diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5461 @@
+{
+  "best_global_step": 6300,
+  "best_metric": 3.5436816215515137,
+  "best_model_checkpoint": "./checkpoints/lstm_h1024_l3_d0.1/checkpoint-6300",
+  "epoch": 0.63,
+  "eval_steps": 50,
+  "global_step": 6300,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001,
+      "grad_norm": 0.3549754023551941,
+      "learning_rate": 1.8e-06,
+      "loss": 60.8128,
+      "step": 10
+    },
+    {
+      "epoch": 0.002,
+      "grad_norm": 0.3771856427192688,
+      "learning_rate": 3.8e-06,
+      "loss": 60.8074,
+      "step": 20
+    },
+    {
+      "epoch": 0.003,
+      "grad_norm": 0.38984280824661255,
+      "learning_rate": 5.8e-06,
+      "loss": 60.7973,
+      "step": 30
+    },
+    {
+      "epoch": 0.004,
+      "grad_norm": 0.3642944097518921,
+      "learning_rate": 7.8e-06,
+      "loss": 60.786,
+      "step": 40
+    },
+    {
+      "epoch": 0.005,
+      "grad_norm": 0.38839685916900635,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 60.7688,
+      "step": 50
+    },
+    {
+      "epoch": 0.005,
+      "eval_loss": 15.188536643981934,
+      "eval_runtime": 6.2292,
+      "eval_samples_per_second": 321.071,
+      "eval_steps_per_second": 20.067,
+      "step": 50
+    },
+    {
+      "epoch": 0.006,
+      "grad_norm": 0.4147929549217224,
+      "learning_rate": 1.18e-05,
+      "loss": 60.7432,
+      "step": 60
+    },
+    {
+      "epoch": 0.007,
+      "grad_norm": 0.4341050088405609,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 60.7053,
+      "step": 70
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 0.5660704374313354,
+      "learning_rate": 1.58e-05,
+      "loss": 60.6374,
+      "step": 80
+    },
+    {
+      "epoch": 0.009,
+      "grad_norm": 1.0963184833526611,
+      "learning_rate": 1.78e-05,
+      "loss": 60.4508,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7994534969329834,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 59.5999,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "eval_loss": 14.58733081817627,
+      "eval_runtime": 6.1941,
+      "eval_samples_per_second": 322.887,
+      "eval_steps_per_second": 20.18,
+      "step": 100
+    },
+    {
+      "epoch": 0.011,
+      "grad_norm": 5.019301414489746,
+      "learning_rate": 2.18e-05,
+      "loss": 57.248,
+      "step": 110
+    },
+    {
+      "epoch": 0.012,
+      "grad_norm": 2.526005744934082,
+      "learning_rate": 2.38e-05,
+      "loss": 55.7676,
+      "step": 120
+    },
+    {
+      "epoch": 0.013,
+      "grad_norm": 2.1965389251708984,
+      "learning_rate": 2.58e-05,
+      "loss": 55.3816,
+      "step": 130
+    },
+    {
+      "epoch": 0.014,
+      "grad_norm": 1.439666748046875,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 54.9717,
+      "step": 140
+    },
+    {
+      "epoch": 0.015,
+      "grad_norm": 1.7099782228469849,
+      "learning_rate": 2.98e-05,
+      "loss": 54.954,
+      "step": 150
+    },
+    {
+      "epoch": 0.015,
+      "eval_loss": 13.733558654785156,
+      "eval_runtime": 6.2158,
+      "eval_samples_per_second": 321.758,
+      "eval_steps_per_second": 20.11,
+      "step": 150
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 1.8854618072509766,
+      "learning_rate": 3.18e-05,
+      "loss": 54.5982,
+      "step": 160
+    },
+    {
+      "epoch": 0.017,
+      "grad_norm": 2.250976085662842,
+      "learning_rate": 3.38e-05,
+      "loss": 54.526,
+      "step": 170
+    },
+    {
+      "epoch": 0.018,
+      "grad_norm": 1.5891501903533936,
+      "learning_rate": 3.58e-05,
+      "loss": 54.5763,
+      "step": 180
+    },
+    {
+      "epoch": 0.019,
+      "grad_norm": 1.5920642614364624,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 54.4684,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5550366640090942,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 54.3639,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 13.6970796585083,
+      "eval_runtime": 6.2966,
+      "eval_samples_per_second": 317.634,
+      "eval_steps_per_second": 19.852,
+      "step": 200
+    },
+    {
+      "epoch": 0.021,
+      "grad_norm": 2.051086187362671,
+      "learning_rate": 4.18e-05,
+      "loss": 54.3304,
+      "step": 210
+    },
+    {
+      "epoch": 0.022,
+      "grad_norm": 1.848323106765747,
+      "learning_rate": 4.38e-05,
+      "loss": 54.744,
+      "step": 220
+    },
+    {
+      "epoch": 0.023,
+      "grad_norm": 1.9148523807525635,
+      "learning_rate": 4.58e-05,
+      "loss": 54.6018,
+      "step": 230
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 3.823683500289917,
+      "learning_rate": 4.78e-05,
+      "loss": 54.7114,
+      "step": 240
+    },
+    {
+      "epoch": 0.025,
+      "grad_norm": 2.885927438735962,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 54.6399,
+      "step": 250
+    },
+    {
+      "epoch": 0.025,
+      "eval_loss": 13.650457382202148,
+      "eval_runtime": 6.1864,
+      "eval_samples_per_second": 323.289,
+      "eval_steps_per_second": 20.206,
+      "step": 250
+    },
+    {
+      "epoch": 0.026,
+      "grad_norm": 4.444192886352539,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 54.3902,
+      "step": 260
+    },
+    {
+      "epoch": 0.027,
+      "grad_norm": 7.780248641967773,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 54.1046,
+      "step": 270
+    },
+    {
+      "epoch": 0.028,
+      "grad_norm": 2.7638983726501465,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 54.0532,
+      "step": 280
+    },
+    {
+      "epoch": 0.029,
+      "grad_norm": 2.145719051361084,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 53.8505,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.996760606765747,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 53.6261,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 13.41512680053711,
+      "eval_runtime": 6.2516,
+      "eval_samples_per_second": 319.917,
+      "eval_steps_per_second": 19.995,
+      "step": 300
+    },
+    {
+      "epoch": 0.031,
+      "grad_norm": 2.4033031463623047,
+      "learning_rate": 6.18e-05,
+      "loss": 53.2967,
+      "step": 310
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 3.834418773651123,
+      "learning_rate": 6.38e-05,
+      "loss": 52.9328,
+      "step": 320
+    },
+    {
+      "epoch": 0.033,
+      "grad_norm": 6.866483211517334,
+      "learning_rate": 6.58e-05,
+      "loss": 52.3613,
+      "step": 330
+    },
+    {
+      "epoch": 0.034,
+      "grad_norm": 3.098707675933838,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 52.4274,
+      "step": 340
+    },
+    {
+      "epoch": 0.035,
+      "grad_norm": 9.082242012023926,
+      "learning_rate": 6.98e-05,
+      "loss": 52.2897,
+      "step": 350
+    },
+    {
+      "epoch": 0.035,
+      "eval_loss": 13.071863174438477,
+      "eval_runtime": 6.1905,
+      "eval_samples_per_second": 323.074,
+      "eval_steps_per_second": 20.192,
+      "step": 350
+    },
+    {
+      "epoch": 0.036,
+      "grad_norm": 2.5560662746429443,
+      "learning_rate": 7.18e-05,
+      "loss": 52.0423,
+      "step": 360
+    },
+    {
+      "epoch": 0.037,
+      "grad_norm": 2.2769477367401123,
+      "learning_rate": 7.38e-05,
+      "loss": 51.3791,
+      "step": 370
+    },
+    {
+      "epoch": 0.038,
+      "grad_norm": 2.816531181335449,
+      "learning_rate": 7.58e-05,
+      "loss": 51.0433,
+      "step": 380
+    },
+    {
+      "epoch": 0.039,
+      "grad_norm": 4.625880718231201,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 49.9025,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 3.8817570209503174,
+      "learning_rate": 7.98e-05,
+      "loss": 49.2651,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 12.226531982421875,
+      "eval_runtime": 6.218,
+      "eval_samples_per_second": 321.647,
+      "eval_steps_per_second": 20.103,
+      "step": 400
+    },
+    {
+      "epoch": 0.041,
+      "grad_norm": 2.100292205810547,
+      "learning_rate": 8.18e-05,
+      "loss": 48.4186,
+      "step": 410
+    },
+    {
+      "epoch": 0.042,
+      "grad_norm": 2.875107765197754,
+      "learning_rate": 8.38e-05,
+      "loss": 47.5267,
+      "step": 420
+    },
+    {
+      "epoch": 0.043,
+      "grad_norm": 3.6629738807678223,
+      "learning_rate": 8.58e-05,
+      "loss": 46.0668,
+      "step": 430
+    },
+    {
+      "epoch": 0.044,
+      "grad_norm": 6.750997543334961,
+      "learning_rate": 8.78e-05,
+      "loss": 45.6252,
+      "step": 440
+    },
+    {
+      "epoch": 0.045,
+      "grad_norm": 4.621273040771484,
+      "learning_rate": 8.98e-05,
+      "loss": 44.6924,
+      "step": 450
+    },
+    {
+      "epoch": 0.045,
+      "eval_loss": 11.125019073486328,
+      "eval_runtime": 6.1835,
+      "eval_samples_per_second": 323.442,
+      "eval_steps_per_second": 20.215,
+      "step": 450
+    },
+    {
+      "epoch": 0.046,
+      "grad_norm": 7.5620856285095215,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 43.9264,
+      "step": 460
+    },
+    {
+      "epoch": 0.047,
+      "grad_norm": 4.797828197479248,
+      "learning_rate": 9.38e-05,
+      "loss": 43.1773,
+      "step": 470
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 6.183513641357422,
+      "learning_rate": 9.58e-05,
+      "loss": 42.4152,
+      "step": 480
+    },
+    {
+      "epoch": 0.049,
+      "grad_norm": 2.3638603687286377,
+      "learning_rate": 9.78e-05,
+      "loss": 41.8333,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.4479477405548096,
+      "learning_rate": 9.98e-05,
+      "loss": 41.2254,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 10.2489013671875,
+      "eval_runtime": 6.1973,
+      "eval_samples_per_second": 322.722,
+      "eval_steps_per_second": 20.17,
+      "step": 500
+    },
+    {
+      "epoch": 0.051,
+      "grad_norm": 3.256962537765503,
+      "learning_rate": 9.911197057469107e-05,
+      "loss": 40.7175,
+      "step": 510
+    },
+    {
+      "epoch": 0.052,
+      "grad_norm": 4.1723175048828125,
+      "learning_rate": 9.815249038111776e-05,
+      "loss": 39.9997,
+      "step": 520
+    },
+    {
+      "epoch": 0.053,
+      "grad_norm": 2.062711477279663,
+      "learning_rate": 9.722034684781694e-05,
+      "loss": 38.5916,
+      "step": 530
+    },
+    {
+      "epoch": 0.054,
+      "grad_norm": 2.106576681137085,
+      "learning_rate": 9.631426606617744e-05,
+      "loss": 37.6645,
+      "step": 540
+    },
+    {
+      "epoch": 0.055,
+      "grad_norm": 3.3942205905914307,
+      "learning_rate": 9.543305571897804e-05,
+      "loss": 36.591,
+      "step": 550
+    },
+    {
+      "epoch": 0.055,
+      "eval_loss": 9.059900283813477,
+      "eval_runtime": 6.1773,
