3.8034539473684212,
      "grad_norm": 0.25934499502182007,
      "learning_rate": 6.19983552631579e-06,
      "loss": 0.2866,
      "step": 9250
    },
    {
      "epoch": 3.807565789473684,
      "grad_norm": 0.5576783418655396,
      "learning_rate": 6.195723684210527e-06,
      "loss": 0.0381,
      "step": 9260
    },
    {
      "epoch": 3.8116776315789473,
      "grad_norm": 0.00027464405866339803,
      "learning_rate": 6.191611842105264e-06,
      "loss": 0.4702,
      "step": 9270
    },
    {
      "epoch": 3.8157894736842106,
      "grad_norm": 56.96987533569336,
      "learning_rate": 6.1875000000000005e-06,
      "loss": 0.3138,
      "step": 9280
    },
    {
      "epoch": 3.8199013157894735,
      "grad_norm": 8.449546813964844,
      "learning_rate": 6.183388157894737e-06,
      "loss": 0.1577,
      "step": 9290
    },
    {
      "epoch": 3.8240131578947367,
      "grad_norm": 0.0012974452693015337,
      "learning_rate": 6.1792763157894745e-06,
      "loss": 0.103,
      "step": 9300
    },
    {
      "epoch": 3.828125,
      "grad_norm": 0.14060595631599426,
      "learning_rate": 6.175164473684211e-06,
      "loss": 0.1491,
      "step": 9310
    },
    {
      "epoch": 3.8322368421052633,
      "grad_norm": 4.392583847045898,
      "learning_rate": 6.171052631578948e-06,
      "loss": 0.0416,
      "step": 9320
    },
    {
      "epoch": 3.8363486842105265,
      "grad_norm": 95.75727844238281,
      "learning_rate": 6.166940789473684e-06,
      "loss": 0.2727,
      "step": 9330
    },
    {
      "epoch": 3.8404605263157894,
      "grad_norm": 1.871319055557251,
      "learning_rate": 6.162828947368421e-06,
      "loss": 0.188,
      "step": 9340
    },
    {
      "epoch": 3.8445723684210527,
      "grad_norm": 14.953452110290527,
      "learning_rate": 6.158717105263159e-06,
      "loss": 0.4324,
      "step": 9350
    },
    {
      "epoch": 3.848684210526316,
      "grad_norm": 55.11267852783203,
      "learning_rate": 6.154605263157896e-06,
      "loss": 0.0427,
      "step": 9360
    },
    {
      "epoch": 3.8527960526315788,
      "grad_norm": 0.2453383207321167,
      "learning_rate": 6.1504934210526325e-06,
      "loss": 0.1925,
      "step": 9370
    },
    {
      "epoch": 3.856907894736842,
      "grad_norm": 0.05007212609052658,
      "learning_rate": 6.146381578947369e-06,
      "loss": 0.2459,
      "step": 9380
    },
    {
      "epoch": 3.8610197368421053,
      "grad_norm": 129.4027862548828,
      "learning_rate": 6.142269736842106e-06,
      "loss": 0.3909,
      "step": 9390
    },
    {
      "epoch": 3.8651315789473686,
      "grad_norm": 0.061377912759780884,
      "learning_rate": 6.138157894736842e-06,
      "loss": 0.1726,
      "step": 9400
    },
    {
      "epoch": 3.8692434210526314,
      "grad_norm": 0.07050663232803345,
      "learning_rate": 6.134046052631579e-06,
      "loss": 0.2633,
      "step": 9410
    },
    {
      "epoch": 3.8733552631578947,
      "grad_norm": 0.820439875125885,
      "learning_rate": 6.129934210526316e-06,
      "loss": 0.3758,
      "step": 9420
    },
    {
      "epoch": 3.877467105263158,
      "grad_norm": 64.52366638183594,
      "learning_rate": 6.125822368421053e-06,
      "loss": 0.0713,
      "step": 9430
    },
    {
      "epoch": 3.8815789473684212,
      "grad_norm": 53.69047927856445,
      "learning_rate": 6.12171052631579e-06,
      "loss": 0.2844,
      "step": 9440
    },
    {
      "epoch": 3.885690789473684,
      "grad_norm": 0.0030439761467278004,
      "learning_rate": 6.117598684210527e-06,
      "loss": 0.1889,
      "step": 9450
    },
    {
      "epoch": 3.8898026315789473,
      "grad_norm": 0.0717514380812645,
      "learning_rate": 6.113486842105264e-06,
      "loss": 0.4461,
      "step": 9460
    },
    {
      "epoch": 3.8939144736842106,
      "grad_norm": 0.015020138584077358,
      "learning_rate": 6.109375000000001e-06,
      "loss": 0.115,
      "step": 9470
    },
    {
      "epoch": 3.8980263157894735,
      "grad_norm": 2.4814159870147705,
      "learning_rate": 6.105263157894738e-06,
      "loss": 0.2168,
      "step": 9480
    },
    {
      "epoch": 3.9021381578947367,
      "grad_norm": 1.2755320072174072,
      "learning_rate": 6.101151315789474e-06,
      "loss": 0.0417,
      "step": 9490
    },
    {
      "epoch": 3.90625,
      "grad_norm": 0.14059193432331085,
      "learning_rate": 6.097039473684211e-06,
      "loss": 0.1493,
      "step": 9500
    },
    {
      "epoch": 3.9103618421052633,
      "grad_norm": 0.1622990518808365,
      "learning_rate": 6.0929276315789476e-06,
      "loss": 0.0046,
      "step": 9510
    },
    {
      "epoch": 3.9144736842105265,
      "grad_norm": 0.36548370122909546,
      "learning_rate": 6.088815789473684e-06,
      "loss": 0.0143,
      "step": 9520
    },
    {
      "epoch": 3.9185855263157894,
      "grad_norm": 6.167750835418701,
      "learning_rate": 6.084703947368421e-06,
      "loss": 0.2583,
      "step": 9530
    },
    {
      "epoch": 3.9226973684210527,
      "grad_norm": 0.017177041620016098,
      "learning_rate": 6.080592105263159e-06,
      "loss": 0.2583,
      "step": 9540
    },
    {
      "epoch": 3.926809210526316,
      "grad_norm": 7.4479875564575195,
      "learning_rate": 6.076480263157896e-06,
      "loss": 0.0245,
      "step": 9550
    },
    {
      "epoch": 3.9309210526315788,
      "grad_norm": 6.392214775085449,
      "learning_rate": 6.072368421052632e-06,
      "loss": 0.0585,
      "step": 9560
    },
    {
      "epoch": 3.935032894736842,
      "grad_norm": 45.556400299072266,
      "learning_rate": 6.068256578947369e-06,
      "loss": 0.1436,
      "step": 9570
    },
    {
      "epoch": 3.9391447368421053,
      "grad_norm": 0.0030166308861225843,
      "learning_rate": 6.0641447368421055e-06,
      "loss": 0.3354,
      "step": 9580
    },
    {
      "epoch": 3.9432565789473686,
      "grad_norm": 0.04616885259747505,
      "learning_rate": 6.060032894736842e-06,
      "loss": 0.3983,
      "step": 9590
    },
    {
      "epoch": 3.9473684210526314,
      "grad_norm": 50.227230072021484,
      "learning_rate": 6.0559210526315796e-06,
      "loss": 0.2884,
      "step": 9600
    },
    {
      "epoch": 3.9514802631578947,
      "grad_norm": 74.54116821289062,
      "learning_rate": 6.051809210526316e-06,
      "loss": 0.0673,
      "step": 9610
    },
    {
      "epoch": 3.955592105263158,
      "grad_norm": 2.7192249298095703,
      "learning_rate": 6.047697368421053e-06,
      "loss": 0.0884,
      "step": 9620
    },
    {
      "epoch": 3.9597039473684212,
      "grad_norm": 9.346713066101074,
      "learning_rate": 6.043585526315789e-06,
      "loss": 0.0085,
      "step": 9630
    },
    {
      "epoch": 3.963815789473684,
      "grad_norm": 67.16130065917969,
      "learning_rate": 6.039473684210527e-06,
      "loss": 0.6519,
      "step": 9640
    },
    {
      "epoch": 3.9679276315789473,
      "grad_norm": 0.004526804201304913,
      "learning_rate": 6.035361842105264e-06,
      "loss": 0.0041,
      "step": 9650
    },
    {
      "epoch": 3.9720394736842106,
      "grad_norm": 22.933185577392578,
      "learning_rate": 6.031250000000001e-06,
      "loss": 0.0321,
      "step": 9660
    },
    {
      "epoch": 3.9761513157894735,
      "grad_norm": 0.031579192727804184,
      "learning_rate": 6.0271381578947375e-06,
      "loss": 0.0978,
      "step": 9670
    },
    {
      "epoch": 3.9802631578947367,
      "grad_norm": 0.30152076482772827,
      "learning_rate": 6.023026315789474e-06,
      "loss": 0.2345,
      "step": 9680
    },
    {
      "epoch": 3.984375,
      "grad_norm": 0.02326991781592369,
      "learning_rate": 6.018914473684211e-06,
      "loss": 0.1673,
      "step": 9690
    },
    {
      "epoch": 3.9884868421052633,
      "grad_norm": 0.11922337114810944,
      "learning_rate": 6.014802631578947e-06,
      "loss": 0.5942,
      "step": 9700
    },
    {
      "epoch": 3.9925986842105265,
      "grad_norm": 0.20456120371818542,
      "learning_rate": 6.010690789473684e-06,
      "loss": 0.2563,
      "step": 9710
    },
    {
      "epoch": 3.9967105263157894,
      "grad_norm": 0.08687052130699158,
      "learning_rate": 6.006578947368421e-06,
      "loss": 0.1033,
      "step": 9720
    },
    {
      "epoch": 4.000822368421052,
      "grad_norm": 0.08968238532543182,
      "learning_rate": 6.002467105263159e-06,
      "loss": 0.1698,
      "step": 9730
    },
    {
      "epoch": 4.004934210526316,
      "grad_norm": 0.6397474408149719,
      "learning_rate": 5.9983552631578955e-06,
      "loss": 0.0484,
      "step": 9740
    },
    {
      "epoch": 4.009046052631579,
      "grad_norm": 4.620758056640625,
      "learning_rate": 5.994243421052632e-06,
      "loss": 0.1483,
      "step": 9750
    },
    {
      "epoch": 4.0131578947368425,
      "grad_norm": 10.348681449890137,
      "learning_rate": 5.990131578947369e-06,
      "loss": 0.0073,
      "step": 9760
    },
    {
      "epoch": 4.017269736842105,
      "grad_norm": 62.68147659301758,
      "learning_rate": 5.986019736842106e-06,
      "loss": 0.0964,
      "step": 9770
    },
    {
      "epoch": 4.021381578947368,
      "grad_norm": 5.771296501159668,
      "learning_rate": 5.981907894736843e-06,
      "loss": 0.0631,
      "step": 9780
    },
    {
      "epoch": 4.025493421052632,
      "grad_norm": 63.21457290649414,
      "learning_rate": 5.977796052631579e-06,
      "loss": 0.3454,
      "step": 9790
    },
    {
      "epoch": 4.029605263157895,
      "grad_norm": 0.0331178717315197,
      "learning_rate": 5.973684210526316e-06,
      "loss": 0.0043,
      "step": 9800
    },
    {
      "epoch": 4.0337171052631575,
      "grad_norm": 13.591999053955078,
      "learning_rate": 5.969572368421053e-06,
      "loss": 0.01,
      "step": 9810
    },
    {
      "epoch": 4.037828947368421,
      "grad_norm": 2.523550033569336,
      "learning_rate": 5.965460526315789e-06,
      "loss": 0.1164,
      "step": 9820
    },
    {
      "epoch": 4.041940789473684,
      "grad_norm": 0.0010173360351473093,
      "learning_rate": 5.9613486842105275e-06,
      "loss": 0.0702,
      "step": 9830
    },
    {
      "epoch": 4.046052631578948,
      "grad_norm": 0.008227844722568989,
      "learning_rate": 5.957236842105264e-06,
      "loss": 0.0444,
      "step": 9840
    },
    {
      "epoch": 4.050164473684211,
      "grad_norm": 51.15441131591797,
      "learning_rate": 5.953125000000001e-06,
      "loss": 0.0465,
      "step": 9850
    },
    {
      "epoch": 4.0542763157894735,
      "grad_norm": 0.08978760987520218,
      "learning_rate": 5.949013157894737e-06,
      "loss": 0.2733,
      "step": 9860
    },
    {
      "epoch": 4.058388157894737,
      "grad_norm": 0.0032470934092998505,
      "learning_rate": 5.944901315789474e-06,
      "loss": 0.0077,
      "step": 9870
    },
    {
      "epoch": 4.0625,
      "grad_norm": 2.342414617538452,
      "learning_rate": 5.9407894736842105e-06,
      "loss": 0.023,
      "step": 9880
    },
    {
      "epoch": 4.066611842105263,
      "grad_norm": 0.2732146978378296,
      "learning_rate": 5.936677631578948e-06,
      "loss": 0.1348,
      "step": 9890
    },
    {
      "epoch": 4.0707236842105265,
      "grad_norm": 0.014328996650874615,
      "learning_rate": 5.932565789473685e-06,
      "loss": 0.0011,
      "step": 9900
    },
    {
      "epoch": 4.074835526315789,
      "grad_norm": 56.14218521118164,
      "learning_rate": 5.928453947368421e-06,
      "loss": 0.1015,
      "step": 9910
    },
    {
      "epoch": 4.078947368421052,
      "grad_norm": 0.04061206430196762,
      "learning_rate": 5.924342105263159e-06,
      "loss": 0.3239,
      "step": 9920
    },
    {
      "epoch": 4.083059210526316,
      "grad_norm": 0.8352648019790649,
      "learning_rate": 5.920230263157895e-06,
      "loss": 0.0036,
      "step": 9930
    },
    {
      "epoch": 4.087171052631579,
      "grad_norm": 8.8486909866333,
      "learning_rate": 5.916118421052633e-06,
      "loss": 0.0076,
      "step": 9940
    },
    {
      "epoch": 4.0912828947368425,
      "grad_norm": 88.49449157714844,
      "learning_rate": 5.912006578947369e-06,
      "loss": 0.179,
      "step": 9950
    },
    {
      "epoch": 4.095394736842105,
      "grad_norm": 2.3278348445892334,
      "learning_rate": 5.907894736842106e-06,
      "loss": 0.1412,
      "step": 9960
    },
    {
      "epoch": 4.099506578947368,
      "grad_norm": 17.677732467651367,
      "learning_rate": 5.9037828947368425e-06,
      "loss": 0.0092,
      "step": 9970
    },
    {
      "epoch": 4.103618421052632,
      "grad_norm": 1.240334391593933,
      "learning_rate": 5.899671052631579e-06,
      "loss": 0.2093,
      "step": 9980
    },
    {
      "epoch": 4.107730263157895,
      "grad_norm": 0.012150059454143047,
      "learning_rate": 5.895559210526316e-06,
      "loss": 0.0079,
      "step": 9990
    },
    {
      "epoch": 4.1118421052631575,
      "grad_norm": 76.60591125488281,
      "learning_rate": 5.891447368421052e-06,
      "loss": 0.2322,
      "step": 10000
    },
    {
      "epoch": 4.115953947368421,
      "grad_norm": 0.41860559582710266,
      "learning_rate": 5.88733552631579e-06,
      "loss": 0.1039,
      "step": 10010
    },
    {
      "epoch": 4.120065789473684,
      "grad_norm": 0.010364804416894913,
      "learning_rate": 5.883223684210527e-06,
      "loss": 0.0324,
      "step": 10020
    },
    {
      "epoch": 4.124177631578948,
      "grad_norm": 14.311861038208008,
      "learning_rate": 5.879111842105264e-06,
      "loss": 0.0699,
      "step": 10030
    },
    {
      "epoch": 4.128289473684211,
      "grad_norm": 0.5761501789093018,
      "learning_rate": 5.8750000000000005e-06,
      "loss": 0.5303,
      "step": 10040
    },
    {
      "epoch": 4.1324013157894735,
      "grad_norm": 1.0431439876556396,
      "learning_rate": 5.870888157894737e-06,
      "loss": 0.0503,
      "step": 10050
    },
    {
      "epoch": 4.136513157894737,
      "grad_norm": 0.0012574722059071064,
      "learning_rate": 5.8667763157894746e-06,
      "loss": 0.0505,
      "step": 10060
    },
    {
      "epoch": 4.140625,
      "grad_norm": 1.2190923690795898,
      "learning_rate": 5.862664473684211e-06,
      "loss": 0.0039,
      "step": 10070
    },
    {
      "epoch": 4.144736842105263,
      "grad_norm": 0.010865393094718456,
      "learning_rate": 5.858552631578948e-06,
      "loss": 0.0199,
      "step": 10080
    },
    {
      "epoch": 4.1488486842105265,
      "grad_norm": 0.5229134559631348,
      "learning_rate": 5.854440789473684e-06,
      "loss": 0.3392,
      "step": 10090
    },
    {
      "epoch": 4.152960526315789,
      "grad_norm": 18.424875259399414,
      "learning_rate": 5.850328947368421e-06,
      "loss": 0.2647,
      "step": 10100
    },
    {
      "epoch": 4.157072368421052,
      "grad_norm": 0.25895851850509644,
      "learning_rate": 5.846217105263159e-06,
      "loss": 0.1165,
      "step": 10110
    },
    {
      "epoch": 4.161184210526316,
      "grad_norm": 53.10039520263672,
      "learning_rate": 5.842105263157896e-06,
      "loss": 0.0722,
      "step": 10120
    },
    {
      "epoch": 4.165296052631579,
      "grad_norm": 0.66398024559021,
      "learning_rate": 5.8379934210526325e-06,
      "loss": 0.1419,
      "step": 10130
    },
    {
      "epoch": 4.1694078947368425,
      "grad_norm": 0.014206175692379475,
      "learning_rate": 5.833881578947369e-06,
      "loss": 0.2225,
      "step": 10140
    },
    {
      "epoch": 4.173519736842105,
      "grad_norm": 5.291435718536377,
      "learning_rate": 5.829769736842106e-06,
      "loss": 0.0228,
      "step": 10150
    },
    {
      "epoch": 4.177631578947368,
      "grad_norm": 3.0082762241363525,
      "learning_rate": 5.825657894736842e-06,
      "loss": 0.2913,
      "step": 10160
    },
    {
      "epoch": 4.181743421052632,
      "grad_norm": 0.08264022320508957,
      "learning_rate": 5.821546052631579e-06,
      "loss": 0.3597,
      "step": 10170
    },
    {
      "epoch": 4.185855263157895,
      "grad_norm": 35.833770751953125,
      "learning_rate": 5.8174342105263156e-06,
      "loss": 0.0419,
      "step": 10180
    },
    {
      "epoch": 4.1899671052631575,
      "grad_norm": 85.25164031982422,
      "learning_rate": 5.813322368421053e-06,
      "loss": 0.2057,
      "step": 10190
    },
    {
      "epoch": 4.194078947368421,
      "grad_norm": 0.0025984339881688356,
      "learning_rate": 5.80921052631579e-06,
      "loss": 0.3559,
      "step": 10200
    },
    {
      "epoch": 4.198190789473684,
      "grad_norm": 0.2660520672798157,
      "learning_rate": 5.805098684210527e-06,
      "loss": 0.0053,
      "step": 10210
    },
    {
      "epoch": 4.202302631578948,
      "grad_norm": 0.7417953014373779,
      "learning_rate": 5.800986842105264e-06,
      "loss": 0.1898,
      "step": 10220
    },
    {
      "epoch": 4.206414473684211,
      "grad_norm": 0.37967509031295776,
      "learning_rate": 5.796875e-06,
      "loss": 0.27,
      "step": 10230
    },
    {
      "epoch": 4.2105263157894735,
      "grad_norm": 0.0646011158823967,
      "learning_rate": 5.792763157894738e-06,
      "loss": 0.5033,
      "step": 10240
    },
    {
      "epoch": 4.214638157894737,
      "grad_norm": 0.042865119874477386,
      "learning_rate": 5.788651315789474e-06,
      "loss": 0.0579,
      "step": 10250
    },
    {
      "epoch": 4.21875,
      "grad_norm": 0.905980110168457,
      "learning_rate": 5.784539473684211e-06,
      "loss": 0.0618,
      "step": 10260
    },
    {
      "epoch": 4.222861842105263,
      "grad_norm": 0.011512444354593754,
      "learning_rate": 5.780427631578948e-06,
      "loss": 0.0534,
      "step": 10270
    },
    {
      "epoch": 4.2269736842105265,
      "grad_norm": 0.009037915617227554,
      "learning_rate": 5.776315789473684e-06,
      "loss": 0.1521,
      "step": 10280
    },
    {
      "epoch": 4.231085526315789,
      "grad_norm": 44.18505859375,
      "learning_rate": 5.772203947368421e-06,
      "loss": 0.0802,
      "step": 10290
    },
    {
      "epoch": 4.235197368421052,
      "grad_norm": 0.24068020284175873,
      "learning_rate": 5.768092105263159e-06,
      "loss": 0.0494,
      "step": 10300
    },
    {
      "epoch": 4.239309210526316,
      "grad_norm": 3.5520458221435547,
      "learning_rate": 5.763980263157896e-06,
      "loss": 0.0715,
      "step": 10310
    },
    {
      "epoch": 4.243421052631579,
      "grad_norm": 0.0010897587053477764,
      "learning_rate": 5.759868421052632e-06,
      "loss": 0.087,
      "step": 10320
    },
    {
      "epoch": 4.2475328947368425,
      "grad_norm": 5.748364448547363,
      "learning_rate": 5.755756578947369e-06,
      "loss": 0.0721,
      "step": 10330
    },
    {
      "epoch": 4.251644736842105,
      "grad_norm": 0.019158218055963516,
      "learning_rate": 5.7516447368421055e-06,
      "loss": 0.458,
      "step": 10340
    },
    {
      "epoch": 4.255756578947368,
      "grad_norm": 0.030008016154170036,
      "learning_rate": 5.747532894736842e-06,
      "loss": 0.052,
      "step": 10350
    },
    {
      "epoch": 4.259868421052632,
      "grad_norm": 37.76634216308594,
      "learning_rate": 5.74342105263158e-06,
      "loss": 0.275,
      "step": 10360
    },
    {
      "epoch": 4.263980263157895,
      "grad_norm": 0.003279157681390643,
      "learning_rate": 5.739309210526316e-06,
      "loss": 0.0036,
      "step": 10370
    },
    {
      "epoch": 4.2680921052631575,
      "grad_norm": 0.0217817984521389,
      "learning_rate": 5.735197368421053e-06,
      "loss": 0.0017,
      "step": 10380
    },
    {
      "epoch": 4.272203947368421,
      "grad_norm": 0.6077308058738708,
      "learning_rate": 5.731085526315789e-06,
      "loss": 0.0008,
      "step": 10390
    },
    {
      "epoch": 4.276315789473684,
      "grad_norm": 0.4642079770565033,
      "learning_rate": 5.726973684210527e-06,
      "loss": 0.2043,
      "step": 10400
    },
    {
      "epoch": 4.280427631578947,
      "grad_norm": 0.002989094937220216,
      "learning_rate": 5.722861842105264e-06,
      "loss": 0.052,
      "step": 10410
    },
    {
      "epoch": 4.284539473684211,
      "grad_norm": 0.0273353923112154,
      "learning_rate": 5.719161184210527e-06,
      "loss": 0.3025,
      "step": 10420
    },
    {
      "epoch": 4.2886513157894735,
      "grad_norm": 77.05641174316406,
      "learning_rate": 5.715049342105263e-06,
      "loss": 0.2568,
      "step": 10430
    },
    {
      "epoch": 4.292763157894737,
      "grad_norm": 0.10438290983438492,
      "learning_rate": 5.7109375e-06,
      "loss": 0.142,
      "step": 10440
    },
    {
      "epoch": 4.296875,
      "grad_norm": 0.05108823999762535,
      "learning_rate": 5.706825657894738e-06,
      "loss": 0.123,
      "step": 10450
    },
    {
      "epoch": 4.300986842105263,
      "grad_norm": 24.811241149902344,
      "learning_rate": 5.702713815789475e-06,
      "loss": 0.0748,
      "step": 10460
    },
    {
      "epoch": 4.3050986842105265,
      "grad_norm": 1.4945974349975586,
      "learning_rate": 5.698601973684211e-06,
      "loss": 0.0022,
      "step": 10470
    },
    {
      "epoch": 4.309210526315789,
      "grad_norm": 0.003993542864918709,
      "learning_rate": 5.694490131578948e-06,
      "loss": 0.0409,
      "step": 10480
    },
    {
      "epoch": 4.313322368421053,
      "grad_norm": 0.3219471573829651,
      "learning_rate": 5.690378289473685e-06,
      "loss": 0.1386,
      "step": 10490
    },
    {
      "epoch": 4.317434210526316,
      "grad_norm": 15.706138610839844,
      "learning_rate": 5.686266447368421e-06,
      "loss": 0.0043,
      "step": 10500
    },
    {
      "epoch": 4.321546052631579,
      "grad_norm": 26.479265213012695,
      "learning_rate": 5.682154605263158e-06,
      "loss": 0.1416,
      "step": 10510
    },
    {
      "epoch": 4.3256578947368425,
      "grad_norm": 0.08380565047264099,
      "learning_rate": 5.678042763157895e-06,
      "loss": 0.0611,
      "step": 10520
    },
    {
      "epoch": 4.329769736842105,
      "grad_norm": 0.5438801050186157,
      "learning_rate": 5.673930921052632e-06,
      "loss": 0.2394,
      "step": 10530
    },
    {
      "epoch": 4.333881578947368,
      "grad_norm": 83.95884704589844,
      "learning_rate": 5.6698190789473685e-06,
      "loss": 0.343,
      "step": 10540
    },
    {
      "epoch": 4.337993421052632,
      "grad_norm": 68.3953857421875,
      "learning_rate": 5.665707236842106e-06,
      "loss": 0.0774,
      "step": 10550
    },
    {
      "epoch": 4.342105263157895,
      "grad_norm": 0.2621876299381256,
      "learning_rate": 5.661595394736843e-06,
      "loss": 0.0171,
      "step": 10560
    },
    {
      "epoch": 4.3462171052631575,
      "grad_norm": 3.258847951889038,
      "learning_rate": 5.65748355263158e-06,
      "loss": 0.0275,
      "step": 10570
    },
    {
      "epoch": 4.350328947368421,
      "grad_norm": 52.767311096191406,
      "learning_rate": 5.653371710526317e-06,
      "loss": 0.0406,
      "step": 10580
    },
    {
      "epoch": 4.354440789473684,
      "grad_norm": 0.07652966678142548,
      "learning_rate": 5.649259868421053e-06,
      "loss": 0.1451,
      "step": 10590
    },
    {
      "epoch": 4.358552631578947,
      "grad_norm": 0.0008912061457522213,
      "learning_rate": 5.64514802631579e-06,
      "loss": 0.007,
      "step": 10600
    },
    {
      "epoch": 4.362664473684211,
      "grad_norm": 0.0011371123837307096,
      "learning_rate": 5.6410361842105265e-06,
      "loss": 0.0695,
      "step": 10610
    },
    {
      "epoch": 4.3667763157894735,
      "grad_norm": 1.0595893859863281,
      "learning_rate": 5.636924342105263e-06,
      "loss": 0.0103,
      "step": 10620
    },
    {
      "epoch": 4.370888157894737,
      "grad_norm": 34.24848937988281,
      "learning_rate": 5.6328125e-06,
      "loss": 0.5059,
      "step": 10630
    },
    {
      "epoch": 4.375,
      "grad_norm": 0.004730758722871542,
      "learning_rate": 5.628700657894738e-06,
      "loss": 0.0572,
      "step": 10640
    },
    {
      "epoch": 4.379111842105263,
      "grad_norm": 0.024553537368774414,
      "learning_rate": 5.624588815789475e-06,
      "loss": 0.0008,
      "step": 10650
    },
    {
      "epoch": 4.3832236842105265,
      "grad_norm": 0.03798150643706322,
      "learning_rate": 5.620476973684211e-06,
      "loss": 0.1627,
      "step": 10660
    },
    {
      "epoch": 4.387335526315789,
      "grad_norm": 1.5321662425994873,
      "learning_rate": 5.616365131578948e-06,
      "loss": 0.266,
      "step": 10670
    },
    {
      "epoch": 4.391447368421053,
      "grad_norm": 0.02827838808298111,
      "learning_rate": 5.6122532894736844e-06,
      "loss": 0.0043,
      "step": 10680
    },
    {
      "epoch": 4.395559210526316,
      "grad_norm": 0.8193231225013733,
      "learning_rate": 5.608141447368422e-06,
      "loss": 0.1079,
      "step": 10690
    },
    {
      "epoch": 4.399671052631579,
      "grad_norm": 0.008944814093410969,
      "learning_rate": 5.6040296052631585e-06,
      "loss": 0.1774,
      "step": 10700
    },
    {
      "epoch": 4.4037828947368425,
      "grad_norm": 71.00315856933594,
      "learning_rate": 5.599917763157895e-06,
      "loss": 0.252,
      "step": 10710
    },
    {
      "epoch": 4.407894736842105,
      "grad_norm": 0.2633829712867737,
      "learning_rate": 5.595805921052632e-06,
      "loss": 0.2704,
      "step": 10720
    },
    {
      "epoch": 4.412006578947368,
      "grad_norm": 0.5669187903404236,
      "learning_rate": 5.591694078947368e-06,
      "loss": 0.1003,
      "step": 10730
    },
    {
      "epoch": 4.416118421052632,
      "grad_norm": 0.027173712849617004,
      "learning_rate": 5.587582236842107e-06,
      "loss": 0.079,
      "step": 10740
    },
    {
      "epoch": 4.420230263157895,
      "grad_norm": 0.06833348423242569,
      "learning_rate": 5.583470394736843e-06,
      "loss": 0.2802,
      "step": 10750
    },
    {
      "epoch": 4.4243421052631575,
      "grad_norm": 77.60299682617188,
      "learning_rate": 5.57935855263158e-06,
      "loss": 0.2194,
      "step": 10760
    },
    {
      "epoch": 4.428453947368421,
      "grad_norm": 0.24349355697631836,
      "learning_rate": 5.5752467105263165e-06,
      "loss": 0.2518,
      "step": 10770
    },
    {
      "epoch": 4.432565789473684,
      "grad_norm": 0.6897732019424438,
      "learning_rate": 5.571134868421053e-06,
      "loss": 0.3367,
      "step": 10780
    },
    {
      "epoch": 4.436677631578947,
      "grad_norm": 4.426616668701172,
      "learning_rate": 5.56702302631579e-06,
      "loss": 0.2116,
      "step": 10790
    },
    {
      "epoch": 4.440789473684211,
      "grad_norm": 0.06270567327737808,
      "learning_rate": 5.562911184210526e-06,
      "loss": 0.0551,
      "step": 10800
    },
    {
      "epoch": 4.4449013157894735,
      "grad_norm": 15.072378158569336,
      "learning_rate": 5.558799342105263e-06,
      "loss": 0.0283,
      "step": 10810
    },
    {
      "epoch": 4.449013157894737,
      "grad_norm": 0.22645829617977142,
      "learning_rate": 5.5546875e-06,
      "loss": 0.0336,
      "step": 10820
    },
    {
      "epoch": 4.453125,
      "grad_norm": 0.595355749130249,
      "learning_rate": 5.550575657894738e-06,
      "loss": 0.1071,
      "step": 10830
    },
    {
      "epoch": 4.457236842105263,
      "grad_norm": 1.1005072593688965,
      "learning_rate": 5.546463815789474e-06,
      "loss": 0.047,
      "step": 10840
    },
    {
      "epoch": 4.4613486842105265,
      "grad_norm": 121.18562316894531,
      "learning_rate": 5.542351973684211e-06,
      "loss": 0.7024,
      "step": 10850
    },
    {
      "epoch": 4.465460526315789,
      "grad_norm": 0.004475445486605167,
      "learning_rate": 5.538240131578948e-06,
      "loss": 0.0434,
      "step": 10860
    },
    {
      "epoch": 4.469572368421053,
      "grad_norm": 0.03283686935901642,
      "learning_rate": 5.534128289473685e-06,
      "loss": 0.0131,
      "step": 10870
    },
    {
      "epoch": 4.473684210526316,
      "grad_norm": 0.7433743476867676,
      "learning_rate": 5.530016447368422e-06,
      "loss": 0.0166,
      "step": 10880
    },
    {
      "epoch": 4.477796052631579,
      "grad_norm": 77.12642669677734,
      "learning_rate": 5.525904605263158e-06,
      "loss": 0.1083,
      "step": 10890
    },
    {
      "epoch": 4.4819078947368425,
      "grad_norm": 0.37571632862091064,
      "learning_rate": 5.521792763157895e-06,
      "loss": 0.0452,
      "step": 10900
    },
    {
      "epoch": 4.486019736842105,
      "grad_norm": 0.03792794793844223,
      "learning_rate": 5.5176809210526315e-06,
      "loss": 0.0042,
      "step": 10910
    },
    {
      "epoch": 4.490131578947368,
      "grad_norm": 0.017094863578677177,
      "learning_rate": 5.513569078947368e-06,
      "loss": 0.1781,
      "step": 10920
    },
    {
      "epoch": 4.494243421052632,
      "grad_norm": 9.493055404163897e-05,
      "learning_rate": 5.509457236842106e-06,
      "loss": 0.5008,
      "step": 10930
    },
    {
      "epoch": 4.498355263157895,
      "grad_norm": 1.660346508026123,
      "learning_rate": 5.505345394736843e-06,
      "loss": 0.0303,
      "step": 10940
    },
    {
      "epoch": 4.5024671052631575,
      "grad_norm": 0.02450186386704445,
      "learning_rate": 5.50123355263158e-06,
      "loss": 0.0511,
      "step": 10950
    },
    {
      "epoch": 4.506578947368421,
      "grad_norm": 0.11601614207029343,
      "learning_rate": 5.497121710526316e-06,
      "loss": 0.0899,
      "step": 10960
    },
    {
      "epoch": 4.510690789473684,
      "grad_norm": 4.531919002532959,
      "learning_rate": 5.493009868421053e-06,
      "loss": 0.0918,
      "step": 10970
    },
    {
      "epoch": 4.514802631578947,
      "grad_norm": 8.091043472290039,
      "learning_rate": 5.4888980263157895e-06,
      "loss": 0.187,
      "step": 10980
    },
    {
      "epoch": 4.518914473684211,
      "grad_norm": 51.77922821044922,
      "learning_rate": 5.484786184210527e-06,
      "loss": 0.0883,
      "step": 10990
    },
    {
      "epoch": 4.5230263157894735,
      "grad_norm": 0.0011709548998624086,
      "learning_rate": 5.4806743421052635e-06,
      "loss": 0.183,
      "step": 11000
    },
    {
      "epoch": 4.527138157894737,
      "grad_norm": 75.65191650390625,
      "learning_rate": 5.4765625e-06,
      "loss": 0.5483,
      "step": 11010
    },
    {
      "epoch": 4.53125,
      "grad_norm": 0.31503334641456604,
      "learning_rate": 5.472450657894738e-06,
      "loss": 0.021,
      "step": 11020
    },
    {
      "epoch": 4.535361842105263,
      "grad_norm": 0.0058042751625180244,
      "learning_rate": 5.468338815789474e-06,
      "loss": 0.0388,
      "step": 11030
    },
    {
      "epoch": 4.5394736842105265,
      "grad_norm": 0.70481938123703,
      "learning_rate": 5.464226973684212e-06,
      "loss": 0.1086,
      "step": 11040
    },
    {
      "epoch": 4.543585526315789,
      "grad_norm": 6.683977127075195,
      "learning_rate": 5.460115131578948e-06,
      "loss": 0.1748,
      "step": 11050
    },
    {
      "epoch": 4.547697368421053,
      "grad_norm": 0.8053579926490784,
      "learning_rate": 5.456003289473685e-06,
      "loss": 0.2015,
      "step": 11060
    },
    {
      "epoch": 4.551809210526316,
      "grad_norm": 0.025869661942124367,
      "learning_rate": 5.4518914473684215e-06,
      "loss": 0.0331,
      "step": 11070
    },
    {
      "epoch": 4.555921052631579,
      "grad_norm": 11.96985912322998,
      "learning_rate": 5.447779605263158e-06,
      "loss": 0.1065,
      "step": 11080
    },
    {
      "epoch": 4.5600328947368425,
      "grad_norm": 0.0028174128383398056,
      "learning_rate": 5.443667763157895e-06,
      "loss": 0.0375,
      "step": 11090
    },
    {
      "epoch": 4.564144736842105,
      "grad_norm": 85.98474884033203,
      "learning_rate": 5.439555921052631e-06,
      "loss": 0.14,
      "step": 11100
    },
    {
      "epoch": 4.568256578947368,
      "grad_norm": 7.298089981079102,
      "learning_rate": 5.435444078947369e-06,
      "loss": 0.1029,
      "step": 11110
    },
    {
      "epoch": 4.572368421052632,
      "grad_norm": 0.0005351307918317616,
      "learning_rate": 5.431332236842106e-06,
      "loss": 0.2296,
      "step": 11120
    },
    {
      "epoch": 4.576480263157895,
      "grad_norm": 0.5227247476577759,
      "learning_rate": 5.427220394736843e-06,
      "loss": 0.0584,
      "step": 11130
    },
    {
      "epoch": 4.5805921052631575,
      "grad_norm": 0.37593796849250793,
      "learning_rate": 5.4231085526315794e-06,
      "loss": 0.0093,
      "step": 11140
    },
    {
      "epoch": 4.584703947368421,
      "grad_norm": 0.3121981918811798,
      "learning_rate": 5.418996710526316e-06,
      "loss": 0.0211,
      "step": 11150
    },
    {
      "epoch": 4.588815789473684,
      "grad_norm": 0.44901901483535767,
      "learning_rate": 5.4148848684210535e-06,
      "loss": 0.0044,
      "step": 11160
    },
    {
      "epoch": 4.592927631578947,
      "grad_norm": 34.624664306640625,
      "learning_rate": 5.41077302631579e-06,
      "loss": 0.1367,
      "step": 11170
    },
    {
      "epoch": 4.597039473684211,
      "grad_norm": 17.65924835205078,
      "learning_rate": 5.406661184210527e-06,
      "loss": 0.419,
      "step": 11180
    },
    {
      "epoch": 4.6011513157894735,
      "grad_norm": 0.12152664363384247,
      "learning_rate": 5.402549342105263e-06,
      "loss": 0.0223,
      "step": 11190
    },
    {
      "epoch": 4.605263157894737,
      "grad_norm": 0.002171924104914069,
      "learning_rate": 5.3984375e-06,
      "loss": 0.0406,
      "step": 11200
    },
    {
      "epoch": 4.609375,
      "grad_norm": 0.030598795041441917,
      "learning_rate": 5.394325657894738e-06,
      "loss": 0.0308,
      "step": 11210
    },
    {
      "epoch": 4.613486842105263,
      "grad_norm": 116.11152648925781,
      "learning_rate": 5.390213815789475e-06,
      "loss": 0.1755,
      "step": 11220
    },
    {
      "epoch": 4.6175986842105265,
      "grad_norm": 0.0010453220456838608,
      "learning_rate": 5.3861019736842114e-06,
      "loss": 0.4576,
      "step": 11230
    },
    {
      "epoch": 4.621710526315789,
      "grad_norm": 0.016688738018274307,
      "learning_rate": 5.381990131578948e-06,
      "loss": 0.2701,
      "step": 11240
    },
    {
      "epoch": 4.625822368421053,
      "grad_norm": 0.0017963089048862457,
      "learning_rate": 5.377878289473685e-06,
      "loss": 0.0009,
      "step": 11250
    },
    {
      "epoch": 4.629934210526316,
      "grad_norm": 0.00248286803252995,
      "learning_rate": 5.373766447368421e-06,
      "loss": 0.2538,
      "step": 11260
    },
    {
      "epoch": 4.634046052631579,
      "grad_norm": 0.01858946494758129,
      "learning_rate": 5.369654605263158e-06,
      "loss": 0.5668,
      "step": 11270
    },
    {
      "epoch": 4.6381578947368425,
      "grad_norm": 84.9400634765625,
      "learning_rate": 5.3655427631578945e-06,
      "loss": 0.1191,
      "step": 11280
    },
    {
      "epoch": 4.642269736842105,
      "grad_norm": 46.583229064941406,
      "learning_rate": 5.361430921052632e-06,
      "loss": 0.0579,
      "step": 11290
    },
    {
      "epoch": 4.646381578947368,
      "grad_norm": 3.9586727619171143,
      "learning_rate": 5.3573190789473686e-06,
      "loss": 0.1239,
      "step": 11300
    },
    {
      "epoch": 4.650493421052632,
      "grad_norm": 1.132432222366333,
      "learning_rate": 5.353207236842106e-06,
      "loss": 0.1484,
      "step": 11310
    },
    {
      "epoch": 4.654605263157895,
      "grad_norm": 2.125668525695801,
      "learning_rate": 5.349095394736843e-06,
      "loss": 0.412,
      "step": 11320
    },
    {
      "epoch": 4.6587171052631575,
      "grad_norm": 0.03645682334899902,
      "learning_rate": 5.344983552631579e-06,
      "loss": 0.0031,
      "step": 11330
    },
    {
      "epoch": 4.662828947368421,
      "grad_norm": 81.14274597167969,
      "learning_rate": 5.340871710526317e-06,
      "loss": 0.0924,
      "step": 11340
    },
    {
      "epoch": 4.666940789473684,
      "grad_norm": 0.13125614821910858,
      "learning_rate": 5.336759868421053e-06,
      "loss": 0.4672,
      "step": 11350
    },
    {
      "epoch": 4.671052631578947,
      "grad_norm": 0.0066694519482553005,
      "learning_rate": 5.33264802631579e-06,
      "loss": 0.0737,
      "step": 11360
    },
    {
      "epoch": 4.675164473684211,
      "grad_norm": 0.05095091834664345,
      "learning_rate": 5.3285361842105265e-06,
      "loss": 0.2657,
      "step": 11370
    },
    {
      "epoch": 4.6792763157894735,
      "grad_norm": 0.019465183839201927,
      "learning_rate": 5.324424342105263e-06,
      "loss": 0.0345,
      "step": 11380
    },
    {
      "epoch": 4.683388157894737,
      "grad_norm": 0.2005145102739334,
      "learning_rate": 5.3203125e-06,
      "loss": 0.0541,
      "step": 11390
    },
    {
      "epoch": 4.6875,
      "grad_norm": 2.549844264984131,
      "learning_rate": 5.316200657894738e-06,
      "loss": 0.2792,
      "step": 11400
    },
    {
      "epoch": 4.691611842105263,
      "grad_norm": 9.314865112304688,
      "learning_rate": 5.312088815789475e-06,
      "loss": 0.0775,
      "step": 11410
    },
    {
      "epoch": 4.6957236842105265,
      "grad_norm": 0.14117684960365295,
      "learning_rate": 5.307976973684211e-06,
      "loss": 0.295,
      "step": 11420
    },
    {
      "epoch": 4.699835526315789,
      "grad_norm": 0.00023204373428598046,
      "learning_rate": 5.303865131578948e-06,
      "loss": 0.1064,
      "step": 11430
    },
    {
      "epoch": 4.703947368421053,
      "grad_norm": 0.059099286794662476,
      "learning_rate": 5.2997532894736845e-06,
      "loss": 0.1719,
      "step": 11440
    },
    {
      "epoch": 4.708059210526316,
      "grad_norm": 1.6177492141723633,
      "learning_rate": 5.295641447368421e-06,
      "loss": 0.0534,
      "step": 11450
    },
    {
      "epoch": 4.712171052631579,
      "grad_norm": 0.029669273644685745,
      "learning_rate": 5.2915296052631585e-06,
      "loss": 0.0223,
      "step": 11460
    },
    {
      "epoch": 4.7162828947368425,
      "grad_norm": 0.011819909326732159,
      "learning_rate": 5.287417763157895e-06,
      "loss": 0.1184,
      "step": 11470
    },
    {
      "epoch": 4.720394736842105,
      "grad_norm": 89.91704559326172,
      "learning_rate": 5.283305921052632e-06,
      "loss": 0.4615,
      "step": 11480
    },
    {
      "epoch": 4.724506578947368,
      "grad_norm": 0.08650683611631393,
      "learning_rate": 5.279194078947368e-06,
      "loss": 0.0019,
      "step": 11490
    },
    {
      "epoch": 4.728618421052632,
      "grad_norm": 0.15594571828842163,
      "learning_rate": 5.275082236842106e-06,
      "loss": 0.0339,
      "step": 11500
    },
    {
      "epoch": 4.732730263157895,
      "grad_norm": 1.9714744091033936,
      "learning_rate": 5.270970394736843e-06,
      "loss": 0.2982,
      "step": 11510
    },
    {
      "epoch": 4.7368421052631575,
      "grad_norm": 0.053652070462703705,
      "learning_rate": 5.26685855263158e-06,
      "loss": 0.1086,
      "step": 11520
    },
    {
      "epoch": 4.740953947368421,
      "grad_norm": 2.379514217376709,
      "learning_rate": 5.2627467105263165e-06,
      "loss": 0.0506,
      "step": 11530
    },
    {
      "epoch": 4.745065789473684,
      "grad_norm": 0.000377872318495065,
      "learning_rate": 5.258634868421053e-06,
      "loss": 0.0937,
      "step": 11540
    },
    {
      "epoch": 4.749177631578947,
      "grad_norm": 1.3539845943450928,
      "learning_rate": 5.25452302631579e-06,
      "loss": 0.0948,
      "step": 11550
    },
    {
      "epoch": 4.753289473684211,
      "grad_norm": 1.4154525995254517,
      "learning_rate": 5.250411184210526e-06,
      "loss": 0.1704,
      "step": 11560
    },
    {
      "epoch": 4.7574013157894735,
      "grad_norm": 0.5577759146690369,
      "learning_rate": 5.246299342105263e-06,
      "loss": 0.0408,
      "step": 11570
    },
    {
      "epoch": 4.761513157894737,
      "grad_norm": 74.89480590820312,
      "learning_rate": 5.2421875e-06,
      "loss": 0.3608,
      "step": 11580
    },
    {
      "epoch": 4.765625,
      "grad_norm": 5.859063625335693,
      "learning_rate": 5.238075657894738e-06,
      "loss": 0.2313,
      "step": 11590
    },
    {
      "epoch": 4.769736842105263,
      "grad_norm": 0.0718579813838005,
      "learning_rate": 5.2339638157894744e-06,
      "loss": 0.0054,
      "step": 11600
    },
    {
      "epoch": 4.7738486842105265,
      "grad_norm": 0.0002502045244909823,
      "learning_rate": 5.229851973684211e-06,
      "loss": 0.0367,
      "step": 11610
    },
    {
      "epoch": 4.777960526315789,
      "grad_norm": 0.042701832950115204,
      "learning_rate": 5.225740131578948e-06,
      "loss": 0.2447,
      "step": 11620
    },
    {
      "epoch": 4.782072368421053,
      "grad_norm": 0.010373505763709545,
      "learning_rate": 5.221628289473685e-06,
      "loss": 0.2749,
      "step": 11630
    },
    {
      "epoch": 4.786184210526316,
      "grad_norm": 0.005021689459681511,
      "learning_rate": 5.217516447368422e-06,
      "loss": 0.0291,
      "step": 11640
    },
    {
      "epoch": 4.790296052631579,
      "grad_norm": 0.009168042801320553,
      "learning_rate": 5.213404605263158e-06,
      "loss": 0.012,
      "step": 11650
    },
    {
      "epoch": 4.7944078947368425,
      "grad_norm": 0.08961399644613266,
      "learning_rate": 5.209292763157895e-06,
      "loss": 0.0174,
      "step": 11660
    },
    {
      "epoch": 4.798519736842105,
      "grad_norm": 6.835995674133301,
      "learning_rate": 5.2051809210526315e-06,
      "loss": 0.306,
      "step": 11670
    },
    {
      "epoch": 4.802631578947368,
      "grad_norm": 110.7245101928711,
      "learning_rate": 5.201069078947368e-06,
      "loss": 0.7956,
      "step": 11680
    },
    {
      "epoch": 4.806743421052632,
      "grad_norm": 85.60328674316406,
      "learning_rate": 5.1969572368421064e-06,
      "loss": 0.1353,
      "step": 11690
    },
    {
      "epoch": 4.810855263157895,
      "grad_norm": 8.091873168945312,
      "learning_rate": 5.192845394736843e-06,
      "loss": 0.0128,
      "step": 11700
    },
    {
      "epoch": 4.8149671052631575,
      "grad_norm": 83.81761932373047,
      "learning_rate": 5.18873355263158e-06,
      "loss": 0.3359,
      "step": 11710
    },
    {
      "epoch": 4.819078947368421,
      "grad_norm": 6.044814109802246,
      "learning_rate": 5.184621710526316e-06,
      "loss": 0.1492,
      "step": 11720
    },
    {
      "epoch": 4.823190789473684,
      "grad_norm": 0.00032899685902521014,
      "learning_rate": 5.180509868421053e-06,
      "loss": 0.1777,
      "step": 11730
    },
    {
      "epoch": 4.827302631578947,
      "grad_norm": 0.09665809571743011,
      "learning_rate": 5.1763980263157895e-06,
      "loss": 0.0136,
      "step": 11740
    },
    {
      "epoch": 4.831414473684211,
      "grad_norm": 0.19958020746707916,
      "learning_rate": 5.172286184210527e-06,
      "loss": 0.234,
      "step": 11750
    },
    {
      "epoch": 4.8355263157894735,
      "grad_norm": 0.044992148876190186,
      "learning_rate": 5.1681743421052636e-06,
      "loss": 0.0143,
      "step": 11760
    },
    {
      "epoch": 4.839638157894737,
      "grad_norm": 0.6200563311576843,
      "learning_rate": 5.1640625e-06,
      "loss": 0.1882,
      "step": 11770
    },
    {
      "epoch": 4.84375,
      "grad_norm": 0.04625256359577179,
      "learning_rate": 5.159950657894738e-06,
      "loss": 0.0057,
      "step": 11780
    },
    {
      "epoch": 4.847861842105263,
      "grad_norm": 0.7069116234779358,
      "learning_rate": 5.155838815789474e-06,
      "loss": 0.3012,
      "step": 11790
    },
    {
      "epoch": 4.8519736842105265,
      "grad_norm": 0.0004912884323857725,
      "learning_rate": 5.151726973684212e-06,
      "loss": 0.0003,
      "step": 11800
    },
    {
      "epoch": 4.856085526315789,
      "grad_norm": 0.00741523178294301,
      "learning_rate": 5.147615131578948e-06,
      "loss": 0.1208,
      "step": 11810
    },
    {
      "epoch": 4.860197368421053,
      "grad_norm": 34.17157745361328,
      "learning_rate": 5.143503289473685e-06,
      "loss": 0.1658,
      "step": 11820
    },
    {
      "epoch": 4.864309210526316,
      "grad_norm": 43.198394775390625,
      "learning_rate": 5.1393914473684215e-06,
      "loss": 0.1518,
      "step": 11830
    },
    {
      "epoch": 4.868421052631579,
      "grad_norm": 0.00559881329536438,
      "learning_rate": 5.135279605263158e-06,
      "loss": 0.1322,
      "step": 11840
    },
    {
      "epoch": 4.8725328947368425,
      "grad_norm": 0.061246518045663834,
      "learning_rate": 5.131167763157895e-06,
      "loss": 0.1169,
      "step": 11850
    },
    {
      "epoch": 4.876644736842105,
      "grad_norm": 2.1810858249664307,
      "learning_rate": 5.127055921052631e-06,
      "loss": 0.0755,
      "step": 11860
    },
    {
      "epoch": 4.880756578947368,
      "grad_norm": 0.13748008012771606,
      "learning_rate": 5.122944078947368e-06,
      "loss": 0.0136,
      "step": 11870
    },
    {
      "epoch": 4.884868421052632,
      "grad_norm": 13.2427339553833,
      "learning_rate": 5.118832236842106e-06,
      "loss": 0.0096,
      "step": 11880
    },
    {
      "epoch": 4.888980263157895,
      "grad_norm": 0.20603924989700317,
      "learning_rate": 5.114720394736843e-06,
      "loss": 0.0793,
      "step": 11890
    },
    {
      "epoch": 4.8930921052631575,
      "grad_norm": 56.690826416015625,
      "learning_rate": 5.1106085526315795e-06,
      "loss": 0.3857,
      "step": 11900
    },
    {
      "epoch": 4.897203947368421,
      "grad_norm": 5.994784832000732,
      "learning_rate": 5.106496710526316e-06,
      "loss": 0.0047,
      "step": 11910
    },
    {
      "epoch": 4.901315789473684,
      "grad_norm": 0.028743823990225792,
      "learning_rate": 5.102384868421053e-06,
      "loss": 0.1121,
      "step": 11920
    },
    {
      "epoch": 4.905427631578947,
      "grad_norm": 0.002641650615260005,
      "learning_rate": 5.09827302631579e-06,
      "loss": 0.0928,
      "step": 11930
    },
    {
      "epoch": 4.909539473684211,
      "grad_norm": 0.11559925973415375,
      "learning_rate": 5.094161184210527e-06,
      "loss": 0.0364,
      "step": 11940
    },
    {
      "epoch": 4.9136513157894735,
      "grad_norm": 0.013286828063428402,
      "learning_rate": 5.090049342105263e-06,
      "loss": 0.2054,
      "step": 11950
    },
    {
      "epoch": 4.917763157894737,
      "grad_norm": 0.14687436819076538,
      "learning_rate": 5.0859375e-06,
      "loss": 0.2326,
      "step": 11960
    },
    {
      "epoch": 4.921875,
      "grad_norm": 0.004031555261462927,
      "learning_rate": 5.081825657894737e-06,
      "loss": 0.0089,
      "step": 11970
    },
    {
      "epoch": 4.925986842105263,
      "grad_norm": 0.21741409599781036,
      "learning_rate": 5.077713815789475e-06,
      "loss": 0.2988,
      "step": 11980
    },
    {
      "epoch": 4.9300986842105265,
      "grad_norm": 0.027137205004692078,
      "learning_rate": 5.0736019736842115e-06,
      "loss": 0.208,
      "step": 11990
    },
    {
      "epoch": 4.934210526315789,
      "grad_norm": 40.1978645324707,
      "learning_rate": 5.069490131578948e-06,
      "loss": 0.0322,
      "step": 12000
    },
    {
      "epoch": 4.938322368421053,
      "grad_norm": 84.67223358154297,
      "learning_rate": 5.065378289473685e-06,
      "loss": 0.2434,
      "step": 12010
    },
    {
      "epoch": 4.942434210526316,
      "grad_norm": 0.4514327347278595,
      "learning_rate": 5.061266447368421e-06,
      "loss": 0.23,
      "step": 12020
    },
    {
      "epoch": 4.946546052631579,
      "grad_norm": 19.516311645507812,
      "learning_rate": 5.057154605263158e-06,
      "loss": 0.1211,
      "step": 12030
    },
    {
      "epoch": 4.9506578947368425,
      "grad_norm": 69.44686126708984,
      "learning_rate": 5.0530427631578945e-06,
      "loss": 0.0526,
      "step": 12040
    },
    {
      "epoch": 4.954769736842105,
      "grad_norm": 0.7357942461967468,
      "learning_rate": 5.048930921052632e-06,
      "loss": 0.096,
      "step": 12050
    },
    {
      "epoch": 4.958881578947368,
      "grad_norm": 55.00428009033203,
      "learning_rate": 5.044819078947369e-06,
      "loss": 0.1763,
      "step": 12060
    },
    {
      "epoch": 4.962993421052632,
      "grad_norm": 0.006061846856027842,
      "learning_rate": 5.040707236842106e-06,
      "loss": 0.2433,
      "step": 12070
    },
    {
      "epoch": 4.967105263157895,
      "grad_norm": 0.01570850983262062,
      "learning_rate": 5.036595394736843e-06,
      "loss": 0.005,
      "step": 12080
    },
    {
      "epoch": 4.9712171052631575,
      "grad_norm": 0.002090919530019164,
      "learning_rate": 5.032483552631579e-06,
      "loss": 0.0092,
      "step": 12090
    },
    {
      "epoch": 4.975328947368421,
      "grad_norm": 0.02324344776570797,
      "learning_rate": 5.028371710526317e-06,
      "loss": 0.2655,
      "step": 12100
    },
    {
      "epoch": 4.979440789473684,
      "grad_norm": 0.023654013872146606,
      "learning_rate": 5.024259868421053e-06,
      "loss": 0.1504,
      "step": 12110
    },
    {
      "epoch": 4.983552631578947,
      "grad_norm": 70.16049194335938,
      "learning_rate": 5.02014802631579e-06,
      "loss": 0.3557,
      "step": 12120
    },
    {
      "epoch": 4.987664473684211,
      "grad_norm": 0.004474622197449207,
      "learning_rate": 5.0160361842105265e-06,
      "loss": 0.266,
      "step": 12130
    },
    {
      "epoch": 4.9917763157894735,
      "grad_norm": 0.035371605306863785,
      "learning_rate": 5.011924342105263e-06,
      "loss": 0.0795,
      "step": 12140
    },
    {
      "epoch": 4.995888157894737,
      "grad_norm": 0.007857566699385643,
      "learning_rate": 5.0078125e-06,
      "loss": 0.0035,
      "step": 12150
    },
    {
      "epoch": 5.0,
      "grad_norm": 118.69224548339844,
      "learning_rate": 5.003700657894738e-06,
      "loss": 0.1007,
      "step": 12160
    },
    {
      "epoch": 5.004111842105263,
      "grad_norm": 3.55021071434021,
      "learning_rate": 4.999588815789474e-06,
      "loss": 0.09,
      "step": 12170
    },
    {
      "epoch": 5.0082236842105265,
      "grad_norm": 2.0292088985443115,
      "learning_rate": 4.9954769736842104e-06,
      "loss": 0.051,
      "step": 12180
    },
    {
      "epoch": 5.012335526315789,
      "grad_norm": 0.41962769627571106,
      "learning_rate": 4.991365131578948e-06,
      "loss": 0.1036,
      "step": 12190
    },
    {
      "epoch": 5.016447368421052,
      "grad_norm": 0.009977146051824093,
      "learning_rate": 4.9872532894736845e-06,
      "loss": 0.0087,
      "step": 12200
    },
    {
      "epoch": 5.020559210526316,
      "grad_norm": 0.0003735716745723039,
      "learning_rate": 4.983141447368421e-06,
      "loss": 0.0022,
      "step": 12210
    },
    {
      "epoch": 5.024671052631579,
      "grad_norm": 16.074596405029297,
      "learning_rate": 4.9790296052631586e-06,
      "loss": 0.0139,
      "step": 12220
    },
    {
      "epoch": 5.0287828947368425,
      "grad_norm": 3.9828500747680664,
      "learning_rate": 4.974917763157895e-06,
      "loss": 0.3339,
      "step": 12230
    },
    {
      "epoch": 5.032894736842105,
      "grad_norm": 54.579383850097656,
      "learning_rate": 4.970805921052632e-06,
      "loss": 0.0736,
      "step": 12240
    },
    {
      "epoch": 5.037006578947368,
      "grad_norm": 0.0028906783554702997,
      "learning_rate": 4.966694078947369e-06,
      "loss": 0.0565,
      "step": 12250
    },
    {
      "epoch": 5.041118421052632,
      "grad_norm": 0.09334980696439743,
      "learning_rate": 4.962582236842106e-06,
      "loss": 0.0103,
      "step": 12260
    },
    {
      "epoch": 5.045230263157895,
      "grad_norm": 0.44651469588279724,
      "learning_rate": 4.9584703947368424e-06,
      "loss": 0.0421,
      "step": 12270
    },
    {
      "epoch": 5.0493421052631575,
      "grad_norm": 0.3897380530834198,
      "learning_rate": 4.95435855263158e-06,
      "loss": 0.0438,
      "step": 12280
    },
    {
      "epoch": 5.053453947368421,
      "grad_norm": 0.2058589607477188,
      "learning_rate": 4.9502467105263165e-06,
      "loss": 0.0285,
      "step": 12290
    },
    {
      "epoch": 5.057565789473684,
      "grad_norm": 7.561193466186523,
      "learning_rate": 4.946134868421053e-06,
      "loss": 0.0042,
      "step": 12300
    },
    {
      "epoch": 5.061677631578948,
      "grad_norm": 47.261566162109375,
      "learning_rate": 4.94202302631579e-06,
      "loss": 0.2775,
      "step": 12310
    },
    {
      "epoch": 5.065789473684211,
      "grad_norm": 45.68752670288086,
      "learning_rate": 4.937911184210526e-06,
      "loss": 0.1912,
      "step": 12320
    },
    {
      "epoch": 5.0699013157894735,
      "grad_norm": 0.008435727097094059,
      "learning_rate": 4.933799342105264e-06,
      "loss": 0.0739,
      "step": 12330
    },
    {
      "epoch": 5.074013157894737,
      "grad_norm": 0.40947651863098145,
      "learning_rate": 4.9296875e-06,
      "loss": 0.1447,
      "step": 12340
    },
    {
      "epoch": 5.078125,
      "grad_norm": 0.009739860892295837,
      "learning_rate": 4.925575657894737e-06,
      "loss": 0.0158,
      "step": 12350
    },
    {
      "epoch": 5.082236842105263,
      "grad_norm": 0.007655106019228697,
      "learning_rate": 4.921463815789474e-06,
      "loss": 0.0416,
      "step": 12360
    },
    {
      "epoch": 5.0863486842105265,
      "grad_norm": 3.085749387741089,
      "learning_rate": 4.917351973684211e-06,
      "loss": 0.0112,
      "step": 12370
    },
    {
      "epoch": 5.090460526315789,
      "grad_norm": 0.8423932194709778,
      "learning_rate": 4.913240131578948e-06,
      "loss": 0.0338,
      "step": 12380
    },
    {
      "epoch": 5.094572368421052,
      "grad_norm": 0.0011507633607834578,
      "learning_rate": 4.909128289473684e-06,
      "loss": 0.0024,
      "step": 12390
    },
    {
      "epoch": 5.098684210526316,
      "grad_norm": 0.06639353185892105,
      "learning_rate": 4.905016447368422e-06,
      "loss": 0.1583,
      "step": 12400
    },
    {
      "epoch": 5.102796052631579,
      "grad_norm": 0.28097525238990784,
      "learning_rate": 4.900904605263158e-06,
      "loss": 0.0059,
      "step": 12410
    },
    {
      "epoch": 5.1069078947368425,
      "grad_norm": 0.0003991532139480114,
      "learning_rate": 4.896792763157895e-06,
      "loss": 0.0997,
      "step": 12420
    },
    {
      "epoch": 5.111019736842105,
      "grad_norm": 0.10429614037275314,
      "learning_rate": 4.892680921052632e-06,
      "loss": 0.1612,
      "step": 12430
    },
    {
      "epoch": 5.115131578947368,
      "grad_norm": 1.887984037399292,
      "learning_rate": 4.888569078947369e-06,
      "loss": 0.0778,
      "step": 12440
    },
    {
      "epoch": 5.119243421052632,
      "grad_norm": 1.0391831398010254,
      "learning_rate": 4.884457236842106e-06,
      "loss": 0.0022,
      "step": 12450
    },
    {
      "epoch": 5.123355263157895,
      "grad_norm": 0.005739231593906879,
      "learning_rate": 4.880345394736842e-06,
      "loss": 0.2004,
      "step": 12460
    },
    {
      "epoch": 5.1274671052631575,
      "grad_norm": 0.3580719828605652,
      "learning_rate": 4.87623355263158e-06,
      "loss": 0.0694,
      "step": 12470
    },
    {
      "epoch": 5.131578947368421,
      "grad_norm": 0.00015461303701158613,
      "learning_rate": 4.872121710526316e-06,
      "loss": 0.1299,
      "step": 12480
    },
    {
      "epoch": 5.135690789473684,
      "grad_norm": 0.012092826887965202,
      "learning_rate": 4.868009868421053e-06,
      "loss": 0.1834,
      "step": 12490
    },
    {
      "epoch": 5.139802631578948,
      "grad_norm": 0.019745737314224243,
      "learning_rate": 4.8638980263157895e-06,
      "loss": 0.0675,
      "step": 12500
    },
    {
      "epoch": 5.143914473684211,
      "grad_norm": 9.999950408935547,
      "learning_rate": 4.859786184210526e-06,
      "loss": 0.0185,
      "step": 12510
    },
    {
      "epoch": 5.1480263157894735,
      "grad_norm": 4.208477973937988,
      "learning_rate": 4.855674342105264e-06,
      "loss": 0.116,
      "step": 12520
    },
    {
      "epoch": 5.152138157894737,
      "grad_norm": 0.0007267239852808416,
      "learning_rate": 4.8515625e-06,
      "loss": 0.0604,
      "step": 12530
    },
    {
      "epoch": 5.15625,
      "grad_norm": 0.0007356388377957046,
      "learning_rate": 4.847450657894738e-06,
      "loss": 0.0038,
      "step": 12540
    },
    {
      "epoch": 5.160361842105263,
      "grad_norm": 76.24391174316406,
      "learning_rate": 4.843338815789474e-06,
      "loss": 0.264,
      "step": 12550
    },
    {
      "epoch": 5.1644736842105265,
      "grad_norm": 0.02224615029990673,
      "learning_rate": 4.839226973684211e-06,
      "loss": 0.0107,
      "step": 12560
    },
    {
      "epoch": 5.168585526315789,
      "grad_norm": 0.872387707233429,
      "learning_rate": 4.835115131578948e-06,
      "loss": 0.0017,
      "step": 12570
    },
    {
      "epoch": 5.172697368421052,
      "grad_norm": 0.055020976811647415,
      "learning_rate": 4.831003289473685e-06,
      "loss": 0.1254,
      "step": 12580
    },
    {
      "epoch": 5.176809210526316,
      "grad_norm": 0.028826193884015083,
      "learning_rate": 4.8268914473684215e-06,
      "loss": 0.0145,
      "step": 12590
    },
    {
      "epoch": 5.180921052631579,
      "grad_norm": 0.12968547642230988,
      "learning_rate": 4.822779605263158e-06,
      "loss": 0.0151,
      "step": 12600
    },
    {
      "epoch": 5.1850328947368425,
      "grad_norm": 0.10976220667362213,
      "learning_rate": 4.818667763157895e-06,
      "loss": 0.0207,
      "step": 12610
    },
    {
      "epoch": 5.189144736842105,
      "grad_norm": 0.003534722840413451,
      "learning_rate": 4.814555921052632e-06,
      "loss": 0.0085,
      "step": 12620
    },
    {
      "epoch": 5.193256578947368,
      "grad_norm": 0.008794968947768211,
      "learning_rate": 4.810444078947369e-06,
      "loss": 0.125,
      "step": 12630
    },
    {
      "epoch": 5.197368421052632,
      "grad_norm": 1.5557653903961182,
      "learning_rate": 4.806332236842105e-06,
      "loss": 0.0224,
      "step": 12640
    },
    {
      "epoch": 5.201480263157895,
      "grad_norm": 0.010107816196978092,
      "learning_rate": 4.802220394736842e-06,
      "loss": 0.1335,
      "step": 12650
    },
    {
      "epoch": 5.2055921052631575,
      "grad_norm": 4.243147373199463,
      "learning_rate": 4.7981085526315795e-06,
      "loss": 0.0065,
      "step": 12660
    },
    {
      "epoch": 5.209703947368421,
      "grad_norm": 0.33208709955215454,
      "learning_rate": 4.793996710526316e-06,
      "loss": 0.1848,
      "step": 12670
    },
    {
      "epoch": 5.213815789473684,
      "grad_norm": 4.119780540466309,
      "learning_rate": 4.789884868421053e-06,
      "loss": 0.0587,
      "step": 12680
    },
    {
      "epoch": 5.217927631578948,
      "grad_norm": 115.96353912353516,
      "learning_rate": 4.78577302631579e-06,
      "loss": 0.4392,
      "step": 12690
    },
    {
      "epoch": 5.222039473684211,
      "grad_norm": 0.0005382675444707274,