+      "eval_samples_per_second": 323.764,
+      "eval_steps_per_second": 20.235,
+      "step": 550
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 5.337911128997803,
+      "learning_rate": 9.457559848219179e-05,
+      "loss": 35.6712,
+      "step": 560
+    },
+    {
+      "epoch": 0.057,
+      "grad_norm": 5.007609844207764,
+      "learning_rate": 9.374084606744877e-05,
+      "loss": 34.8073,
+      "step": 570
+    },
+    {
+      "epoch": 0.058,
+      "grad_norm": 3.1866860389709473,
+      "learning_rate": 9.292781383291611e-05,
+      "loss": 34.288,
+      "step": 580
+    },
+    {
+      "epoch": 0.059,
+      "grad_norm": 3.2338063716888428,
+      "learning_rate": 9.213557589959345e-05,
+      "loss": 33.4251,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.3791048526763916,
+      "learning_rate": 9.136326071794409e-05,
+      "loss": 32.8373,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 8.064367294311523,
+      "eval_runtime": 6.1993,
+      "eval_samples_per_second": 322.618,
+      "eval_steps_per_second": 20.164,
+      "step": 600
+    },
+    {
+      "epoch": 0.061,
+      "grad_norm": 2.501295804977417,
+      "learning_rate": 9.061004703659374e-05,
+      "loss": 31.4265,
+      "step": 610
+    },
+    {
+      "epoch": 0.062,
+      "grad_norm": 2.6768312454223633,
+      "learning_rate": 8.987516023070193e-05,
+      "loss": 30.8398,
+      "step": 620
+    },
+    {
+      "epoch": 0.063,
+      "grad_norm": 2.1591920852661133,
+      "learning_rate": 8.915786895268651e-05,
+      "loss": 29.9915,
+      "step": 630
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 1.987754464149475,
+      "learning_rate": 8.84574820723792e-05,
+      "loss": 28.8786,
+      "step": 640
+    },
+    {
+      "epoch": 0.065,
+      "grad_norm": 3.341383695602417,
+      "learning_rate": 8.777334587751072e-05,
+      "loss": 27.9284,
+      "step": 650
+    },
+    {
+      "epoch": 0.065,
+      "eval_loss": 6.836698532104492,
+      "eval_runtime": 6.2313,
+      "eval_samples_per_second": 320.959,
+      "eval_steps_per_second": 20.06,
+      "step": 650
+    },
+    {
+      "epoch": 0.066,
+      "grad_norm": 3.247847318649292,
+      "learning_rate": 8.710484150874758e-05,
+      "loss": 26.8209,
+      "step": 660
+    },
+    {
+      "epoch": 0.067,
+      "grad_norm": 2.0966415405273438,
+      "learning_rate": 8.645138260640511e-05,
+      "loss": 26.5482,
+      "step": 670
+    },
+    {
+      "epoch": 0.068,
+      "grad_norm": 2.398731231689453,
+      "learning_rate": 8.581241314849611e-05,
+      "loss": 25.8964,
+      "step": 680
+    },
+    {
+      "epoch": 0.069,
+      "grad_norm": 3.018005132675171,
+      "learning_rate": 8.51874054619982e-05,
+      "loss": 24.7768,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.627410411834717,
+      "learning_rate": 8.457585839117285e-05,
+      "loss": 23.7828,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 6.068182468414307,
+      "eval_runtime": 6.1922,
+      "eval_samples_per_second": 322.987,
+      "eval_steps_per_second": 20.187,
+      "step": 700
+    },
+    {
+      "epoch": 0.071,
+      "grad_norm": 1.910626769065857,
+      "learning_rate": 8.397729560848631e-05,
+      "loss": 24.4096,
+      "step": 710
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 1.7598278522491455,
+      "learning_rate": 8.339126405519483e-05,
+      "loss": 23.4098,
+      "step": 720
+    },
+    {
+      "epoch": 0.073,
+      "grad_norm": 2.161638021469116,
+      "learning_rate": 8.281733249999222e-05,
+      "loss": 22.4748,
+      "step": 730
+    },
+    {
+      "epoch": 0.074,
+      "grad_norm": 2.6153981685638428,
+      "learning_rate": 8.225509020529976e-05,
+      "loss": 22.4988,
+      "step": 740
+    },
+    {
+      "epoch": 0.075,
+      "grad_norm": 2.6121160984039307,
+      "learning_rate": 8.170414569182506e-05,
+      "loss": 21.9317,
+      "step": 750
+    },
+    {
+      "epoch": 0.075,
+      "eval_loss": 5.473199844360352,
+      "eval_runtime": 6.1666,
+      "eval_samples_per_second": 324.327,
+      "eval_steps_per_second": 20.27,
+      "step": 750
+    },
+    {
+      "epoch": 0.076,
+      "grad_norm": 2.71346378326416,
+      "learning_rate": 8.116412559294567e-05,
+      "loss": 21.6393,
+      "step": 760
+    },
+    {
+      "epoch": 0.077,
+      "grad_norm": 1.7239511013031006,
+      "learning_rate": 8.063467359130037e-05,
+      "loss": 21.3508,
+      "step": 770
+    },
+    {
+      "epoch": 0.078,
+      "grad_norm": 1.9020029306411743,
+      "learning_rate": 8.011544943070565e-05,
+      "loss": 20.6425,
+      "step": 780
+    },
+    {
+      "epoch": 0.079,
+      "grad_norm": 1.9728212356567383,
+      "learning_rate": 7.960612799717214e-05,
+      "loss": 20.3836,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0105056762695312,
+      "learning_rate": 7.910639846338163e-05,
+      "loss": 20.1252,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 5.180527210235596,
+      "eval_runtime": 6.2098,
+      "eval_samples_per_second": 322.073,
+      "eval_steps_per_second": 20.13,
+      "step": 800
+    },
+    {
+      "epoch": 0.081,
+      "grad_norm": 1.8437271118164062,
+      "learning_rate": 7.861596349150974e-05,
+      "loss": 20.5089,
+      "step": 810
+    },
+    {
+      "epoch": 0.082,
+      "grad_norm": 1.525156855583191,
+      "learning_rate": 7.813453848974926e-05,
+      "loss": 20.143,
+      "step": 820
+    },
+    {
+      "epoch": 0.083,
+      "grad_norm": 1.9869450330734253,
+      "learning_rate": 7.766185091831061e-05,
+      "loss": 20.4813,
+      "step": 830
+    },
+    {
+      "epoch": 0.084,
+      "grad_norm": 1.3000235557556152,
+      "learning_rate": 7.719763964105497e-05,
+      "loss": 19.9456,
+      "step": 840
+    },
+    {
+      "epoch": 0.085,
+      "grad_norm": 1.4937679767608643,
+      "learning_rate": 7.674165431925523e-05,
+      "loss": 19.5407,
+      "step": 850
+    },
+    {
+      "epoch": 0.085,
+      "eval_loss": 4.915063858032227,
+      "eval_runtime": 6.1969,
+      "eval_samples_per_second": 322.74,
+      "eval_steps_per_second": 20.171,
+      "step": 850
+    },
+    {
+      "epoch": 0.086,
+      "grad_norm": 1.616244912147522,
+      "learning_rate": 7.629365484428845e-05,
+      "loss": 19.4324,
+      "step": 860
+    },
+    {
+      "epoch": 0.087,
+      "grad_norm": 1.8139533996582031,
+      "learning_rate": 7.585341080633831e-05,
+      "loss": 19.2495,
+      "step": 870
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 1.4117282629013062,
+      "learning_rate": 7.542070099643788e-05,
+      "loss": 19.334,
+      "step": 880
+    },
+    {
+      "epoch": 0.089,
+      "grad_norm": 1.3447221517562866,
+      "learning_rate": 7.499531293940736e-05,
+      "loss": 18.7034,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8889520168304443,
+      "learning_rate": 7.457704245544709e-05,
+      "loss": 18.094,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 4.761510848999023,
+      "eval_runtime": 6.2364,
+      "eval_samples_per_second": 320.698,
+      "eval_steps_per_second": 20.044,
+      "step": 900
+    },
+    {
+      "epoch": 0.091,
+      "grad_norm": 1.7895255088806152,
+      "learning_rate": 7.41656932483308e-05,
+      "loss": 18.8197,
+      "step": 910
+    },
+    {
+      "epoch": 0.092,
+      "grad_norm": 3.055880069732666,
+      "learning_rate": 7.376107651831263e-05,
+      "loss": 18.4797,
+      "step": 920
+    },
+    {
+      "epoch": 0.093,
+      "grad_norm": 1.4833000898361206,
+      "learning_rate": 7.336301059801394e-05,
+      "loss": 18.7489,
+      "step": 930
+    },
+    {
+      "epoch": 0.094,
+      "grad_norm": 1.4441015720367432,
+      "learning_rate": 7.297132060969499e-05,
+      "loss": 17.7985,
+      "step": 940
+    },
+    {
+      "epoch": 0.095,
+      "grad_norm": 1.970638394355774,
+      "learning_rate": 7.258583814244268e-05,
+      "loss": 17.9167,
+      "step": 950
+    },
+    {
+      "epoch": 0.095,
+      "eval_loss": 4.611672878265381,
+      "eval_runtime": 6.2185,
+      "eval_samples_per_second": 321.621,
+      "eval_steps_per_second": 20.101,
+      "step": 950
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 1.66256582736969,
+      "learning_rate": 7.220640094792103e-05,
+      "loss": 18.063,
+      "step": 960
+    },
+    {
+      "epoch": 0.097,
+      "grad_norm": 1.8178454637527466,
+      "learning_rate": 7.183285265343593e-05,
+      "loss": 18.5062,
+      "step": 970
+    },
+    {
+      "epoch": 0.098,
+      "grad_norm": 1.5309795141220093,
+      "learning_rate": 7.14650424911616e-05,
+      "loss": 18.2083,
+      "step": 980
+    },
+    {
+      "epoch": 0.099,
+      "grad_norm": 1.72297203540802,
+      "learning_rate": 7.110282504246376e-05,
+      "loss": 17.6364,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.8936313390731812,
+      "learning_rate": 7.074605999633481e-05,
+      "loss": 17.6297,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 4.519549369812012,
+      "eval_runtime": 6.233,
+      "eval_samples_per_second": 320.874,
+      "eval_steps_per_second": 20.055,
+      "step": 1000
+    },
+    {
+      "epoch": 0.101,
+      "grad_norm": 1.5446372032165527,
+      "learning_rate": 7.03946119210298e-05,
+      "loss": 17.7636,
+      "step": 1010
+    },
+    {
+      "epoch": 0.102,
+      "grad_norm": 1.8526910543441772,
+      "learning_rate": 7.004835004805859e-05,
+      "loss": 17.4462,
+      "step": 1020
+    },
+    {
+      "epoch": 0.103,
+      "grad_norm": 1.521323323249817,
+      "learning_rate": 6.970714806775237e-05,
+      "loss": 18.2182,
+      "step": 1030
+    },
+    {
+      "epoch": 0.104,
+      "grad_norm": 1.8971052169799805,
+      "learning_rate": 6.937088393567812e-05,
+      "loss": 17.8118,
+      "step": 1040
+    },
+    {
+      "epoch": 0.105,
+      "grad_norm": 2.105436325073242,
+      "learning_rate": 6.903943968922749e-05,
+      "loss": 17.0171,
+      "step": 1050
+    },
+    {
+      "epoch": 0.105,
+      "eval_loss": 4.442924499511719,
+      "eval_runtime": 6.2358,
+      "eval_samples_per_second": 320.727,
+      "eval_steps_per_second": 20.045,
+      "step": 1050
+    },
+    {
+      "epoch": 0.106,
+      "grad_norm": 1.4423048496246338,
+      "learning_rate": 6.871270127375409e-05,
+      "loss": 17.8339,
+      "step": 1060
+    },
+    {
+      "epoch": 0.107,
+      "grad_norm": 1.2477686405181885,
+      "learning_rate": 6.839055837767724e-05,
+      "loss": 17.8775,
+      "step": 1070
+    },
+    {
+      "epoch": 0.108,
+      "grad_norm": 1.5547492504119873,
+      "learning_rate": 6.807290427601058e-05,
+      "loss": 16.7857,
+      "step": 1080
+    },
+    {
+      "epoch": 0.109,
+      "grad_norm": 2.1341335773468018,