      "learning_rate": 4.781661184210527e-06,
      "loss": 0.0101,
      "step": 12700
    },
    {
      "epoch": 5.2261513157894735,
      "grad_norm": 0.5530701279640198,
      "learning_rate": 4.777549342105263e-06,
      "loss": 0.0904,
      "step": 12710
    },
    {
      "epoch": 5.230263157894737,
      "grad_norm": 86.44254302978516,
      "learning_rate": 4.773437500000001e-06,
      "loss": 0.135,
      "step": 12720
    },
    {
      "epoch": 5.234375,
      "grad_norm": 2.6025938987731934,
      "learning_rate": 4.7693256578947374e-06,
      "loss": 0.0513,
      "step": 12730
    },
    {
      "epoch": 5.238486842105263,
      "grad_norm": 0.04528487101197243,
      "learning_rate": 4.765213815789474e-06,
      "loss": 0.1211,
      "step": 12740
    },
    {
      "epoch": 5.2425986842105265,
      "grad_norm": 0.049920398741960526,
      "learning_rate": 4.761101973684211e-06,
      "loss": 0.2927,
      "step": 12750
    },
    {
      "epoch": 5.246710526315789,
      "grad_norm": 0.231965571641922,
      "learning_rate": 4.756990131578948e-06,
      "loss": 0.3086,
      "step": 12760
    },
    {
      "epoch": 5.250822368421053,
      "grad_norm": 0.4021833539009094,
      "learning_rate": 4.752878289473685e-06,
      "loss": 0.1136,
      "step": 12770
    },
    {
      "epoch": 5.254934210526316,
      "grad_norm": 79.92125701904297,
      "learning_rate": 4.748766447368421e-06,
      "loss": 0.08,
      "step": 12780
    },
    {
      "epoch": 5.259046052631579,
      "grad_norm": 0.0020701682660728693,
      "learning_rate": 4.744654605263158e-06,
      "loss": 0.0191,
      "step": 12790
    },
    {
      "epoch": 5.2631578947368425,
      "grad_norm": 0.015794718638062477,
      "learning_rate": 4.7405427631578945e-06,
      "loss": 0.0095,
      "step": 12800
    },
    {
      "epoch": 5.267269736842105,
      "grad_norm": 0.31466972827911377,
      "learning_rate": 4.736430921052632e-06,
      "loss": 0.0093,
      "step": 12810
    },
    {
      "epoch": 5.271381578947368,
      "grad_norm": 0.1875465214252472,
      "learning_rate": 4.732319078947369e-06,
      "loss": 0.1396,
      "step": 12820
    },
    {
      "epoch": 5.275493421052632,
      "grad_norm": 12.725687026977539,
      "learning_rate": 4.728207236842105e-06,
      "loss": 0.1678,
      "step": 12830
    },
    {
      "epoch": 5.279605263157895,
      "grad_norm": 0.07018950581550598,
      "learning_rate": 4.724095394736843e-06,
      "loss": 0.083,
      "step": 12840
    },
    {
      "epoch": 5.2837171052631575,
      "grad_norm": 7.318544864654541,
      "learning_rate": 4.719983552631579e-06,
      "loss": 0.0284,
      "step": 12850
    },
    {
      "epoch": 5.287828947368421,
      "grad_norm": 2.0364904403686523,
      "learning_rate": 4.715871710526317e-06,
      "loss": 0.1169,
      "step": 12860
    },
    {
      "epoch": 5.291940789473684,
      "grad_norm": 0.018038108944892883,
      "learning_rate": 4.711759868421053e-06,
      "loss": 0.0011,
      "step": 12870
    },
    {
      "epoch": 5.296052631578947,
      "grad_norm": 0.15174317359924316,
      "learning_rate": 4.70764802631579e-06,
      "loss": 0.0515,
      "step": 12880
    },
    {
      "epoch": 5.300164473684211,
      "grad_norm": 0.06006835028529167,
      "learning_rate": 4.7035361842105266e-06,
      "loss": 0.0028,
      "step": 12890
    },
    {
      "epoch": 5.3042763157894735,
      "grad_norm": 0.21908903121948242,
      "learning_rate": 4.699424342105264e-06,
      "loss": 0.0261,
      "step": 12900
    },
    {
      "epoch": 5.308388157894737,
      "grad_norm": 3.5568766593933105,
      "learning_rate": 4.695312500000001e-06,
      "loss": 0.3875,
      "step": 12910
    },
    {
      "epoch": 5.3125,
      "grad_norm": 0.43564674258232117,
      "learning_rate": 4.691200657894737e-06,
      "loss": 0.0769,
      "step": 12920
    },
    {
      "epoch": 5.316611842105263,
      "grad_norm": 0.14090363681316376,
      "learning_rate": 4.687088815789474e-06,
      "loss": 0.035,
      "step": 12930
    },
    {
      "epoch": 5.3207236842105265,
      "grad_norm": 0.14797407388687134,
      "learning_rate": 4.6829769736842105e-06,
      "loss": 0.015,
      "step": 12940
    },
    {
      "epoch": 5.324835526315789,
      "grad_norm": 9.737833976745605,
      "learning_rate": 4.678865131578948e-06,
      "loss": 0.1553,
      "step": 12950
    },
    {
      "epoch": 5.328947368421053,
      "grad_norm": 0.6963122487068176,
      "learning_rate": 4.6747532894736845e-06,
      "loss": 0.0025,
      "step": 12960
    },
    {
      "epoch": 5.333059210526316,
      "grad_norm": 7.62711763381958,
      "learning_rate": 4.670641447368421e-06,
      "loss": 0.0051,
      "step": 12970
    },
    {
      "epoch": 5.337171052631579,
      "grad_norm": 0.015172763727605343,
      "learning_rate": 4.666529605263158e-06,
      "loss": 0.08,
      "step": 12980
    },
    {
      "epoch": 5.3412828947368425,
      "grad_norm": 0.47709551453590393,
      "learning_rate": 4.662417763157895e-06,
      "loss": 0.0008,
      "step": 12990
    },
    {
      "epoch": 5.345394736842105,
      "grad_norm": 8.82782660482917e-06,
      "learning_rate": 4.658305921052632e-06,
      "loss": 0.0129,
      "step": 13000
    },
    {
      "epoch": 5.349506578947368,
      "grad_norm": 55.73831558227539,
      "learning_rate": 4.654194078947369e-06,
      "loss": 0.1756,
      "step": 13010
    },
    {
      "epoch": 5.353618421052632,
      "grad_norm": 0.013916180469095707,
      "learning_rate": 4.650082236842106e-06,
      "loss": 0.0494,
      "step": 13020
    },
    {
      "epoch": 5.357730263157895,
      "grad_norm": 0.01592426560819149,
      "learning_rate": 4.6459703947368425e-06,
      "loss": 0.0407,
      "step": 13030
    },
    {
      "epoch": 5.3618421052631575,
      "grad_norm": 65.08041381835938,
      "learning_rate": 4.64185855263158e-06,
      "loss": 0.0859,
      "step": 13040
    },
    {
      "epoch": 5.365953947368421,
      "grad_norm": 0.08960430324077606,
      "learning_rate": 4.6377467105263165e-06,
      "loss": 0.0019,
      "step": 13050
    },
    {
      "epoch": 5.370065789473684,
      "grad_norm": 14.894381523132324,
      "learning_rate": 4.633634868421053e-06,
      "loss": 0.0202,
      "step": 13060
    },
    {
      "epoch": 5.374177631578947,
      "grad_norm": 0.005502131301909685,
      "learning_rate": 4.62952302631579e-06,
      "loss": 0.0053,
      "step": 13070
    },
    {
      "epoch": 5.378289473684211,
      "grad_norm": 21.16437339782715,
      "learning_rate": 4.625411184210526e-06,
      "loss": 0.1385,
      "step": 13080
    },
    {
      "epoch": 5.3824013157894735,
      "grad_norm": 0.05989899858832359,
      "learning_rate": 4.621299342105264e-06,
      "loss": 0.0147,
      "step": 13090
    },
    {
      "epoch": 5.386513157894737,
      "grad_norm": 0.002133729634806514,
      "learning_rate": 4.6171875e-06,
      "loss": 0.0007,
      "step": 13100
    },
    {
      "epoch": 5.390625,
      "grad_norm": 47.23904037475586,
      "learning_rate": 4.613075657894737e-06,
      "loss": 0.1596,
      "step": 13110
    },
    {
      "epoch": 5.394736842105263,
      "grad_norm": 0.6729401350021362,
      "learning_rate": 4.608963815789474e-06,
      "loss": 0.0944,
      "step": 13120
    },
    {
      "epoch": 5.3988486842105265,
      "grad_norm": 0.04317113384604454,
      "learning_rate": 4.60485197368421e-06,
      "loss": 0.2098,
      "step": 13130
    },
    {
      "epoch": 5.402960526315789,
      "grad_norm": 0.1802137941122055,
      "learning_rate": 4.600740131578948e-06,
      "loss": 0.2664,
      "step": 13140
    },
    {
      "epoch": 5.407072368421053,
      "grad_norm": 1.5689358711242676,
      "learning_rate": 4.596628289473684e-06,
      "loss": 0.2259,
      "step": 13150
    },
    {
      "epoch": 5.411184210526316,
      "grad_norm": 0.18609057366847992,
      "learning_rate": 4.592516447368422e-06,
      "loss": 0.0157,
      "step": 13160
    },
    {
      "epoch": 5.415296052631579,
      "grad_norm": 116.86294555664062,
      "learning_rate": 4.588404605263158e-06,
      "loss": 0.1214,
      "step": 13170
    },
    {
      "epoch": 5.4194078947368425,
      "grad_norm": 1.2058242559432983,
      "learning_rate": 4.584292763157895e-06,
      "loss": 0.3339,
      "step": 13180
    },
    {
      "epoch": 5.423519736842105,
      "grad_norm": 2.2884137630462646,
      "learning_rate": 4.5801809210526324e-06,
      "loss": 0.0364,
      "step": 13190
    },
    {
      "epoch": 5.427631578947368,
      "grad_norm": 109.472900390625,
      "learning_rate": 4.576069078947369e-06,
      "loss": 0.2061,
      "step": 13200
    },
    {
      "epoch": 5.431743421052632,
      "grad_norm": 83.02965545654297,
      "learning_rate": 4.571957236842106e-06,
      "loss": 0.3225,
      "step": 13210
    },
    {
      "epoch": 5.435855263157895,
      "grad_norm": 4.199236869812012,
      "learning_rate": 4.567845394736842e-06,
      "loss": 0.0084,
      "step": 13220
    },
    {
      "epoch": 5.4399671052631575,
      "grad_norm": 70.33255004882812,
      "learning_rate": 4.56373355263158e-06,
      "loss": 0.0816,
      "step": 13230
    },
    {
      "epoch": 5.444078947368421,
      "grad_norm": 0.0003307022852823138,
      "learning_rate": 4.559621710526316e-06,
      "loss": 0.0943,
      "step": 13240
    },
    {
      "epoch": 5.448190789473684,
      "grad_norm": 0.011274745687842369,
      "learning_rate": 4.555509868421053e-06,
      "loss": 0.0331,
      "step": 13250
    },
    {
      "epoch": 5.452302631578947,
      "grad_norm": 0.04053517058491707,
      "learning_rate": 4.5513980263157895e-06,
      "loss": 0.0392,
      "step": 13260
    },
    {
      "epoch": 5.456414473684211,
      "grad_norm": 29.87954330444336,
      "learning_rate": 4.547286184210526e-06,
      "loss": 0.0388,
      "step": 13270
    },
    {
      "epoch": 5.4605263157894735,
      "grad_norm": 0.06873561441898346,
      "learning_rate": 4.543174342105264e-06,
      "loss": 0.0063,
      "step": 13280
    },
    {
      "epoch": 5.464638157894737,
      "grad_norm": 71.52073669433594,
      "learning_rate": 4.5390625e-06,
      "loss": 0.233,
      "step": 13290
    },
    {
      "epoch": 5.46875,
      "grad_norm": 1.1586495637893677,
      "learning_rate": 4.534950657894737e-06,
      "loss": 0.2069,
      "step": 13300
    },
    {
      "epoch": 5.472861842105263,
      "grad_norm": 4.539996147155762,
      "learning_rate": 4.530838815789474e-06,
      "loss": 0.0605,
      "step": 13310
    },
    {
      "epoch": 5.4769736842105265,
      "grad_norm": 0.469481498003006,
      "learning_rate": 4.526726973684211e-06,
      "loss": 0.0862,
      "step": 13320
    },
    {
      "epoch": 5.481085526315789,
      "grad_norm": 0.10244907438755035,
      "learning_rate": 4.522615131578948e-06,
      "loss": 0.191,
      "step": 13330
    },
    {
      "epoch": 5.485197368421053,
      "grad_norm": 0.026087487116456032,
      "learning_rate": 4.518503289473685e-06,
      "loss": 0.0111,
      "step": 13340
    },
    {
      "epoch": 5.489309210526316,
      "grad_norm": 0.050827547907829285,
      "learning_rate": 4.5143914473684216e-06,
      "loss": 0.0014,
      "step": 13350
    },
    {
      "epoch": 5.493421052631579,
      "grad_norm": 7.369386730715632e-05,
      "learning_rate": 4.510279605263158e-06,
      "loss": 0.1305,
      "step": 13360
    },
    {
      "epoch": 5.4975328947368425,
      "grad_norm": 0.5626145601272583,
      "learning_rate": 4.506167763157895e-06,
      "loss": 0.0828,
      "step": 13370
    },
    {
      "epoch": 5.501644736842105,
      "grad_norm": 83.1645278930664,
      "learning_rate": 4.502055921052632e-06,
      "loss": 0.1359,
      "step": 13380
    },
    {
      "epoch": 5.505756578947368,
      "grad_norm": 0.1569296270608902,
      "learning_rate": 4.497944078947369e-06,
      "loss": 0.0781,
      "step": 13390
    },
    {
      "epoch": 5.509868421052632,
      "grad_norm": 83.7970962524414,
      "learning_rate": 4.4938322368421055e-06,
      "loss": 0.1579,
      "step": 13400
    },
    {
      "epoch": 5.513980263157895,
      "grad_norm": 41.769187927246094,
      "learning_rate": 4.489720394736842e-06,
      "loss": 0.019,
      "step": 13410
    },
    {
      "epoch": 5.5180921052631575,
      "grad_norm": 0.12914538383483887,
      "learning_rate": 4.4856085526315795e-06,
      "loss": 0.0873,
      "step": 13420
    },
    {
      "epoch": 5.522203947368421,
      "grad_norm": 0.0012703635729849339,
      "learning_rate": 4.481496710526316e-06,
      "loss": 0.1208,
      "step": 13430
    },
    {
      "epoch": 5.526315789473684,
      "grad_norm": 14.571756362915039,
      "learning_rate": 4.477384868421053e-06,
      "loss": 0.0554,
      "step": 13440
    },
    {
      "epoch": 5.530427631578947,
      "grad_norm": 0.10961569845676422,
      "learning_rate": 4.473273026315789e-06,
      "loss": 0.2124,
      "step": 13450
    },
    {
      "epoch": 5.534539473684211,
      "grad_norm": 0.089634008705616,
      "learning_rate": 4.469161184210527e-06,
      "loss": 0.0901,
      "step": 13460
    },
    {
      "epoch": 5.5386513157894735,
      "grad_norm": 0.0024442391004413366,
      "learning_rate": 4.465049342105263e-06,
      "loss": 0.158,
      "step": 13470
    },
    {
      "epoch": 5.542763157894737,
      "grad_norm": 0.005259633529931307,
      "learning_rate": 4.460937500000001e-06,
      "loss": 0.1263,
      "step": 13480
    },
    {
      "epoch": 5.546875,
      "grad_norm": 0.020105471834540367,
      "learning_rate": 4.4568256578947375e-06,
      "loss": 0.0782,
      "step": 13490
    },
    {
      "epoch": 5.550986842105263,
      "grad_norm": 8.300959587097168,
      "learning_rate": 4.452713815789474e-06,
      "loss": 0.01,
      "step": 13500
    },
    {
      "epoch": 5.5550986842105265,
      "grad_norm": 0.009316924028098583,
      "learning_rate": 4.448601973684211e-06,
      "loss": 0.0182,
      "step": 13510
    },
    {
      "epoch": 5.559210526315789,
      "grad_norm": 0.0011465359712019563,
      "learning_rate": 4.444490131578948e-06,
      "loss": 0.0434,
      "step": 13520
    },
    {
      "epoch": 5.563322368421053,
      "grad_norm": 0.6433992981910706,
      "learning_rate": 4.440378289473685e-06,
      "loss": 0.0082,
      "step": 13530
    },
    {
      "epoch": 5.567434210526316,
      "grad_norm": 6.921369552612305,
      "learning_rate": 4.436266447368421e-06,
      "loss": 0.2029,
      "step": 13540
    },
    {
      "epoch": 5.571546052631579,
      "grad_norm": 0.5579603314399719,
      "learning_rate": 4.432154605263158e-06,
      "loss": 0.0043,
      "step": 13550
    },
    {
      "epoch": 5.5756578947368425,
      "grad_norm": 1.0528669357299805,
      "learning_rate": 4.4280427631578946e-06,
      "loss": 0.0507,
      "step": 13560
    },
    {
      "epoch": 5.579769736842105,
      "grad_norm": 0.24581840634346008,
      "learning_rate": 4.423930921052632e-06,
      "loss": 0.0037,
      "step": 13570
    },
    {
      "epoch": 5.583881578947368,
      "grad_norm": 0.12832553684711456,
      "learning_rate": 4.419819078947369e-06,
      "loss": 0.0006,
      "step": 13580
    },
    {
      "epoch": 5.587993421052632,
      "grad_norm": 0.013015109114348888,
      "learning_rate": 4.415707236842105e-06,
      "loss": 0.0041,
      "step": 13590
    },
    {
      "epoch": 5.592105263157895,
      "grad_norm": 0.06528721004724503,
      "learning_rate": 4.411595394736843e-06,
      "loss": 0.344,
      "step": 13600
    },
    {
      "epoch": 5.5962171052631575,
      "grad_norm": 7.076974391937256,
      "learning_rate": 4.407483552631579e-06,
      "loss": 0.0548,
      "step": 13610
    },
    {
      "epoch": 5.600328947368421,
      "grad_norm": 0.002825407776981592,
      "learning_rate": 4.403371710526316e-06,
      "loss": 0.2002,
      "step": 13620
    },
    {
      "epoch": 5.604440789473684,
      "grad_norm": 0.4160158336162567,
      "learning_rate": 4.399259868421053e-06,
      "loss": 0.0354,
      "step": 13630
    },
    {
      "epoch": 5.608552631578947,
      "grad_norm": 0.05299230292439461,
      "learning_rate": 4.39514802631579e-06,
      "loss": 0.0077,
      "step": 13640
    },
    {
      "epoch": 5.612664473684211,
      "grad_norm": 0.13743530213832855,
      "learning_rate": 4.391036184210527e-06,
      "loss": 0.0826,
      "step": 13650
    },
    {
      "epoch": 5.6167763157894735,
      "grad_norm": 2.4112481696647592e-05,
      "learning_rate": 4.386924342105264e-06,
      "loss": 0.0542,
      "step": 13660
    },
    {
      "epoch": 5.620888157894737,
      "grad_norm": 0.04345851391553879,
      "learning_rate": 4.382812500000001e-06,
      "loss": 0.0395,
      "step": 13670
    },
    {
      "epoch": 5.625,
      "grad_norm": 0.7097082138061523,
      "learning_rate": 4.378700657894737e-06,
      "loss": 0.0448,
      "step": 13680
    },
    {
      "epoch": 5.629111842105263,
      "grad_norm": 13.222297668457031,
      "learning_rate": 4.374588815789474e-06,
      "loss": 0.1012,
      "step": 13690
    },
    {
      "epoch": 5.6332236842105265,
      "grad_norm": 0.02765653096139431,
      "learning_rate": 4.3704769736842105e-06,
      "loss": 0.0137,
      "step": 13700
    },
    {
      "epoch": 5.637335526315789,
      "grad_norm": 1.4771395921707153,
      "learning_rate": 4.366365131578948e-06,
      "loss": 0.0751,
      "step": 13710
    },
    {
      "epoch": 5.641447368421053,
      "grad_norm": 0.03940972313284874,
      "learning_rate": 4.3622532894736845e-06,
      "loss": 0.0257,
      "step": 13720
    },
    {
      "epoch": 5.645559210526316,
      "grad_norm": 0.011548848822712898,
      "learning_rate": 4.358141447368421e-06,
      "loss": 0.0014,
      "step": 13730
    },
    {
      "epoch": 5.649671052631579,
      "grad_norm": 24.877702713012695,
      "learning_rate": 4.354029605263158e-06,
      "loss": 0.3987,
      "step": 13740
    },
    {
      "epoch": 5.6537828947368425,
      "grad_norm": 0.5792186856269836,
      "learning_rate": 4.349917763157895e-06,
      "loss": 0.002,
      "step": 13750
    },
    {
      "epoch": 5.657894736842105,
      "grad_norm": 2.246400833129883,
      "learning_rate": 4.345805921052632e-06,
      "loss": 0.0552,
      "step": 13760
    },
    {
      "epoch": 5.662006578947368,
      "grad_norm": 91.90821075439453,
      "learning_rate": 4.3416940789473684e-06,
      "loss": 0.1837,
      "step": 13770
    },
    {
      "epoch": 5.666118421052632,
      "grad_norm": 0.14475248754024506,
      "learning_rate": 4.337582236842106e-06,
      "loss": 0.0574,
      "step": 13780
    },
    {
      "epoch": 5.670230263157895,
      "grad_norm": 1.058624267578125,
      "learning_rate": 4.3334703947368425e-06,
      "loss": 0.1577,
      "step": 13790
    },
    {
      "epoch": 5.6743421052631575,
      "grad_norm": 0.0024513236712664366,
      "learning_rate": 4.32935855263158e-06,
      "loss": 0.0034,
      "step": 13800
    },
    {
      "epoch": 5.678453947368421,
      "grad_norm": 30.24159049987793,
      "learning_rate": 4.3252467105263166e-06,
      "loss": 0.0203,
      "step": 13810
    },
    {
      "epoch": 5.682565789473684,
      "grad_norm": 0.013055126182734966,
      "learning_rate": 4.321134868421053e-06,
      "loss": 0.0296,
      "step": 13820
    },
    {
      "epoch": 5.686677631578947,
      "grad_norm": 0.004864171613007784,
      "learning_rate": 4.31702302631579e-06,
      "loss": 0.0186,
      "step": 13830
    },
    {
      "epoch": 5.690789473684211,
      "grad_norm": 0.012071618810296059,
      "learning_rate": 4.312911184210526e-06,
      "loss": 0.0178,
      "step": 13840
    },
    {
      "epoch": 5.6949013157894735,
      "grad_norm": 0.709204375743866,
      "learning_rate": 4.308799342105264e-06,
      "loss": 0.0024,
      "step": 13850
    },
    {
      "epoch": 5.699013157894737,
      "grad_norm": 0.6648589968681335,
      "learning_rate": 4.3046875000000004e-06,
      "loss": 0.0004,
      "step": 13860
    },
    {
      "epoch": 5.703125,
      "grad_norm": 0.01644590124487877,
      "learning_rate": 4.300575657894737e-06,
      "loss": 0.0129,
      "step": 13870
    },
    {
      "epoch": 5.707236842105263,
      "grad_norm": 72.33856201171875,
      "learning_rate": 4.296463815789474e-06,
      "loss": 0.3645,
      "step": 13880
    },
    {
      "epoch": 5.7113486842105265,
      "grad_norm": 12.403470993041992,
      "learning_rate": 4.29235197368421e-06,
      "loss": 0.008,
      "step": 13890
    },
    {
      "epoch": 5.715460526315789,
      "grad_norm": 0.0032954413909465075,
      "learning_rate": 4.288240131578948e-06,
      "loss": 0.0349,
      "step": 13900
    },
    {
      "epoch": 5.719572368421053,
      "grad_norm": 63.751197814941406,
      "learning_rate": 4.284128289473684e-06,
      "loss": 0.0786,
      "step": 13910
    },
    {
      "epoch": 5.723684210526316,
      "grad_norm": 0.05703187361359596,
      "learning_rate": 4.280016447368421e-06,
      "loss": 0.0003,
      "step": 13920
    },
    {
      "epoch": 5.727796052631579,
      "grad_norm": 2.2651917934417725,
      "learning_rate": 4.275904605263158e-06,
      "loss": 0.0292,
      "step": 13930
    },
    {
      "epoch": 5.7319078947368425,
      "grad_norm": 0.029149610549211502,
      "learning_rate": 4.271792763157895e-06,
      "loss": 0.2247,
      "step": 13940
    },
    {
      "epoch": 5.736019736842105,
      "grad_norm": 0.013435146771371365,
      "learning_rate": 4.2676809210526325e-06,
      "loss": 0.1923,
      "step": 13950
    },
    {
      "epoch": 5.740131578947368,
      "grad_norm": 6.325374852167442e-05,
      "learning_rate": 4.263569078947369e-06,
      "loss": 0.0346,
      "step": 13960
    },
    {
      "epoch": 5.744243421052632,
      "grad_norm": 0.08834278583526611,
      "learning_rate": 4.259457236842106e-06,
      "loss": 0.0478,
      "step": 13970
    },
    {
      "epoch": 5.748355263157895,
      "grad_norm": 0.433755099773407,
      "learning_rate": 4.255345394736842e-06,
      "loss": 0.1382,
      "step": 13980
    },
    {
      "epoch": 5.7524671052631575,
      "grad_norm": 9.808202743530273,
      "learning_rate": 4.25123355263158e-06,
      "loss": 0.1213,
      "step": 13990
    },
    {
      "epoch": 5.756578947368421,
      "grad_norm": 0.3236713707447052,
      "learning_rate": 4.247121710526316e-06,
      "loss": 0.0581,
      "step": 14000
    },
    {
      "epoch": 5.760690789473684,
      "grad_norm": 0.0836513340473175,
      "learning_rate": 4.243009868421053e-06,
      "loss": 0.0769,
      "step": 14010
    },
    {
      "epoch": 5.764802631578947,
      "grad_norm": 0.01956997998058796,
      "learning_rate": 4.2388980263157896e-06,
      "loss": 0.0916,
      "step": 14020
    },
    {
      "epoch": 5.768914473684211,
      "grad_norm": 0.290725439786911,
      "learning_rate": 4.234786184210526e-06,
      "loss": 0.0507,
      "step": 14030
    },
    {
      "epoch": 5.7730263157894735,
      "grad_norm": 123.39529418945312,
      "learning_rate": 4.230674342105264e-06,
      "loss": 0.1676,
      "step": 14040
    },
    {
      "epoch": 5.777138157894737,
      "grad_norm": 1.4887871742248535,
      "learning_rate": 4.2265625e-06,
      "loss": 0.0391,
      "step": 14050
    },
    {
      "epoch": 5.78125,
      "grad_norm": 0.020826542750000954,
      "learning_rate": 4.222450657894737e-06,
      "loss": 0.0302,
      "step": 14060
    },
    {
      "epoch": 5.785361842105263,
      "grad_norm": 0.0018061811570078135,
      "learning_rate": 4.218338815789474e-06,
      "loss": 0.0297,
      "step": 14070
    },
    {
      "epoch": 5.7894736842105265,
      "grad_norm": 83.14344024658203,
      "learning_rate": 4.214226973684211e-06,
      "loss": 0.1815,
      "step": 14080
    },
    {
      "epoch": 5.793585526315789,
      "grad_norm": 0.048979900777339935,
      "learning_rate": 4.2101151315789475e-06,
      "loss": 0.015,
      "step": 14090
    },
    {
      "epoch": 5.797697368421053,
      "grad_norm": 0.09402741491794586,
      "learning_rate": 4.206003289473685e-06,
      "loss": 0.1655,
      "step": 14100
    },
    {
      "epoch": 5.801809210526316,
      "grad_norm": 0.022667359560728073,
      "learning_rate": 4.201891447368422e-06,
      "loss": 0.0383,
      "step": 14110
    },
    {
      "epoch": 5.805921052631579,
      "grad_norm": 0.004963419865816832,
      "learning_rate": 4.197779605263158e-06,
      "loss": 0.0758,
      "step": 14120
    },
    {
      "epoch": 5.8100328947368425,
      "grad_norm": 22.636215209960938,
      "learning_rate": 4.193667763157895e-06,
      "loss": 0.1386,
      "step": 14130
    },
    {
      "epoch": 5.814144736842105,
      "grad_norm": 0.4905577003955841,
      "learning_rate": 4.189555921052632e-06,
      "loss": 0.0355,
      "step": 14140
    },
    {
      "epoch": 5.818256578947368,
      "grad_norm": 0.06438548117876053,
      "learning_rate": 4.185444078947369e-06,
      "loss": 0.0085,
      "step": 14150
    },
    {
      "epoch": 5.822368421052632,
      "grad_norm": 27.258342742919922,
      "learning_rate": 4.1813322368421055e-06,
      "loss": 0.0948,
      "step": 14160
    },
    {
      "epoch": 5.826480263157895,
      "grad_norm": 0.06388669461011887,
      "learning_rate": 4.177220394736842e-06,
      "loss": 0.0785,
      "step": 14170
    },
    {
      "epoch": 5.8305921052631575,
      "grad_norm": 0.03271382674574852,
      "learning_rate": 4.1731085526315795e-06,
      "loss": 0.0534,
      "step": 14180
    },
    {
      "epoch": 5.834703947368421,
      "grad_norm": 0.18359790742397308,
      "learning_rate": 4.168996710526316e-06,
      "loss": 0.0073,
      "step": 14190
    },
    {
      "epoch": 5.838815789473684,
      "grad_norm": 0.005616027396172285,
      "learning_rate": 4.164884868421053e-06,
      "loss": 0.2368,
      "step": 14200
    },
    {
      "epoch": 5.842927631578947,
      "grad_norm": 0.2086600810289383,
      "learning_rate": 4.160773026315789e-06,
      "loss": 0.0106,
      "step": 14210
    },
    {
      "epoch": 5.847039473684211,
      "grad_norm": 18.045181274414062,
      "learning_rate": 4.156661184210527e-06,
      "loss": 0.2462,
      "step": 14220
    },
    {
      "epoch": 5.8511513157894735,
      "grad_norm": 4.4964518547058105,
      "learning_rate": 4.1525493421052634e-06,
      "loss": 0.3903,
      "step": 14230
    },
    {
      "epoch": 5.855263157894737,
      "grad_norm": 0.012703659944236279,
      "learning_rate": 4.1484375e-06,
      "loss": 0.0317,
      "step": 14240
    },
    {
      "epoch": 5.859375,
      "grad_norm": 2.445992946624756,
      "learning_rate": 4.1443256578947375e-06,
      "loss": 0.2889,
      "step": 14250
    },
    {
      "epoch": 5.863486842105263,
      "grad_norm": 1.1592912673950195,
      "learning_rate": 4.140213815789474e-06,
      "loss": 0.1458,
      "step": 14260
    },
    {
      "epoch": 5.8675986842105265,
      "grad_norm": 0.46567994356155396,
      "learning_rate": 4.136101973684211e-06,
      "loss": 0.0028,
      "step": 14270
    },
    {
      "epoch": 5.871710526315789,
      "grad_norm": 96.94190216064453,
      "learning_rate": 4.131990131578948e-06,
      "loss": 0.4599,
      "step": 14280
    },
    {
      "epoch": 5.875822368421053,
      "grad_norm": 0.22972775995731354,
      "learning_rate": 4.127878289473685e-06,
      "loss": 0.0446,
      "step": 14290
    },
    {
      "epoch": 5.879934210526316,
      "grad_norm": 0.0005772469448857009,
      "learning_rate": 4.123766447368421e-06,
      "loss": 0.0015,
      "step": 14300
    },
    {
      "epoch": 5.884046052631579,
      "grad_norm": 0.45011672377586365,
      "learning_rate": 4.119654605263158e-06,
      "loss": 0.0278,
      "step": 14310
    },
    {
      "epoch": 5.8881578947368425,
      "grad_norm": 0.02912212163209915,
      "learning_rate": 4.115542763157895e-06,
      "loss": 0.0088,
      "step": 14320
    },
    {
      "epoch": 5.892269736842105,
      "grad_norm": 1.1713041067123413,
      "learning_rate": 4.111430921052632e-06,
      "loss": 0.0243,
      "step": 14330
    },
    {
      "epoch": 5.896381578947368,
      "grad_norm": 59.63473892211914,
      "learning_rate": 4.107319078947369e-06,
      "loss": 0.0342,
      "step": 14340
    },
    {
      "epoch": 5.900493421052632,
      "grad_norm": 62.537635803222656,
      "learning_rate": 4.103207236842105e-06,
      "loss": 0.06,
      "step": 14350
    },
    {
      "epoch": 5.904605263157895,
      "grad_norm": 6.495685374829918e-05,
      "learning_rate": 4.099095394736842e-06,
      "loss": 0.0735,
      "step": 14360
    },
    {
      "epoch": 5.9087171052631575,
      "grad_norm": 3.6121654510498047,
      "learning_rate": 4.094983552631579e-06,
      "loss": 0.0735,
      "step": 14370
    },
    {
      "epoch": 5.912828947368421,
      "grad_norm": 0.03086622804403305,
      "learning_rate": 4.090871710526316e-06,
      "loss": 0.0034,
      "step": 14380
    },
    {
      "epoch": 5.916940789473684,
      "grad_norm": 0.07952877134084702,
      "learning_rate": 4.086759868421053e-06,
      "loss": 0.0433,
      "step": 14390
    },
    {
      "epoch": 5.921052631578947,
      "grad_norm": 0.419914573431015,
      "learning_rate": 4.08264802631579e-06,
      "loss": 0.1109,
      "step": 14400
    },
    {
      "epoch": 5.925164473684211,
      "grad_norm": 0.00708195473998785,
      "learning_rate": 4.078536184210527e-06,
      "loss": 0.0013,
      "step": 14410
    },
    {
      "epoch": 5.9292763157894735,
      "grad_norm": 0.025563985109329224,
      "learning_rate": 4.07483552631579e-06,
      "loss": 0.0536,