+      "learning_rate": 6.775963568181182e-05,
+      "loss": 17.0301,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.172020673751831,
+      "learning_rate": 6.74506526050837e-05,
+      "loss": 17.62,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 4.378727912902832,
+      "eval_runtime": 6.2451,
+      "eval_samples_per_second": 320.249,
+      "eval_steps_per_second": 20.016,
+      "step": 1100
+    },
+    {
+      "epoch": 0.111,
+      "grad_norm": 1.4341588020324707,
+      "learning_rate": 6.714585821868878e-05,
+      "loss": 17.2087,
+      "step": 1110
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 1.8107764720916748,
+      "learning_rate": 6.68451587308695e-05,
+      "loss": 17.0762,
+      "step": 1120
+    },
+    {
+      "epoch": 0.113,
+      "grad_norm": 1.6209523677825928,
+      "learning_rate": 6.654846326399234e-05,
+      "loss": 17.8408,
+      "step": 1130
+    },
+    {
+      "epoch": 0.114,
+      "grad_norm": 1.4772510528564453,
+      "learning_rate": 6.625568373916034e-05,
+      "loss": 17.497,
+      "step": 1140
+    },
+    {
+      "epoch": 0.115,
+      "grad_norm": 1.7777224779129028,
+      "learning_rate": 6.596673476636102e-05,
+      "loss": 17.0651,
+      "step": 1150
+    },
+    {
+      "epoch": 0.115,
+      "eval_loss": 4.317691326141357,
+      "eval_runtime": 6.1999,
+      "eval_samples_per_second": 322.587,
+      "eval_steps_per_second": 20.162,
+      "step": 1150
+    },
+    {
+      "epoch": 0.116,
+      "grad_norm": 1.7212860584259033,
+      "learning_rate": 6.568153353983866e-05,
+      "loss": 17.0681,
+      "step": 1160
+    },
+    {
+      "epoch": 0.117,
+      "grad_norm": 1.2618902921676636,
+      "learning_rate": 6.53999997384e-05,
+      "loss": 16.4462,
+      "step": 1170
+    },
+    {
+      "epoch": 0.118,
+      "grad_norm": 1.5486454963684082,
+      "learning_rate": 6.512205543038029e-05,
+      "loss": 17.3137,
+      "step": 1180
+    },
+    {
+      "epoch": 0.119,
+      "grad_norm": 1.424975037574768,
+      "learning_rate": 6.48476249830151e-05,
+      "loss": 17.0415,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4205785989761353,
+      "learning_rate": 6.457663497597783e-05,
+      "loss": 16.9146,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 4.270516395568848,
+      "eval_runtime": 6.2291,
+      "eval_samples_per_second": 321.073,
+      "eval_steps_per_second": 20.067,
+      "step": 1200
+    },
+    {
+      "epoch": 0.121,
+      "grad_norm": 1.6844050884246826,
+      "learning_rate": 6.430901411885911e-05,
+      "loss": 17.1692,
+      "step": 1210
+    },
+    {
+      "epoch": 0.122,
+      "grad_norm": 1.607489824295044,
+      "learning_rate": 6.40446931723768e-05,
+      "loss": 16.319,
+      "step": 1220
+    },
+    {
+      "epoch": 0.123,
+      "grad_norm": 1.5149353742599487,
+      "learning_rate": 6.378360487311965e-05,
+      "loss": 16.5953,
+      "step": 1230
+    },
+    {
+      "epoch": 0.124,
+      "grad_norm": 1.5084587335586548,
+      "learning_rate": 6.352568386163805e-05,
+      "loss": 17.0408,
+      "step": 1240
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 1.342761516571045,
+      "learning_rate": 6.327086661370808e-05,
+      "loss": 16.8208,
+      "step": 1250
+    },
+    {
+      "epoch": 0.125,
+      "eval_loss": 4.237093448638916,
+      "eval_runtime": 6.2362,
+      "eval_samples_per_second": 320.71,
+      "eval_steps_per_second": 20.044,
+      "step": 1250
+    },
+    {
+      "epoch": 0.126,
+      "grad_norm": 1.3136533498764038,
+      "learning_rate": 6.301909137460409e-05,
+      "loss": 16.7708,
+      "step": 1260
+    },
+    {
+      "epoch": 0.127,
+      "grad_norm": 1.4737874269485474,
+      "learning_rate": 6.277029809622579e-05,
+      "loss": 17.0393,
+      "step": 1270
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 1.409523367881775,
+      "learning_rate": 6.252442837693433e-05,
+      "loss": 16.7568,
+      "step": 1280
+    },
+    {
+      "epoch": 0.129,
+      "grad_norm": 1.3414747714996338,
+      "learning_rate": 6.22814254039606e-05,
+      "loss": 16.3892,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1919307708740234,
+      "learning_rate": 6.204123389825647e-05,
+      "loss": 16.6787,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13,
+      "eval_loss": 4.198209762573242,
+      "eval_runtime": 6.2223,
+      "eval_samples_per_second": 321.426,
+      "eval_steps_per_second": 20.089,
+      "step": 1300
+    },
+    {
+      "epoch": 0.131,
+      "grad_norm": 1.3193895816802979,
+      "learning_rate": 6.180380006166808e-05,
+      "loss": 16.7276,
+      "step": 1310
+    },
+    {
+      "epoch": 0.132,
+      "grad_norm": 1.4551233053207397,
+      "learning_rate": 6.156907152631576e-05,
+      "loss": 16.5019,
+      "step": 1320
+    },
+    {
+      "epoch": 0.133,
+      "grad_norm": 1.6626161336898804,
+      "learning_rate": 6.133699730607301e-05,
+      "loss": 16.7516,
+      "step": 1330
+    },
+    {
+      "epoch": 0.134,
+      "grad_norm": 1.442935824394226,
+      "learning_rate": 6.110752775004223e-05,
+      "loss": 16.5704,
+      "step": 1340
+    },
+    {
+      "epoch": 0.135,
+      "grad_norm": 1.667401909828186,
+      "learning_rate": 6.088061449793082e-05,
+      "loss": 16.4191,
+      "step": 1350
+    },
+    {
+      "epoch": 0.135,
+      "eval_loss": 4.162066459655762,
+      "eval_runtime": 6.2174,
+      "eval_samples_per_second": 321.679,
+      "eval_steps_per_second": 20.105,
+      "step": 1350
+    },
+    {
+      "epoch": 0.136,
+      "grad_norm": 1.3251672983169556,
+      "learning_rate": 6.065621043723658e-05,
+      "loss": 16.1052,
+      "step": 1360
+    },
+    {
+      "epoch": 0.137,
+      "grad_norm": 1.4112827777862549,
+      "learning_rate": 6.043426966215649e-05,
+      "loss": 16.4661,
+      "step": 1370
+    },
+    {
+      "epoch": 0.138,
+      "grad_norm": 1.4340181350708008,
+      "learning_rate": 6.021474743413714e-05,
+      "loss": 16.2721,
+      "step": 1380
+    },
+    {
+      "epoch": 0.139,
+      "grad_norm": 1.4311639070510864,
+      "learning_rate": 5.999760014399041e-05,
+      "loss": 16.1731,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.649147868156433,
+      "learning_rate": 5.978278527550084e-05,
+      "loss": 16.3638,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 4.139286518096924,
+      "eval_runtime": 6.2422,
+      "eval_samples_per_second": 320.402,
+      "eval_steps_per_second": 20.025,
+      "step": 1400
+    },
+    {
+      "epoch": 0.141,
+      "grad_norm": 1.2151726484298706,
+      "learning_rate": 5.957026137045648e-05,
+      "loss": 16.2606,
+      "step": 1410
+    },
+    {
+      "epoch": 0.142,
+      "grad_norm": 1.3587441444396973,
+      "learning_rate": 5.935998799503725e-05,
+      "loss": 16.083,
+      "step": 1420
+    },
+    {
+      "epoch": 0.143,
+      "grad_norm": 1.181125283241272,
+      "learning_rate": 5.91519257074994e-05,
+      "loss": 16.117,
+      "step": 1430
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 1.076505184173584,
+      "learning_rate": 5.8946036027097295e-05,
+      "loss": 15.7447,
+      "step": 1440
+    },
+    {
+      "epoch": 0.145,
+      "grad_norm": 1.402579665184021,
+      "learning_rate": 5.8742281404186785e-05,
+      "loss": 16.4264,
+      "step": 1450
+    },
+    {
+      "epoch": 0.145,
+      "eval_loss": 4.133431434631348,
+      "eval_runtime": 6.1793,
+      "eval_samples_per_second": 323.661,
+      "eval_steps_per_second": 20.229,
+      "step": 1450
+    },
+    {
+      "epoch": 0.146,
+      "grad_norm": 1.0925723314285278,
+      "learning_rate": 5.8540625191457576e-05,
+      "loss": 16.6002,
+      "step": 1460
+    },
+    {
+      "epoch": 0.147,
+      "grad_norm": 1.100009799003601,
+      "learning_rate": 5.834103161624459e-05,
+      "loss": 16.3612,
+      "step": 1470
+    },
+    {
+      "epoch": 0.148,
+      "grad_norm": 1.91799795627594,
+      "learning_rate": 5.8143465753870694e-05,
+      "loss": 16.4593,
+      "step": 1480
+    },
+    {
+      "epoch": 0.149,
+      "grad_norm": 1.2950007915496826,
+      "learning_rate": 5.7947893501975715e-05,
+      "loss": 15.9469,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.4627280235290527,
+      "learning_rate": 5.7754281555789e-05,
+      "loss": 16.009,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 4.096825122833252,
+      "eval_runtime": 6.2109,
+      "eval_samples_per_second": 322.016,
+      "eval_steps_per_second": 20.126,
+      "step": 1500
+    },
+    {
+      "epoch": 0.151,
+      "grad_norm": 1.588494062423706,
+      "learning_rate": 5.756259738430475e-05,
+      "loss": 15.8375,
+      "step": 1510
+    },
+    {
+      "epoch": 0.152,
+      "grad_norm": 1.4053016901016235,
+      "learning_rate": 5.7372809207321355e-05,
+      "loss": 16.3388,
+      "step": 1520
+    },
+    {
+      "epoch": 0.153,
+      "grad_norm": 1.3402225971221924,
+      "learning_rate": 5.71848859733081e-05,
+      "loss": 16.3163,
+      "step": 1530
+    },
+    {
+      "epoch": 0.154,
+      "grad_norm": 1.195488691329956,
+      "learning_rate": 5.699879733806412e-05,
+      "loss": 16.056,
+      "step": 1540
+    },
+    {
+      "epoch": 0.155,
+      "grad_norm": 1.3463131189346313,
+      "learning_rate": 5.681451364413635e-05,
+      "loss": 16.0528,
+      "step": 1550
+    },
+    {
+      "epoch": 0.155,
+      "eval_loss": 4.0687785148620605,
+      "eval_runtime": 6.2217,
+      "eval_samples_per_second": 321.456,
+      "eval_steps_per_second": 20.091,
+      "step": 1550
+    },
+    {
+      "epoch": 0.156,
+      "grad_norm": 1.0611999034881592,
+      "learning_rate": 5.663200590096471e-05,
+      "loss": 15.7964,
+      "step": 1560
+    },
+    {
+      "epoch": 0.157,
+      "grad_norm": 1.2196369171142578,
+      "learning_rate": 5.645124576572452e-05,
+      "loss": 16.2541,
+      "step": 1570
+    },
+    {
+      "epoch": 0.158,
+      "grad_norm": 1.5234003067016602,
+      "learning_rate": 5.627220552483715e-05,
+      "loss": 15.9411,
+      "step": 1580
+    },
+    {
+      "epoch": 0.159,
+      "grad_norm": 1.2116581201553345,
+      "learning_rate": 5.609485807612173e-05,
+      "loss": 16.2016,
+      "step": 1590
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.3541016578674316,
+      "learning_rate": 5.591917691156175e-05,
+      "loss": 16.0676,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 4.064485549926758,
+      "eval_runtime": 6.2348,
+      "eval_samples_per_second": 320.778,
+      "eval_steps_per_second": 20.049,
+      "step": 1600
+    },
+    {
+      "epoch": 0.161,
+      "grad_norm": 1.320177674293518,
+      "learning_rate": 5.5745136100661674e-05,
+      "loss": 15.6234,
+      "step": 1610
+    },
+    {
+      "epoch": 0.162,
+      "grad_norm": 1.6600571870803833,
+      "learning_rate": 5.557271027436971e-05,
+      "loss": 15.4577,
+      "step": 1620
+    },
+    {
+      "epoch": 0.163,
+      "grad_norm": 1.194664478302002,
+      "learning_rate": 5.540187460954447e-05,
+      "loss": 15.8888,
+      "step": 1630
+    },
+    {
+      "epoch": 0.164,
+      "grad_norm": 1.5018142461776733,
+      "learning_rate": 5.523260481394348e-05,
+      "loss": 15.5123,
+      "step": 1640
+    },
+    {
+      "epoch": 0.165,
+      "grad_norm": 1.5969749689102173,
+      "learning_rate": 5.506487711171322e-05,
+      "loss": 15.4382,
+      "step": 1650
+    },
+    {
+      "epoch": 0.165,
+      "eval_loss": 4.040594577789307,
+      "eval_runtime": 6.2211,
+      "eval_samples_per_second": 321.486,
+      "eval_steps_per_second": 20.093,
+      "step": 1650
+    },
+    {
+      "epoch": 0.166,
+      "grad_norm": 1.6020385026931763,
+      "learning_rate": 5.489866822936095e-05,
+      "loss": 16.4235,
+      "step": 1660
+    },
+    {
+      "epoch": 0.167,
+      "grad_norm": 1.2778631448745728,
+      "learning_rate": 5.4733955382189484e-05,
+      "loss": 15.8776,
+      "step": 1670
+    },
+    {
+      "epoch": 0.168,
+      "grad_norm": 1.5798327922821045,
+      "learning_rate": 5.457071626117703e-05,
+      "loss": 15.4616,
+      "step": 1680
+    },
+    {
+      "epoch": 0.169,
+      "grad_norm": 1.3043429851531982,
+      "learning_rate": 5.440892902028488e-05,
+      "loss": 16.2131,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.4358928203582764,
+      "learning_rate": 5.424857226417659e-05,
+      "loss": 15.7224,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17,
+      "eval_loss": 4.0164995193481445,
+      "eval_runtime": 6.1969,
+      "eval_samples_per_second": 322.742,
+      "eval_steps_per_second": 20.171,
+      "step": 1700
+    },
+    {
+      "epoch": 0.171,
+      "grad_norm": 1.1965007781982422,
+      "learning_rate": 5.408962503633292e-05,
+      "loss": 15.134,
+      "step": 1710
+    },
+    {
+      "epoch": 0.172,
+      "grad_norm": 1.2486337423324585,
+      "learning_rate": 5.39320668075478e-05,
+      "loss": 16.0313,
+      "step": 1720
+    },
+    {
+      "epoch": 0.173,
+      "grad_norm": 1.285971999168396,
+      "learning_rate": 5.3775877464790436e-05,
+      "loss": 15.5165,
+      "step": 1730
+    },
+    {
+      "epoch": 0.174,
+      "grad_norm": 1.28692626953125,
+      "learning_rate": 5.362103730042052e-05,
+      "loss": 15.7254,
+      "step": 1740
+    },
+    {
+      "epoch": 0.175,
+      "grad_norm": 1.1224619150161743,
+      "learning_rate": 5.346752700174288e-05,
+      "loss": 15.3909,
+      "step": 1750
+    },
+    {
+      "epoch": 0.175,
+      "eval_loss": 4.034187316894531,
+      "eval_runtime": 6.2137,
+      "eval_samples_per_second": 321.872,
+      "eval_steps_per_second": 20.117,
+      "step": 1750
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 1.2035942077636719,
+      "learning_rate": 5.331532764088928e-05,
+      "loss": 15.7032,
+      "step": 1760
+    },
+    {
+      "epoch": 0.177,
+      "grad_norm": 1.5655755996704102,
+      "learning_rate": 5.316442066501519e-05,
+      "loss": 15.5386,
+      "step": 1770
+    },
+    {
+      "epoch": 0.178,
+      "grad_norm": 1.363478660583496,
+      "learning_rate": 5.30147878868001e-05,
+      "loss": 15.736,
+      "step": 1780
+    },
+    {
+      "epoch": 0.179,
+      "grad_norm": 1.1897183656692505,
+      "learning_rate": 5.2866411475240354e-05,
+      "loss": 15.4589,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.4313431978225708,
+      "learning_rate": 5.2719273946723746e-05,
+      "loss": 15.9495,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 3.990009307861328,
+      "eval_runtime": 6.2278,
+      "eval_samples_per_second": 321.142,
+      "eval_steps_per_second": 20.071,
+      "step": 1800
+    },
+    {
+      "epoch": 0.181,
+      "grad_norm": 1.5434056520462036,
+      "learning_rate": 5.257335815637598e-05,
+      "loss": 15.8692,
+      "step": 1810
+    },
+    {
+      "epoch": 0.182,
+      "grad_norm": 1.4785411357879639,
+      "learning_rate": 5.242864728966902e-05,
+      "loss": 16.1013,
+      "step": 1820
+    },
+    {
+      "epoch": 0.183,
+      "grad_norm": 1.264227032661438,
+      "learning_rate": 5.2285124854282266e-05,
+      "loss": 15.9116,
+      "step": 1830
+    },
+    {
+      "epoch": 0.184,
+      "grad_norm": 1.4657548666000366,
+      "learning_rate": 5.2142774672207326e-05,
+      "loss": 15.8593,
+      "step": 1840
+    },
+    {
+      "epoch": 0.185,
+      "grad_norm": 1.4138890504837036,
+      "learning_rate": 5.200158087208814e-05,
+      "loss": 15.9445,
+      "step": 1850
+    },
+    {
+      "epoch": 0.185,
+      "eval_loss": 3.9676108360290527,
+      "eval_runtime": 6.205,
+      "eval_samples_per_second": 322.322,
+      "eval_steps_per_second": 20.145,
+      "step": 1850
+    },
+    {
+      "epoch": 0.186,
+      "grad_norm": 1.3209441900253296,
+      "learning_rate": 5.186152788178785e-05,
+      "loss": 15.2406,
+      "step": 1860
+    },
+    {
+      "epoch": 0.187,
+      "grad_norm": 1.2953730821609497,
+      "learning_rate": 5.172260042117486e-05,
+      "loss": 15.5566,
+      "step": 1870
+    },
+    {
+      "epoch": 0.188,
+      "grad_norm": 1.1381759643554688,
+      "learning_rate": 5.1584783495120195e-05,
+      "loss": 16.0945,
+      "step": 1880
+    },
+    {
+      "epoch": 0.189,
+      "grad_norm": 1.1710163354873657,
+      "learning_rate": 5.1448062386699125e-05,
+      "loss": 15.9861,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.3093607425689697,
+      "learning_rate": 5.1312422650589934e-05,
+      "loss": 15.7921,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 3.962754487991333,
+      "eval_runtime": 6.2302,
+      "eval_samples_per_second": 321.018,
+      "eval_steps_per_second": 20.064,
+      "step": 1900
+    },
+    {
+      "epoch": 0.191,
+      "grad_norm": 1.4274228811264038,
+      "learning_rate": 5.117785010666307e-05,
+      "loss": 15.6979,
+      "step": 1910
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 1.5318870544433594,
+      "learning_rate": 5.104433083375434e-05,
+      "loss": 15.4107,
+      "step": 1920
+    },
+    {
+      "epoch": 0.193,
+      "grad_norm": 1.5562711954116821,
+      "learning_rate": 5.091185116361582e-05,
+      "loss": 15.5957,
+      "step": 1930
+    },
+    {
+      "epoch": 0.194,
+      "grad_norm": 1.5389816761016846,
+      "learning_rate": 5.0780397675038636e-05,
+      "loss": 15.5794,
+      "step": 1940
+    },
+    {
+      "epoch": 0.195,
+      "grad_norm": 1.1989519596099854,
+      "learning_rate": 5.0649957188141786e-05,
+      "loss": 15.3292,
+      "step": 1950
+    },
+    {
+      "epoch": 0.195,
+      "eval_loss": 3.9427313804626465,
+      "eval_runtime": 6.2462,
+      "eval_samples_per_second": 320.195,
+      "eval_steps_per_second": 20.012,
+      "step": 1950
+    },
+    {
+      "epoch": 0.196,
+      "grad_norm": 1.64460027217865,
+      "learning_rate": 5.052051675882142e-05,
+      "loss": 15.4911,
+      "step": 1960
+    },
+    {
+      "epoch": 0.197,
+      "grad_norm": 1.2332730293273926,
+      "learning_rate": 5.0392063673355584e-05,
+      "loss": 15.5501,
+      "step": 1970
+    },
+    {
+      "epoch": 0.198,
+      "grad_norm": 1.463189959526062,
+      "learning_rate": 5.026458544315881e-05,
+      "loss": 15.6168,
+      "step": 1980
+    },
+    {
+      "epoch": 0.199,
+      "grad_norm": 1.3763878345489502,
+      "learning_rate": 5.013806979968224e-05,
+      "loss": 15.4993,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.0232411623001099,
+      "learning_rate": 5.0012504689453974e-05,
+      "loss": 15.5746,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 3.9369847774505615,
+      "eval_runtime": 6.2415,
+      "eval_samples_per_second": 320.436,
+      "eval_steps_per_second": 20.027,
+      "step": 2000
+    },
+    {
+      "epoch": 0.201,
+      "grad_norm": 1.4828171730041504,
+      "learning_rate": 4.988787826925559e-05,
+      "loss": 15.9385,
+      "step": 2010
+    },
+    {
+      "epoch": 0.202,
+      "grad_norm": 1.041183590888977,
+      "learning_rate": 4.9764178901430006e-05,
+      "loss": 15.8048,
+      "step": 2020
+    },
+    {
+      "epoch": 0.203,
+      "grad_norm": 1.3011362552642822,
+      "learning_rate": 4.9641395149316935e-05,
+      "loss": 15.6496,
+      "step": 2030
+    },
+    {
+      "epoch": 0.204,
+      "grad_norm": 1.38265860080719,
+      "learning_rate": 4.951951577281135e-05,
+      "loss": 15.4042,
+      "step": 2040
+    },
+    {
+      "epoch": 0.205,
+      "grad_norm": 1.2859934568405151,
+      "learning_rate": 4.939852972404162e-05,
+      "loss": 15.8823,
+      "step": 2050
+    },
+    {
+      "epoch": 0.205,
+      "eval_loss": 3.9209940433502197,
+      "eval_runtime": 7.0368,
+      "eval_samples_per_second": 284.218,
+      "eval_steps_per_second": 17.764,
+      "step": 2050
+    },
+    {
+      "epoch": 0.206,
+      "grad_norm": 1.7220138311386108,
+      "learning_rate": 4.927842614316289e-05,
+      "loss": 15.5192,
+      "step": 2060
+    },
+    {
+      "epoch": 0.207,
+      "grad_norm": 1.396269679069519,
+      "learning_rate": 4.9159194354262706e-05,
+      "loss": 15.5069,
+      "step": 2070
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 1.146694540977478,
+      "learning_rate": 4.904082386137498e-05,
+      "loss": 15.3511,
+      "step": 2080
+    },
+    {
+      "epoch": 0.209,
+      "grad_norm": 1.231331467628479,
+      "learning_rate": 4.892330434459896e-05,
+      "loss": 15.0852,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.0101878643035889,
+      "learning_rate": 4.880662565632016e-05,
+      "loss": 15.1614,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 3.9238545894622803,
+      "eval_runtime": 6.2013,
+      "eval_samples_per_second": 322.513,
+      "eval_steps_per_second": 20.157,
+      "step": 2100
+    },
+    {
+      "epoch": 0.211,
+      "grad_norm": 1.128787875175476,
+      "learning_rate": 4.869077781752983e-05,
+      "loss": 14.8515,
+      "step": 2110
+    },
+    {
+      "epoch": 0.212,
+      "grad_norm": 1.4294734001159668,
+      "learning_rate": 4.857575101424013e-05,
+      "loss": 14.8838,
+      "step": 2120
+    },
+    {
+      "epoch": 0.213,
+      "grad_norm": 1.6185399293899536,
+      "learning_rate": 4.846153559399206e-05,
+      "loss": 15.0193,
+      "step": 2130
+    },
+    {
+      "epoch": 0.214,
+      "grad_norm": 1.1443840265274048,
+      "learning_rate": 4.834812206245318e-05,
+      "loss": 15.2785,
+      "step": 2140
+    },
+    {
+      "epoch": 0.215,