      "step": 14420
    },
    {
      "epoch": 5.933388157894737,
      "grad_norm": 0.0034198074135929346,
      "learning_rate": 4.070723684210527e-06,
      "loss": 0.0419,
      "step": 14430
    },
    {
      "epoch": 5.9375,
      "grad_norm": 0.005447972100228071,
      "learning_rate": 4.066611842105264e-06,
      "loss": 0.0912,
      "step": 14440
    },
    {
      "epoch": 5.941611842105263,
      "grad_norm": 0.10684799402952194,
      "learning_rate": 4.0625000000000005e-06,
      "loss": 0.026,
      "step": 14450
    },
    {
      "epoch": 5.9457236842105265,
      "grad_norm": 0.13118505477905273,
      "learning_rate": 4.058388157894737e-06,
      "loss": 0.0021,
      "step": 14460
    },
    {
      "epoch": 5.949835526315789,
      "grad_norm": 0.44725626707077026,
      "learning_rate": 4.054276315789474e-06,
      "loss": 0.0009,
      "step": 14470
    },
    {
      "epoch": 5.953947368421053,
      "grad_norm": 0.5507727861404419,
      "learning_rate": 4.050164473684211e-06,
      "loss": 0.0008,
      "step": 14480
    },
    {
      "epoch": 5.958059210526316,
      "grad_norm": 0.012128915637731552,
      "learning_rate": 4.046052631578948e-06,
      "loss": 0.0067,
      "step": 14490
    },
    {
      "epoch": 5.962171052631579,
      "grad_norm": 6.004103660583496,
      "learning_rate": 4.041940789473684e-06,
      "loss": 0.1394,
      "step": 14500
    },
    {
      "epoch": 5.9662828947368425,
      "grad_norm": 0.0007618535892106593,
      "learning_rate": 4.037828947368421e-06,
      "loss": 0.0646,
      "step": 14510
    },
    {
      "epoch": 5.970394736842105,
      "grad_norm": 0.003796122968196869,
      "learning_rate": 4.033717105263158e-06,
      "loss": 0.1546,
      "step": 14520
    },
    {
      "epoch": 5.974506578947368,
      "grad_norm": 92.0025863647461,
      "learning_rate": 4.029605263157895e-06,
      "loss": 0.2674,
      "step": 14530
    },
    {
      "epoch": 5.978618421052632,
      "grad_norm": 0.24422402679920197,
      "learning_rate": 4.025493421052632e-06,
      "loss": 0.3426,
      "step": 14540
    },
    {
      "epoch": 5.982730263157895,
      "grad_norm": 100.8232650756836,
      "learning_rate": 4.021381578947368e-06,
      "loss": 0.1401,
      "step": 14550
    },
    {
      "epoch": 5.9868421052631575,
      "grad_norm": 0.010361550375819206,
      "learning_rate": 4.017269736842106e-06,
      "loss": 0.1157,
      "step": 14560
    },
    {
      "epoch": 5.990953947368421,
      "grad_norm": 2.4274368286132812,
      "learning_rate": 4.013157894736842e-06,
      "loss": 0.0128,
      "step": 14570
    },
    {
      "epoch": 5.995065789473684,
      "grad_norm": 33.48968505859375,
      "learning_rate": 4.00904605263158e-06,
      "loss": 0.0646,
      "step": 14580
    },
    {
      "epoch": 5.999177631578947,
      "grad_norm": 0.3265540897846222,
      "learning_rate": 4.004934210526316e-06,
      "loss": 0.0041,
      "step": 14590
    },
    {
      "epoch": 6.003289473684211,
      "grad_norm": 0.12049008160829544,
      "learning_rate": 4.000822368421053e-06,
      "loss": 0.2218,
      "step": 14600
    },
    {
      "epoch": 6.0074013157894735,
      "grad_norm": 0.025286437943577766,
      "learning_rate": 3.99671052631579e-06,
      "loss": 0.0038,
      "step": 14610
    },
    {
      "epoch": 6.011513157894737,
      "grad_norm": 0.015311812050640583,
      "learning_rate": 3.992598684210527e-06,
      "loss": 0.002,
      "step": 14620
    },
    {
      "epoch": 6.015625,
      "grad_norm": 0.2929839491844177,
      "learning_rate": 3.988486842105264e-06,
      "loss": 0.0652,
      "step": 14630
    },
    {
      "epoch": 6.019736842105263,
      "grad_norm": 0.006915655452758074,
      "learning_rate": 3.984375e-06,
      "loss": 0.0182,
      "step": 14640
    },
    {
      "epoch": 6.0238486842105265,
      "grad_norm": 0.03598955273628235,
      "learning_rate": 3.980263157894737e-06,
      "loss": 0.0327,
      "step": 14650
    },
    {
      "epoch": 6.027960526315789,
      "grad_norm": 0.007388933561742306,
      "learning_rate": 3.9761513157894735e-06,
      "loss": 0.085,
      "step": 14660
    },
    {
      "epoch": 6.032072368421052,
      "grad_norm": 0.2169804871082306,
      "learning_rate": 3.972039473684211e-06,
      "loss": 0.1496,
      "step": 14670
    },
    {
      "epoch": 6.036184210526316,
      "grad_norm": 3.28409743309021,
      "learning_rate": 3.9679276315789476e-06,
      "loss": 0.0039,
      "step": 14680
    },
    {
      "epoch": 6.040296052631579,
      "grad_norm": 6.0432586669921875,
      "learning_rate": 3.963815789473684e-06,
      "loss": 0.0479,
      "step": 14690
    },
    {
      "epoch": 6.0444078947368425,
      "grad_norm": 5.428047180175781,
      "learning_rate": 3.959703947368422e-06,
      "loss": 0.0167,
      "step": 14700
    },
    {
      "epoch": 6.048519736842105,
      "grad_norm": 3.8892929553985596,
      "learning_rate": 3.955592105263158e-06,
      "loss": 0.0089,
      "step": 14710
    },
    {
      "epoch": 6.052631578947368,
      "grad_norm": 13.23945426940918,
      "learning_rate": 3.951480263157895e-06,
      "loss": 0.0926,
      "step": 14720
    },
    {
      "epoch": 6.056743421052632,
      "grad_norm": 0.0022840045858174562,
      "learning_rate": 3.947368421052632e-06,
      "loss": 0.0017,
      "step": 14730
    },
    {
      "epoch": 6.060855263157895,
      "grad_norm": 19.486738204956055,
      "learning_rate": 3.943256578947369e-06,
      "loss": 0.013,
      "step": 14740
    },
    {
      "epoch": 6.0649671052631575,
      "grad_norm": 0.0012240969808772206,
      "learning_rate": 3.9391447368421055e-06,
      "loss": 0.0591,
      "step": 14750
    },
    {
      "epoch": 6.069078947368421,
      "grad_norm": 7.582399845123291,
      "learning_rate": 3.935032894736843e-06,
      "loss": 0.0043,
      "step": 14760
    },
    {
      "epoch": 6.073190789473684,
      "grad_norm": 2.2928051948547363,
      "learning_rate": 3.93092105263158e-06,
      "loss": 0.0007,
      "step": 14770
    },
    {
      "epoch": 6.077302631578948,
      "grad_norm": 0.0408819355070591,
      "learning_rate": 3.926809210526316e-06,
      "loss": 0.1468,
      "step": 14780
    },
    {
      "epoch": 6.081414473684211,
      "grad_norm": 0.3422407805919647,
      "learning_rate": 3.922697368421053e-06,
      "loss": 0.0169,
      "step": 14790
    },
    {
      "epoch": 6.0855263157894735,
      "grad_norm": 0.14110057055950165,
      "learning_rate": 3.918585526315789e-06,
      "loss": 0.0263,
      "step": 14800
    },
    {
      "epoch": 6.089638157894737,
      "grad_norm": 0.011704709380865097,
      "learning_rate": 3.914473684210527e-06,
      "loss": 0.011,
      "step": 14810
    },
    {
      "epoch": 6.09375,
      "grad_norm": 0.21069417893886566,
      "learning_rate": 3.9103618421052635e-06,
      "loss": 0.0939,
      "step": 14820
    },
    {
      "epoch": 6.097861842105263,
      "grad_norm": 0.0002515464148018509,
      "learning_rate": 3.90625e-06,
      "loss": 0.1361,
      "step": 14830
    },
    {
      "epoch": 6.1019736842105265,
      "grad_norm": 0.18480344116687775,
      "learning_rate": 3.902138157894737e-06,
      "loss": 0.1573,
      "step": 14840
    },
    {
      "epoch": 6.106085526315789,
      "grad_norm": 0.015395952388644218,
      "learning_rate": 3.898026315789474e-06,
      "loss": 0.0741,
      "step": 14850
    },
    {
      "epoch": 6.110197368421052,
      "grad_norm": 0.2790496349334717,
      "learning_rate": 3.893914473684211e-06,
      "loss": 0.0041,
      "step": 14860
    },
    {
      "epoch": 6.114309210526316,
      "grad_norm": 0.017724014818668365,
      "learning_rate": 3.889802631578947e-06,
      "loss": 0.2005,
      "step": 14870
    },
    {
      "epoch": 6.118421052631579,
      "grad_norm": 0.0019499589689075947,
      "learning_rate": 3.885690789473685e-06,
      "loss": 0.0004,
      "step": 14880
    },
    {
      "epoch": 6.1225328947368425,
      "grad_norm": 0.21595421433448792,
      "learning_rate": 3.8815789473684214e-06,
      "loss": 0.0384,
      "step": 14890
    },
    {
      "epoch": 6.126644736842105,
      "grad_norm": 19.680784225463867,
      "learning_rate": 3.877467105263158e-06,
      "loss": 0.0066,
      "step": 14900
    },
    {
      "epoch": 6.130756578947368,
      "grad_norm": 0.0334845706820488,
      "learning_rate": 3.8733552631578955e-06,
      "loss": 0.0008,
      "step": 14910
    },
    {
      "epoch": 6.134868421052632,
      "grad_norm": 0.09054317325353622,
      "learning_rate": 3.869243421052632e-06,
      "loss": 0.0209,
      "step": 14920
    },
    {
      "epoch": 6.138980263157895,
      "grad_norm": 0.06103743240237236,
      "learning_rate": 3.865131578947369e-06,
      "loss": 0.3223,
      "step": 14930
    },
    {
      "epoch": 6.1430921052631575,
      "grad_norm": 0.028684088960289955,
      "learning_rate": 3.861019736842105e-06,
      "loss": 0.0032,
      "step": 14940
    },
    {
      "epoch": 6.147203947368421,
      "grad_norm": 0.00259952899068594,
      "learning_rate": 3.856907894736843e-06,
      "loss": 0.0267,
      "step": 14950
    },
    {
      "epoch": 6.151315789473684,
      "grad_norm": 0.0002748616971075535,
      "learning_rate": 3.852796052631579e-06,
      "loss": 0.094,
      "step": 14960
    },
    {
      "epoch": 6.155427631578948,
      "grad_norm": 53.90045166015625,
      "learning_rate": 3.848684210526316e-06,
      "loss": 0.0521,
      "step": 14970
    },
    {
      "epoch": 6.159539473684211,
      "grad_norm": 0.017857911065220833,
      "learning_rate": 3.844572368421053e-06,
      "loss": 0.0002,
      "step": 14980
    },
    {
      "epoch": 6.1636513157894735,
      "grad_norm": 4.2133588790893555,
      "learning_rate": 3.840460526315789e-06,
      "loss": 0.2049,
      "step": 14990
    },
    {
      "epoch": 6.167763157894737,
      "grad_norm": 0.0010838140733540058,
      "learning_rate": 3.836348684210527e-06,
      "loss": 0.1498,
      "step": 15000
    },
    {
      "epoch": 6.171875,
      "grad_norm": 0.25620415806770325,
      "learning_rate": 3.832236842105263e-06,
      "loss": 0.0029,
      "step": 15010
    },
    {
      "epoch": 6.175986842105263,
      "grad_norm": 102.45184326171875,
      "learning_rate": 3.828125000000001e-06,
      "loss": 0.1873,
      "step": 15020
    },
    {
      "epoch": 6.1800986842105265,
      "grad_norm": 0.4820612967014313,
      "learning_rate": 3.824013157894737e-06,
      "loss": 0.2001,
      "step": 15030
    },
    {
      "epoch": 6.184210526315789,
      "grad_norm": 64.73114776611328,
      "learning_rate": 3.819901315789474e-06,
      "loss": 0.0622,
      "step": 15040
    },
    {
      "epoch": 6.188322368421052,
      "grad_norm": 0.00912266131490469,
      "learning_rate": 3.815789473684211e-06,
      "loss": 0.0231,
      "step": 15050
    },
    {
      "epoch": 6.192434210526316,
      "grad_norm": 1.8277114629745483,
      "learning_rate": 3.8116776315789476e-06,
      "loss": 0.0014,
      "step": 15060
    },
    {
      "epoch": 6.196546052631579,
      "grad_norm": 0.056886494159698486,
      "learning_rate": 3.8075657894736846e-06,
      "loss": 0.0015,
      "step": 15070
    },
    {
      "epoch": 6.2006578947368425,
      "grad_norm": 0.10024416446685791,
      "learning_rate": 3.8034539473684212e-06,
      "loss": 0.0224,
      "step": 15080
    },
    {
      "epoch": 6.204769736842105,
      "grad_norm": 42.96096420288086,
      "learning_rate": 3.799342105263158e-06,
      "loss": 0.0894,
      "step": 15090
    },
    {
      "epoch": 6.208881578947368,
      "grad_norm": 0.007689566817134619,
      "learning_rate": 3.7952302631578953e-06,
      "loss": 0.1738,
      "step": 15100
    },
    {
      "epoch": 6.212993421052632,
      "grad_norm": 0.0625908225774765,
      "learning_rate": 3.791118421052632e-06,
      "loss": 0.1453,
      "step": 15110
    },
    {
      "epoch": 6.217105263157895,
      "grad_norm": 23.779741287231445,
      "learning_rate": 3.7870065789473685e-06,
      "loss": 0.0153,
      "step": 15120
    },
    {
      "epoch": 6.2212171052631575,
      "grad_norm": 0.0019070090493187308,
      "learning_rate": 3.7828947368421055e-06,
      "loss": 0.1492,
      "step": 15130
    },
    {
      "epoch": 6.225328947368421,
      "grad_norm": 0.03249938786029816,
      "learning_rate": 3.7787828947368426e-06,
      "loss": 0.1902,
      "step": 15140
    },
    {
      "epoch": 6.229440789473684,
      "grad_norm": 0.09956735372543335,
      "learning_rate": 3.7746710526315796e-06,
      "loss": 0.0199,
      "step": 15150
    },
    {
      "epoch": 6.233552631578948,
      "grad_norm": 0.005886862985789776,
      "learning_rate": 3.770559210526316e-06,
      "loss": 0.0007,
      "step": 15160
    },
    {
      "epoch": 6.237664473684211,
      "grad_norm": 3.7049739360809326,
      "learning_rate": 3.766447368421053e-06,
      "loss": 0.004,
      "step": 15170
    },
    {
      "epoch": 6.2417763157894735,
      "grad_norm": 53.633445739746094,
      "learning_rate": 3.7623355263157894e-06,
      "loss": 0.5929,
      "step": 15180
    },
    {
      "epoch": 6.245888157894737,
      "grad_norm": 1.0991520881652832,
      "learning_rate": 3.758223684210527e-06,
      "loss": 0.0172,
      "step": 15190
    },
    {
      "epoch": 6.25,
      "grad_norm": 0.020138980820775032,
      "learning_rate": 3.7541118421052635e-06,
      "loss": 0.0978,
      "step": 15200
    },
    {
      "epoch": 6.254111842105263,
      "grad_norm": 0.02193092741072178,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.0205,
      "step": 15210
    },
    {
      "epoch": 6.2582236842105265,
      "grad_norm": 0.04015963524580002,
      "learning_rate": 3.745888157894737e-06,
      "loss": 0.0017,
      "step": 15220
    },
    {
      "epoch": 6.262335526315789,
      "grad_norm": 3.5081162452697754,
      "learning_rate": 3.7417763157894737e-06,
      "loss": 0.0017,
      "step": 15230
    },
    {
      "epoch": 6.266447368421053,
      "grad_norm": 0.0039045389275997877,
      "learning_rate": 3.737664473684211e-06,
      "loss": 0.0078,
      "step": 15240
    },
    {
      "epoch": 6.270559210526316,
      "grad_norm": 0.2885434627532959,
      "learning_rate": 3.733552631578948e-06,
      "loss": 0.0171,
      "step": 15250
    },
    {
      "epoch": 6.274671052631579,
      "grad_norm": 0.028393974527716637,
      "learning_rate": 3.7294407894736844e-06,
      "loss": 0.0788,
      "step": 15260
    },
    {
      "epoch": 6.2787828947368425,
      "grad_norm": 116.33313751220703,
      "learning_rate": 3.725328947368421e-06,
      "loss": 0.0904,
      "step": 15270
    },
    {
      "epoch": 6.282894736842105,
      "grad_norm": 0.009480939246714115,
      "learning_rate": 3.721217105263158e-06,
      "loss": 0.0602,
      "step": 15280
    },
    {
      "epoch": 6.287006578947368,
      "grad_norm": 2.470651865005493,
      "learning_rate": 3.717105263157895e-06,
      "loss": 0.0058,
      "step": 15290
    },
    {
      "epoch": 6.291118421052632,
      "grad_norm": 0.00010311637015547603,
      "learning_rate": 3.712993421052632e-06,
      "loss": 0.1697,
      "step": 15300
    },
    {
      "epoch": 6.295230263157895,
      "grad_norm": 2.6846992113860324e-06,
      "learning_rate": 3.7088815789473687e-06,
      "loss": 0.0135,
      "step": 15310
    },
    {
      "epoch": 6.2993421052631575,
      "grad_norm": 0.005294320173561573,
      "learning_rate": 3.7047697368421053e-06,
      "loss": 0.0015,
      "step": 15320
    },
    {
      "epoch": 6.303453947368421,
      "grad_norm": 0.05376331880688667,
      "learning_rate": 3.7006578947368428e-06,
      "loss": 0.2444,
      "step": 15330
    },
    {
      "epoch": 6.307565789473684,
      "grad_norm": 0.4994964897632599,
      "learning_rate": 3.6965460526315794e-06,
      "loss": 0.0348,
      "step": 15340
    },
    {
      "epoch": 6.311677631578947,
      "grad_norm": 0.04438924789428711,
      "learning_rate": 3.692434210526316e-06,
      "loss": 0.1409,
      "step": 15350
    },
    {
      "epoch": 6.315789473684211,
      "grad_norm": 2.267488718032837,
      "learning_rate": 3.688322368421053e-06,
      "loss": 0.01,
      "step": 15360
    },
    {
      "epoch": 6.3199013157894735,
      "grad_norm": 65.40839385986328,
      "learning_rate": 3.6842105263157896e-06,
      "loss": 0.1391,
      "step": 15370
    },
    {
      "epoch": 6.324013157894737,
      "grad_norm": 1.6076836585998535,
      "learning_rate": 3.6800986842105267e-06,
      "loss": 0.0277,
      "step": 15380
    },
    {
      "epoch": 6.328125,
      "grad_norm": 71.12222290039062,
      "learning_rate": 3.6759868421052637e-06,
      "loss": 0.3136,
      "step": 15390
    },
    {
      "epoch": 6.332236842105263,
      "grad_norm": 0.27795493602752686,
      "learning_rate": 3.6718750000000003e-06,
      "loss": 0.0337,
      "step": 15400
    },
    {
      "epoch": 6.3363486842105265,
      "grad_norm": 0.005055265035480261,
      "learning_rate": 3.667763157894737e-06,
      "loss": 0.0913,
      "step": 15410
    },
    {
      "epoch": 6.340460526315789,
      "grad_norm": 1.20720374584198,
      "learning_rate": 3.663651315789474e-06,
      "loss": 0.049,
      "step": 15420
    },
    {
      "epoch": 6.344572368421053,
      "grad_norm": 0.020946979522705078,
      "learning_rate": 3.659539473684211e-06,
      "loss": 0.0702,
      "step": 15430
    },
    {
      "epoch": 6.348684210526316,
      "grad_norm": 0.006545315962284803,
      "learning_rate": 3.6554276315789476e-06,
      "loss": 0.0369,
      "step": 15440
    },
    {
      "epoch": 6.352796052631579,
      "grad_norm": 0.05248495563864708,
      "learning_rate": 3.6513157894736846e-06,
      "loss": 0.0086,
      "step": 15450
    },
    {
      "epoch": 6.3569078947368425,
      "grad_norm": 19.267728805541992,
      "learning_rate": 3.6472039473684212e-06,
      "loss": 0.0154,
      "step": 15460
    },
    {
      "epoch": 6.361019736842105,
      "grad_norm": 0.8632445931434631,
      "learning_rate": 3.643092105263158e-06,
      "loss": 0.0369,
      "step": 15470
    },
    {
      "epoch": 6.365131578947368,
      "grad_norm": 13.235368728637695,
      "learning_rate": 3.6389802631578953e-06,
      "loss": 0.0172,
      "step": 15480
    },
    {
      "epoch": 6.369243421052632,
      "grad_norm": 0.0018560757162049413,
      "learning_rate": 3.634868421052632e-06,
      "loss": 0.0082,
      "step": 15490
    },
    {
      "epoch": 6.373355263157895,
      "grad_norm": 0.04792603850364685,
      "learning_rate": 3.6307565789473685e-06,
      "loss": 0.1249,
      "step": 15500
    },
    {
      "epoch": 6.3774671052631575,
      "grad_norm": 66.24674224853516,
      "learning_rate": 3.6266447368421055e-06,
      "loss": 0.0234,
      "step": 15510
    },
    {
      "epoch": 6.381578947368421,
      "grad_norm": 1.533945918083191,
      "learning_rate": 3.6225328947368426e-06,
      "loss": 0.0262,
      "step": 15520
    },
    {
      "epoch": 6.385690789473684,
      "grad_norm": 0.4053722620010376,
      "learning_rate": 3.618421052631579e-06,
      "loss": 0.0028,
      "step": 15530
    },
    {
      "epoch": 6.389802631578947,
      "grad_norm": 0.1870436817407608,
      "learning_rate": 3.6143092105263162e-06,
      "loss": 0.2569,
      "step": 15540
    },
    {
      "epoch": 6.393914473684211,
      "grad_norm": 0.31710654497146606,
      "learning_rate": 3.610197368421053e-06,
      "loss": 0.0286,
      "step": 15550
    },
    {
      "epoch": 6.3980263157894735,
      "grad_norm": 0.01970168575644493,
      "learning_rate": 3.6060855263157894e-06,
      "loss": 0.0044,
      "step": 15560
    },
    {
      "epoch": 6.402138157894737,
      "grad_norm": 0.22997547686100006,
      "learning_rate": 3.601973684210527e-06,
      "loss": 0.1054,
      "step": 15570
    },
    {
      "epoch": 6.40625,
      "grad_norm": 0.16296066343784332,
      "learning_rate": 3.5978618421052635e-06,
      "loss": 0.1698,
      "step": 15580
    },
    {
      "epoch": 6.410361842105263,
      "grad_norm": 0.0006991025293245912,
      "learning_rate": 3.59375e-06,
      "loss": 0.1825,
      "step": 15590
    },
    {
      "epoch": 6.4144736842105265,
      "grad_norm": 0.07622674107551575,
      "learning_rate": 3.589638157894737e-06,
      "loss": 0.1282,
      "step": 15600
    },
    {
      "epoch": 6.418585526315789,
      "grad_norm": 0.5237715244293213,
      "learning_rate": 3.5855263157894737e-06,
      "loss": 0.0007,
      "step": 15610
    },
    {
      "epoch": 6.422697368421053,
      "grad_norm": 0.9545769691467285,
      "learning_rate": 3.581414473684211e-06,
      "loss": 0.0279,
      "step": 15620
    },
    {
      "epoch": 6.426809210526316,
      "grad_norm": 0.0007075117900967598,
      "learning_rate": 3.577302631578948e-06,
      "loss": 0.2943,
      "step": 15630
    },
    {
      "epoch": 6.430921052631579,
      "grad_norm": 0.005581384990364313,
      "learning_rate": 3.5731907894736844e-06,
      "loss": 0.0005,
      "step": 15640
    },
    {
      "epoch": 6.4350328947368425,
      "grad_norm": 0.3551817238330841,
      "learning_rate": 3.569078947368421e-06,
      "loss": 0.1535,
      "step": 15650
    },
    {
      "epoch": 6.439144736842105,
      "grad_norm": 0.0016296847024932504,
      "learning_rate": 3.564967105263158e-06,
      "loss": 0.0004,
      "step": 15660
    },
    {
      "epoch": 6.443256578947368,
      "grad_norm": 0.0016560270451009274,
      "learning_rate": 3.560855263157895e-06,
      "loss": 0.0101,
      "step": 15670
    },
    {
      "epoch": 6.447368421052632,
      "grad_norm": 0.1265086531639099,
      "learning_rate": 3.556743421052632e-06,
      "loss": 0.0003,
      "step": 15680
    },
    {
      "epoch": 6.451480263157895,
      "grad_norm": 0.37500542402267456,
      "learning_rate": 3.5526315789473687e-06,
      "loss": 0.014,
      "step": 15690
    },
    {
      "epoch": 6.4555921052631575,
      "grad_norm": 0.041531626135110855,
      "learning_rate": 3.5485197368421053e-06,
      "loss": 0.0201,
      "step": 15700
    },
    {
      "epoch": 6.459703947368421,
      "grad_norm": 60.48843765258789,
      "learning_rate": 3.544407894736843e-06,
      "loss": 0.1084,
      "step": 15710
    },
    {
      "epoch": 6.463815789473684,
      "grad_norm": 0.0003151536511722952,
      "learning_rate": 3.5402960526315794e-06,
      "loss": 0.005,
      "step": 15720
    },
    {
      "epoch": 6.467927631578947,
      "grad_norm": 1.3518043756484985,
      "learning_rate": 3.536184210526316e-06,
      "loss": 0.002,
      "step": 15730
    },
    {
      "epoch": 6.472039473684211,
      "grad_norm": 0.019122233614325523,
      "learning_rate": 3.5320723684210526e-06,
      "loss": 0.0592,
      "step": 15740
    },
    {
      "epoch": 6.4761513157894735,
      "grad_norm": 0.022713489830493927,
      "learning_rate": 3.5279605263157897e-06,
      "loss": 0.0002,
      "step": 15750
    },
    {
      "epoch": 6.480263157894737,
      "grad_norm": 0.03588182106614113,
      "learning_rate": 3.5238486842105267e-06,
      "loss": 0.1249,
      "step": 15760
    },
    {
      "epoch": 6.484375,
      "grad_norm": 38.98967742919922,
      "learning_rate": 3.5197368421052637e-06,
      "loss": 0.0222,
      "step": 15770
    },
    {
      "epoch": 6.488486842105263,
      "grad_norm": 0.08748682588338852,
      "learning_rate": 3.5156250000000003e-06,
      "loss": 0.0345,
      "step": 15780
    },
    {
      "epoch": 6.4925986842105265,
      "grad_norm": 0.46851471066474915,
      "learning_rate": 3.511513157894737e-06,
      "loss": 0.0363,
      "step": 15790
    },
    {
      "epoch": 6.496710526315789,
      "grad_norm": 1.2130106687545776,
      "learning_rate": 3.5074013157894735e-06,
      "loss": 0.0027,
      "step": 15800
    },
    {
      "epoch": 6.500822368421053,
      "grad_norm": 0.11882048100233078,
      "learning_rate": 3.503289473684211e-06,
      "loss": 0.0275,
      "step": 15810
    },
    {
      "epoch": 6.504934210526316,
      "grad_norm": 0.19122253358364105,
      "learning_rate": 3.4991776315789476e-06,
      "loss": 0.0953,
      "step": 15820
    },
    {
      "epoch": 6.509046052631579,
      "grad_norm": 10.092044830322266,
      "learning_rate": 3.4950657894736846e-06,
      "loss": 0.1668,
      "step": 15830
    },
    {
      "epoch": 6.5131578947368425,
      "grad_norm": 0.1250755488872528,
      "learning_rate": 3.4909539473684212e-06,
      "loss": 0.0107,
      "step": 15840
    },
    {
      "epoch": 6.517269736842105,
      "grad_norm": 0.006768290884792805,
      "learning_rate": 3.486842105263158e-06,
      "loss": 0.0964,
      "step": 15850
    },
    {
      "epoch": 6.521381578947368,
      "grad_norm": 0.003787360619753599,
      "learning_rate": 3.4827302631578953e-06,
      "loss": 0.0818,
      "step": 15860
    },
    {
      "epoch": 6.525493421052632,
      "grad_norm": 0.004058108199387789,
      "learning_rate": 3.478618421052632e-06,
      "loss": 0.0004,
      "step": 15870
    },
    {
      "epoch": 6.529605263157895,
      "grad_norm": 1.6648340225219727,
      "learning_rate": 3.4745065789473685e-06,
      "loss": 0.0036,
      "step": 15880
    },
    {
      "epoch": 6.5337171052631575,
      "grad_norm": 0.0452425591647625,
      "learning_rate": 3.4703947368421056e-06,
      "loss": 0.0892,
      "step": 15890
    },
    {
      "epoch": 6.537828947368421,
      "grad_norm": 0.038435883820056915,
      "learning_rate": 3.4662828947368426e-06,
      "loss": 0.109,
      "step": 15900
    },
    {
      "epoch": 6.541940789473684,
      "grad_norm": 0.17975053191184998,
      "learning_rate": 3.462171052631579e-06,
      "loss": 0.0345,
      "step": 15910
    },
    {
      "epoch": 6.546052631578947,
      "grad_norm": 0.018029751256108284,
      "learning_rate": 3.4580592105263162e-06,
      "loss": 0.2231,
      "step": 15920
    },
    {
      "epoch": 6.550164473684211,
      "grad_norm": 100.19669342041016,
      "learning_rate": 3.453947368421053e-06,
      "loss": 0.264,
      "step": 15930
    },
    {
      "epoch": 6.5542763157894735,
      "grad_norm": 0.0011808631243184209,
      "learning_rate": 3.4498355263157895e-06,
      "loss": 0.0064,
      "step": 15940
    },
    {
      "epoch": 6.558388157894737,
      "grad_norm": 0.053727325052022934,
      "learning_rate": 3.445723684210527e-06,
      "loss": 0.0457,
      "step": 15950
    },
    {
      "epoch": 6.5625,
      "grad_norm": 0.07882308214902878,
      "learning_rate": 3.4416118421052635e-06,
      "loss": 0.0161,
      "step": 15960
    },
    {
      "epoch": 6.566611842105263,
      "grad_norm": 0.0010185951832681894,
      "learning_rate": 3.4375e-06,
      "loss": 0.3123,
      "step": 15970
    },
    {
      "epoch": 6.5707236842105265,
      "grad_norm": 0.03667188435792923,
      "learning_rate": 3.433388157894737e-06,
      "loss": 0.0293,
      "step": 15980
    },
    {
      "epoch": 6.574835526315789,
      "grad_norm": 0.4083603322505951,
      "learning_rate": 3.4292763157894738e-06,
      "loss": 0.0157,
      "step": 15990
    },
    {
      "epoch": 6.578947368421053,
      "grad_norm": 0.5004896521568298,
      "learning_rate": 3.4251644736842112e-06,
      "loss": 0.0088,
      "step": 16000
    },
    {
      "epoch": 6.583059210526316,
      "grad_norm": 0.7514389753341675,
      "learning_rate": 3.421052631578948e-06,
      "loss": 0.0133,
      "step": 16010
    },
    {
      "epoch": 6.587171052631579,
      "grad_norm": 7.884642601013184,
      "learning_rate": 3.4169407894736844e-06,
      "loss": 0.0301,
      "step": 16020
    },
    {
      "epoch": 6.5912828947368425,
      "grad_norm": 0.028393534943461418,
      "learning_rate": 3.412828947368421e-06,
      "loss": 0.0043,
      "step": 16030
    },
    {
      "epoch": 6.595394736842105,
      "grad_norm": 6.707031726837158,
      "learning_rate": 3.408717105263158e-06,
      "loss": 0.0033,
      "step": 16040
    },
    {
      "epoch": 6.599506578947368,
      "grad_norm": 0.5291396379470825,
      "learning_rate": 3.404605263157895e-06,
      "loss": 0.0036,
      "step": 16050
    },
    {
      "epoch": 6.603618421052632,
      "grad_norm": 0.9197802543640137,
      "learning_rate": 3.4004934210526317e-06,
      "loss": 0.0796,
      "step": 16060
    },
    {
      "epoch": 6.607730263157895,
      "grad_norm": 0.025339746847748756,
      "learning_rate": 3.3963815789473687e-06,
      "loss": 0.012,
      "step": 16070
    },
    {
      "epoch": 6.6118421052631575,
      "grad_norm": 0.10476145893335342,
      "learning_rate": 3.3922697368421054e-06,
      "loss": 0.0095,
      "step": 16080
    },
    {
      "epoch": 6.615953947368421,
      "grad_norm": 15.861358642578125,
      "learning_rate": 3.388157894736843e-06,
      "loss": 0.0038,
      "step": 16090
    },
    {
      "epoch": 6.620065789473684,
      "grad_norm": 0.002507624914869666,
      "learning_rate": 3.3840460526315794e-06,
      "loss": 0.2463,
      "step": 16100
    },
    {
      "epoch": 6.624177631578947,
      "grad_norm": 46.91939926147461,
      "learning_rate": 3.379934210526316e-06,
      "loss": 0.1331,
      "step": 16110
    },
    {
      "epoch": 6.628289473684211,
      "grad_norm": 77.30012512207031,
      "learning_rate": 3.3758223684210526e-06,
      "loss": 0.1453,
      "step": 16120
    },
    {
      "epoch": 6.6324013157894735,
      "grad_norm": 0.0014009532751515508,
      "learning_rate": 3.3717105263157897e-06,
      "loss": 0.0051,
      "step": 16130
    },
    {
      "epoch": 6.636513157894737,
      "grad_norm": 1.946133017539978,
      "learning_rate": 3.3675986842105267e-06,