+      "grad_norm": 1.3191642761230469,
+      "learning_rate": 4.8235501080102624e-05,
+      "loss": 15.1923,
+      "step": 2150
+    },
+    {
+      "epoch": 0.215,
+      "eval_loss": 3.9039673805236816,
+      "eval_runtime": 6.2499,
+      "eval_samples_per_second": 320.003,
+      "eval_steps_per_second": 20.0,
+      "step": 2150
+    },
+    {
+      "epoch": 0.216,
+      "grad_norm": 1.6260229349136353,
+      "learning_rate": 4.812366345900056e-05,
+      "loss": 15.2295,
+      "step": 2160
+    },
+    {
+      "epoch": 0.217,
+      "grad_norm": 1.3886921405792236,
+      "learning_rate": 4.801260015963979e-05,
+      "loss": 15.149,
+      "step": 2170
+    },
+    {
+      "epoch": 0.218,
+      "grad_norm": 1.159067988395691,
+      "learning_rate": 4.790230228787671e-05,
+      "loss": 14.3841,
+      "step": 2180
+    },
+    {
+      "epoch": 0.219,
+      "grad_norm": 1.0690653324127197,
+      "learning_rate": 4.779276109193975e-05,
+      "loss": 14.8772,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.275758147239685,
+      "learning_rate": 4.7683967959512366e-05,
+      "loss": 15.0286,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22,
+      "eval_loss": 3.9052722454071045,
+      "eval_runtime": 6.1987,
+      "eval_samples_per_second": 322.65,
+      "eval_steps_per_second": 20.166,
+      "step": 2200
+    },
+    {
+      "epoch": 0.221,
+      "grad_norm": 1.3399527072906494,
+      "learning_rate": 4.757591441488914e-05,
+      "loss": 15.691,
+      "step": 2210
+    },
+    {
+      "epoch": 0.222,
+      "grad_norm": 1.5562835931777954,
+      "learning_rate": 4.746859211620213e-05,
+      "loss": 15.8526,
+      "step": 2220
+    },
+    {
+      "epoch": 0.223,
+      "grad_norm": 1.3359469175338745,
+      "learning_rate": 4.7361992852715864e-05,
+      "loss": 15.4991,
+      "step": 2230
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 1.1983004808425903,
+      "learning_rate": 4.7256108542188905e-05,
+      "loss": 15.2238,
+      "step": 2240
+    },
+    {
+      "epoch": 0.225,
+      "grad_norm": 1.349929690361023,
+      "learning_rate": 4.715093122829988e-05,
+      "loss": 15.3992,
+      "step": 2250
+    },
+    {
+      "epoch": 0.225,
+      "eval_loss": 3.880711317062378,
+      "eval_runtime": 6.2762,
+      "eval_samples_per_second": 318.664,
+      "eval_steps_per_second": 19.916,
+      "step": 2250
+    },
+    {
+      "epoch": 0.226,
+      "grad_norm": 1.2029528617858887,
+      "learning_rate": 4.704645307813639e-05,
+      "loss": 15.5645,
+      "step": 2260
+    },
+    {
+      "epoch": 0.227,
+      "grad_norm": 1.2777271270751953,
+      "learning_rate": 4.6942666379744746e-05,
+      "loss": 15.4002,
+      "step": 2270
+    },
+    {
+      "epoch": 0.228,
+      "grad_norm": 1.5176602602005005,
+      "learning_rate": 4.683956353973895e-05,
+      "loss": 15.6002,
+      "step": 2280
+    },
+    {
+      "epoch": 0.229,
+      "grad_norm": 1.6124011278152466,
+      "learning_rate": 4.6737137080967207e-05,
+      "loss": 15.335,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.8375682830810547,
+      "learning_rate": 4.663537964023428e-05,
+      "loss": 15.6535,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 3.8743247985839844,
+      "eval_runtime": 6.2459,
+      "eval_samples_per_second": 320.209,
+      "eval_steps_per_second": 20.013,
+      "step": 2300
+    },
+    {
+      "epoch": 0.231,
+      "grad_norm": 1.488021969795227,
+      "learning_rate": 4.653428396607817e-05,
+      "loss": 15.6247,
+      "step": 2310
+    },
+    {
+      "epoch": 0.232,
+      "grad_norm": 1.5195178985595703,
+      "learning_rate": 4.643384291659964e-05,
+      "loss": 15.2655,
+      "step": 2320
+    },
+    {
+      "epoch": 0.233,
+      "grad_norm": 1.240750789642334,
+      "learning_rate": 4.6334049457342925e-05,
+      "loss": 14.4992,
+      "step": 2330
+    },
+    {
+      "epoch": 0.234,
+      "grad_norm": 1.3504952192306519,
+      "learning_rate": 4.623489665922651e-05,
+      "loss": 15.07,
+      "step": 2340
+    },
+    {
+      "epoch": 0.235,
+      "grad_norm": 1.385376214981079,
+      "learning_rate": 4.613637769652221e-05,
+      "loss": 15.1405,
+      "step": 2350
+    },
+    {
+      "epoch": 0.235,
+      "eval_loss": 3.872030735015869,
+      "eval_runtime": 6.2626,
+      "eval_samples_per_second": 319.357,
+      "eval_steps_per_second": 19.96,
+      "step": 2350
+    },
+    {
+      "epoch": 0.236,
+      "grad_norm": 1.143419861793518,
+      "learning_rate": 4.603848584488156e-05,
+      "loss": 14.7073,
+      "step": 2360
+    },
+    {
+      "epoch": 0.237,
+      "grad_norm": 1.4339746236801147,
+      "learning_rate": 4.594121447940805e-05,
+      "loss": 14.9982,
+      "step": 2370
+    },
+    {
+      "epoch": 0.238,
+      "grad_norm": 1.5718977451324463,
+      "learning_rate": 4.58445570727739e-05,
+      "loss": 14.811,
+      "step": 2380
+    },
+    {
+      "epoch": 0.239,
+      "grad_norm": 1.4219948053359985,
+      "learning_rate": 4.574850719338033e-05,
+      "loss": 15.0809,
+      "step": 2390
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4401710033416748,
+      "learning_rate": 4.5653058503559986e-05,
+      "loss": 15.3598,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 3.85036563873291,
+      "eval_runtime": 6.2378,
+      "eval_samples_per_second": 320.628,
+      "eval_steps_per_second": 20.039,
+      "step": 2400
+    },
+    {
+      "epoch": 0.241,
+      "grad_norm": 1.5921422243118286,
+      "learning_rate": 4.555820475782052e-05,
+      "loss": 15.2551,
+      "step": 2410
+    },
+    {
+      "epoch": 0.242,
+      "grad_norm": 1.5687378644943237,
+      "learning_rate": 4.5463939801128054e-05,
+      "loss": 15.1761,
+      "step": 2420
+    },
+    {
+      "epoch": 0.243,
+      "grad_norm": 1.3394169807434082,
+      "learning_rate": 4.537025756722961e-05,
+      "loss": 15.4282,
+      "step": 2430
+    },
+    {
+      "epoch": 0.244,
+      "grad_norm": 1.3350077867507935,
+      "learning_rate": 4.527715207701344e-05,
+      "loss": 15.8026,
+      "step": 2440
+    },
+    {
+      "epoch": 0.245,
+      "grad_norm": 1.2768425941467285,
+      "learning_rate": 4.51846174369061e-05,
+      "loss": 15.2332,
+      "step": 2450
+    },
+    {
+      "epoch": 0.245,
+      "eval_loss": 3.8509819507598877,
+      "eval_runtime": 6.2712,
+      "eval_samples_per_second": 318.92,
+      "eval_steps_per_second": 19.933,
+      "step": 2450
+    },
+    {
+      "epoch": 0.246,
+      "grad_norm": 1.13394033908844,
+      "learning_rate": 4.5092647837305585e-05,
+      "loss": 14.7858,
+      "step": 2460
+    },
+    {
+      "epoch": 0.247,
+      "grad_norm": 1.117067575454712,
+      "learning_rate": 4.500123755104922e-05,
+      "loss": 15.0364,
+      "step": 2470
+    },
+    {
+      "epoch": 0.248,
+      "grad_norm": 1.2777060270309448,
+      "learning_rate": 4.4910380931915706e-05,
+      "loss": 15.4375,
+      "step": 2480
+    },
+    {
+      "epoch": 0.249,
+      "grad_norm": 1.1972999572753906,
+      "learning_rate": 4.4820072413160295e-05,
+      "loss": 15.2635,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.1700772047042847,
+      "learning_rate": 4.4730306506082105e-05,
+      "loss": 15.3477,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 3.8322501182556152,
+      "eval_runtime": 6.2555,
+      "eval_samples_per_second": 319.718,
+      "eval_steps_per_second": 19.982,
+      "step": 2500
+    },
+    {
+      "epoch": 0.251,
+      "grad_norm": 1.5145503282546997,
+      "learning_rate": 4.4641077798623075e-05,
+      "loss": 14.9236,
+      "step": 2510
+    },
+    {
+      "epoch": 0.252,
+      "grad_norm": 1.3210171461105347,
+      "learning_rate": 4.455238095399737e-05,
+      "loss": 14.8855,
+      "step": 2520
+    },
+    {
+      "epoch": 0.253,
+      "grad_norm": 1.3452820777893066,
+      "learning_rate": 4.446421070935069e-05,
+      "loss": 15.3636,
+      "step": 2530
+    },
+    {
+      "epoch": 0.254,
+      "grad_norm": 1.3298344612121582,
+      "learning_rate": 4.437656187444866e-05,
+      "loss": 15.6724,
+      "step": 2540
+    },
+    {
+      "epoch": 0.255,
+      "grad_norm": 1.1179172992706299,
+      "learning_rate": 4.428942933039357e-05,
+      "loss": 15.4457,
+      "step": 2550
+    },
+    {
+      "epoch": 0.255,
+      "eval_loss": 3.829561471939087,
+      "eval_runtime": 8.4255,
+      "eval_samples_per_second": 237.373,
+      "eval_steps_per_second": 14.836,
+      "step": 2550
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 1.2972711324691772,
+      "learning_rate": 4.420280802836862e-05,
+      "loss": 14.9027,
+      "step": 2560
+    },
+    {
+      "epoch": 0.257,
+      "grad_norm": 1.156786322593689,
+      "learning_rate": 4.4116692988409206e-05,
+      "loss": 14.8428,
+      "step": 2570
+    },
+    {
+      "epoch": 0.258,
+      "grad_norm": 1.2626519203186035,
+      "learning_rate": 4.4031079298200316e-05,
+      "loss": 15.1462,
+      "step": 2580
+    },
+    {
+      "epoch": 0.259,
+      "grad_norm": 1.174372673034668,
+      "learning_rate": 4.394596211189963e-05,
+      "loss": 15.5364,
+      "step": 2590
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.359827995300293,
+      "learning_rate": 4.3861336648985394e-05,
+      "loss": 15.2199,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 3.823606014251709,
+      "eval_runtime": 6.2399,
+      "eval_samples_per_second": 320.517,
+      "eval_steps_per_second": 20.032,
+      "step": 2600
+    },
+    {
+      "epoch": 0.261,
+      "grad_norm": 1.157862901687622,
+      "learning_rate": 4.3777198193128756e-05,
+      "loss": 14.998,
+      "step": 2610
+    },
+    {
+      "epoch": 0.262,
+      "grad_norm": 1.2194920778274536,
+      "learning_rate": 4.369354209108969e-05,
+      "loss": 14.8853,
+      "step": 2620
+    },
+    {
+      "epoch": 0.263,
+      "grad_norm": 1.1531281471252441,
+      "learning_rate": 4.3610363751636137e-05,
+      "loss": 14.8103,
+      "step": 2630
+    },
+    {
+      "epoch": 0.264,
+      "grad_norm": 0.967728316783905,
+      "learning_rate": 4.352765864448559e-05,
+      "loss": 14.9321,
+      "step": 2640
+    },
+    {
+      "epoch": 0.265,
+      "grad_norm": 1.277216911315918,
+      "learning_rate": 4.344542229926874e-05,
+      "loss": 14.5667,
+      "step": 2650
+    },
+    {
+      "epoch": 0.265,
+      "eval_loss": 3.816464900970459,
+      "eval_runtime": 6.1924,
+      "eval_samples_per_second": 322.979,
+      "eval_steps_per_second": 20.186,
+      "step": 2650
+    },
+    {
+      "epoch": 0.266,
+      "grad_norm": 1.4289199113845825,
+      "learning_rate": 4.336365030451462e-05,
+      "loss": 15.0527,
+      "step": 2660
+    },