      "loss": 0.1491,
      "step": 16140
    },
    {
      "epoch": 6.640625,
      "grad_norm": 44.568275451660156,
      "learning_rate": 3.3634868421052637e-06,
      "loss": 0.0212,
      "step": 16150
    },
    {
      "epoch": 6.644736842105263,
      "grad_norm": 46.899925231933594,
      "learning_rate": 3.3593750000000003e-06,
      "loss": 0.0164,
      "step": 16160
    },
    {
      "epoch": 6.6488486842105265,
      "grad_norm": 55.38142776489258,
      "learning_rate": 3.355263157894737e-06,
      "loss": 0.027,
      "step": 16170
    },
    {
      "epoch": 6.652960526315789,
      "grad_norm": 44.49644088745117,
      "learning_rate": 3.3511513157894736e-06,
      "loss": 0.0183,
      "step": 16180
    },
    {
      "epoch": 6.657072368421053,
      "grad_norm": 9.747554577188566e-05,
      "learning_rate": 3.347039473684211e-06,
      "loss": 0.0015,
      "step": 16190
    },
    {
      "epoch": 6.661184210526316,
      "grad_norm": 7.189190364442766e-05,
      "learning_rate": 3.3429276315789476e-06,
      "loss": 0.121,
      "step": 16200
    },
    {
      "epoch": 6.665296052631579,
      "grad_norm": 0.22454024851322174,
      "learning_rate": 3.3388157894736847e-06,
      "loss": 0.0552,
      "step": 16210
    },
    {
      "epoch": 6.6694078947368425,
      "grad_norm": 0.09667879343032837,
      "learning_rate": 3.3347039473684213e-06,
      "loss": 0.1034,
      "step": 16220
    },
    {
      "epoch": 6.673519736842105,
      "grad_norm": 0.00030903020524419844,
      "learning_rate": 3.330592105263158e-06,
      "loss": 0.0216,
      "step": 16230
    },
    {
      "epoch": 6.677631578947368,
      "grad_norm": 0.0006744477432221174,
      "learning_rate": 3.3264802631578953e-06,
      "loss": 0.029,
      "step": 16240
    },
    {
      "epoch": 6.681743421052632,
      "grad_norm": 0.0019801906310021877,
      "learning_rate": 3.322368421052632e-06,
      "loss": 0.0155,
      "step": 16250
    },
    {
      "epoch": 6.685855263157895,
      "grad_norm": 0.041232798248529434,
      "learning_rate": 3.3182565789473685e-06,
      "loss": 0.001,
      "step": 16260
    },
    {
      "epoch": 6.6899671052631575,
      "grad_norm": 0.022349294275045395,
      "learning_rate": 3.314144736842105e-06,
      "loss": 0.0077,
      "step": 16270
    },
    {
      "epoch": 6.694078947368421,
      "grad_norm": 0.00042771780863404274,
      "learning_rate": 3.3100328947368426e-06,
      "loss": 0.0002,
      "step": 16280
    },
    {
      "epoch": 6.698190789473684,
      "grad_norm": 0.32936057448387146,
      "learning_rate": 3.3059210526315792e-06,
      "loss": 0.001,
      "step": 16290
    },
    {
      "epoch": 6.702302631578947,
      "grad_norm": 1.45549738407135,
      "learning_rate": 3.3018092105263162e-06,
      "loss": 0.0842,
      "step": 16300
    },
    {
      "epoch": 6.706414473684211,
      "grad_norm": 75.73883056640625,
      "learning_rate": 3.297697368421053e-06,
      "loss": 0.0506,
      "step": 16310
    },
    {
      "epoch": 6.7105263157894735,
      "grad_norm": 0.0010807811049744487,
      "learning_rate": 3.2935855263157895e-06,
      "loss": 0.0477,
      "step": 16320
    },
    {
      "epoch": 6.714638157894737,
      "grad_norm": 8.679156303405762,
      "learning_rate": 3.289473684210527e-06,
      "loss": 0.0093,
      "step": 16330
    },
    {
      "epoch": 6.71875,
      "grad_norm": 0.07848314195871353,
      "learning_rate": 3.2853618421052635e-06,
      "loss": 0.0009,
      "step": 16340
    },
    {
      "epoch": 6.722861842105263,
      "grad_norm": 0.0014197869459167123,
      "learning_rate": 3.28125e-06,
      "loss": 0.1809,
      "step": 16350
    },
    {
      "epoch": 6.7269736842105265,
      "grad_norm": 0.10961536318063736,
      "learning_rate": 3.277138157894737e-06,
      "loss": 0.0595,
      "step": 16360
    },
    {
      "epoch": 6.731085526315789,
      "grad_norm": 0.028819315135478973,
      "learning_rate": 3.2730263157894738e-06,
      "loss": 0.035,
      "step": 16370
    },
    {
      "epoch": 6.735197368421053,
      "grad_norm": 0.12262340635061264,
      "learning_rate": 3.268914473684211e-06,
      "loss": 0.3158,
      "step": 16380
    },
    {
      "epoch": 6.739309210526316,
      "grad_norm": 0.001396475243382156,
      "learning_rate": 3.264802631578948e-06,
      "loss": 0.0022,
      "step": 16390
    },
    {
      "epoch": 6.743421052631579,
      "grad_norm": 0.009128180332481861,
      "learning_rate": 3.2606907894736844e-06,
      "loss": 0.0893,
      "step": 16400
    },
    {
      "epoch": 6.7475328947368425,
      "grad_norm": 0.012526153586804867,
      "learning_rate": 3.256578947368421e-06,
      "loss": 0.0013,
      "step": 16410
    },
    {
      "epoch": 6.751644736842105,
      "grad_norm": 0.8247601389884949,
      "learning_rate": 3.252467105263158e-06,
      "loss": 0.0028,
      "step": 16420
    },
    {
      "epoch": 6.755756578947368,
      "grad_norm": 0.003459361381828785,
      "learning_rate": 3.248355263157895e-06,
      "loss": 0.0008,
      "step": 16430
    },
    {
      "epoch": 6.759868421052632,
      "grad_norm": 0.006843214388936758,
      "learning_rate": 3.2446546052631583e-06,
      "loss": 0.0352,
      "step": 16440
    },
    {
      "epoch": 6.763980263157895,
      "grad_norm": 0.20814478397369385,
      "learning_rate": 3.240542763157895e-06,
      "loss": 0.0066,
      "step": 16450
    },
    {
      "epoch": 6.7680921052631575,
      "grad_norm": 0.029610252007842064,
      "learning_rate": 3.236430921052632e-06,
      "loss": 0.0001,
      "step": 16460
    },
    {
      "epoch": 6.772203947368421,
      "grad_norm": 0.811451256275177,
      "learning_rate": 3.2323190789473686e-06,
      "loss": 0.0861,
      "step": 16470
    },
    {
      "epoch": 6.776315789473684,
      "grad_norm": 0.24721278250217438,
      "learning_rate": 3.2282072368421056e-06,
      "loss": 0.016,
      "step": 16480
    },
    {
      "epoch": 6.780427631578947,
      "grad_norm": 85.86888122558594,
      "learning_rate": 3.2240953947368426e-06,
      "loss": 0.2855,
      "step": 16490
    },
    {
      "epoch": 6.784539473684211,
      "grad_norm": 0.6534148454666138,
      "learning_rate": 3.2199835526315792e-06,
      "loss": 0.0448,
      "step": 16500
    },
    {
      "epoch": 6.7886513157894735,
      "grad_norm": 0.04840293154120445,
      "learning_rate": 3.215871710526316e-06,
      "loss": 0.0451,
      "step": 16510
    },
    {
      "epoch": 6.792763157894737,
      "grad_norm": 0.07276813685894012,
      "learning_rate": 3.211759868421053e-06,
      "loss": 0.0015,
      "step": 16520
    },
    {
      "epoch": 6.796875,
      "grad_norm": 0.16435430943965912,
      "learning_rate": 3.20764802631579e-06,
      "loss": 0.0069,
      "step": 16530
    },
    {
      "epoch": 6.800986842105263,
      "grad_norm": 11.423922538757324,
      "learning_rate": 3.2035361842105265e-06,
      "loss": 0.0218,
      "step": 16540
    },
    {
      "epoch": 6.8050986842105265,
      "grad_norm": 10.006841659545898,
      "learning_rate": 3.1994243421052636e-06,
      "loss": 0.0037,
      "step": 16550
    },
    {
      "epoch": 6.809210526315789,
      "grad_norm": 0.2944813072681427,
      "learning_rate": 3.1953125e-06,
      "loss": 0.0209,
      "step": 16560
    },
    {
      "epoch": 6.813322368421053,
      "grad_norm": 0.027711883187294006,
      "learning_rate": 3.1912006578947368e-06,
      "loss": 0.346,
      "step": 16570
    },
    {
      "epoch": 6.817434210526316,
      "grad_norm": 0.01937483809888363,
      "learning_rate": 3.1870888157894742e-06,
      "loss": 0.018,
      "step": 16580
    },
    {
      "epoch": 6.821546052631579,
      "grad_norm": 0.11694393306970596,
      "learning_rate": 3.182976973684211e-06,
      "loss": 0.0014,
      "step": 16590
    },
    {
      "epoch": 6.8256578947368425,
      "grad_norm": 0.011637609452009201,
      "learning_rate": 3.1788651315789474e-06,
      "loss": 0.0743,
      "step": 16600
    },
    {
      "epoch": 6.829769736842105,
      "grad_norm": 0.0008141595753841102,
      "learning_rate": 3.1747532894736845e-06,
      "loss": 0.009,
      "step": 16610
    },
    {
      "epoch": 6.833881578947368,
      "grad_norm": 0.004597888328135014,
      "learning_rate": 3.170641447368421e-06,
      "loss": 0.0006,
      "step": 16620
    },
    {
      "epoch": 6.837993421052632,
      "grad_norm": 2.2619147300720215,
      "learning_rate": 3.1665296052631585e-06,
      "loss": 0.0054,
      "step": 16630
    },
    {
      "epoch": 6.842105263157895,
      "grad_norm": 0.02606726624071598,
      "learning_rate": 3.162417763157895e-06,
      "loss": 0.1062,
      "step": 16640
    },
    {
      "epoch": 6.8462171052631575,
      "grad_norm": 0.62782222032547,
      "learning_rate": 3.1583059210526318e-06,
      "loss": 0.1222,
      "step": 16650
    },
    {
      "epoch": 6.850328947368421,
      "grad_norm": 2.099074125289917,
      "learning_rate": 3.1541940789473684e-06,
      "loss": 0.0009,
      "step": 16660
    },
    {
      "epoch": 6.854440789473684,
      "grad_norm": 0.028837570920586586,
      "learning_rate": 3.150082236842106e-06,
      "loss": 0.3018,
      "step": 16670
    },
    {
      "epoch": 6.858552631578947,
      "grad_norm": 0.07627784460783005,
      "learning_rate": 3.1459703947368424e-06,
      "loss": 0.0038,
      "step": 16680
    },
    {
      "epoch": 6.862664473684211,
      "grad_norm": 0.03027813322842121,
      "learning_rate": 3.141858552631579e-06,
      "loss": 0.009,
      "step": 16690
    },
    {
      "epoch": 6.8667763157894735,
      "grad_norm": 0.024021049961447716,
      "learning_rate": 3.137746710526316e-06,
      "loss": 0.0797,
      "step": 16700
    },
    {
      "epoch": 6.870888157894737,
      "grad_norm": 0.0008662066538818181,
      "learning_rate": 3.1336348684210527e-06,
      "loss": 0.0039,
      "step": 16710
    },
    {
      "epoch": 6.875,
      "grad_norm": 0.003181145526468754,
      "learning_rate": 3.12952302631579e-06,
      "loss": 0.1774,
      "step": 16720
    },
    {
      "epoch": 6.879111842105263,
      "grad_norm": 0.35272935032844543,
      "learning_rate": 3.1254111842105267e-06,
      "loss": 0.0915,
      "step": 16730
    },
    {
      "epoch": 6.8832236842105265,
      "grad_norm": 5.994056701660156,
      "learning_rate": 3.1212993421052634e-06,
      "loss": 0.1812,
      "step": 16740
    },
    {
      "epoch": 6.887335526315789,
      "grad_norm": 0.1723736673593521,
      "learning_rate": 3.1171875e-06,
      "loss": 0.0865,
      "step": 16750
    },
    {
      "epoch": 6.891447368421053,
      "grad_norm": 0.045521121472120285,
      "learning_rate": 3.113075657894737e-06,
      "loss": 0.0032,
      "step": 16760
    },
    {
      "epoch": 6.895559210526316,
      "grad_norm": 25.4680233001709,
      "learning_rate": 3.108963815789474e-06,
      "loss": 0.0145,
      "step": 16770
    },
    {
      "epoch": 6.899671052631579,
      "grad_norm": 42.64589309692383,
      "learning_rate": 3.104851973684211e-06,
      "loss": 0.3033,
      "step": 16780
    },
    {
      "epoch": 6.9037828947368425,
      "grad_norm": 0.0010777267161756754,
      "learning_rate": 3.1007401315789477e-06,
      "loss": 0.0036,
      "step": 16790
    },
    {
      "epoch": 6.907894736842105,
      "grad_norm": 0.05889376997947693,
      "learning_rate": 3.0966282894736843e-06,
      "loss": 0.006,
      "step": 16800
    },
    {
      "epoch": 6.912006578947368,
      "grad_norm": 2.3862788677215576,
      "learning_rate": 3.092516447368421e-06,
      "loss": 0.0118,
      "step": 16810
    },
    {
      "epoch": 6.916118421052632,
      "grad_norm": 0.4769994020462036,
      "learning_rate": 3.0884046052631583e-06,
      "loss": 0.1088,
      "step": 16820
    },
    {
      "epoch": 6.920230263157895,
      "grad_norm": 11.872325897216797,
      "learning_rate": 3.084292763157895e-06,
      "loss": 0.0055,
      "step": 16830
    },
    {
      "epoch": 6.9243421052631575,
      "grad_norm": 1.2848358154296875,
      "learning_rate": 3.080180921052632e-06,
      "loss": 0.0026,
      "step": 16840
    },
    {
      "epoch": 6.928453947368421,
      "grad_norm": 0.01796242967247963,
      "learning_rate": 3.0760690789473686e-06,
      "loss": 0.0634,
      "step": 16850
    },
    {
      "epoch": 6.932565789473684,
      "grad_norm": 1.0784401893615723,
      "learning_rate": 3.0719572368421056e-06,
      "loss": 0.0024,
      "step": 16860
    },
    {
      "epoch": 6.936677631578947,
      "grad_norm": 1.7011362314224243,
      "learning_rate": 3.0678453947368426e-06,
      "loss": 0.0013,
      "step": 16870
    },
    {
      "epoch": 6.940789473684211,
      "grad_norm": 0.2864052653312683,
      "learning_rate": 3.0637335526315793e-06,
      "loss": 0.1147,
      "step": 16880
    },
    {
      "epoch": 6.9449013157894735,
      "grad_norm": 0.4255252480506897,
      "learning_rate": 3.059621710526316e-06,
      "loss": 0.0022,
      "step": 16890
    },
    {
      "epoch": 6.949013157894737,
      "grad_norm": 0.007732719648629427,
      "learning_rate": 3.0555098684210525e-06,
      "loss": 0.0004,
      "step": 16900
    },
    {
      "epoch": 6.953125,
      "grad_norm": 1.8991835117340088,
      "learning_rate": 3.05139802631579e-06,
      "loss": 0.0041,
      "step": 16910
    },
    {
      "epoch": 6.957236842105263,
      "grad_norm": 0.32865986227989197,
      "learning_rate": 3.0472861842105265e-06,
      "loss": 0.0413,
      "step": 16920
    },
    {
      "epoch": 6.9613486842105265,
      "grad_norm": 2.4880166053771973,
      "learning_rate": 3.0431743421052636e-06,
      "loss": 0.0635,
      "step": 16930
    },
    {
      "epoch": 6.965460526315789,
      "grad_norm": 17.090551376342773,
      "learning_rate": 3.0390625e-06,
      "loss": 0.2446,
      "step": 16940
    },
    {
      "epoch": 6.969572368421053,
      "grad_norm": 0.25987255573272705,
      "learning_rate": 3.0349506578947368e-06,
      "loss": 0.3464,
      "step": 16950
    },
    {
      "epoch": 6.973684210526316,
      "grad_norm": 0.030250955373048782,
      "learning_rate": 3.0308388157894742e-06,
      "loss": 0.1271,
      "step": 16960
    },
    {
      "epoch": 6.977796052631579,
      "grad_norm": 1.7025675773620605,
      "learning_rate": 3.026726973684211e-06,
      "loss": 0.0192,
      "step": 16970
    },
    {
      "epoch": 6.9819078947368425,
      "grad_norm": 0.005293203983455896,
      "learning_rate": 3.0226151315789475e-06,
      "loss": 0.0584,
      "step": 16980
    },
    {
      "epoch": 6.986019736842105,
      "grad_norm": 0.010041214525699615,
      "learning_rate": 3.0185032894736845e-06,
      "loss": 0.1611,
      "step": 16990
    },
    {
      "epoch": 6.990131578947368,
      "grad_norm": 0.4960373342037201,
      "learning_rate": 3.014391447368421e-06,
      "loss": 0.0782,
      "step": 17000
    },
    {
      "epoch": 6.994243421052632,
      "grad_norm": 0.0012437421828508377,
      "learning_rate": 3.010279605263158e-06,
      "loss": 0.0094,
      "step": 17010
    },
    {
      "epoch": 6.998355263157895,
      "grad_norm": 0.10035240650177002,
      "learning_rate": 3.006167763157895e-06,
      "loss": 0.0813,
      "step": 17020
    },
    {
      "epoch": 7.0024671052631575,
      "grad_norm": 0.008433901704847813,
      "learning_rate": 3.0020559210526318e-06,
      "loss": 0.0069,
      "step": 17030
    },
    {
      "epoch": 7.006578947368421,
      "grad_norm": 0.015171228908002377,
      "learning_rate": 2.9979440789473684e-06,
      "loss": 0.0541,
      "step": 17040
    },
    {
      "epoch": 7.010690789473684,
      "grad_norm": 0.0029152543283998966,
      "learning_rate": 2.993832236842106e-06,
      "loss": 0.0029,
      "step": 17050
    },
    {
      "epoch": 7.014802631578948,
      "grad_norm": 60.9827880859375,
      "learning_rate": 2.9897203947368424e-06,
      "loss": 0.0344,
      "step": 17060
    },
    {
      "epoch": 7.018914473684211,
      "grad_norm": 0.023271074518561363,
      "learning_rate": 2.985608552631579e-06,
      "loss": 0.0913,
      "step": 17070
    },
    {
      "epoch": 7.0230263157894735,
      "grad_norm": 1.9478614330291748,
      "learning_rate": 2.981496710526316e-06,
      "loss": 0.0033,
      "step": 17080
    },
    {
      "epoch": 7.027138157894737,
      "grad_norm": 0.0011102440766990185,
      "learning_rate": 2.9773848684210527e-06,
      "loss": 0.0003,
      "step": 17090
    },
    {
      "epoch": 7.03125,
      "grad_norm": 0.15013249218463898,
      "learning_rate": 2.97327302631579e-06,
      "loss": 0.037,
      "step": 17100
    },
    {
      "epoch": 7.035361842105263,
      "grad_norm": 27.406702041625977,
      "learning_rate": 2.9691611842105268e-06,
      "loss": 0.0286,
      "step": 17110
    },
    {
      "epoch": 7.0394736842105265,
      "grad_norm": 21.793682098388672,
      "learning_rate": 2.9650493421052634e-06,
      "loss": 0.007,
      "step": 17120
    },
    {
      "epoch": 7.043585526315789,
      "grad_norm": 17.29696273803711,
      "learning_rate": 2.9609375e-06,
      "loss": 0.0202,
      "step": 17130
    },
    {
      "epoch": 7.047697368421052,
      "grad_norm": 0.0060730380937457085,
      "learning_rate": 2.956825657894737e-06,
      "loss": 0.1069,
      "step": 17140
    },
    {
      "epoch": 7.051809210526316,
      "grad_norm": 15.568187713623047,
      "learning_rate": 2.952713815789474e-06,
      "loss": 0.007,
      "step": 17150
    },
    {
      "epoch": 7.055921052631579,
      "grad_norm": 0.008214758709073067,
      "learning_rate": 2.948601973684211e-06,
      "loss": 0.0032,
      "step": 17160
    },
    {
      "epoch": 7.0600328947368425,
      "grad_norm": 0.024761829525232315,
      "learning_rate": 2.9444901315789477e-06,
      "loss": 0.0378,
      "step": 17170
    },
    {
      "epoch": 7.064144736842105,
      "grad_norm": 2.6922731194645166e-05,
      "learning_rate": 2.9403782894736843e-06,
      "loss": 0.0021,
      "step": 17180
    },
    {
      "epoch": 7.068256578947368,
      "grad_norm": 0.38131505250930786,
      "learning_rate": 2.936266447368421e-06,
      "loss": 0.0004,
      "step": 17190
    },
    {
      "epoch": 7.072368421052632,
      "grad_norm": 12.381378173828125,
      "learning_rate": 2.9321546052631584e-06,
      "loss": 0.0054,
      "step": 17200
    },
    {
      "epoch": 7.076480263157895,
      "grad_norm": 0.8414236903190613,
      "learning_rate": 2.928042763157895e-06,
      "loss": 0.0006,
      "step": 17210
    },
    {
      "epoch": 7.0805921052631575,
      "grad_norm": 3.8690497875213623,
      "learning_rate": 2.9239309210526316e-06,
      "loss": 0.0018,
      "step": 17220
    },
    {
      "epoch": 7.084703947368421,
      "grad_norm": 0.09639491885900497,
      "learning_rate": 2.9198190789473686e-06,
      "loss": 0.1549,
      "step": 17230
    },
    {
      "epoch": 7.088815789473684,
      "grad_norm": 1.0091197490692139,
      "learning_rate": 2.9157072368421056e-06,
      "loss": 0.1204,
      "step": 17240
    },
    {
      "epoch": 7.092927631578948,
      "grad_norm": 0.03747154399752617,
      "learning_rate": 2.9115953947368427e-06,
      "loss": 0.0481,
      "step": 17250
    },
    {
      "epoch": 7.097039473684211,
      "grad_norm": 0.06905113905668259,
      "learning_rate": 2.9074835526315793e-06,
      "loss": 0.1313,
      "step": 17260
    },
    {
      "epoch": 7.1011513157894735,
      "grad_norm": 76.20909881591797,
      "learning_rate": 2.903371710526316e-06,
      "loss": 0.1095,
      "step": 17270
    },
    {
      "epoch": 7.105263157894737,
      "grad_norm": 0.022789154201745987,
      "learning_rate": 2.8992598684210525e-06,
      "loss": 0.0382,
      "step": 17280
    },
    {
      "epoch": 7.109375,
      "grad_norm": 0.2253035455942154,
      "learning_rate": 2.89514802631579e-06,
      "loss": 0.0017,
      "step": 17290
    },
    {
      "epoch": 7.113486842105263,
      "grad_norm": 0.5477920174598694,
      "learning_rate": 2.8910361842105266e-06,
      "loss": 0.0379,
      "step": 17300
    },
    {
      "epoch": 7.1175986842105265,
      "grad_norm": 0.000436304573668167,
      "learning_rate": 2.8869243421052636e-06,
      "loss": 0.2187,
      "step": 17310
    },
    {
      "epoch": 7.121710526315789,
      "grad_norm": 0.0398329421877861,
      "learning_rate": 2.8828125e-06,
      "loss": 0.0004,
      "step": 17320
    },
    {
      "epoch": 7.125822368421052,
      "grad_norm": 0.03360723704099655,
      "learning_rate": 2.878700657894737e-06,
      "loss": 0.0016,
      "step": 17330
    },
    {
      "epoch": 7.129934210526316,
      "grad_norm": 0.07153233885765076,
      "learning_rate": 2.8745888157894743e-06,
      "loss": 0.0569,
      "step": 17340
    },
    {
      "epoch": 7.134046052631579,
      "grad_norm": 89.36862182617188,
      "learning_rate": 2.870476973684211e-06,
      "loss": 0.0574,
      "step": 17350
    },
    {
      "epoch": 7.1381578947368425,
      "grad_norm": 8.033440589904785,
      "learning_rate": 2.8663651315789475e-06,
      "loss": 0.0031,
      "step": 17360
    },
    {
      "epoch": 7.142269736842105,
      "grad_norm": 0.07727915793657303,
      "learning_rate": 2.8622532894736845e-06,
      "loss": 0.0011,
      "step": 17370
    },
    {
      "epoch": 7.146381578947368,
      "grad_norm": 0.718957781791687,
      "learning_rate": 2.858141447368421e-06,
      "loss": 0.0052,
      "step": 17380
    },
    {
      "epoch": 7.150493421052632,
      "grad_norm": 0.12261500954627991,
      "learning_rate": 2.854029605263158e-06,
      "loss": 0.1109,
      "step": 17390
    },
    {
      "epoch": 7.154605263157895,
      "grad_norm": 0.0148479538038373,
      "learning_rate": 2.849917763157895e-06,
      "loss": 0.0013,
      "step": 17400
    },
    {
      "epoch": 7.1587171052631575,
      "grad_norm": 0.01821858249604702,
      "learning_rate": 2.8458059210526318e-06,
      "loss": 0.0625,
      "step": 17410
    },
    {
      "epoch": 7.162828947368421,
      "grad_norm": 0.0008887009462341666,
      "learning_rate": 2.8416940789473684e-06,
      "loss": 0.022,
      "step": 17420
    },
    {
      "epoch": 7.166940789473684,
      "grad_norm": 0.004944947548210621,
      "learning_rate": 2.837582236842106e-06,
      "loss": 0.0002,
      "step": 17430
    },
    {
      "epoch": 7.171052631578948,
      "grad_norm": 0.042470525950193405,
      "learning_rate": 2.8334703947368425e-06,
      "loss": 0.0414,
      "step": 17440
    },
    {
      "epoch": 7.175164473684211,
      "grad_norm": 0.09561196714639664,
      "learning_rate": 2.8297697368421057e-06,
      "loss": 0.1359,
      "step": 17450
    },
    {
      "epoch": 7.1792763157894735,
      "grad_norm": 0.0038538984954357147,
      "learning_rate": 2.8256578947368423e-06,
      "loss": 0.0158,
      "step": 17460
    },
    {
      "epoch": 7.183388157894737,
      "grad_norm": 0.00038770309765823185,
      "learning_rate": 2.8215460526315793e-06,
      "loss": 0.1407,
      "step": 17470
    },
    {
      "epoch": 7.1875,
      "grad_norm": 0.02665071003139019,
      "learning_rate": 2.817434210526316e-06,
      "loss": 0.2346,
      "step": 17480
    },
    {
      "epoch": 7.191611842105263,
      "grad_norm": 0.09127389639616013,
      "learning_rate": 2.813322368421053e-06,
      "loss": 0.0522,
      "step": 17490
    },
    {
      "epoch": 7.1957236842105265,
      "grad_norm": 0.006527309771627188,
      "learning_rate": 2.80921052631579e-06,
      "loss": 0.0368,
      "step": 17500
    },
    {
      "epoch": 7.199835526315789,
      "grad_norm": 0.01165294460952282,
      "learning_rate": 2.8050986842105266e-06,
      "loss": 0.0099,
      "step": 17510
    },
    {
      "epoch": 7.203947368421052,
      "grad_norm": 0.053283970803022385,
      "learning_rate": 2.800986842105263e-06,
      "loss": 0.0007,
      "step": 17520
    },
    {
      "epoch": 7.208059210526316,
      "grad_norm": 0.0038871471770107746,
      "learning_rate": 2.796875e-06,
      "loss": 0.0005,
      "step": 17530
    },
    {
      "epoch": 7.212171052631579,
      "grad_norm": 0.072207972407341,
      "learning_rate": 2.7927631578947373e-06,
      "loss": 0.1669,
      "step": 17540
    },
    {
      "epoch": 7.2162828947368425,
      "grad_norm": 21.420900344848633,
      "learning_rate": 2.788651315789474e-06,
      "loss": 0.1412,
      "step": 17550
    },
    {
      "epoch": 7.220394736842105,
      "grad_norm": 0.31269371509552,
      "learning_rate": 2.784539473684211e-06,
      "loss": 0.0002,
      "step": 17560
    },
    {
      "epoch": 7.224506578947368,
      "grad_norm": 0.005971134640276432,
      "learning_rate": 2.7804276315789475e-06,
      "loss": 0.0058,
      "step": 17570
    },
    {
      "epoch": 7.228618421052632,
      "grad_norm": 0.02348402515053749,
      "learning_rate": 2.776315789473684e-06,
      "loss": 0.0503,
      "step": 17580
    },
    {
      "epoch": 7.232730263157895,
      "grad_norm": 0.0007353632245212793,
      "learning_rate": 2.7722039473684216e-06,
      "loss": 0.1017,
      "step": 17590
    },
    {
      "epoch": 7.2368421052631575,
      "grad_norm": 0.0007847705273889005,
      "learning_rate": 2.768092105263158e-06,
      "loss": 0.0003,
      "step": 17600
    },
    {
      "epoch": 7.240953947368421,
      "grad_norm": 0.3893216848373413,
      "learning_rate": 2.7639802631578948e-06,
      "loss": 0.056,
      "step": 17610
    },
    {
      "epoch": 7.245065789473684,
      "grad_norm": 0.0016018354799598455,
      "learning_rate": 2.759868421052632e-06,
      "loss": 0.0792,
      "step": 17620
    },
    {
      "epoch": 7.249177631578948,
      "grad_norm": 0.056312862783670425,
      "learning_rate": 2.755756578947369e-06,
      "loss": 0.3022,
      "step": 17630
    },
    {
      "epoch": 7.253289473684211,
      "grad_norm": 1.7281204462051392,
      "learning_rate": 2.7516447368421055e-06,
      "loss": 0.0159,
      "step": 17640
    },
    {
      "epoch": 7.2574013157894735,
      "grad_norm": 0.3130502700805664,
      "learning_rate": 2.7475328947368425e-06,
      "loss": 0.002,
      "step": 17650
    },
    {
      "epoch": 7.261513157894737,
      "grad_norm": 47.450931549072266,
      "learning_rate": 2.743421052631579e-06,
      "loss": 0.0829,
      "step": 17660
    },
    {
      "epoch": 7.265625,
      "grad_norm": 26.058509826660156,
      "learning_rate": 2.7393092105263157e-06,
      "loss": 0.0375,
      "step": 17670
    },
    {
      "epoch": 7.269736842105263,
      "grad_norm": 0.39331546425819397,
      "learning_rate": 2.735197368421053e-06,
      "loss": 0.0011,
      "step": 17680
    },
    {
      "epoch": 7.2738486842105265,
      "grad_norm": 6.446929910453036e-05,
      "learning_rate": 2.7310855263157898e-06,
      "loss": 0.0001,
      "step": 17690
    },
    {
      "epoch": 7.277960526315789,
      "grad_norm": 0.02226644940674305,
      "learning_rate": 2.7269736842105264e-06,
      "loss": 0.032,
      "step": 17700
    },
    {
      "epoch": 7.282072368421053,
      "grad_norm": 69.86528778076172,
      "learning_rate": 2.7228618421052634e-06,
      "loss": 0.0477,
      "step": 17710
    },
    {
      "epoch": 7.286184210526316,
      "grad_norm": 0.06220513954758644,
      "learning_rate": 2.71875e-06,
      "loss": 0.0539,
      "step": 17720
    },
    {
      "epoch": 7.290296052631579,
      "grad_norm": 66.28457641601562,
      "learning_rate": 2.7146381578947375e-06,
      "loss": 0.0337,
      "step": 17730
    },
    {
      "epoch": 7.2944078947368425,
      "grad_norm": 0.001533765229396522,
      "learning_rate": 2.710526315789474e-06,
      "loss": 0.0013,
      "step": 17740
    },
    {
      "epoch": 7.298519736842105,
      "grad_norm": 0.00953629519790411,
      "learning_rate": 2.7064144736842107e-06,
      "loss": 0.0065,
      "step": 17750
    },
    {
      "epoch": 7.302631578947368,
      "grad_norm": 0.0008454016642645001,
      "learning_rate": 2.7023026315789473e-06,
      "loss": 0.0112,
      "step": 17760
    },
    {
      "epoch": 7.306743421052632,
      "grad_norm": 81.20194244384766,
      "learning_rate": 2.6981907894736843e-06,
      "loss": 0.3135,
      "step": 17770
    },
    {
      "epoch": 7.310855263157895,
      "grad_norm": 0.013238240964710712,
      "learning_rate": 2.6940789473684214e-06,
      "loss": 0.0044,
      "step": 17780
    },
    {
      "epoch": 7.3149671052631575,
      "grad_norm": 0.06490977853536606,
      "learning_rate": 2.6899671052631584e-06,
      "loss": 0.0383,
      "step": 17790
    },
    {
      "epoch": 7.319078947368421,
      "grad_norm": 21.222938537597656,
      "learning_rate": 2.685855263157895e-06,
      "loss": 0.0087,
      "step": 17800
    },
    {
      "epoch": 7.323190789473684,
      "grad_norm": 0.05870138481259346,
      "learning_rate": 2.6817434210526316e-06,
      "loss": 0.0385,
      "step": 17810
    },
    {
      "epoch": 7.327302631578947,
      "grad_norm": 0.10026119649410248,
      "learning_rate": 2.677631578947369e-06,
      "loss": 0.009,
      "step": 17820
    },
    {
      "epoch": 7.331414473684211,
      "grad_norm": 0.018641915172338486,
      "learning_rate": 2.6735197368421057e-06,
      "loss": 0.0752,
      "step": 17830
    },
    {
      "epoch": 7.3355263157894735,
      "grad_norm": 25.40374755859375,
      "learning_rate": 2.6694078947368423e-06,
      "loss": 0.0081,
      "step": 17840
    },
    {
      "epoch": 7.339638157894737,
      "grad_norm": 0.01974809169769287,
      "learning_rate": 2.665296052631579e-06,
      "loss": 0.0384,
      "step": 17850
    },
    {
      "epoch": 7.34375,
      "grad_norm": 0.013562479056417942,
      "learning_rate": 2.661184210526316e-06,
      "loss": 0.0005,
      "step": 17860
    },
    {
      "epoch": 7.347861842105263,
      "grad_norm": 0.0006354886572808027,
      "learning_rate": 2.657072368421053e-06,
      "loss": 0.0303,
      "step": 17870
    },
    {
      "epoch": 7.3519736842105265,
      "grad_norm": 0.045913152396678925,
      "learning_rate": 2.65296052631579e-06,
      "loss": 0.0004,
      "step": 17880
    },
    {
      "epoch": 7.356085526315789,
      "grad_norm": 0.5759618878364563,
      "learning_rate": 2.6488486842105266e-06,
      "loss": 0.0016,
      "step": 17890
    },
    {
      "epoch": 7.360197368421053,
      "grad_norm": 0.031576257199048996,
      "learning_rate": 2.644736842105263e-06,
      "loss": 0.0129,
      "step": 17900
    },
    {
      "epoch": 7.364309210526316,
      "grad_norm": 0.002541609574109316,
      "learning_rate": 2.640625e-06,
      "loss": 0.0144,
      "step": 17910
    },
    {
      "epoch": 7.368421052631579,
      "grad_norm": 11.867419242858887,
      "learning_rate": 2.6365131578947373e-06,
      "loss": 0.0057,
      "step": 17920
    },
    {
      "epoch": 7.3725328947368425,
      "grad_norm": 0.06471031159162521,
      "learning_rate": 2.632401315789474e-06,
      "loss": 0.03,
      "step": 17930
    },
    {
      "epoch": 7.376644736842105,
      "grad_norm": 0.12563598155975342,
      "learning_rate": 2.628289473684211e-06,
      "loss": 0.0092,
      "step": 17940
    },
    {
      "epoch": 7.380756578947368,
      "grad_norm": 0.03712340444326401,
      "learning_rate": 2.6241776315789475e-06,
      "loss": 0.0716,
      "step": 17950
    },
    {
      "epoch": 7.384868421052632,
      "grad_norm": 0.05920384079217911,
      "learning_rate": 2.620065789473684e-06,
      "loss": 0.0106,
      "step": 17960
    },
    {
      "epoch": 7.388980263157895,
      "grad_norm": 20.164339065551758,
      "learning_rate": 2.6159539473684216e-06,
      "loss": 0.2053,
      "step": 17970
    },
    {
      "epoch": 7.3930921052631575,
      "grad_norm": 0.33210375905036926,
      "learning_rate": 2.611842105263158e-06,
      "loss": 0.0211,
      "step": 17980
    },
    {
      "epoch": 7.397203947368421,
      "grad_norm": 0.42094624042510986,
      "learning_rate": 2.607730263157895e-06,
      "loss": 0.0053,
      "step": 17990
    },
    {
      "epoch": 7.401315789473684,
      "grad_norm": 5.063699722290039,
      "learning_rate": 2.603618421052632e-06,
      "loss": 0.1868,
      "step": 18000
    },
    {
      "epoch": 7.405427631578947,
      "grad_norm": 0.022125262767076492,
      "learning_rate": 2.599506578947369e-06,
      "loss": 0.1222,
      "step": 18010
    },
    {
      "epoch": 7.409539473684211,
      "grad_norm": 0.017482653260231018,
      "learning_rate": 2.5953947368421055e-06,
      "loss": 0.056,
      "step": 18020
    },
    {
      "epoch": 7.4136513157894735,
      "grad_norm": 0.00026524471468292177,
      "learning_rate": 2.5912828947368425e-06,
      "loss": 0.0495,
      "step": 18030
    },
    {
      "epoch": 7.417763157894737,
      "grad_norm": 59.36870574951172,
      "learning_rate": 2.587171052631579e-06,
      "loss": 0.0286,
      "step": 18040
    },
    {
      "epoch": 7.421875,
      "grad_norm": 0.0069809104315936565,
      "learning_rate": 2.5830592105263157e-06,
      "loss": 0.0115,
      "step": 18050
    },
    {
      "epoch": 7.425986842105263,
      "grad_norm": 0.05242922902107239,
      "learning_rate": 2.578947368421053e-06,
      "loss": 0.0008,
      "step": 18060
    },
    {
      "epoch": 7.4300986842105265,
      "grad_norm": 0.04204375296831131,
      "learning_rate": 2.5748355263157898e-06,
      "loss": 0.0118,
      "step": 18070
    },
    {
      "epoch": 7.434210526315789,
      "grad_norm": 0.1886630803346634,
      "learning_rate": 2.5707236842105264e-06,
      "loss": 0.0003,
      "step": 18080
    },
    {
      "epoch": 7.438322368421053,
      "grad_norm": 0.8755178451538086,
      "learning_rate": 2.5666118421052634e-06,
      "loss": 0.0237,
      "step": 18090
    },
    {
      "epoch": 7.442434210526316,
      "grad_norm": 8.999763488769531,
      "learning_rate": 2.5625e-06,
      "loss": 0.0263,
      "step": 18100
    },
    {
      "epoch": 7.446546052631579,
      "grad_norm": 0.20764632523059845,
      "learning_rate": 2.558388157894737e-06,
      "loss": 0.0012,
      "step": 18110
    },
    {
      "epoch": 7.4506578947368425,
      "grad_norm": 0.0067911092191934586,
      "learning_rate": 2.554276315789474e-06,
      "loss": 0.1123,
      "step": 18120
    },
    {
      "epoch": 7.454769736842105,
      "grad_norm": 0.9484823346138,
      "learning_rate": 2.5501644736842107e-06,
      "loss": 0.034,
      "step": 18130
    },
    {
      "epoch": 7.458881578947368,
      "grad_norm": 0.0008993869996629655,
      "learning_rate": 2.5460526315789473e-06,
      "loss": 0.0037,
      "step": 18140
    },
    {
      "epoch": 7.462993421052632,
      "grad_norm": 0.0014803741360083222,
      "learning_rate": 2.5419407894736843e-06,
      "loss": 0.5715,
      "step": 18150
    },
    {
      "epoch": 7.467105263157895,
      "grad_norm": 0.00023283193877432495,
      "learning_rate": 2.5378289473684214e-06,
      "loss": 0.0531,
      "step": 18160
    },
    {
      "epoch": 7.4712171052631575,
      "grad_norm": 0.002006068592891097,
      "learning_rate": 2.533717105263158e-06,
      "loss": 0.0066,
      "step": 18170
    },
    {
      "epoch": 7.475328947368421,
      "grad_norm": 0.025551754981279373,
      "learning_rate": 2.529605263157895e-06,
      "loss": 0.0429,
      "step": 18180
    },
    {
      "epoch": 7.479440789473684,
      "grad_norm": 0.07115461677312851,
      "learning_rate": 2.5254934210526316e-06,
      "loss": 0.0069,
      "step": 18190
    },
    {
      "epoch": 7.483552631578947,
      "grad_norm": 0.29560112953186035,
      "learning_rate": 2.521381578947369e-06,
      "loss": 0.0207,
      "step": 18200
    },
    {
      "epoch": 7.487664473684211,
      "grad_norm": 0.2443089634180069,
      "learning_rate": 2.5172697368421057e-06,
      "loss": 0.0065,
      "step": 18210
    },
    {
      "epoch": 7.4917763157894735,
      "grad_norm": 0.015790261328220367,
      "learning_rate": 2.5131578947368423e-06,
      "loss": 0.1727,
      "step": 18220
    },
    {
      "epoch": 7.495888157894737,
      "grad_norm": 0.0037406887859106064,
      "learning_rate": 2.509046052631579e-06,
      "loss": 0.0239,
      "step": 18230
    },
    {
      "epoch": 7.5,
      "grad_norm": 0.0033218811731785536,
      "learning_rate": 2.504934210526316e-06,
      "loss": 0.0043,
      "step": 18240
    },
    {
      "epoch": 7.504111842105263,
      "grad_norm": 4.251826763153076,
      "learning_rate": 2.500822368421053e-06,
      "loss": 0.0046,
      "step": 18250
    },
    {
      "epoch": 7.5082236842105265,
      "grad_norm": 0.12638850510120392,
      "learning_rate": 2.49671052631579e-06,
      "loss": 0.001,
      "step": 18260
    },
    {
      "epoch": 7.512335526315789,
      "grad_norm": 0.43785837292671204,
      "learning_rate": 2.4925986842105266e-06,
      "loss": 0.0017,
      "step": 18270
    },
    {
      "epoch": 7.516447368421053,
      "grad_norm": 95.06336212158203,
      "learning_rate": 2.4884868421052632e-06,
      "loss": 0.3436,
      "step": 18280
    },
    {
      "epoch": 7.520559210526316,
      "grad_norm": 0.06603921949863434,
      "learning_rate": 2.4843750000000002e-06,
      "loss": 0.0048,
      "step": 18290
    },
    {
      "epoch": 7.524671052631579,
      "grad_norm": 29.374561309814453,
      "learning_rate": 2.480263157894737e-06,
      "loss": 0.1906,
      "step": 18300
    },
    {
      "epoch": 7.5287828947368425,
      "grad_norm": 0.49131685495376587,
      "learning_rate": 2.476151315789474e-06,
      "loss": 0.0013,
      "step": 18310
    },
    {
      "epoch": 7.532894736842105,
      "grad_norm": 0.01167258806526661,
      "learning_rate": 2.472039473684211e-06,
      "loss": 0.0053,
      "step": 18320
    },
    {
      "epoch": 7.537006578947368,
      "grad_norm": 0.024763192981481552,
      "learning_rate": 2.4679276315789475e-06,
      "loss": 0.0126,
      "step": 18330
    },
    {
      "epoch": 7.541118421052632,
      "grad_norm": 0.02219541370868683,
      "learning_rate": 2.4638157894736846e-06,
      "loss": 0.0045,
      "step": 18340
    },
    {
      "epoch": 7.545230263157895,
      "grad_norm": 0.0026065174024552107,
      "learning_rate": 2.459703947368421e-06,
      "loss": 0.0004,
      "step": 18350
    },
    {
      "epoch": 7.5493421052631575,
      "grad_norm": 2.9072539806365967,
      "learning_rate": 2.455592105263158e-06,
      "loss": 0.1266,
      "step": 18360
    },
    {
      "epoch": 7.553453947368421,
      "grad_norm": 0.2119685709476471,
      "learning_rate": 2.451480263157895e-06,
      "loss": 0.0018,
      "step": 18370
    },
    {
      "epoch": 7.557565789473684,
      "grad_norm": 0.914875864982605,
      "learning_rate": 2.447368421052632e-06,
      "loss": 0.1778,
      "step": 18380
    },
    {
      "epoch": 7.561677631578947,
      "grad_norm": 2.1064213342469884e-06,
      "learning_rate": 2.4432565789473684e-06,
      "loss": 0.0,
      "step": 18390
    },
    {
      "epoch": 7.565789473684211,
      "grad_norm": 0.06288709491491318,
      "learning_rate": 2.4391447368421055e-06,
      "loss": 0.001,
      "step": 18400
    },
    {
      "epoch": 7.5699013157894735,
      "grad_norm": 0.0016081187641248107,
      "learning_rate": 2.4350328947368425e-06,
      "loss": 0.1154,
      "step": 18410
    },
    {
      "epoch": 7.574013157894737,
      "grad_norm": 0.10475052893161774,
      "learning_rate": 2.430921052631579e-06,
      "loss": 0.0003,
      "step": 18420
    },
    {
      "epoch": 7.578125,
      "grad_norm": 0.010752451606094837,
      "learning_rate": 2.426809210526316e-06,
      "loss": 0.0367,
      "step": 18430
    },
    {
      "epoch": 7.582236842105263,
      "grad_norm": 0.016824735328555107,
      "learning_rate": 2.4226973684210528e-06,
      "loss": 0.0011,
      "step": 18440
    },
    {
      "epoch": 7.5863486842105265,
      "grad_norm": 0.20019060373306274,
      "learning_rate": 2.41858552631579e-06,
      "loss": 0.0791,
      "step": 18450
    },
    {
      "epoch": 7.590460526315789,
      "grad_norm": 0.0019403524929657578,
      "learning_rate": 2.4144736842105264e-06,
      "loss": 0.0171,
      "step": 18460
    },
    {
      "epoch": 7.594572368421053,
      "grad_norm": 0.0023165084421634674,
      "learning_rate": 2.4103618421052634e-06,
      "loss": 0.1922,
      "step": 18470
    },
    {
      "epoch": 7.598684210526316,
      "grad_norm": 40.954689025878906,
      "learning_rate": 2.40625e-06,
      "loss": 0.0143,
      "step": 18480
    },
    {
      "epoch": 7.602796052631579,
      "grad_norm": 0.0013854552526026964,
      "learning_rate": 2.402138157894737e-06,
      "loss": 0.0003,
      "step": 18490
    },
    {
      "epoch": 7.6069078947368425,
      "grad_norm": 0.0016220198012888432,
      "learning_rate": 2.398026315789474e-06,
      "loss": 0.4475,
      "step": 18500
    },
    {
      "epoch": 7.611019736842105,
      "grad_norm": 0.048832572996616364,
      "learning_rate": 2.3939144736842107e-06,
      "loss": 0.0003,
      "step": 18510
    },
    {
      "epoch": 7.615131578947368,
      "grad_norm": 5.190423488616943,
      "learning_rate": 2.3898026315789473e-06,
      "loss": 0.0027,
      "step": 18520
    },
    {
      "epoch": 7.619243421052632,
      "grad_norm": 0.9031723737716675,
      "learning_rate": 2.3856907894736844e-06,
      "loss": 0.0028,
      "step": 18530
    },
    {
      "epoch": 7.623355263157895,
      "grad_norm": 0.02315152995288372,
      "learning_rate": 2.381578947368421e-06,
      "loss": 0.1872,
      "step": 18540
    },
    {
      "epoch": 7.6274671052631575,
      "grad_norm": 0.025038328021764755,
      "learning_rate": 2.377467105263158e-06,
      "loss": 0.0054,
      "step": 18550
    },
    {
      "epoch": 7.631578947368421,
      "grad_norm": 0.27843013405799866,
      "learning_rate": 2.373355263157895e-06,
      "loss": 0.0178,
      "step": 18560
    },
    {
      "epoch": 7.635690789473684,
      "grad_norm": 0.24566157162189484,
      "learning_rate": 2.369243421052632e-06,
      "loss": 0.0026,
      "step": 18570
    },
    {
      "epoch": 7.639802631578947,
      "grad_norm": 0.00402486976236105,
      "learning_rate": 2.3651315789473687e-06,
      "loss": 0.0115,
      "step": 18580
    },
    {
      "epoch": 7.643914473684211,
      "grad_norm": 0.11018090695142746,
      "learning_rate": 2.3610197368421053e-06,
      "loss": 0.0342,
      "step": 18590
    },
    {
      "epoch": 7.6480263157894735,
      "grad_norm": 0.10057216137647629,
      "learning_rate": 2.3569078947368423e-06,
      "loss": 0.0003,
      "step": 18600
    },
    {
      "epoch": 7.652138157894737,
      "grad_norm": 0.00027278883499093354,
      "learning_rate": 2.352796052631579e-06,
      "loss": 0.0003,
      "step": 18610
    },
    {
      "epoch": 7.65625,
      "grad_norm": 56.96162796020508,
      "learning_rate": 2.348684210526316e-06,
      "loss": 0.0221,
      "step": 18620
    },
    {
      "epoch": 7.660361842105263,
      "grad_norm": 0.32547372579574585,
      "learning_rate": 2.344572368421053e-06,
      "loss": 0.0014,
      "step": 18630
    },
    {
      "epoch": 7.6644736842105265,
      "grad_norm": 0.0017805839888751507,
      "learning_rate": 2.3404605263157896e-06,
      "loss": 0.001,
      "step": 18640
    },
    {
      "epoch": 7.668585526315789,
      "grad_norm": 0.003879399737343192,
      "learning_rate": 2.3363486842105266e-06,
      "loss": 0.0014,
      "step": 18650
    },
    {
      "epoch": 7.672697368421053,
      "grad_norm": 0.0014848548453301191,
      "learning_rate": 2.3322368421052632e-06,
      "loss": 0.0116,
      "step": 18660
    },
    {
      "epoch": 7.676809210526316,
      "grad_norm": 12.23731517791748,
      "learning_rate": 2.3281250000000003e-06,
      "loss": 0.0269,
      "step": 18670
    },
    {
      "epoch": 7.680921052631579,
      "grad_norm": 0.0011852079769596457,
      "learning_rate": 2.324013157894737e-06,
      "loss": 0.0164,
      "step": 18680
    },
    {
      "epoch": 7.6850328947368425,
      "grad_norm": 0.19763517379760742,
      "learning_rate": 2.319901315789474e-06,
      "loss": 0.0009,
      "step": 18690
    },
    {
      "epoch": 7.689144736842105,
      "grad_norm": 0.48111191391944885,
      "learning_rate": 2.3157894736842105e-06,
      "loss": 0.0044,
      "step": 18700
    },
    {
      "epoch": 7.693256578947368,
      "grad_norm": 35.88668441772461,
      "learning_rate": 2.3116776315789475e-06,
      "loss": 0.018,
      "step": 18710
    },
    {
      "epoch": 7.697368421052632,
      "grad_norm": 0.4510961174964905,
      "learning_rate": 2.3075657894736846e-06,
      "loss": 0.0201,
      "step": 18720
    },
    {
      "epoch": 7.701480263157895,
      "grad_norm": 0.5701677799224854,
      "learning_rate": 2.303453947368421e-06,
      "loss": 0.0481,
      "step": 18730
    },
    {
      "epoch": 7.7055921052631575,
      "grad_norm": 0.023179659619927406,
      "learning_rate": 2.2993421052631582e-06,
      "loss": 0.0016,
      "step": 18740
    },
    {
      "epoch": 7.709703947368421,
      "grad_norm": 22.315807342529297,
      "learning_rate": 2.295230263157895e-06,
      "loss": 0.0161,
      "step": 18750
    },
    {
      "epoch": 7.713815789473684,
      "grad_norm": 0.0018141778418794274,
      "learning_rate": 2.291118421052632e-06,
      "loss": 0.054,
      "step": 18760
    },
    {
      "epoch": 7.717927631578947,
      "grad_norm": 0.030514366924762726,
      "learning_rate": 2.2870065789473685e-06,
      "loss": 0.0029,
      "step": 18770
    },
    {
      "epoch": 7.722039473684211,
      "grad_norm": 55.70794677734375,
      "learning_rate": 2.2828947368421055e-06,
      "loss": 0.0208,
      "step": 18780
    },
    {
      "epoch": 7.7261513157894735,
      "grad_norm": 2.907081127166748,
      "learning_rate": 2.2787828947368425e-06,
      "loss": 0.1681,
      "step": 18790
    },
    {
      "epoch": 7.730263157894737,
      "grad_norm": 6.280221462249756,
      "learning_rate": 2.274671052631579e-06,
      "loss": 0.0125,
      "step": 18800
    },
    {
      "epoch": 7.734375,
      "grad_norm": 0.02076088637113571,
      "learning_rate": 2.270559210526316e-06,
      "loss": 0.0321,
      "step": 18810
    },
    {
      "epoch": 7.738486842105263,
      "grad_norm": 119.19054412841797,
      "learning_rate": 2.2664473684210528e-06,
      "loss": 0.2288,
      "step": 18820
    },
    {
      "epoch": 7.7425986842105265,
      "grad_norm": 0.0002872719196602702,
      "learning_rate": 2.26233552631579e-06,
      "loss": 0.0028,
      "step": 18830
    },
    {
      "epoch": 7.746710526315789,
      "grad_norm": 0.004000790882855654,
      "learning_rate": 2.2582236842105264e-06,
      "loss": 0.0001,
      "step": 18840
    },
    {
      "epoch": 7.750822368421053,
      "grad_norm": 8.02085018157959,
      "learning_rate": 2.2541118421052634e-06,
      "loss": 0.1302,
      "step": 18850
    },
    {
      "epoch": 7.754934210526316,
      "grad_norm": 0.011402469128370285,
      "learning_rate": 2.25e-06,
      "loss": 0.1687,
      "step": 18860
    },
    {
      "epoch": 7.759046052631579,
      "grad_norm": 0.03788752481341362,
      "learning_rate": 2.245888157894737e-06,
      "loss": 0.0934,
      "step": 18870
    },
    {
      "epoch": 7.7631578947368425,
      "grad_norm": 0.01945221796631813,
      "learning_rate": 2.241776315789474e-06,
      "loss": 0.0411,
      "step": 18880
    },
    {
      "epoch": 7.767269736842105,
      "grad_norm": 0.009651098400354385,
      "learning_rate": 2.2376644736842107e-06,
      "loss": 0.0012,
      "step": 18890
    },
    {
      "epoch": 7.771381578947368,
      "grad_norm": 0.5178136825561523,
      "learning_rate": 2.2335526315789473e-06,
      "loss": 0.0013,
      "step": 18900
    },
    {
      "epoch": 7.775493421052632,
      "grad_norm": 0.00043934083078056574,
      "learning_rate": 2.2294407894736844e-06,
      "loss": 0.0033,
      "step": 18910
    },
    {
      "epoch": 7.779605263157895,
      "grad_norm": 0.4319761097431183,
      "learning_rate": 2.225328947368421e-06,
      "loss": 0.0091,
      "step": 18920
    },
    {
      "epoch": 7.7837171052631575,
      "grad_norm": 0.0020860438235104084,
      "learning_rate": 2.221217105263158e-06,
      "loss": 0.0012,
      "step": 18930
    },
    {
      "epoch": 7.787828947368421,
      "grad_norm": 0.00016367706120945513,
      "learning_rate": 2.217105263157895e-06,
      "loss": 0.0004,
      "step": 18940
    },
    {
      "epoch": 7.791940789473684,
      "grad_norm": 0.043262697756290436,
      "learning_rate": 2.212993421052632e-06,
      "loss": 0.0086,
      "step": 18950
    },
    {
      "epoch": 7.796052631578947,
      "grad_norm": 0.04485470429062843,
      "learning_rate": 2.2088815789473687e-06,
      "loss": 0.0615,
      "step": 18960
    },
    {
      "epoch": 7.800164473684211,
      "grad_norm": 0.005395176820456982,
      "learning_rate": 2.2047697368421053e-06,
      "loss": 0.0035,
      "step": 18970
    },
    {
      "epoch": 7.8042763157894735,
      "grad_norm": 0.009486041963100433,
      "learning_rate": 2.2006578947368423e-06,
      "loss": 0.0013,
      "step": 18980
    },
    {
      "epoch": 7.808388157894737,
      "grad_norm": 0.0067543457262218,
      "learning_rate": 2.196546052631579e-06,
      "loss": 0.0006,
      "step": 18990
    },
    {
      "epoch": 7.8125,
      "grad_norm": 4.130396366119385,
      "learning_rate": 2.192434210526316e-06,
      "loss": 0.1369,
      "step": 19000
    },
    {
      "epoch": 7.816611842105263,
      "grad_norm": 0.011135872453451157,
      "learning_rate": 2.1883223684210526e-06,
      "loss": 0.0014,
      "step": 19010
    },
    {
      "epoch": 7.8207236842105265,
      "grad_norm": 11.25094223022461,
      "learning_rate": 2.1842105263157896e-06,
      "loss": 0.0195,
      "step": 19020
    },
    {
      "epoch": 7.824835526315789,
      "grad_norm": 0.0633421316742897,
      "learning_rate": 2.1800986842105266e-06,
      "loss": 0.0052,
      "step": 19030
    },
    {
      "epoch": 7.828947368421053,
      "grad_norm": 1.6162339448928833,
      "learning_rate": 2.1759868421052632e-06,
      "loss": 0.0023,
      "step": 19040
    },
    {
      "epoch": 7.833059210526316,
      "grad_norm": 0.2471773475408554,
      "learning_rate": 2.1718750000000003e-06,
      "loss": 0.0224,
      "step": 19050
    },
    {
      "epoch": 7.837171052631579,
      "grad_norm": 0.005709551740437746,
      "learning_rate": 2.167763157894737e-06,
      "loss": 0.051,
      "step": 19060
    },
    {
      "epoch": 7.8412828947368425,
      "grad_norm": 0.0015900834696367383,
      "learning_rate": 2.163651315789474e-06,
      "loss": 0.0017,
      "step": 19070
    },
    {
      "epoch": 7.845394736842105,
      "grad_norm": 0.2822861969470978,
      "learning_rate": 2.1595394736842105e-06,
      "loss": 0.0009,
      "step": 19080
    },
    {
      "epoch": 7.849506578947368,
      "grad_norm": 0.12618885934352875,
      "learning_rate": 2.1554276315789476e-06,
      "loss": 0.0029,
      "step": 19090
    },
    {
      "epoch": 7.853618421052632,
      "grad_norm": 0.14742045104503632,
      "learning_rate": 2.1513157894736846e-06,
      "loss": 0.0061,
      "step": 19100
    },
    {
      "epoch": 7.857730263157895,
      "grad_norm": 0.021756095811724663,
      "learning_rate": 2.147203947368421e-06,
      "loss": 0.0007,
      "step": 19110
    },
    {
      "epoch": 7.8618421052631575,
      "grad_norm": 0.06931746006011963,
      "learning_rate": 2.1430921052631582e-06,
      "loss": 0.0055,
      "step": 19120
    },
    {
      "epoch": 7.865953947368421,
      "grad_norm": 0.003343462711200118,
      "learning_rate": 2.138980263157895e-06,
      "loss": 0.0014,
      "step": 19130
    },
    {
      "epoch": 7.870065789473684,
      "grad_norm": 0.07024644315242767,
      "learning_rate": 2.134868421052632e-06,
      "loss": 0.0004,
      "step": 19140
    },
    {
      "epoch": 7.874177631578947,
      "grad_norm": 0.19473795592784882,
      "learning_rate": 2.1307565789473685e-06,
      "loss": 0.0041,
      "step": 19150
    },
    {
      "epoch": 7.878289473684211,
      "grad_norm": 0.02276478335261345,
      "learning_rate": 2.1266447368421055e-06,
      "loss": 0.0011,
      "step": 19160
    },
    {
      "epoch": 7.8824013157894735,
      "grad_norm": 0.04341767355799675,
      "learning_rate": 2.122532894736842e-06,
      "loss": 0.0002,
      "step": 19170
    },
    {
      "epoch": 7.886513157894737,
      "grad_norm": 0.0044844504445791245,
      "learning_rate": 2.118421052631579e-06,
      "loss": 0.0388,
      "step": 19180
    },
    {
      "epoch": 7.890625,
      "grad_norm": 2.5301058292388916,
      "learning_rate": 2.114309210526316e-06,
      "loss": 0.0032,
      "step": 19190
    },
    {
      "epoch": 7.894736842105263,
      "grad_norm": 4.735191345214844,
      "learning_rate": 2.110197368421053e-06,
      "loss": 0.2667,
      "step": 19200
    },
    {
      "epoch": 7.8988486842105265,
      "grad_norm": 0.2628767788410187,
      "learning_rate": 2.10608552631579e-06,
      "loss": 0.0011,
      "step": 19210
    },
    {
      "epoch": 7.902960526315789,
      "grad_norm": 0.0044731805101037025,
      "learning_rate": 2.1019736842105264e-06,
      "loss": 0.0383,
      "step": 19220
    },
    {
      "epoch": 7.907072368421053,
      "grad_norm": 0.020919961854815483,
      "learning_rate": 2.097861842105263e-06,
      "loss": 0.0099,
      "step": 19230
    },
    {
      "epoch": 7.911184210526316,
      "grad_norm": 2.660637255758047e-05,
      "learning_rate": 2.09375e-06,
      "loss": 0.0002,
      "step": 19240
    },
    {
      "epoch": 7.915296052631579,
      "grad_norm": 0.4476158320903778,
      "learning_rate": 2.089638157894737e-06,
      "loss": 0.0502,
      "step": 19250
    },
    {
      "epoch": 7.9194078947368425,
      "grad_norm": 0.040128808468580246,
      "learning_rate": 2.085526315789474e-06,
      "loss": 0.001,
      "step": 19260
    },
    {
      "epoch": 7.923519736842105,
      "grad_norm": 0.007745147682726383,
      "learning_rate": 2.0814144736842107e-06,
      "loss": 0.0959,
      "step": 19270
    },
    {
      "epoch": 7.927631578947368,
      "grad_norm": 0.05349845439195633,
      "learning_rate": 2.0773026315789474e-06,
      "loss": 0.0123,
      "step": 19280
    },
    {
      "epoch": 7.931743421052632,
      "grad_norm": 3.0623538494110107,
      "learning_rate": 2.0731907894736844e-06,
      "loss": 0.0075,
      "step": 19290
    },
    {
      "epoch": 7.935855263157895,
      "grad_norm": 0.04638973996043205,
      "learning_rate": 2.069078947368421e-06,
      "loss": 0.0008,
      "step": 19300
    },
    {
      "epoch": 7.9399671052631575,
      "grad_norm": 0.6016738414764404,
      "learning_rate": 2.064967105263158e-06,
      "loss": 0.0009,
      "step": 19310
    },
    {
      "epoch": 7.944078947368421,
      "grad_norm": 7.008052349090576,
      "learning_rate": 2.060855263157895e-06,
      "loss": 0.0126,
      "step": 19320
    },
    {
      "epoch": 7.948190789473684,
      "grad_norm": 1.100488305091858,
      "learning_rate": 2.0567434210526317e-06,
      "loss": 0.0007,
      "step": 19330
    },
    {
      "epoch": 7.952302631578947,
      "grad_norm": 0.02694018743932247,
      "learning_rate": 2.0526315789473687e-06,
      "loss": 0.001,
      "step": 19340
    },
    {
      "epoch": 7.956414473684211,
      "grad_norm": 0.015289338305592537,
      "learning_rate": 2.0485197368421053e-06,
      "loss": 0.0756,
      "step": 19350
    },
    {
      "epoch": 7.9605263157894735,
      "grad_norm": 0.0007966504781506956,
      "learning_rate": 2.0444078947368423e-06,
      "loss": 0.1056,
      "step": 19360
    },
    {
      "epoch": 7.964638157894737,
      "grad_norm": 0.19730961322784424,
      "learning_rate": 2.040296052631579e-06,
      "loss": 0.0108,
      "step": 19370
    },
    {
      "epoch": 7.96875,
      "grad_norm": 98.96208953857422,
      "learning_rate": 2.036184210526316e-06,
      "loss": 0.1418,
      "step": 19380
    },
    {
      "epoch": 7.972861842105263,
      "grad_norm": 0.01025812141597271,
      "learning_rate": 2.0320723684210526e-06,
      "loss": 0.0054,
      "step": 19390
    },
    {
      "epoch": 7.9769736842105265,
      "grad_norm": 0.009373322129249573,
      "learning_rate": 2.0279605263157896e-06,
      "loss": 0.0506,
      "step": 19400
    },
    {
      "epoch": 7.981085526315789,
      "grad_norm": 1.8693686723709106,
      "learning_rate": 2.0238486842105266e-06,
      "loss": 0.0203,
      "step": 19410
    },
    {
      "epoch": 7.985197368421053,
      "grad_norm": 0.08877656608819962,
      "learning_rate": 2.0197368421052633e-06,
      "loss": 0.0003,
      "step": 19420
    },
    {
      "epoch": 7.989309210526316,
      "grad_norm": 0.06994573026895523,
      "learning_rate": 2.0156250000000003e-06,
      "loss": 0.0022,
      "step": 19430
    },
    {
      "epoch": 7.993421052631579,
      "grad_norm": 0.05470941215753555,
      "learning_rate": 2.011513157894737e-06,
      "loss": 0.0115,
      "step": 19440
    },
    {
      "epoch": 7.9975328947368425,
      "grad_norm": 0.00017169667989946902,
      "learning_rate": 2.007401315789474e-06,
      "loss": 0.004,
      "step": 19450
    },
    {
      "epoch": 8.001644736842104,
      "grad_norm": 0.0003484611224848777,
      "learning_rate": 2.0032894736842105e-06,
      "loss": 0.1084,
      "step": 19460
    },
    {
      "epoch": 8.005756578947368,
      "grad_norm": 0.6738888621330261,
      "learning_rate": 1.9991776315789476e-06,
      "loss": 0.0307,
      "step": 19470
    },
    {
      "epoch": 8.009868421052632,
      "grad_norm": 8.662660598754883,
      "learning_rate": 1.9950657894736846e-06,
      "loss": 0.0083,
      "step": 19480
    },
    {
      "epoch": 8.013980263157896,
      "grad_norm": 0.003018350340425968,
      "learning_rate": 1.990953947368421e-06,
      "loss": 0.0003,
      "step": 19490
    },
    {
      "epoch": 8.018092105263158,
      "grad_norm": 0.002044851426035166,
      "learning_rate": 1.9868421052631582e-06,
      "loss": 0.0002,
      "step": 19500
    },
    {
      "epoch": 8.022203947368421,
      "grad_norm": 0.031014088541269302,
      "learning_rate": 1.982730263157895e-06,
      "loss": 0.0001,
      "step": 19510
    },
    {
      "epoch": 8.026315789473685,
      "grad_norm": 0.923111081123352,
      "learning_rate": 1.978618421052632e-06,
      "loss": 0.038,
      "step": 19520
    },
    {
      "epoch": 8.030427631578947,
      "grad_norm": 0.00033148875809274614,
      "learning_rate": 1.9745065789473685e-06,
      "loss": 0.0548,
      "step": 19530
    },
    {
      "epoch": 8.03453947368421,
      "grad_norm": 0.002517040353268385,
      "learning_rate": 1.970394736842105e-06,
      "loss": 0.0792,
      "step": 19540
    },
    {
      "epoch": 8.038651315789474,
      "grad_norm": 0.9935305714607239,
      "learning_rate": 1.966282894736842e-06,
      "loss": 0.0134,
      "step": 19550
    },
    {
      "epoch": 8.042763157894736,
      "grad_norm": 0.0008426306885667145,
      "learning_rate": 1.962171052631579e-06,
      "loss": 0.0069,
      "step": 19560
    },
    {
      "epoch": 8.046875,
      "grad_norm": 0.13046151399612427,
      "learning_rate": 1.958059210526316e-06,
      "loss": 0.0019,
      "step": 19570
    },
    {