+    {
+      "epoch": 0.267,
+      "grad_norm": 1.1268882751464844,
+      "learning_rate": 4.3282338306656564e-05,
+      "loss": 14.8036,
+      "step": 2670
+    },
+    {
+      "epoch": 0.268,
+      "grad_norm": 1.1748231649398804,
+      "learning_rate": 4.32014820090587e-05,
+      "loss": 14.4974,
+      "step": 2680
+    },
+    {
+      "epoch": 0.269,
+      "grad_norm": 1.355406403541565,
+      "learning_rate": 4.312107717106231e-05,
+      "loss": 14.7474,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.3552896976470947,
+      "learning_rate": 4.30411196070517e-05,
+      "loss": 15.0418,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 3.800353527069092,
+      "eval_runtime": 6.2378,
+      "eval_samples_per_second": 320.624,
+      "eval_steps_per_second": 20.039,
+      "step": 2700
+    },
+    {
+      "epoch": 0.271,
+      "grad_norm": 1.2844239473342896,
+      "learning_rate": 4.296160518553892e-05,
+      "loss": 15.5347,
+      "step": 2710
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 1.2744359970092773,
+      "learning_rate": 4.288252982826728e-05,
+      "loss": 15.3665,
+      "step": 2720
+    },
+    {
+      "epoch": 0.273,
+      "grad_norm": 1.216260313987732,
+      "learning_rate": 4.2803889509332595e-05,
+      "loss": 14.944,
+      "step": 2730
+    },
+    {
+      "epoch": 0.274,
+      "grad_norm": 1.2304832935333252,
+      "learning_rate": 4.2725680254322424e-05,
+      "loss": 14.9361,
+      "step": 2740
+    },
+    {
+      "epoch": 0.275,
+      "grad_norm": 1.1939611434936523,
+      "learning_rate": 4.2647898139472306e-05,
+      "loss": 15.0726,
+      "step": 2750
+    },
+    {
+      "epoch": 0.275,
+      "eval_loss": 3.7999727725982666,
+      "eval_runtime": 6.2305,
+      "eval_samples_per_second": 321.001,
+      "eval_steps_per_second": 20.063,
+      "step": 2750
+    },
+    {
+      "epoch": 0.276,
+      "grad_norm": 1.5539189577102661,
+      "learning_rate": 4.2570539290839e-05,
+      "loss": 14.7788,
+      "step": 2760
+    },
+    {
+      "epoch": 0.277,
+      "grad_norm": 1.3662588596343994,
+      "learning_rate": 4.2493599883490024e-05,
+      "loss": 15.1841,
+      "step": 2770
+    },
+    {
+      "epoch": 0.278,
+      "grad_norm": 1.1410843133926392,
+      "learning_rate": 4.241707614070937e-05,
+      "loss": 14.9416,
+      "step": 2780
+    },
+    {
+      "epoch": 0.279,
+      "grad_norm": 1.1068998575210571,
+      "learning_rate": 4.234096433321879e-05,
+      "loss": 14.4817,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.329091191291809,
+      "learning_rate": 4.226526077841448e-05,
+      "loss": 14.5635,
+      "step": 2800
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 3.810391902923584,
+      "eval_runtime": 6.2132,
+      "eval_samples_per_second": 321.895,
+      "eval_steps_per_second": 20.118,
+      "step": 2800
+    },
+    {
+      "epoch": 0.281,
+      "grad_norm": 1.2415845394134521,
+      "learning_rate": 4.218996183961868e-05,
+      "loss": 14.6589,
+      "step": 2810
+    },
+    {
+      "epoch": 0.282,
+      "grad_norm": 1.3501887321472168,
+      "learning_rate": 4.2115063925345885e-05,
+      "loss": 15.0743,
+      "step": 2820
+    },
+    {
+      "epoch": 0.283,
+      "grad_norm": 0.9185730218887329,
+      "learning_rate": 4.204056348858339e-05,
+      "loss": 15.1358,
+      "step": 2830
+    },
+    {
+      "epoch": 0.284,
+      "grad_norm": 1.2723404169082642,
+      "learning_rate": 4.196645702608569e-05,
+      "loss": 14.9579,
+      "step": 2840
+    },
+    {
+      "epoch": 0.285,
+      "grad_norm": 1.2337239980697632,
+      "learning_rate": 4.18927410776826e-05,
+      "loss": 15.1503,
+      "step": 2850
+    },
+    {
+      "epoch": 0.285,
+      "eval_loss": 3.800377607345581,
+      "eval_runtime": 10.3739,
+      "eval_samples_per_second": 192.791,
+      "eval_steps_per_second": 12.049,
+      "step": 2850
+    },
+    {
+      "epoch": 0.286,
+      "grad_norm": 1.2774711847305298,
+      "learning_rate": 4.181941222560067e-05,
+      "loss": 15.5372,
+      "step": 2860
+    },
+    {
+      "epoch": 0.287,
+      "grad_norm": 1.3295443058013916,
+      "learning_rate": 4.1746467093797576e-05,
+      "loss": 14.8642,
+      "step": 2870
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 1.6138548851013184,
+      "learning_rate": 4.167390234730933e-05,
+      "loss": 14.1073,
+      "step": 2880
+    },
+    {
+      "epoch": 0.289,
+      "grad_norm": 1.4765127897262573,
+      "learning_rate": 4.1601714691609825e-05,
+      "loss": 15.0385,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.415253758430481,
+      "learning_rate": 4.152990087198261e-05,
+      "loss": 14.8696,
+      "step": 2900
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 3.7786402702331543,
+      "eval_runtime": 6.24,
+      "eval_samples_per_second": 320.514,
+      "eval_steps_per_second": 20.032,
+      "step": 2900
+    },
+    {
+      "epoch": 0.291,
+      "grad_norm": 1.3082252740859985,
+      "learning_rate": 4.145845767290457e-05,
+      "loss": 14.56,
+      "step": 2910
+    },
+    {
+      "epoch": 0.292,
+      "grad_norm": 1.446778416633606,
+      "learning_rate": 4.1387381917441095e-05,
+      "loss": 14.8571,
+      "step": 2920
+    },
+    {
+      "epoch": 0.293,
+      "grad_norm": 1.2211068868637085,
+      "learning_rate": 4.131667046665284e-05,
+      "loss": 14.6906,
+      "step": 2930
+    },
+    {
+      "epoch": 0.294,
+      "grad_norm": 1.1490166187286377,
+      "learning_rate": 4.1246320219013365e-05,
+      "loss": 14.5224,
+      "step": 2940
+    },
+    {
+      "epoch": 0.295,
+      "grad_norm": 0.9107897281646729,
+      "learning_rate": 4.117632810983782e-05,
+      "loss": 14.4431,
+      "step": 2950
+    },
+    {
+      "epoch": 0.295,
+      "eval_loss": 3.7727034091949463,
+      "eval_runtime": 6.217,
+      "eval_samples_per_second": 321.698,
+      "eval_steps_per_second": 20.106,
+      "step": 2950
+    },
+    {
+      "epoch": 0.296,
+      "grad_norm": 1.0651347637176514,
+      "learning_rate": 4.1106691110722134e-05,
+      "loss": 13.9779,
+      "step": 2960
+    },
+    {
+      "epoch": 0.297,
+      "grad_norm": 1.1543402671813965,
+      "learning_rate": 4.10374062289927e-05,
+      "loss": 14.586,
+      "step": 2970
+    },
+    {
+      "epoch": 0.298,
+      "grad_norm": 1.2242450714111328,
+      "learning_rate": 4.096847050716615e-05,
+      "loss": 14.2348,
+      "step": 2980
+    },
+    {
+      "epoch": 0.299,
+      "grad_norm": 1.3821468353271484,
+      "learning_rate": 4.089988102241916e-05,
+      "loss": 14.2154,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4441609382629395,
+      "learning_rate": 4.083163488606789e-05,
+      "loss": 14.9342,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 3.770812749862671,
+      "eval_runtime": 6.2111,
+      "eval_samples_per_second": 322.006,
+      "eval_steps_per_second": 20.125,
+      "step": 3000
+    },
+    {
+      "epoch": 0.301,
+      "grad_norm": 1.5218119621276855,
+      "learning_rate": 4.076372924305703e-05,
+      "loss": 14.6337,
+      "step": 3010
+    },
+    {
+      "epoch": 0.302,
+      "grad_norm": 1.123818039894104,
+      "learning_rate": 4.0696161271458113e-05,
+      "loss": 14.4631,
+      "step": 3020
+    },
+    {
+      "epoch": 0.303,
+      "grad_norm": 1.3742114305496216,
+      "learning_rate": 4.0628928181976975e-05,
+      "loss": 15.2754,
+      "step": 3030
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 1.4702446460723877,
+      "learning_rate": 4.0562027217470115e-05,
+      "loss": 15.2532,
+      "step": 3040
+    },
+    {
+      "epoch": 0.305,
+      "grad_norm": 1.0198601484298706,
+      "learning_rate": 4.0495455652469793e-05,
+      "loss": 15.0958,
+      "step": 3050
+    },
+    {
+      "epoch": 0.305,
+      "eval_loss": 3.7610201835632324,
+      "eval_runtime": 6.2419,
+      "eval_samples_per_second": 320.415,
+      "eval_steps_per_second": 20.026,
+      "step": 3050
+    },
+    {
+      "epoch": 0.306,
+      "grad_norm": 1.2997148036956787,
+      "learning_rate": 4.0429210792717695e-05,
+      "loss": 15.122,
+      "step": 3060
+    },
+    {
+      "epoch": 0.307,
+      "grad_norm": 1.367250919342041,
+      "learning_rate": 4.03632899747069e-05,
+      "loss": 14.633,
+      "step": 3070
+    },
+    {
+      "epoch": 0.308,
+      "grad_norm": 1.1207269430160522,
+      "learning_rate": 4.0297690565232124e-05,
+      "loss": 14.3819,
+      "step": 3080
+    },
+    {
+      "epoch": 0.309,
+      "grad_norm": 1.5955203771591187,
+      "learning_rate": 4.0232409960947924e-05,
+      "loss": 14.4362,
+      "step": 3090
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3510704040527344,
+      "learning_rate": 4.0167445587934724e-05,
+      "loss": 14.9327,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 3.744568347930908,
+      "eval_runtime": 6.2372,
+      "eval_samples_per_second": 320.657,
+      "eval_steps_per_second": 20.041,
+      "step": 3100
+    },
+    {
+      "epoch": 0.311,
+      "grad_norm": 1.2507836818695068,
+      "learning_rate": 4.0102794901272596e-05,
+      "loss": 14.4538,
+      "step": 3110
+    },
+    {
+      "epoch": 0.312,
+      "grad_norm": 1.300366997718811,
+      "learning_rate": 4.00384553846225e-05,
+      "loss": 14.086,
+      "step": 3120
+    },
+    {
+      "epoch": 0.313,
+      "grad_norm": 1.5376310348510742,
+      "learning_rate": 3.997442454981493e-05,
+      "loss": 14.7837,
+      "step": 3130
+    },
+    {
+      "epoch": 0.314,
+      "grad_norm": 1.0232264995574951,
+      "learning_rate": 3.991069993644577e-05,
+      "loss": 14.8787,
+      "step": 3140
+    },
+    {
+      "epoch": 0.315,
+      "grad_norm": 1.519584059715271,
+      "learning_rate": 3.984727911147916e-05,
+      "loss": 15.0554,
+      "step": 3150
+    },
+    {
+      "epoch": 0.315,
+      "eval_loss": 3.735930919647217,
+      "eval_runtime": 6.2326,
+      "eval_samples_per_second": 320.894,
+      "eval_steps_per_second": 20.056,
+      "step": 3150
+    },
+    {
+      "epoch": 0.316,
+      "grad_norm": 1.5088673830032349,
+      "learning_rate": 3.978415966885739e-05,
+      "loss": 15.0307,
+      "step": 3160
+    },
+    {
+      "epoch": 0.317,
+      "grad_norm": 1.2005739212036133,
+      "learning_rate": 3.972133922911742e-05,
+      "loss": 14.6702,
+      "step": 3170
+    },
+    {
+      "epoch": 0.318,
+      "grad_norm": 1.2146950960159302,
+      "learning_rate": 3.9658815439014244e-05,
+      "loss": 15.1964,
+      "step": 3180
+    },
+    {
+      "epoch": 0.319,
+      "grad_norm": 1.1482326984405518,
+      "learning_rate": 3.959658597115049e-05,
+      "loss": 15.2801,
+      "step": 3190