      "epoch": 8.050986842105264,
      "grad_norm": 0.00037748689646832645,
      "learning_rate": 1.953947368421053e-06,
      "loss": 0.074,
      "step": 19580
    },
    {
      "epoch": 8.055098684210526,
      "grad_norm": 0.11450521647930145,
      "learning_rate": 1.94983552631579e-06,
      "loss": 0.0003,
      "step": 19590
    },
    {
      "epoch": 8.05921052631579,
      "grad_norm": 0.0015347315929830074,
      "learning_rate": 1.9457236842105264e-06,
      "loss": 0.0054,
      "step": 19600
    },
    {
      "epoch": 8.063322368421053,
      "grad_norm": 0.004102109465748072,
      "learning_rate": 1.941611842105263e-06,
      "loss": 0.001,
      "step": 19610
    },
    {
      "epoch": 8.067434210526315,
      "grad_norm": 0.041049521416425705,
      "learning_rate": 1.9375e-06,
      "loss": 0.0076,
      "step": 19620
    },
    {
      "epoch": 8.071546052631579,
      "grad_norm": 0.012618621811270714,
      "learning_rate": 1.933388157894737e-06,
      "loss": 0.0229,
      "step": 19630
    },
    {
      "epoch": 8.075657894736842,
      "grad_norm": 0.0814553052186966,
      "learning_rate": 1.929276315789474e-06,
      "loss": 0.0004,
      "step": 19640
    },
    {
      "epoch": 8.079769736842104,
      "grad_norm": 0.4970387816429138,
      "learning_rate": 1.9251644736842108e-06,
      "loss": 0.0231,
      "step": 19650
    },
    {
      "epoch": 8.083881578947368,
      "grad_norm": 0.002543308772146702,
      "learning_rate": 1.9210526315789474e-06,
      "loss": 0.0002,
      "step": 19660
    },
    {
      "epoch": 8.087993421052632,
      "grad_norm": 0.0017268903320655227,
      "learning_rate": 1.9169407894736844e-06,
      "loss": 0.0163,
      "step": 19670
    },
    {
      "epoch": 8.092105263157896,
      "grad_norm": 0.0731857493519783,
      "learning_rate": 1.912828947368421e-06,
      "loss": 0.0018,
      "step": 19680
    },
    {
      "epoch": 8.096217105263158,
      "grad_norm": 0.10746968537569046,
      "learning_rate": 1.908717105263158e-06,
      "loss": 0.1044,
      "step": 19690
    },
    {
      "epoch": 8.100328947368421,
      "grad_norm": 0.45797082781791687,
      "learning_rate": 1.9046052631578949e-06,
      "loss": 0.0738,
      "step": 19700
    },
    {
      "epoch": 8.104440789473685,
      "grad_norm": 0.005883471108973026,
      "learning_rate": 1.9004934210526319e-06,
      "loss": 0.0015,
      "step": 19710
    },
    {
      "epoch": 8.108552631578947,
      "grad_norm": 0.41784408688545227,
      "learning_rate": 1.8963815789473685e-06,
      "loss": 0.002,
      "step": 19720
    },
    {
      "epoch": 8.11266447368421,
      "grad_norm": 1.8776366710662842,
      "learning_rate": 1.8922697368421053e-06,
      "loss": 0.0038,
      "step": 19730
    },
    {
      "epoch": 8.116776315789474,
      "grad_norm": 1.0636307001113892,
      "learning_rate": 1.8881578947368423e-06,
      "loss": 0.0006,
      "step": 19740
    },
    {
      "epoch": 8.120888157894736,
      "grad_norm": 7.0736284255981445,
      "learning_rate": 1.884046052631579e-06,
      "loss": 0.0024,
      "step": 19750
    },
    {
      "epoch": 8.125,
      "grad_norm": 0.3112788498401642,
      "learning_rate": 1.879934210526316e-06,
      "loss": 0.0278,
      "step": 19760
    },
    {
      "epoch": 8.129111842105264,
      "grad_norm": 0.15028250217437744,
      "learning_rate": 1.8758223684210528e-06,
      "loss": 0.0057,
      "step": 19770
    },
    {
      "epoch": 8.133223684210526,
      "grad_norm": 0.0035646697506308556,
      "learning_rate": 1.8717105263157898e-06,
      "loss": 0.0009,
      "step": 19780
    },
    {
      "epoch": 8.13733552631579,
      "grad_norm": 27.537662506103516,
      "learning_rate": 1.8675986842105265e-06,
      "loss": 0.0144,
      "step": 19790
    },
    {
      "epoch": 8.141447368421053,
      "grad_norm": 0.49430692195892334,
      "learning_rate": 1.8634868421052633e-06,
      "loss": 0.0003,
      "step": 19800
    },
    {
      "epoch": 8.145559210526315,
      "grad_norm": 0.18590670824050903,
      "learning_rate": 1.8593750000000003e-06,
      "loss": 0.0008,
      "step": 19810
    },
    {
      "epoch": 8.149671052631579,
      "grad_norm": 0.012526645325124264,
      "learning_rate": 1.855263157894737e-06,
      "loss": 0.0032,
      "step": 19820
    },
    {
      "epoch": 8.153782894736842,
      "grad_norm": 0.0013973083114251494,
      "learning_rate": 1.851151315789474e-06,
      "loss": 0.0001,
      "step": 19830
    },
    {
      "epoch": 8.157894736842104,
      "grad_norm": 0.0001499800564488396,
      "learning_rate": 1.8470394736842108e-06,
      "loss": 0.0018,
      "step": 19840
    },
    {
      "epoch": 8.162006578947368,
      "grad_norm": 0.009103098884224892,
      "learning_rate": 1.8429276315789474e-06,
      "loss": 0.0002,
      "step": 19850
    },
    {
      "epoch": 8.166118421052632,
      "grad_norm": 0.07066304981708527,
      "learning_rate": 1.8388157894736844e-06,
      "loss": 0.0012,
      "step": 19860
    },
    {
      "epoch": 8.170230263157896,
      "grad_norm": 0.0303801316767931,
      "learning_rate": 1.834703947368421e-06,
      "loss": 0.0002,
      "step": 19870
    },
    {
      "epoch": 8.174342105263158,
      "grad_norm": 3.0752358436584473,
      "learning_rate": 1.830592105263158e-06,
      "loss": 0.5168,
      "step": 19880
    },
    {
      "epoch": 8.178453947368421,
      "grad_norm": 0.01663016527891159,
      "learning_rate": 1.8264802631578949e-06,
      "loss": 0.0171,
      "step": 19890
    },
    {
      "epoch": 8.182565789473685,
      "grad_norm": 0.00019705640443135053,
      "learning_rate": 1.822368421052632e-06,
      "loss": 0.1079,
      "step": 19900
    },
    {
      "epoch": 8.186677631578947,
      "grad_norm": 107.32159423828125,
      "learning_rate": 1.8182565789473685e-06,
      "loss": 0.1808,
      "step": 19910
    },
    {
      "epoch": 8.19078947368421,
      "grad_norm": 0.07139640301465988,
      "learning_rate": 1.8141447368421053e-06,
      "loss": 0.0005,
      "step": 19920
    },
    {
      "epoch": 8.194901315789474,
      "grad_norm": 0.007935700938105583,
      "learning_rate": 1.8100328947368424e-06,
      "loss": 0.0021,
      "step": 19930
    },
    {
      "epoch": 8.199013157894736,
      "grad_norm": 0.6481422781944275,
      "learning_rate": 1.805921052631579e-06,
      "loss": 0.0008,
      "step": 19940
    },
    {
      "epoch": 8.203125,
      "grad_norm": 24.086261749267578,
      "learning_rate": 1.801809210526316e-06,
      "loss": 0.0123,
      "step": 19950
    },
    {
      "epoch": 8.207236842105264,
      "grad_norm": 0.04818257689476013,
      "learning_rate": 1.7976973684210528e-06,
      "loss": 0.0459,
      "step": 19960
    },
    {
      "epoch": 8.211348684210526,
      "grad_norm": 0.0009188010590150952,
      "learning_rate": 1.7935855263157898e-06,
      "loss": 0.0034,
      "step": 19970
    },
    {
      "epoch": 8.21546052631579,
      "grad_norm": 0.01342172920703888,
      "learning_rate": 1.7894736842105265e-06,
      "loss": 0.0011,
      "step": 19980
    },
    {
      "epoch": 8.219572368421053,
      "grad_norm": 0.01993461884558201,
      "learning_rate": 1.7853618421052633e-06,
      "loss": 0.0011,
      "step": 19990
    },
    {
      "epoch": 8.223684210526315,
      "grad_norm": 0.0007997810025699437,
      "learning_rate": 1.78125e-06,
      "loss": 0.1823,
      "step": 20000
    },
    {
      "epoch": 8.227796052631579,
      "grad_norm": 0.00039553025271743536,
      "learning_rate": 1.777138157894737e-06,
      "loss": 0.0475,
      "step": 20010
    },
    {
      "epoch": 8.231907894736842,
      "grad_norm": 0.029472004622220993,
      "learning_rate": 1.773026315789474e-06,
      "loss": 0.001,
      "step": 20020
    },
    {
      "epoch": 8.236019736842104,
      "grad_norm": 0.13089366257190704,
      "learning_rate": 1.7689144736842106e-06,
      "loss": 0.0053,
      "step": 20030
    },
    {
      "epoch": 8.240131578947368,
      "grad_norm": 0.07195531576871872,
      "learning_rate": 1.7648026315789474e-06,
      "loss": 0.0747,
      "step": 20040
    },
    {
      "epoch": 8.244243421052632,
      "grad_norm": 0.00015038340643513948,
      "learning_rate": 1.7606907894736844e-06,
      "loss": 0.0335,
      "step": 20050
    },
    {
      "epoch": 8.248355263157896,
      "grad_norm": 0.014339683577418327,
      "learning_rate": 1.756578947368421e-06,
      "loss": 0.0147,
      "step": 20060
    },
    {
      "epoch": 8.252467105263158,
      "grad_norm": 0.04160633683204651,
      "learning_rate": 1.752467105263158e-06,
      "loss": 0.0161,
      "step": 20070
    },
    {
      "epoch": 8.256578947368421,
      "grad_norm": 0.09322536736726761,
      "learning_rate": 1.7483552631578949e-06,
      "loss": 0.0025,
      "step": 20080
    },
    {
      "epoch": 8.260690789473685,
      "grad_norm": 0.003090464510023594,
      "learning_rate": 1.744243421052632e-06,
      "loss": 0.0136,
      "step": 20090
    },
    {
      "epoch": 8.264802631578947,
      "grad_norm": 0.0023041388485580683,
      "learning_rate": 1.7401315789473685e-06,
      "loss": 0.0058,
      "step": 20100
    },
    {
      "epoch": 8.26891447368421,
      "grad_norm": 0.02266848273575306,
      "learning_rate": 1.7360197368421053e-06,
      "loss": 0.0002,
      "step": 20110
    },
    {
      "epoch": 8.273026315789474,
      "grad_norm": 0.042311668395996094,
      "learning_rate": 1.7319078947368424e-06,
      "loss": 0.0184,
      "step": 20120
    },
    {
      "epoch": 8.277138157894736,
      "grad_norm": 0.032821740955114365,
      "learning_rate": 1.727796052631579e-06,
      "loss": 0.0104,
      "step": 20130
    },
    {
      "epoch": 8.28125,
      "grad_norm": 4.365161418914795,
      "learning_rate": 1.723684210526316e-06,
      "loss": 0.0067,
      "step": 20140
    },
    {
      "epoch": 8.285361842105264,
      "grad_norm": 0.2567031979560852,
      "learning_rate": 1.7195723684210528e-06,
      "loss": 0.0941,
      "step": 20150
    },
    {
      "epoch": 8.289473684210526,
      "grad_norm": 0.00967695377767086,
      "learning_rate": 1.7154605263157896e-06,
      "loss": 0.001,
      "step": 20160
    },
    {
      "epoch": 8.29358552631579,
      "grad_norm": 0.005732494406402111,
      "learning_rate": 1.7113486842105265e-06,
      "loss": 0.1364,
      "step": 20170
    },
    {
      "epoch": 8.297697368421053,
      "grad_norm": 1.0150928497314453,
      "learning_rate": 1.7072368421052633e-06,
      "loss": 0.1152,
      "step": 20180
    },
    {
      "epoch": 8.301809210526315,
      "grad_norm": 141.66786193847656,
      "learning_rate": 1.703125e-06,
      "loss": 0.12,
      "step": 20190
    },
    {
      "epoch": 8.305921052631579,
      "grad_norm": 7.221472263336182,
      "learning_rate": 1.699013157894737e-06,
      "loss": 0.0027,
      "step": 20200
    },
    {
      "epoch": 8.310032894736842,
      "grad_norm": 22.781036376953125,
      "learning_rate": 1.694901315789474e-06,
      "loss": 0.0124,
      "step": 20210
    },
    {
      "epoch": 8.314144736842104,
      "grad_norm": 0.007851810194551945,
      "learning_rate": 1.6907894736842106e-06,
      "loss": 0.0007,
      "step": 20220
    },
    {
      "epoch": 8.318256578947368,
      "grad_norm": 0.5739917755126953,
      "learning_rate": 1.6866776315789474e-06,
      "loss": 0.0041,
      "step": 20230
    },
    {
      "epoch": 8.322368421052632,
      "grad_norm": 1.0938937664031982,
      "learning_rate": 1.6825657894736844e-06,
      "loss": 0.156,
      "step": 20240
    },
    {
      "epoch": 8.326480263157896,
      "grad_norm": 0.008032879792153835,
      "learning_rate": 1.678453947368421e-06,
      "loss": 0.0599,
      "step": 20250
    },
    {
      "epoch": 8.330592105263158,
      "grad_norm": 0.003187471767887473,
      "learning_rate": 1.674342105263158e-06,
      "loss": 0.0016,
      "step": 20260
    },
    {
      "epoch": 8.334703947368421,
      "grad_norm": 1.5649436712265015,
      "learning_rate": 1.6702302631578949e-06,
      "loss": 0.0017,
      "step": 20270
    },
    {
      "epoch": 8.338815789473685,
      "grad_norm": 3.473937511444092,
      "learning_rate": 1.666118421052632e-06,
      "loss": 0.0016,
      "step": 20280
    },
    {
      "epoch": 8.342927631578947,
      "grad_norm": 0.0011312034912407398,
      "learning_rate": 1.6620065789473685e-06,
      "loss": 0.1365,
      "step": 20290
    },
    {
      "epoch": 8.34703947368421,
      "grad_norm": 0.022491563111543655,
      "learning_rate": 1.6578947368421053e-06,
      "loss": 0.0005,
      "step": 20300
    },
    {
      "epoch": 8.351151315789474,
      "grad_norm": 0.01682455465197563,
      "learning_rate": 1.6537828947368424e-06,
      "loss": 0.1908,
      "step": 20310
    },
    {
      "epoch": 8.355263157894736,
      "grad_norm": 0.15820525586605072,
      "learning_rate": 1.649671052631579e-06,
      "loss": 0.0302,
      "step": 20320
    },
    {
      "epoch": 8.359375,
      "grad_norm": 0.035599756985902786,
      "learning_rate": 1.645559210526316e-06,
      "loss": 0.0491,
      "step": 20330
    },
    {
      "epoch": 8.363486842105264,
      "grad_norm": 0.0006870443467050791,
      "learning_rate": 1.6414473684210528e-06,
      "loss": 0.0453,
      "step": 20340
    },
    {
      "epoch": 8.367598684210526,
      "grad_norm": 0.017046842724084854,
      "learning_rate": 1.6373355263157897e-06,
      "loss": 0.001,
      "step": 20350
    },
    {
      "epoch": 8.37171052631579,
      "grad_norm": 0.02666572667658329,
      "learning_rate": 1.6332236842105265e-06,
      "loss": 0.2276,
      "step": 20360
    },
    {
      "epoch": 8.375822368421053,
      "grad_norm": 2.208775520324707,
      "learning_rate": 1.6291118421052633e-06,
      "loss": 0.0019,
      "step": 20370
    },
    {
      "epoch": 8.379934210526315,
      "grad_norm": 0.1765185445547104,
      "learning_rate": 1.6250000000000001e-06,
      "loss": 0.0006,
      "step": 20380
    },
    {
      "epoch": 8.384046052631579,
      "grad_norm": 0.0014591231010854244,
      "learning_rate": 1.620888157894737e-06,
      "loss": 0.3287,
      "step": 20390
    },
    {
      "epoch": 8.388157894736842,
      "grad_norm": 0.01978524774312973,
      "learning_rate": 1.616776315789474e-06,
      "loss": 0.0006,
      "step": 20400
    },
    {
      "epoch": 8.392269736842104,
      "grad_norm": 0.1453421711921692,
      "learning_rate": 1.6126644736842106e-06,
      "loss": 0.0253,
      "step": 20410
    },
    {
      "epoch": 8.396381578947368,
      "grad_norm": 0.0024396302178502083,
      "learning_rate": 1.6085526315789474e-06,
      "loss": 0.0255,
      "step": 20420
    },
    {
      "epoch": 8.400493421052632,
      "grad_norm": 0.03476029261946678,
      "learning_rate": 1.6044407894736844e-06,
      "loss": 0.1261,
      "step": 20430
    },
    {
      "epoch": 8.404605263157896,
      "grad_norm": 0.0002798949717544019,
      "learning_rate": 1.600328947368421e-06,
      "loss": 0.0139,
      "step": 20440
    },
    {
      "epoch": 8.408717105263158,
      "grad_norm": 0.0015076054260134697,
      "learning_rate": 1.596217105263158e-06,
      "loss": 0.0027,
      "step": 20450
    },
    {
      "epoch": 8.412828947368421,
      "grad_norm": 6.669851779937744,
      "learning_rate": 1.5921052631578949e-06,
      "loss": 0.0024,
      "step": 20460
    },
    {
      "epoch": 8.416940789473685,
      "grad_norm": 0.0023337227758020163,
      "learning_rate": 1.587993421052632e-06,
      "loss": 0.0006,
      "step": 20470
    },
    {
      "epoch": 8.421052631578947,
      "grad_norm": 0.1198020726442337,
      "learning_rate": 1.5838815789473685e-06,
      "loss": 0.0007,
      "step": 20480
    },
    {
      "epoch": 8.42516447368421,
      "grad_norm": 0.021439597010612488,
      "learning_rate": 1.5797697368421053e-06,
      "loss": 0.0002,
      "step": 20490
    },
    {
      "epoch": 8.429276315789474,
      "grad_norm": 0.09605780243873596,
      "learning_rate": 1.5756578947368424e-06,
      "loss": 0.0013,
      "step": 20500
    },
    {
      "epoch": 8.433388157894736,
      "grad_norm": 0.0040932162664830685,
      "learning_rate": 1.571546052631579e-06,
      "loss": 0.0022,
      "step": 20510
    },
    {
      "epoch": 8.4375,
      "grad_norm": 0.00021242848015390337,
      "learning_rate": 1.567434210526316e-06,
      "loss": 0.0001,
      "step": 20520
    },
    {
      "epoch": 8.441611842105264,
      "grad_norm": 0.016136594116687775,
      "learning_rate": 1.5633223684210526e-06,
      "loss": 0.0033,
      "step": 20530
    },
    {
      "epoch": 8.445723684210526,
      "grad_norm": 1.1913484334945679,
      "learning_rate": 1.5592105263157897e-06,
      "loss": 0.0005,
      "step": 20540
    },
    {
      "epoch": 8.44983552631579,
      "grad_norm": 0.0010520926443859935,
      "learning_rate": 1.5550986842105265e-06,
      "loss": 0.0196,
      "step": 20550
    },
    {
      "epoch": 8.453947368421053,
      "grad_norm": 0.05620012804865837,
      "learning_rate": 1.550986842105263e-06,
      "loss": 0.222,
      "step": 20560
    },
    {
      "epoch": 8.458059210526315,
      "grad_norm": 0.35795560479164124,
      "learning_rate": 1.5468750000000001e-06,
      "loss": 0.0986,
      "step": 20570
    },
    {
      "epoch": 8.462171052631579,
      "grad_norm": 146.11349487304688,
      "learning_rate": 1.542763157894737e-06,
      "loss": 0.1901,
      "step": 20580
    },
    {
      "epoch": 8.466282894736842,
      "grad_norm": 0.03306944668292999,
      "learning_rate": 1.538651315789474e-06,
      "loss": 0.0003,
      "step": 20590
    },
    {
      "epoch": 8.470394736842104,
      "grad_norm": 0.0032949261367321014,
      "learning_rate": 1.5345394736842106e-06,
      "loss": 0.0021,
      "step": 20600
    },
    {
      "epoch": 8.474506578947368,
      "grad_norm": 0.044259537011384964,
      "learning_rate": 1.5304276315789474e-06,
      "loss": 0.0033,
      "step": 20610
    },
    {
      "epoch": 8.478618421052632,
      "grad_norm": 0.09277831763029099,
      "learning_rate": 1.5263157894736844e-06,
      "loss": 0.055,
      "step": 20620
    },
    {
      "epoch": 8.482730263157896,
      "grad_norm": 0.0008916630758903921,
      "learning_rate": 1.522203947368421e-06,
      "loss": 0.1617,
      "step": 20630
    },
    {
      "epoch": 8.486842105263158,
      "grad_norm": 1.467398762702942,
      "learning_rate": 1.518092105263158e-06,
      "loss": 0.001,
      "step": 20640
    },
    {
      "epoch": 8.490953947368421,
      "grad_norm": 0.1042017787694931,
      "learning_rate": 1.5139802631578949e-06,
      "loss": 0.0006,
      "step": 20650
    },
    {
      "epoch": 8.495065789473685,
      "grad_norm": 0.010952174663543701,
      "learning_rate": 1.5098684210526317e-06,
      "loss": 0.0084,
      "step": 20660
    },
    {
      "epoch": 8.499177631578947,
      "grad_norm": 0.05859874561429024,
      "learning_rate": 1.5057565789473685e-06,
      "loss": 0.0698,
      "step": 20670
    },
    {
      "epoch": 8.50328947368421,
      "grad_norm": 0.03631851449608803,
      "learning_rate": 1.5016447368421054e-06,
      "loss": 0.0006,
      "step": 20680
    },
    {
      "epoch": 8.507401315789474,
      "grad_norm": 0.0005391056765802205,
      "learning_rate": 1.4975328947368422e-06,
      "loss": 0.0158,
      "step": 20690
    },
    {
      "epoch": 8.511513157894736,
      "grad_norm": 0.050333429127931595,
      "learning_rate": 1.493421052631579e-06,
      "loss": 0.1081,
      "step": 20700
    },
    {
      "epoch": 8.515625,
      "grad_norm": 0.0007912613800726831,
      "learning_rate": 1.489309210526316e-06,
      "loss": 0.1647,
      "step": 20710
    },
    {
      "epoch": 8.519736842105264,
      "grad_norm": 0.006539842579513788,
      "learning_rate": 1.4851973684210526e-06,
      "loss": 0.0026,
      "step": 20720
    },
    {
      "epoch": 8.523848684210526,
      "grad_norm": 0.04584597051143646,
      "learning_rate": 1.4810855263157897e-06,
      "loss": 0.0003,
      "step": 20730
    },
    {
      "epoch": 8.52796052631579,
      "grad_norm": 0.15439699590206146,
      "learning_rate": 1.4769736842105265e-06,
      "loss": 0.0212,
      "step": 20740
    },
    {
      "epoch": 8.532072368421053,
      "grad_norm": 0.0626300796866417,
      "learning_rate": 1.472861842105263e-06,
      "loss": 0.003,
      "step": 20750
    },
    {
      "epoch": 8.536184210526315,
      "grad_norm": 33.395931243896484,
      "learning_rate": 1.4687500000000001e-06,
      "loss": 0.0136,
      "step": 20760
    },
    {
      "epoch": 8.540296052631579,
      "grad_norm": 0.22012284398078918,
      "learning_rate": 1.464638157894737e-06,
      "loss": 0.014,
      "step": 20770
    },
    {
      "epoch": 8.544407894736842,
      "grad_norm": 6.601381755899638e-05,
      "learning_rate": 1.460526315789474e-06,
      "loss": 0.0032,
      "step": 20780
    },
    {
      "epoch": 8.548519736842106,
      "grad_norm": 0.009945002384483814,
      "learning_rate": 1.4564144736842106e-06,
      "loss": 0.0742,
      "step": 20790
    },
    {
      "epoch": 8.552631578947368,
      "grad_norm": 0.006764125544577837,
      "learning_rate": 1.4523026315789474e-06,
      "loss": 0.0733,
      "step": 20800
    },
    {
      "epoch": 8.556743421052632,
      "grad_norm": 0.08154771476984024,
      "learning_rate": 1.4481907894736844e-06,
      "loss": 0.004,
      "step": 20810
    },
    {
      "epoch": 8.560855263157894,
      "grad_norm": 0.04067627713084221,
      "learning_rate": 1.444078947368421e-06,
      "loss": 0.2454,
      "step": 20820
    },
    {
      "epoch": 8.564967105263158,
      "grad_norm": 0.002863495144993067,
      "learning_rate": 1.439967105263158e-06,
      "loss": 0.0003,
      "step": 20830
    },
    {
      "epoch": 8.569078947368421,
      "grad_norm": 0.7650733590126038,
      "learning_rate": 1.435855263157895e-06,
      "loss": 0.0005,
      "step": 20840
    },
    {
      "epoch": 8.573190789473685,
      "grad_norm": 0.06450064480304718,
      "learning_rate": 1.4317434210526317e-06,
      "loss": 0.002,
      "step": 20850
    },
    {
      "epoch": 8.577302631578947,
      "grad_norm": 0.0002420953824184835,
      "learning_rate": 1.4276315789473685e-06,
      "loss": 0.0409,
      "step": 20860
    },
    {
      "epoch": 8.58141447368421,
      "grad_norm": 9.50124958762899e-05,
      "learning_rate": 1.4235197368421054e-06,
      "loss": 0.0002,
      "step": 20870
    },
    {
      "epoch": 8.585526315789474,
      "grad_norm": 0.24137666821479797,
      "learning_rate": 1.4194078947368422e-06,
      "loss": 0.0215,
      "step": 20880
    },
    {
      "epoch": 8.589638157894736,
      "grad_norm": 0.05245298892259598,
      "learning_rate": 1.415296052631579e-06,
      "loss": 0.0008,
      "step": 20890
    },
    {
      "epoch": 8.59375,
      "grad_norm": 0.0008668911759741604,
      "learning_rate": 1.411184210526316e-06,
      "loss": 0.0003,
      "step": 20900
    },
    {
      "epoch": 8.597861842105264,
      "grad_norm": 0.22282694280147552,
      "learning_rate": 1.4070723684210526e-06,
      "loss": 0.1511,
      "step": 20910
    },
    {
      "epoch": 8.601973684210526,
      "grad_norm": 25.438827514648438,
      "learning_rate": 1.4029605263157897e-06,
      "loss": 0.009,
      "step": 20920
    },
    {
      "epoch": 8.60608552631579,
      "grad_norm": 0.2636086642742157,
      "learning_rate": 1.3988486842105265e-06,
      "loss": 0.0316,
      "step": 20930
    },
    {
      "epoch": 8.610197368421053,
      "grad_norm": 2.275233507156372,
      "learning_rate": 1.394736842105263e-06,
      "loss": 0.0008,
      "step": 20940
    },
    {
      "epoch": 8.614309210526315,
      "grad_norm": 43.522274017333984,
      "learning_rate": 1.3906250000000001e-06,
      "loss": 0.0207,
      "step": 20950
    },
    {
      "epoch": 8.618421052631579,
      "grad_norm": 0.004544149152934551,
      "learning_rate": 1.386513157894737e-06,
      "loss": 0.0002,
      "step": 20960
    },
    {
      "epoch": 8.622532894736842,
      "grad_norm": 2.2549304962158203,
      "learning_rate": 1.382401315789474e-06,
      "loss": 0.1462,
      "step": 20970
    },
    {
      "epoch": 8.626644736842106,
      "grad_norm": 0.8890215754508972,
      "learning_rate": 1.3782894736842106e-06,
      "loss": 0.1217,
      "step": 20980
    },
    {
      "epoch": 8.630756578947368,
      "grad_norm": 0.0007637037779204547,
      "learning_rate": 1.3741776315789474e-06,
      "loss": 0.0372,
      "step": 20990
    },
    {
      "epoch": 8.634868421052632,
      "grad_norm": 0.01935545913875103,
      "learning_rate": 1.3700657894736844e-06,
      "loss": 0.0096,
      "step": 21000
    },
    {
      "epoch": 8.638980263157894,
      "grad_norm": 0.0277896448969841,
      "learning_rate": 1.365953947368421e-06,
      "loss": 0.0056,
      "step": 21010
    },
    {
      "epoch": 8.643092105263158,
      "grad_norm": 0.15072134137153625,
      "learning_rate": 1.361842105263158e-06,
      "loss": 0.0026,
      "step": 21020
    },
    {
      "epoch": 8.647203947368421,
      "grad_norm": 0.7178854942321777,
      "learning_rate": 1.357730263157895e-06,
      "loss": 0.1058,
      "step": 21030
    },
    {
      "epoch": 8.651315789473685,
      "grad_norm": 0.42835232615470886,
      "learning_rate": 1.3536184210526317e-06,
      "loss": 0.0004,
      "step": 21040
    },
    {
      "epoch": 8.655427631578947,
      "grad_norm": 11.795758247375488,
      "learning_rate": 1.3495065789473685e-06,
      "loss": 0.2595,
      "step": 21050
    },
    {
      "epoch": 8.65953947368421,
      "grad_norm": 0.3830679953098297,
      "learning_rate": 1.3453947368421052e-06,
      "loss": 0.0013,
      "step": 21060
    },
    {
      "epoch": 8.663651315789474,
      "grad_norm": 0.09519729763269424,
      "learning_rate": 1.3412828947368422e-06,
      "loss": 0.0049,
      "step": 21070
    },
    {
      "epoch": 8.667763157894736,
      "grad_norm": 0.0007457378087565303,
      "learning_rate": 1.337171052631579e-06,
      "loss": 0.0002,
      "step": 21080
    },
    {
      "epoch": 8.671875,
      "grad_norm": 0.0007768406067043543,
      "learning_rate": 1.333059210526316e-06,
      "loss": 0.0149,
      "step": 21090
    },
    {
      "epoch": 8.675986842105264,
      "grad_norm": 0.034300968050956726,
      "learning_rate": 1.3289473684210526e-06,
      "loss": 0.0046,
      "step": 21100
    },
    {
      "epoch": 8.680098684210526,
      "grad_norm": 0.009960854426026344,
      "learning_rate": 1.3248355263157897e-06,
      "loss": 0.0029,
      "step": 21110
    },
    {
      "epoch": 8.68421052631579,
      "grad_norm": 0.00154775264672935,
      "learning_rate": 1.3207236842105265e-06,
      "loss": 0.0327,
      "step": 21120
    },
    {
      "epoch": 8.688322368421053,
      "grad_norm": 0.113314189016819,
      "learning_rate": 1.3166118421052631e-06,
      "loss": 0.0044,
      "step": 21130
    },
    {
      "epoch": 8.692434210526315,
      "grad_norm": 0.014615275897085667,
      "learning_rate": 1.3125000000000001e-06,
      "loss": 0.0044,
      "step": 21140
    },
    {
      "epoch": 8.696546052631579,
      "grad_norm": 1.0225980281829834,
      "learning_rate": 1.308388157894737e-06,
      "loss": 0.0018,
      "step": 21150
    },
    {
      "epoch": 8.700657894736842,
      "grad_norm": 0.0005192348035052419,
      "learning_rate": 1.304276315789474e-06,
      "loss": 0.0028,
      "step": 21160
    },
    {
      "epoch": 8.704769736842106,
      "grad_norm": 21.393903732299805,
      "learning_rate": 1.3001644736842106e-06,
      "loss": 0.0061,
      "step": 21170
    },
    {
      "epoch": 8.708881578947368,
      "grad_norm": 0.007193740922957659,
      "learning_rate": 1.2960526315789474e-06,
      "loss": 0.0025,
      "step": 21180
    },
    {
      "epoch": 8.712993421052632,
      "grad_norm": 0.0013441045302897692,
      "learning_rate": 1.2919407894736842e-06,
      "loss": 0.0651,
      "step": 21190
    },
    {
      "epoch": 8.717105263157894,
      "grad_norm": 0.05820111557841301,
      "learning_rate": 1.287828947368421e-06,
      "loss": 0.0483,
      "step": 21200
    },
    {
      "epoch": 8.721217105263158,
      "grad_norm": 0.032725583761930466,
      "learning_rate": 1.283717105263158e-06,
      "loss": 0.0002,
      "step": 21210
    },
    {
      "epoch": 8.725328947368421,
      "grad_norm": 0.18405325710773468,
      "learning_rate": 1.2796052631578947e-06,
      "loss": 0.1587,
      "step": 21220
    },
    {
      "epoch": 8.729440789473685,
      "grad_norm": 0.3414626717567444,
      "learning_rate": 1.2754934210526317e-06,
      "loss": 0.0078,
      "step": 21230
    },
    {
      "epoch": 8.733552631578947,
      "grad_norm": 0.01579272747039795,
      "learning_rate": 1.2713815789473686e-06,
      "loss": 0.0097,
      "step": 21240
    },
    {
      "epoch": 8.73766447368421,
      "grad_norm": 0.008699091151356697,
      "learning_rate": 1.2672697368421052e-06,
      "loss": 0.0042,
      "step": 21250
    },
    {
      "epoch": 8.741776315789474,
      "grad_norm": 0.8667317032814026,
      "learning_rate": 1.2631578947368422e-06,
      "loss": 0.0009,
      "step": 21260
    },
    {
      "epoch": 8.745888157894736,
      "grad_norm": 1.3391830921173096,
      "learning_rate": 1.259046052631579e-06,
      "loss": 0.0064,
      "step": 21270
    },
    {
      "epoch": 8.75,
      "grad_norm": 0.04508015140891075,
      "learning_rate": 1.254934210526316e-06,
      "loss": 0.1151,
      "step": 21280
    },
    {
      "epoch": 8.754111842105264,
      "grad_norm": 48.28853988647461,
      "learning_rate": 1.2508223684210527e-06,