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.9639500975608826,
+      "learning_rate": 3.953464852361267e-05,
+      "loss": 14.9487,
+      "step": 3200
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 3.7448246479034424,
+      "eval_runtime": 6.4314,
+      "eval_samples_per_second": 310.973,
+      "eval_steps_per_second": 19.436,
+      "step": 3200
+    },
+    {
+      "epoch": 0.321,
+      "grad_norm": 1.2903177738189697,
+      "learning_rate": 3.9473000819613454e-05,
+      "loss": 14.7958,
+      "step": 3210
+    },
+    {
+      "epoch": 0.322,
+      "grad_norm": 1.5427738428115845,
+      "learning_rate": 3.941164060714018e-05,
+      "loss": 14.3846,
+      "step": 3220
+    },
+    {
+      "epoch": 0.323,
+      "grad_norm": 1.1496590375900269,
+      "learning_rate": 3.9350565658609336e-05,
+      "loss": 14.7951,
+      "step": 3230
+    },
+    {
+      "epoch": 0.324,
+      "grad_norm": 1.3798539638519287,
+      "learning_rate": 3.928977377052684e-05,
+      "loss": 14.4979,
+      "step": 3240
+    },
+    {
+      "epoch": 0.325,
+      "grad_norm": 1.2108612060546875,
+      "learning_rate": 3.922926276315421e-05,
+      "loss": 14.7947,
+      "step": 3250
+    },
+    {
+      "epoch": 0.325,
+      "eval_loss": 3.73313307762146,
+      "eval_runtime": 6.239,
+      "eval_samples_per_second": 320.563,
+      "eval_steps_per_second": 20.035,
+      "step": 3250
+    },
+    {
+      "epoch": 0.326,
+      "grad_norm": 1.2491366863250732,
+      "learning_rate": 3.916903048018023e-05,
+      "loss": 15.05,
+      "step": 3260
+    },
+    {
+      "epoch": 0.327,
+      "grad_norm": 1.5242091417312622,
+      "learning_rate": 3.910907478839825e-05,
+      "loss": 15.0895,
+      "step": 3270
+    },
+    {
+      "epoch": 0.328,
+      "grad_norm": 1.006120204925537,
+      "learning_rate": 3.904939357738885e-05,
+      "loss": 15.0252,
+      "step": 3280
+    },
+    {
+      "epoch": 0.329,
+      "grad_norm": 1.4161890745162964,
+      "learning_rate": 3.8989984759207835e-05,
+      "loss": 14.7367,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.2856372594833374,
+      "learning_rate": 3.893084626807942e-05,
+      "loss": 14.7509,
+      "step": 3300
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 3.728989601135254,
+      "eval_runtime": 6.2598,
+      "eval_samples_per_second": 319.499,
+      "eval_steps_per_second": 19.969,
+      "step": 3300
+    },
+    {
+      "epoch": 0.331,
+      "grad_norm": 1.3783438205718994,
+      "learning_rate": 3.887197606009451e-05,
+      "loss": 14.5589,
+      "step": 3310
+    },
+    {
+      "epoch": 0.332,
+      "grad_norm": 1.2106647491455078,
+      "learning_rate": 3.881337211291406e-05,
+      "loss": 14.1562,
+      "step": 3320
+    },
+    {
+      "epoch": 0.333,
+      "grad_norm": 1.212920069694519,
+      "learning_rate": 3.875503242547721e-05,
+      "loss": 14.6982,
+      "step": 3330
+    },
+    {
+      "epoch": 0.334,
+      "grad_norm": 1.0120047330856323,
+      "learning_rate": 3.869695501771434e-05,
+      "loss": 14.6513,
+      "step": 3340
+    },
+    {
+      "epoch": 0.335,
+      "grad_norm": 1.1318042278289795,
+      "learning_rate": 3.8639137930264805e-05,
+      "loss": 14.9368,
+      "step": 3350
+    },
+    {
+      "epoch": 0.335,
+      "eval_loss": 3.7159037590026855,
+      "eval_runtime": 6.2602,
+      "eval_samples_per_second": 319.479,
+      "eval_steps_per_second": 19.967,
+      "step": 3350
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 1.1732110977172852,
+      "learning_rate": 3.858157922419927e-05,
+      "loss": 14.6541,
+      "step": 3360
+    },
+    {
+      "epoch": 0.337,
+      "grad_norm": 1.2244794368743896,
+      "learning_rate": 3.8524276980746626e-05,
+      "loss": 14.7344,
+      "step": 3370
+    },
+    {
+      "epoch": 0.338,
+      "grad_norm": 1.1306333541870117,
+      "learning_rate": 3.8467229301025334e-05,
+      "loss": 14.5686,
+      "step": 3380
+    },
+    {
+      "epoch": 0.339,
+      "grad_norm": 1.3868407011032104,
+      "learning_rate": 3.8410434305779104e-05,
+      "loss": 14.664,
+      "step": 3390
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.124047875404358,
+      "learning_rate": 3.835389013511689e-05,
+      "loss": 14.7722,
+      "step": 3400
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 3.7149546146392822,
+      "eval_runtime": 7.4746,
+      "eval_samples_per_second": 267.574,
+      "eval_steps_per_second": 16.723,
+      "step": 3400
+    },
+    {
+      "epoch": 0.341,
+      "grad_norm": 1.340169072151184,
+      "learning_rate": 3.829759494825711e-05,
+      "loss": 14.5645,
+      "step": 3410
+    },
+    {
+      "epoch": 0.342,
+      "grad_norm": 1.1147572994232178,
+      "learning_rate": 3.82415469232759e-05,
+      "loss": 14.568,
+      "step": 3420
+    },
+    {
+      "epoch": 0.343,
+      "grad_norm": 0.9722641110420227,
+      "learning_rate": 3.8185744256859485e-05,
+      "loss": 14.4606,
+      "step": 3430
+    },
+    {
+      "epoch": 0.344,
+      "grad_norm": 1.1598691940307617,
+      "learning_rate": 3.813018516406046e-05,
+      "loss": 14.4404,
+      "step": 3440
+    },
+    {
+      "epoch": 0.345,
+      "grad_norm": 1.1491451263427734,
+      "learning_rate": 3.807486787805802e-05,
+      "loss": 14.3155,
+      "step": 3450
+    },
+    {
+      "epoch": 0.345,
+      "eval_loss": 3.7217018604278564,
+      "eval_runtime": 10.2678,
+      "eval_samples_per_second": 194.784,
+      "eval_steps_per_second": 12.174,
+      "step": 3450
+    },
+    {
+      "epoch": 0.346,
+      "grad_norm": 1.2273186445236206,
+      "learning_rate": 3.801979064992194e-05,
+      "loss": 14.5827,
+      "step": 3460
+    },
+    {
+      "epoch": 0.347,
+      "grad_norm": 1.2851728200912476,
+      "learning_rate": 3.796495174838033e-05,
+      "loss": 14.7868,
+      "step": 3470
+    },
+    {
+      "epoch": 0.348,
+      "grad_norm": 1.3409245014190674,
+      "learning_rate": 3.7910349459591094e-05,
+      "loss": 14.4866,
+      "step": 3480
+    },
+    {
+      "epoch": 0.349,
+      "grad_norm": 1.512802004814148,
+      "learning_rate": 3.785598208691693e-05,
+      "loss": 14.3369,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.238440752029419,
+      "learning_rate": 3.780184795070394e-05,
+      "loss": 14.376,
+      "step": 3500
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 3.7125015258789062,
+      "eval_runtime": 6.2266,
+      "eval_samples_per_second": 321.203,
+      "eval_steps_per_second": 20.075,
+      "step": 3500
+    },
+    {
+      "epoch": 0.351,
+      "grad_norm": 1.2165943384170532,
+      "learning_rate": 3.7747945388063626e-05,
+      "loss": 14.7182,
+      "step": 3510
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 1.1595171689987183,
+      "learning_rate": 3.7694272752658386e-05,
+      "loss": 14.8565,
+      "step": 3520
+    },
+    {
+      "epoch": 0.353,
+      "grad_norm": 1.294655442237854,
+      "learning_rate": 3.764082841449032e-05,
+      "loss": 15.2285,
+      "step": 3530
+    },
+    {
+      "epoch": 0.354,
+      "grad_norm": 1.2549623250961304,
+      "learning_rate": 3.758761075969328e-05,
+      "loss": 14.7526,
+      "step": 3540
+    },
+    {
+      "epoch": 0.355,
+      "grad_norm": 1.194077968597412,
+      "learning_rate": 3.7534618190328195e-05,
+      "loss": 14.485,
+      "step": 3550
+    },
+    {
+      "epoch": 0.355,
+      "eval_loss": 3.7080776691436768,
+      "eval_runtime": 6.2647,
+      "eval_samples_per_second": 319.25,
+      "eval_steps_per_second": 19.953,
+      "step": 3550
+    },
+    {
+      "epoch": 0.356,
+      "grad_norm": 1.0828840732574463,
+      "learning_rate": 3.748184912418159e-05,
+      "loss": 14.3297,
+      "step": 3560
+    },
+    {
+      "epoch": 0.357,
+      "grad_norm": 1.0104490518569946,
+      "learning_rate": 3.742930199456709e-05,
+      "loss": 14.2306,
+      "step": 3570
+    },
+    {
+      "epoch": 0.358,
+      "grad_norm": 1.086410641670227,
+      "learning_rate": 3.7376975250130133e-05,
+      "loss": 14.0265,
+      "step": 3580
+    },
+    {
+      "epoch": 0.359,
+      "grad_norm": 1.0463649034500122,
+      "learning_rate": 3.732486735465553e-05,
+      "loss": 13.999,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.2992103099822998,
+      "learning_rate": 3.727297678687811e-05,
+      "loss": 14.3544,
+      "step": 3600
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 3.7135140895843506,
+      "eval_runtime": 6.2372,
+      "eval_samples_per_second": 320.659,
+      "eval_steps_per_second": 20.041,
+      "step": 3600
+    },
+    {
+      "epoch": 0.361,
+      "grad_norm": 1.1995879411697388,
+      "learning_rate": 3.7221302040296116e-05,
+      "loss": 14.5238,
+      "step": 3610
+    },
+    {
+      "epoch": 0.362,
+      "grad_norm": 1.3045077323913574,
+      "learning_rate": 3.7169841622987545e-05,
+      "loss": 14.9165,
+      "step": 3620
+    },
+    {
+      "epoch": 0.363,
+      "grad_norm": 1.0523598194122314,
+      "learning_rate": 3.711859405742924e-05,
+      "loss": 14.4459,
+      "step": 3630
+    },
+    {
+      "epoch": 0.364,
+      "grad_norm": 1.0935602188110352,
+      "learning_rate": 3.706755788031864e-05,
+      "loss": 14.0357,
+      "step": 3640
+    },
+    {
+      "epoch": 0.365,
+      "grad_norm": 1.342834711074829,
+      "learning_rate": 3.701673164239839e-05,
+      "loss": 14.2318,
+      "step": 3650
+    },
+    {
+      "epoch": 0.365,
+      "eval_loss": 3.708907127380371,
+      "eval_runtime": 6.2576,
+      "eval_samples_per_second": 319.613,
+      "eval_steps_per_second": 19.976,
+      "step": 3650
+    },
+    {
+      "epoch": 0.366,
+      "grad_norm": 1.0739957094192505,
+      "learning_rate": 3.696611390828343e-05,
+      "loss": 14.4032,
+      "step": 3660
+    },
+    {
+      "epoch": 0.367,
+      "grad_norm": 1.2032370567321777,
+      "learning_rate": 3.691570325629073e-05,
+      "loss": 14.6113,
+      "step": 3670
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 1.155275583267212,
+      "learning_rate": 3.6865498278271596e-05,
+      "loss": 14.2557,
+      "step": 3680
+    },
+    {
+      "epoch": 0.369,
+      "grad_norm": 1.1695798635482788,
+      "learning_rate": 3.681549757944646e-05,
+      "loss": 14.816,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.3971259593963623,
+      "learning_rate": 3.676569977824213e-05,
+      "loss": 14.6923,
+      "step": 3700
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 3.6910042762756348,
+      "eval_runtime": 6.2304,
+      "eval_samples_per_second": 321.008,
+      "eval_steps_per_second": 20.063,
+      "step": 3700
+    },
+    {
+      "epoch": 0.371,
+      "grad_norm": 1.255376935005188,