      "loss": 0.0162,
      "step": 21290
    },
    {
      "epoch": 8.758223684210526,
      "grad_norm": 0.0023160665296018124,
      "learning_rate": 1.2467105263157895e-06,
      "loss": 0.0003,
      "step": 21300
    },
    {
      "epoch": 8.76233552631579,
      "grad_norm": 2.6674623489379883,
      "learning_rate": 1.2425986842105265e-06,
      "loss": 0.0155,
      "step": 21310
    },
    {
      "epoch": 8.766447368421053,
      "grad_norm": 0.00609904620796442,
      "learning_rate": 1.2384868421052633e-06,
      "loss": 0.0086,
      "step": 21320
    },
    {
      "epoch": 8.770559210526315,
      "grad_norm": 0.02606067806482315,
      "learning_rate": 1.2343750000000001e-06,
      "loss": 0.003,
      "step": 21330
    },
    {
      "epoch": 8.774671052631579,
      "grad_norm": 0.06156745180487633,
      "learning_rate": 1.230263157894737e-06,
      "loss": 0.0029,
      "step": 21340
    },
    {
      "epoch": 8.778782894736842,
      "grad_norm": 0.0004447989631444216,
      "learning_rate": 1.2261513157894738e-06,
      "loss": 0.0738,
      "step": 21350
    },
    {
      "epoch": 8.782894736842106,
      "grad_norm": 0.028903665021061897,
      "learning_rate": 1.2220394736842106e-06,
      "loss": 0.0179,
      "step": 21360
    },
    {
      "epoch": 8.787006578947368,
      "grad_norm": 0.000259620719589293,
      "learning_rate": 1.2179276315789474e-06,
      "loss": 0.0001,
      "step": 21370
    },
    {
      "epoch": 8.791118421052632,
      "grad_norm": 0.00042439354001544416,
      "learning_rate": 1.2138157894736842e-06,
      "loss": 0.0192,
      "step": 21380
    },
    {
      "epoch": 8.795230263157894,
      "grad_norm": 1.8762305974960327,
      "learning_rate": 1.2097039473684213e-06,
      "loss": 0.0011,
      "step": 21390
    },
    {
      "epoch": 8.799342105263158,
      "grad_norm": 0.0003334373177494854,
      "learning_rate": 1.2055921052631579e-06,
      "loss": 0.0928,
      "step": 21400
    },
    {
      "epoch": 8.803453947368421,
      "grad_norm": 3.652294874191284,
      "learning_rate": 1.2014802631578947e-06,
      "loss": 0.1625,
      "step": 21410
    },
    {
      "epoch": 8.807565789473685,
      "grad_norm": 0.31108805537223816,
      "learning_rate": 1.1973684210526317e-06,
      "loss": 0.2264,
      "step": 21420
    },
    {
      "epoch": 8.811677631578947,
      "grad_norm": 0.024107256904244423,
      "learning_rate": 1.1932565789473686e-06,
      "loss": 0.0295,
      "step": 21430
    },
    {
      "epoch": 8.81578947368421,
      "grad_norm": 0.04832738637924194,
      "learning_rate": 1.1891447368421054e-06,
      "loss": 0.0005,
      "step": 21440
    },
    {
      "epoch": 8.819901315789474,
      "grad_norm": 0.0019105199025943875,
      "learning_rate": 1.1850328947368422e-06,
      "loss": 0.08,
      "step": 21450
    },
    {
      "epoch": 8.824013157894736,
      "grad_norm": 0.08762315660715103,
      "learning_rate": 1.180921052631579e-06,
      "loss": 0.0007,
      "step": 21460
    },
    {
      "epoch": 8.828125,
      "grad_norm": 0.7652416825294495,
      "learning_rate": 1.1772203947368422e-06,
      "loss": 0.0521,
      "step": 21470
    },
    {
      "epoch": 8.832236842105264,
      "grad_norm": 0.46437129378318787,
      "learning_rate": 1.173108552631579e-06,
      "loss": 0.0101,
      "step": 21480
    },
    {
      "epoch": 8.836348684210526,
      "grad_norm": 0.013920686207711697,
      "learning_rate": 1.1689967105263159e-06,
      "loss": 0.0024,
      "step": 21490
    },
    {
      "epoch": 8.84046052631579,
      "grad_norm": 0.11752483248710632,
      "learning_rate": 1.1648848684210527e-06,
      "loss": 0.0021,
      "step": 21500
    },
    {
      "epoch": 8.844572368421053,
      "grad_norm": 0.28659138083457947,
      "learning_rate": 1.1607730263157895e-06,
      "loss": 0.0015,
      "step": 21510
    },
    {
      "epoch": 8.848684210526315,
      "grad_norm": 0.02271687239408493,
      "learning_rate": 1.1566611842105265e-06,
      "loss": 0.0097,
      "step": 21520
    },
    {
      "epoch": 8.852796052631579,
      "grad_norm": 0.3271775543689728,
      "learning_rate": 1.1525493421052634e-06,
      "loss": 0.0306,
      "step": 21530
    },
    {
      "epoch": 8.856907894736842,
      "grad_norm": 0.0006081332103349268,
      "learning_rate": 1.1484375e-06,
      "loss": 0.0006,
      "step": 21540
    },
    {
      "epoch": 8.861019736842106,
      "grad_norm": 0.009489807300269604,
      "learning_rate": 1.144325657894737e-06,
      "loss": 0.0002,
      "step": 21550
    },
    {
      "epoch": 8.865131578947368,
      "grad_norm": 22.21544075012207,
      "learning_rate": 1.1402138157894738e-06,
      "loss": 0.0096,
      "step": 21560
    },
    {
      "epoch": 8.869243421052632,
      "grad_norm": 3.277864379924722e-05,
      "learning_rate": 1.1361019736842106e-06,
      "loss": 0.0056,
      "step": 21570
    },
    {
      "epoch": 8.873355263157894,
      "grad_norm": 0.003241215832531452,
      "learning_rate": 1.1319901315789475e-06,
      "loss": 0.0592,
      "step": 21580
    },
    {
      "epoch": 8.877467105263158,
      "grad_norm": 0.15321673452854156,
      "learning_rate": 1.1278782894736843e-06,
      "loss": 0.0005,
      "step": 21590
    },
    {
      "epoch": 8.881578947368421,
      "grad_norm": 0.0789317786693573,
      "learning_rate": 1.123766447368421e-06,
      "loss": 0.0005,
      "step": 21600
    },
    {
      "epoch": 8.885690789473685,
      "grad_norm": 70.32683563232422,
      "learning_rate": 1.119654605263158e-06,
      "loss": 0.1845,
      "step": 21610
    },
    {
      "epoch": 8.889802631578947,
      "grad_norm": 0.24425408244132996,
      "learning_rate": 1.1155427631578947e-06,
      "loss": 0.0718,
      "step": 21620
    },
    {
      "epoch": 8.89391447368421,
      "grad_norm": 0.005757387727499008,
      "learning_rate": 1.1114309210526318e-06,
      "loss": 0.0035,
      "step": 21630
    },
    {
      "epoch": 8.898026315789474,
      "grad_norm": 37.92292785644531,
      "learning_rate": 1.1073190789473686e-06,
      "loss": 0.0205,
      "step": 21640
    },
    {
      "epoch": 8.902138157894736,
      "grad_norm": 1.8165327310562134,
      "learning_rate": 1.1032072368421054e-06,
      "loss": 0.0105,
      "step": 21650
    },
    {
      "epoch": 8.90625,
      "grad_norm": 0.013413684442639351,
      "learning_rate": 1.0990953947368422e-06,
      "loss": 0.0092,
      "step": 21660
    },
    {
      "epoch": 8.910361842105264,
      "grad_norm": 0.0035779434256255627,
      "learning_rate": 1.094983552631579e-06,
      "loss": 0.0016,
      "step": 21670
    },
    {
      "epoch": 8.914473684210526,
      "grad_norm": 0.03039056807756424,
      "learning_rate": 1.0908717105263159e-06,
      "loss": 0.0226,
      "step": 21680
    },
    {
      "epoch": 8.91858552631579,
      "grad_norm": 0.0575411431491375,
      "learning_rate": 1.0867598684210527e-06,
      "loss": 0.0301,
      "step": 21690
    },
    {
      "epoch": 8.922697368421053,
      "grad_norm": 0.02829873189330101,
      "learning_rate": 1.0826480263157895e-06,
      "loss": 0.0019,
      "step": 21700
    },
    {
      "epoch": 8.926809210526315,
      "grad_norm": 0.5007913112640381,
      "learning_rate": 1.0785361842105265e-06,
      "loss": 0.0004,
      "step": 21710
    },
    {
      "epoch": 8.930921052631579,
      "grad_norm": 0.0031358154956251383,
      "learning_rate": 1.0744243421052634e-06,
      "loss": 0.0004,
      "step": 21720
    },
    {
      "epoch": 8.935032894736842,
      "grad_norm": 0.01469179056584835,
      "learning_rate": 1.0703125e-06,
      "loss": 0.0301,
      "step": 21730
    },
    {
      "epoch": 8.939144736842106,
      "grad_norm": 28.088376998901367,
      "learning_rate": 1.0662006578947368e-06,
      "loss": 0.0274,
      "step": 21740
    },
    {
      "epoch": 8.943256578947368,
      "grad_norm": 0.07739987969398499,
      "learning_rate": 1.0620888157894738e-06,
      "loss": 0.0008,
      "step": 21750
    },
    {
      "epoch": 8.947368421052632,
      "grad_norm": 0.11320428550243378,
      "learning_rate": 1.0579769736842106e-06,
      "loss": 0.0008,
      "step": 21760
    },
    {
      "epoch": 8.951480263157894,
      "grad_norm": 0.1822323352098465,
      "learning_rate": 1.0538651315789475e-06,
      "loss": 0.0009,
      "step": 21770
    },
    {
      "epoch": 8.955592105263158,
      "grad_norm": 0.020420599728822708,
      "learning_rate": 1.0497532894736843e-06,
      "loss": 0.1801,
      "step": 21780
    },
    {
      "epoch": 8.959703947368421,
      "grad_norm": 0.0003280929522588849,
      "learning_rate": 1.045641447368421e-06,
      "loss": 0.0017,
      "step": 21790
    },
    {
      "epoch": 8.963815789473685,
      "grad_norm": 0.02803770639002323,
      "learning_rate": 1.041529605263158e-06,
      "loss": 0.0004,
      "step": 21800
    },
    {
      "epoch": 8.967927631578947,
      "grad_norm": 0.03750944510102272,
      "learning_rate": 1.0374177631578947e-06,
      "loss": 0.0144,
      "step": 21810
    },
    {
      "epoch": 8.97203947368421,
      "grad_norm": 0.004547052551060915,
      "learning_rate": 1.0333059210526316e-06,
      "loss": 0.0019,
      "step": 21820
    },
    {
      "epoch": 8.976151315789474,
      "grad_norm": 0.030044863000512123,
      "learning_rate": 1.0291940789473686e-06,
      "loss": 0.0018,
      "step": 21830
    },
    {
      "epoch": 8.980263157894736,
      "grad_norm": 0.09664493799209595,
      "learning_rate": 1.0250822368421054e-06,
      "loss": 0.0016,
      "step": 21840
    },
    {
      "epoch": 8.984375,
      "grad_norm": 0.0004916651523672044,
      "learning_rate": 1.0209703947368422e-06,
      "loss": 0.0146,
      "step": 21850
    },
    {
      "epoch": 8.988486842105264,
      "grad_norm": 0.013362167403101921,
      "learning_rate": 1.016858552631579e-06,
      "loss": 0.0009,
      "step": 21860
    },
    {
      "epoch": 8.992598684210526,
      "grad_norm": 0.001229934743605554,
      "learning_rate": 1.0127467105263159e-06,
      "loss": 0.0507,
      "step": 21870
    },
    {
      "epoch": 8.99671052631579,
      "grad_norm": 0.0040124147199094296,
      "learning_rate": 1.0086348684210527e-06,
      "loss": 0.0034,
      "step": 21880
    },
    {
      "epoch": 9.000822368421053,
      "grad_norm": 0.1037806048989296,
      "learning_rate": 1.0045230263157895e-06,
      "loss": 0.0411,
      "step": 21890
    },
    {
      "epoch": 9.004934210526315,
      "grad_norm": 0.000802923459559679,
      "learning_rate": 1.0004111842105263e-06,
      "loss": 0.0003,
      "step": 21900
    },
    {
      "epoch": 9.009046052631579,
      "grad_norm": 0.000517787819262594,
      "learning_rate": 9.962993421052634e-07,
      "loss": 0.0003,
      "step": 21910
    },
    {
      "epoch": 9.013157894736842,
      "grad_norm": 0.00036574091063812375,
      "learning_rate": 9.921875e-07,
      "loss": 0.0568,
      "step": 21920
    },
    {
      "epoch": 9.017269736842104,
      "grad_norm": 2.1960201263427734,
      "learning_rate": 9.880756578947368e-07,
      "loss": 0.0348,
      "step": 21930
    },
    {
      "epoch": 9.021381578947368,
      "grad_norm": 0.002095481613650918,
      "learning_rate": 9.839638157894738e-07,
      "loss": 0.0066,
      "step": 21940
    },
    {
      "epoch": 9.025493421052632,
      "grad_norm": 0.0013046304229646921,
      "learning_rate": 9.798519736842106e-07,
      "loss": 0.0002,
      "step": 21950
    },
    {
      "epoch": 9.029605263157896,
      "grad_norm": 0.1555413454771042,
      "learning_rate": 9.757401315789475e-07,
      "loss": 0.0024,
      "step": 21960
    },
    {
      "epoch": 9.033717105263158,
      "grad_norm": 1.8907396793365479,
      "learning_rate": 9.716282894736843e-07,
      "loss": 0.001,
      "step": 21970
    },
    {
      "epoch": 9.037828947368421,
      "grad_norm": 2.040466785430908,
      "learning_rate": 9.675164473684211e-07,
      "loss": 0.0006,
      "step": 21980
    },
    {
      "epoch": 9.041940789473685,
      "grad_norm": 0.0005111195496283472,
      "learning_rate": 9.63404605263158e-07,
      "loss": 0.0011,
      "step": 21990
    },
    {
      "epoch": 9.046052631578947,
      "grad_norm": 0.012304543517529964,
      "learning_rate": 9.592927631578947e-07,
      "loss": 0.0001,
      "step": 22000
    },
    {
      "epoch": 9.05016447368421,
      "grad_norm": 0.05926978588104248,
      "learning_rate": 9.551809210526316e-07,
      "loss": 0.0003,
      "step": 22010
    },
    {
      "epoch": 9.054276315789474,
      "grad_norm": 0.0007231761119328439,
      "learning_rate": 9.510690789473685e-07,
      "loss": 0.0357,
      "step": 22020
    },
    {
      "epoch": 9.058388157894736,
      "grad_norm": 0.002011606004089117,
      "learning_rate": 9.469572368421054e-07,
      "loss": 0.0177,
      "step": 22030
    },
    {
      "epoch": 9.0625,
      "grad_norm": 0.09335087239742279,
      "learning_rate": 9.428453947368422e-07,
      "loss": 0.0051,
      "step": 22040
    },
    {
      "epoch": 9.066611842105264,
      "grad_norm": 0.008873417042195797,
      "learning_rate": 9.38733552631579e-07,
      "loss": 0.001,
      "step": 22050
    },
    {
      "epoch": 9.070723684210526,
      "grad_norm": 1.5110753774642944,
      "learning_rate": 9.346217105263159e-07,
      "loss": 0.0517,
      "step": 22060
    },
    {
      "epoch": 9.07483552631579,
      "grad_norm": 0.16701540350914001,
      "learning_rate": 9.305098684210527e-07,
      "loss": 0.0404,
      "step": 22070
    },
    {
      "epoch": 9.078947368421053,
      "grad_norm": 0.0011018244549632072,
      "learning_rate": 9.263980263157895e-07,
      "loss": 0.102,
      "step": 22080
    },
    {
      "epoch": 9.083059210526315,
      "grad_norm": 0.0009757328662090003,
      "learning_rate": 9.222861842105264e-07,
      "loss": 0.0002,
      "step": 22090
    },
    {
      "epoch": 9.087171052631579,
      "grad_norm": 0.0028861812315881252,
      "learning_rate": 9.181743421052633e-07,
      "loss": 0.0392,
      "step": 22100
    },
    {
      "epoch": 9.091282894736842,
      "grad_norm": 0.01574259251356125,
      "learning_rate": 9.140625e-07,
      "loss": 0.0057,
      "step": 22110
    },
    {
      "epoch": 9.095394736842104,
      "grad_norm": 0.04924844950437546,
      "learning_rate": 9.099506578947369e-07,
      "loss": 0.0002,
      "step": 22120
    },
    {
      "epoch": 9.099506578947368,
      "grad_norm": 28.6784725189209,
      "learning_rate": 9.058388157894737e-07,
      "loss": 0.0155,
      "step": 22130
    },
    {
      "epoch": 9.103618421052632,
      "grad_norm": 0.0016250410117208958,
      "learning_rate": 9.017269736842107e-07,
      "loss": 0.0122,
      "step": 22140
    },
    {
      "epoch": 9.107730263157896,
      "grad_norm": 0.0031066699884831905,
      "learning_rate": 8.976151315789475e-07,
      "loss": 0.0001,
      "step": 22150
    },
    {
      "epoch": 9.111842105263158,
      "grad_norm": 0.30667412281036377,
      "learning_rate": 8.935032894736843e-07,
      "loss": 0.0006,
      "step": 22160
    },
    {
      "epoch": 9.115953947368421,
      "grad_norm": 0.000589480740018189,
      "learning_rate": 8.89391447368421e-07,
      "loss": 0.0004,
      "step": 22170
    },
    {
      "epoch": 9.120065789473685,
      "grad_norm": 0.02490841969847679,
      "learning_rate": 8.852796052631579e-07,
      "loss": 0.0282,
      "step": 22180
    },
    {
      "epoch": 9.124177631578947,
      "grad_norm": 0.004660974722355604,
      "learning_rate": 8.811677631578948e-07,
      "loss": 0.0033,
      "step": 22190
    },
    {
      "epoch": 9.12828947368421,
      "grad_norm": 0.15241557359695435,
      "learning_rate": 8.770559210526317e-07,
      "loss": 0.0019,
      "step": 22200
    },
    {
      "epoch": 9.132401315789474,
      "grad_norm": 0.08407646417617798,
      "learning_rate": 8.729440789473685e-07,
      "loss": 0.008,
      "step": 22210
    },
    {
      "epoch": 9.136513157894736,
      "grad_norm": 0.28385263681411743,
      "learning_rate": 8.688322368421054e-07,
      "loss": 0.0022,
      "step": 22220
    },
    {
      "epoch": 9.140625,
      "grad_norm": 0.00033503732993267477,
      "learning_rate": 8.647203947368422e-07,
      "loss": 0.0004,
      "step": 22230
    },
    {
      "epoch": 9.144736842105264,
      "grad_norm": 0.01926404982805252,
      "learning_rate": 8.60608552631579e-07,
      "loss": 0.0566,
      "step": 22240
    },
    {
      "epoch": 9.148848684210526,
      "grad_norm": 26.195524215698242,
      "learning_rate": 8.564967105263158e-07,
      "loss": 0.0118,
      "step": 22250
    },
    {
      "epoch": 9.15296052631579,
      "grad_norm": 0.1406288743019104,
      "learning_rate": 8.523848684210527e-07,
      "loss": 0.3019,
      "step": 22260
    },
    {
      "epoch": 9.157072368421053,
      "grad_norm": 0.5480867028236389,
      "learning_rate": 8.482730263157895e-07,
      "loss": 0.0158,
      "step": 22270
    },
    {
      "epoch": 9.161184210526315,
      "grad_norm": 0.13859370350837708,
      "learning_rate": 8.441611842105265e-07,
      "loss": 0.0009,
      "step": 22280
    },
    {
      "epoch": 9.165296052631579,
      "grad_norm": 0.0005784053937532008,
      "learning_rate": 8.400493421052633e-07,
      "loss": 0.0032,
      "step": 22290
    },
    {
      "epoch": 9.169407894736842,
      "grad_norm": 8.995039939880371,
      "learning_rate": 8.359375e-07,
      "loss": 0.0035,
      "step": 22300
    },
    {
      "epoch": 9.173519736842104,
      "grad_norm": 0.01929420232772827,
      "learning_rate": 8.318256578947369e-07,
      "loss": 0.0166,
      "step": 22310
    },
    {
      "epoch": 9.177631578947368,
      "grad_norm": 0.20890386402606964,
      "learning_rate": 8.277138157894737e-07,
      "loss": 0.0017,
      "step": 22320
    },
    {
      "epoch": 9.181743421052632,
      "grad_norm": 0.1153140589594841,
      "learning_rate": 8.236019736842106e-07,
      "loss": 0.2221,
      "step": 22330
    },
    {
      "epoch": 9.185855263157896,
      "grad_norm": 2.0475165843963623,
      "learning_rate": 8.194901315789475e-07,
      "loss": 0.0247,
      "step": 22340
    },
    {
      "epoch": 9.189967105263158,
      "grad_norm": 0.14061927795410156,
      "learning_rate": 8.153782894736843e-07,
      "loss": 0.0015,
      "step": 22350
    },
    {
      "epoch": 9.194078947368421,
      "grad_norm": 0.0035043880343437195,
      "learning_rate": 8.11266447368421e-07,
      "loss": 0.0017,
      "step": 22360
    },
    {
      "epoch": 9.198190789473685,
      "grad_norm": 61.89847183227539,
      "learning_rate": 8.071546052631579e-07,
      "loss": 0.0711,
      "step": 22370
    },
    {
      "epoch": 9.202302631578947,
      "grad_norm": 1.4339483976364136,
      "learning_rate": 8.030427631578948e-07,
      "loss": 0.0583,
      "step": 22380
    },
    {
      "epoch": 9.20641447368421,
      "grad_norm": 0.03647860139608383,
      "learning_rate": 7.989309210526317e-07,
      "loss": 0.0021,
      "step": 22390
    },
    {
      "epoch": 9.210526315789474,
      "grad_norm": 0.0065275998786091805,
      "learning_rate": 7.948190789473685e-07,
      "loss": 0.0,
      "step": 22400
    },
    {
      "epoch": 9.214638157894736,
      "grad_norm": 0.012942095287144184,
      "learning_rate": 7.907072368421053e-07,
      "loss": 0.0008,
      "step": 22410
    },
    {
      "epoch": 9.21875,
      "grad_norm": 0.003749582916498184,
      "learning_rate": 7.865953947368423e-07,
      "loss": 0.0407,
      "step": 22420
    },
    {
      "epoch": 9.222861842105264,
      "grad_norm": 0.019273262470960617,
      "learning_rate": 7.82483552631579e-07,
      "loss": 0.0023,
      "step": 22430
    },
    {
      "epoch": 9.226973684210526,
      "grad_norm": 0.02620495669543743,
      "learning_rate": 7.783717105263158e-07,
      "loss": 0.0824,
      "step": 22440
    },
    {
      "epoch": 9.23108552631579,
      "grad_norm": 0.13408267498016357,
      "learning_rate": 7.742598684210527e-07,
      "loss": 0.004,
      "step": 22450
    },
    {
      "epoch": 9.235197368421053,
      "grad_norm": 0.07048587501049042,
      "learning_rate": 7.701480263157895e-07,
      "loss": 0.0011,
      "step": 22460
    },
    {
      "epoch": 9.239309210526315,
      "grad_norm": 0.2926476299762726,
      "learning_rate": 7.660361842105265e-07,
      "loss": 0.0032,
      "step": 22470
    },
    {
      "epoch": 9.243421052631579,
      "grad_norm": 0.004648135043680668,
      "learning_rate": 7.619243421052633e-07,
      "loss": 0.0001,
      "step": 22480
    },
    {
      "epoch": 9.247532894736842,
      "grad_norm": 0.007466164883226156,
      "learning_rate": 7.578125e-07,
      "loss": 0.0185,
      "step": 22490
    },
    {
      "epoch": 9.251644736842104,
      "grad_norm": 0.011168756522238255,
      "learning_rate": 7.537006578947368e-07,
      "loss": 0.0004,
      "step": 22500
    },
    {
      "epoch": 9.255756578947368,
      "grad_norm": 0.06408285349607468,
      "learning_rate": 7.495888157894737e-07,
      "loss": 0.0035,
      "step": 22510
    },
    {
      "epoch": 9.259868421052632,
      "grad_norm": 0.09527569264173508,
      "learning_rate": 7.454769736842106e-07,
      "loss": 0.0415,
      "step": 22520
    },
    {
      "epoch": 9.263980263157896,
      "grad_norm": 0.009627602994441986,
      "learning_rate": 7.413651315789475e-07,
      "loss": 0.0782,
      "step": 22530
    },
    {
      "epoch": 9.268092105263158,
      "grad_norm": 0.08369173854589462,
      "learning_rate": 7.372532894736843e-07,
      "loss": 0.0088,
      "step": 22540
    },
    {
      "epoch": 9.272203947368421,
      "grad_norm": 0.15961483120918274,
      "learning_rate": 7.33141447368421e-07,
      "loss": 0.0029,
      "step": 22550
    },
    {
      "epoch": 9.276315789473685,
      "grad_norm": 0.049691326916217804,
      "learning_rate": 7.29029605263158e-07,
      "loss": 0.0059,
      "step": 22560
    },
    {
      "epoch": 9.280427631578947,
      "grad_norm": 0.022236308082938194,
      "learning_rate": 7.249177631578948e-07,
      "loss": 0.0048,
      "step": 22570
    },
    {
      "epoch": 9.28453947368421,
      "grad_norm": 0.0647999495267868,
      "learning_rate": 7.208059210526316e-07,
      "loss": 0.0474,
      "step": 22580
    },
    {
      "epoch": 9.288651315789474,
      "grad_norm": 0.0038779708556830883,
      "learning_rate": 7.166940789473685e-07,
      "loss": 0.0001,
      "step": 22590
    },
    {
      "epoch": 9.292763157894736,
      "grad_norm": 0.05114118382334709,
      "learning_rate": 7.125822368421053e-07,
      "loss": 0.0007,
      "step": 22600
    },
    {
      "epoch": 9.296875,
      "grad_norm": 28.422271728515625,
      "learning_rate": 7.084703947368423e-07,
      "loss": 0.0144,
      "step": 22610
    },
    {
      "epoch": 9.300986842105264,
      "grad_norm": 0.00263079721480608,
      "learning_rate": 7.04358552631579e-07,
      "loss": 0.0036,
      "step": 22620
    },
    {
      "epoch": 9.305098684210526,
      "grad_norm": 0.002628036541864276,
      "learning_rate": 7.002467105263158e-07,
      "loss": 0.0002,
      "step": 22630
    },
    {
      "epoch": 9.30921052631579,
      "grad_norm": 0.0005118601256981492,
      "learning_rate": 6.961348684210527e-07,
      "loss": 0.0019,
      "step": 22640
    },
    {
      "epoch": 9.313322368421053,
      "grad_norm": 0.04837176576256752,
      "learning_rate": 6.920230263157895e-07,
      "loss": 0.0003,
      "step": 22650
    },
    {
      "epoch": 9.317434210526315,
      "grad_norm": 0.004665362648665905,
      "learning_rate": 6.879111842105264e-07,
      "loss": 0.0317,
      "step": 22660
    },
    {
      "epoch": 9.321546052631579,
      "grad_norm": 0.0012209968408569694,
      "learning_rate": 6.837993421052633e-07,
      "loss": 0.0001,
      "step": 22670
    },
    {
      "epoch": 9.325657894736842,
      "grad_norm": 0.07622142881155014,
      "learning_rate": 6.796875e-07,
      "loss": 0.0073,
      "step": 22680
    },
    {
      "epoch": 9.329769736842104,
      "grad_norm": 0.00021652725990861654,
      "learning_rate": 6.755756578947368e-07,
      "loss": 0.0001,
      "step": 22690
    },
    {
      "epoch": 9.333881578947368,
      "grad_norm": 0.0072752246633172035,
      "learning_rate": 6.714638157894738e-07,
      "loss": 0.0042,
      "step": 22700
    },
    {
      "epoch": 9.337993421052632,
      "grad_norm": 0.013057149946689606,
      "learning_rate": 6.673519736842106e-07,
      "loss": 0.0013,
      "step": 22710
    },
    {
      "epoch": 9.342105263157896,
      "grad_norm": 1.3909385204315186,
      "learning_rate": 6.632401315789475e-07,
      "loss": 0.1189,
      "step": 22720
    },
    {
      "epoch": 9.346217105263158,
      "grad_norm": 0.005942532327026129,
      "learning_rate": 6.591282894736843e-07,
      "loss": 0.0,
      "step": 22730
    },
    {
      "epoch": 9.350328947368421,
      "grad_norm": 0.0395037904381752,
      "learning_rate": 6.55016447368421e-07,
      "loss": 0.0002,
      "step": 22740
    },
    {
      "epoch": 9.354440789473685,
      "grad_norm": 0.003000462194904685,
      "learning_rate": 6.50904605263158e-07,
      "loss": 0.0054,
      "step": 22750
    },
    {
      "epoch": 9.358552631578947,
      "grad_norm": 0.0012729447335004807,
      "learning_rate": 6.467927631578948e-07,
      "loss": 0.0,
      "step": 22760
    },
    {
      "epoch": 9.36266447368421,
      "grad_norm": 0.07837924361228943,
      "learning_rate": 6.426809210526316e-07,
      "loss": 0.0873,
      "step": 22770
    },
    {
      "epoch": 9.366776315789474,
      "grad_norm": 0.05736375227570534,
      "learning_rate": 6.385690789473685e-07,
      "loss": 0.4035,
      "step": 22780
    },
    {
      "epoch": 9.370888157894736,
      "grad_norm": 0.004519197158515453,
      "learning_rate": 6.344572368421053e-07,
      "loss": 0.0047,
      "step": 22790
    },
    {
      "epoch": 9.375,
      "grad_norm": 1.6480873823165894,
      "learning_rate": 6.303453947368423e-07,
      "loss": 0.0017,
      "step": 22800
    },
    {
      "epoch": 9.379111842105264,
      "grad_norm": 0.0007083399686962366,
      "learning_rate": 6.26233552631579e-07,
      "loss": 0.0004,
      "step": 22810
    },
    {
      "epoch": 9.383223684210526,
      "grad_norm": 81.98469543457031,
      "learning_rate": 6.221217105263158e-07,
      "loss": 0.2035,
      "step": 22820
    },
    {
      "epoch": 9.38733552631579,
      "grad_norm": 1.910955786705017,
      "learning_rate": 6.180098684210527e-07,
      "loss": 0.0008,
      "step": 22830
    },
    {
      "epoch": 9.391447368421053,
      "grad_norm": 0.5536595582962036,
      "learning_rate": 6.138980263157896e-07,
      "loss": 0.0185,
      "step": 22840
    },
    {
      "epoch": 9.395559210526315,
      "grad_norm": 0.0020302587654441595,
      "learning_rate": 6.097861842105264e-07,
      "loss": 0.0563,
      "step": 22850
    },
    {
      "epoch": 9.399671052631579,
      "grad_norm": 0.025456195697188377,
      "learning_rate": 6.056743421052632e-07,
      "loss": 0.0001,
      "step": 22860
    },
    {
      "epoch": 9.403782894736842,
      "grad_norm": 0.043300170451402664,
      "learning_rate": 6.015625e-07,
      "loss": 0.0002,
      "step": 22870
    },
    {
      "epoch": 9.407894736842104,
      "grad_norm": 0.5046695470809937,
      "learning_rate": 5.974506578947368e-07,
      "loss": 0.0049,
      "step": 22880
    },
    {
      "epoch": 9.412006578947368,
      "grad_norm": 0.00038293583202175796,
      "learning_rate": 5.933388157894738e-07,
      "loss": 0.0003,
      "step": 22890
    },
    {
      "epoch": 9.416118421052632,
      "grad_norm": 0.027482980862259865,
      "learning_rate": 5.892269736842106e-07,
      "loss": 0.0003,
      "step": 22900
    },
    {
      "epoch": 9.420230263157896,
      "grad_norm": 0.08259539306163788,
      "learning_rate": 5.851151315789474e-07,
      "loss": 0.0004,
      "step": 22910
    },
    {
      "epoch": 9.424342105263158,
      "grad_norm": 0.002156503265723586,
      "learning_rate": 5.810032894736842e-07,
      "loss": 0.0002,
      "step": 22920
    },
    {
      "epoch": 9.428453947368421,
      "grad_norm": 0.06109001114964485,
      "learning_rate": 5.768914473684211e-07,
      "loss": 0.0004,
      "step": 22930
    },
    {
      "epoch": 9.432565789473685,
      "grad_norm": 1.3648306131362915,
      "learning_rate": 5.72779605263158e-07,
      "loss": 0.0007,
      "step": 22940
    },
    {
      "epoch": 9.436677631578947,
      "grad_norm": 0.038669224828481674,
      "learning_rate": 5.686677631578948e-07,
      "loss": 0.0002,
      "step": 22950
    },
    {
      "epoch": 9.44078947368421,
      "grad_norm": 0.06159086897969246,
      "learning_rate": 5.645559210526316e-07,
      "loss": 0.0124,
      "step": 22960
    },
    {
      "epoch": 9.444901315789474,
      "grad_norm": 1.0390788316726685,
      "learning_rate": 5.604440789473685e-07,
      "loss": 0.0027,
      "step": 22970
    },
    {
      "epoch": 9.449013157894736,
      "grad_norm": 0.0944819375872612,
      "learning_rate": 5.563322368421052e-07,
      "loss": 0.0005,
      "step": 22980
    },
    {
      "epoch": 9.453125,
      "grad_norm": 0.00038404561928473413,
      "learning_rate": 5.522203947368422e-07,
      "loss": 0.0125,
      "step": 22990
    },
    {
      "epoch": 9.457236842105264,
      "grad_norm": 0.001499609206803143,
      "learning_rate": 5.48108552631579e-07,
      "loss": 0.1189,
      "step": 23000
    }
  ],
  "logging_steps": 10,
  "max_steps": 24320,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}