+      "learning_rate": 3.6716103506131446e-05,
+      "loss": 14.0136,
+      "step": 3710
+    },
+    {
+      "epoch": 0.372,
+      "grad_norm": 1.3371446132659912,
+      "learning_rate": 3.6666707407475314e-05,
+      "loss": 14.5684,
+      "step": 3720
+    },
+    {
+      "epoch": 0.373,
+      "grad_norm": 1.2590174674987793,
+      "learning_rate": 3.661751013936708e-05,
+      "loss": 14.4847,
+      "step": 3730
+    },
+    {
+      "epoch": 0.374,
+      "grad_norm": 1.210468053817749,
+      "learning_rate": 3.656851037147919e-05,
+      "loss": 14.6346,
+      "step": 3740
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 1.449440836906433,
+      "learning_rate": 3.65197067859121e-05,
+      "loss": 14.9902,
+      "step": 3750
+    },
+    {
+      "epoch": 0.375,
+      "eval_loss": 3.7001209259033203,
+      "eval_runtime": 6.3384,
+      "eval_samples_per_second": 315.535,
+      "eval_steps_per_second": 19.721,
+      "step": 3750
+    },
+    {
+      "epoch": 0.376,
+      "grad_norm": 1.287975549697876,
+      "learning_rate": 3.6471098077045404e-05,
+      "loss": 15.1399,
+      "step": 3760
+    },
+    {
+      "epoch": 0.377,
+      "grad_norm": 1.300921082496643,
+      "learning_rate": 3.642268295139107e-05,
+      "loss": 14.7123,
+      "step": 3770
+    },
+    {
+      "epoch": 0.378,
+      "grad_norm": 1.0219051837921143,
+      "learning_rate": 3.6374460127448995e-05,
+      "loss": 14.4742,
+      "step": 3780
+    },
+    {
+      "epoch": 0.379,
+      "grad_norm": 1.0501158237457275,
+      "learning_rate": 3.632642833556441e-05,
+      "loss": 14.4047,
+      "step": 3790
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4280771017074585,
+      "learning_rate": 3.627858631778756e-05,
+      "loss": 14.627,
+      "step": 3800
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 3.6846299171447754,
+      "eval_runtime": 6.2689,
+      "eval_samples_per_second": 319.034,
+      "eval_steps_per_second": 19.94,
+      "step": 3800
+    },
+    {
+      "epoch": 0.381,
+      "grad_norm": 1.6403334140777588,
+      "learning_rate": 3.623093282773527e-05,
+      "loss": 14.751,
+      "step": 3810
+    },
+    {
+      "epoch": 0.382,
+      "grad_norm": 1.3171409368515015,
+      "learning_rate": 3.6183466630454636e-05,
+      "loss": 14.4652,
+      "step": 3820
+    },
+    {
+      "epoch": 0.383,
+      "grad_norm": 1.3603215217590332,
+      "learning_rate": 3.6136186502288535e-05,
+      "loss": 14.1708,
+      "step": 3830
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 1.2372150421142578,
+      "learning_rate": 3.608909123074314e-05,
+      "loss": 14.2904,
+      "step": 3840
+    },
+    {
+      "epoch": 0.385,
+      "grad_norm": 1.4466358423233032,
+      "learning_rate": 3.6042179614357375e-05,
+      "loss": 14.2487,
+      "step": 3850
+    },
+    {
+      "epoch": 0.385,
+      "eval_loss": 3.687807083129883,
+      "eval_runtime": 6.2515,
+      "eval_samples_per_second": 319.922,
+      "eval_steps_per_second": 19.995,
+      "step": 3850
+    },
+    {
+      "epoch": 0.386,
+      "grad_norm": 1.2979004383087158,
+      "learning_rate": 3.5995450462574126e-05,
+      "loss": 14.723,
+      "step": 3860
+    },
+    {
+      "epoch": 0.387,
+      "grad_norm": 1.2666685581207275,
+      "learning_rate": 3.594890259561335e-05,
+      "loss": 14.6014,
+      "step": 3870
+    },
+    {
+      "epoch": 0.388,
+      "grad_norm": 1.3407564163208008,
+      "learning_rate": 3.5902534844346976e-05,
+      "loss": 14.5131,
+      "step": 3880
+    },
+    {
+      "epoch": 0.389,
+      "grad_norm": 1.1121426820755005,
+      "learning_rate": 3.5856346050175565e-05,
+      "loss": 14.7588,
+      "step": 3890
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9877298474311829,
+      "learning_rate": 3.581033506490671e-05,
+      "loss": 14.5174,
+      "step": 3900
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 3.682448148727417,
+      "eval_runtime": 6.2568,
+      "eval_samples_per_second": 319.654,
+      "eval_steps_per_second": 19.978,
+      "step": 3900
+    },
+    {
+      "epoch": 0.391,
+      "grad_norm": 1.2272289991378784,
+      "learning_rate": 3.576450075063519e-05,
+      "loss": 14.6118,
+      "step": 3910
+    },
+    {
+      "epoch": 0.392,
+      "grad_norm": 1.2810298204421997,
+      "learning_rate": 3.571884197962469e-05,
+      "loss": 14.4994,
+      "step": 3920
+    },
+    {
+      "epoch": 0.393,
+      "grad_norm": 1.2651106119155884,
+      "learning_rate": 3.567335763419138e-05,
+      "loss": 14.7608,
+      "step": 3930
+    },
+    {
+      "epoch": 0.394,
+      "grad_norm": 1.2401210069656372,
+      "learning_rate": 3.562804660658888e-05,
+      "loss": 14.2945,
+      "step": 3940
+    },
+    {
+      "epoch": 0.395,
+      "grad_norm": 1.401061773300171,
+      "learning_rate": 3.5582907798895035e-05,
+      "loss": 14.6083,
+      "step": 3950
+    },
+    {
+      "epoch": 0.395,
+      "eval_loss": 3.66426420211792,
+      "eval_runtime": 6.2582,
+      "eval_samples_per_second": 319.58,
+      "eval_steps_per_second": 19.974,
+      "step": 3950
+    },
+    {
+      "epoch": 0.396,
+      "grad_norm": 1.347406268119812,
+      "learning_rate": 3.55379401229001e-05,
+      "loss": 14.2524,
+      "step": 3960
+    },
+    {
+      "epoch": 0.397,
+      "grad_norm": 1.1248286962509155,
+      "learning_rate": 3.5493142499996666e-05,
+      "loss": 14.7505,
+      "step": 3970
+    },
+    {
+      "epoch": 0.398,
+      "grad_norm": 1.3772094249725342,
+      "learning_rate": 3.544851386107085e-05,
+      "loss": 14.7012,
+      "step": 3980
+    },
+    {
+      "epoch": 0.399,
+      "grad_norm": 1.454031229019165,
+      "learning_rate": 3.540405314639526e-05,
+      "loss": 14.6776,
+      "step": 3990
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.1234098672866821,
+      "learning_rate": 3.535975930552322e-05,
+      "loss": 14.6502,
+      "step": 4000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 3.6687095165252686,
+      "eval_runtime": 6.2384,
+      "eval_samples_per_second": 320.595,
+      "eval_steps_per_second": 20.037,
+      "step": 4000
+    },
+    {
+      "epoch": 0.401,
+      "grad_norm": 1.2931911945343018,
+      "learning_rate": 3.531563129718458e-05,
+      "loss": 14.8039,
+      "step": 4010
+    },
+    {
+      "epoch": 0.402,
+      "grad_norm": 1.2497082948684692,
+      "learning_rate": 3.527166808918287e-05,
+      "loss": 14.8149,
+      "step": 4020
+    },
+    {
+      "epoch": 0.403,
+      "grad_norm": 1.4773260354995728,
+      "learning_rate": 3.522786865829391e-05,
+      "loss": 14.7316,
+      "step": 4030
+    },
+    {
+      "epoch": 0.404,
+      "grad_norm": 1.3319817781448364,
+      "learning_rate": 3.51842319901658e-05,
+      "loss": 14.2236,
+      "step": 4040
+    },
+    {
+      "epoch": 0.405,
+      "grad_norm": 1.094673991203308,
+      "learning_rate": 3.5140757079220216e-05,
+      "loss": 14.4414,
+      "step": 4050
+    },
+    {
+      "epoch": 0.405,
+      "eval_loss": 3.6671693325042725,
+      "eval_runtime": 6.278,
+      "eval_samples_per_second": 318.575,
+      "eval_steps_per_second": 19.911,
+      "step": 4050
+    },
+    {
+      "epoch": 0.406,
+      "grad_norm": 1.2040002346038818,
+      "learning_rate": 3.509744292855511e-05,
+      "loss": 14.3051,
+      "step": 4060
+    },
+    {
+      "epoch": 0.407,
+      "grad_norm": 1.0459996461868286,
+      "learning_rate": 3.505428854984869e-05,
+      "loss": 14.4803,
+      "step": 4070
+    },
+    {
+      "epoch": 0.408,
+      "grad_norm": 1.1268916130065918,
+      "learning_rate": 3.5011292963264705e-05,
+      "loss": 14.8989,
+      "step": 4080
+    },
+    {
+      "epoch": 0.409,
+      "grad_norm": 1.1444742679595947,
+      "learning_rate": 3.496845519735901e-05,
+      "loss": 14.7764,
+      "step": 4090
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5272349119186401,
+      "learning_rate": 3.492577428898734e-05,
+      "loss": 14.3487,
+      "step": 4100
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 3.6532514095306396,
+      "eval_runtime": 6.2351,
+      "eval_samples_per_second": 320.767,
+      "eval_steps_per_second": 20.048,
+      "step": 4100
+    },
+    {
+      "epoch": 0.411,
+      "grad_norm": 1.4240920543670654,
+      "learning_rate": 3.4883249283214416e-05,
+      "loss": 14.7764,
+      "step": 4110
+    },
+    {
+      "epoch": 0.412,
+      "grad_norm": 1.325265884399414,
+      "learning_rate": 3.4840879233224196e-05,
+      "loss": 14.2673,
+      "step": 4120
+    },
+    {
+      "epoch": 0.413,
+      "grad_norm": 1.321703314781189,
+      "learning_rate": 3.4798663200231265e-05,
+      "loss": 13.9183,
+      "step": 4130
+    },
+    {
+      "epoch": 0.414,
+      "grad_norm": 1.1967962980270386,
+      "learning_rate": 3.475660025339355e-05,
+      "loss": 14.1222,
+      "step": 4140
+    },
+    {
+      "epoch": 0.415,
+      "grad_norm": 1.0946017503738403,
+      "learning_rate": 3.471468946972612e-05,
+      "loss": 14.6894,
+      "step": 4150
+    },
+    {
+      "epoch": 0.415,
+      "eval_loss": 3.6469974517822266,
+      "eval_runtime": 6.2794,
+      "eval_samples_per_second": 318.5,
+      "eval_steps_per_second": 19.906,
+      "step": 4150
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 1.444520354270935,
+      "learning_rate": 3.467292993401603e-05,
+      "loss": 14.8196,
+      "step": 4160
+    },
+    {
+      "epoch": 0.417,
+      "grad_norm": 1.612980842590332,
+      "learning_rate": 3.4631320738738494e-05,
+      "loss": 13.9702,
+      "step": 4170
+    },
+    {
+      "epoch": 0.418,
+      "grad_norm": 1.541488528251648,
+      "learning_rate": 3.458986098397395e-05,
+      "loss": 14.4179,
+      "step": 4180
+    },
+    {
+      "epoch": 0.419,
+      "grad_norm": 1.2451997995376587,
+      "learning_rate": 3.45485497773264e-05,
+      "loss": 14.3002,
+      "step": 4190
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0385589599609375,
+      "learning_rate": 3.450738623384265e-05,
+      "loss": 14.2547,
+      "step": 4200
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 3.649686098098755,
+      "eval_runtime": 6.2648,
+      "eval_samples_per_second": 319.246,
+      "eval_steps_per_second": 19.953,
+      "step": 4200
+    },
+    {
+      "epoch": 0.421,
+      "grad_norm": 1.3172301054000854,
+      "learning_rate": 3.4466369475932744e-05,
+      "loss": 14.6115,
+      "step": 4210
+    },
+    {
+      "epoch": 0.422,
+      "grad_norm": 1.1311509609222412,
+      "learning_rate": 3.442549863329138e-05,
+      "loss": 14.6898,
+      "step": 4220
+    },
+    {
+      "epoch": 0.423,
+      "grad_norm": 1.1822139024734497,
+      "learning_rate": 3.438477284282031e-05,
+      "loss": 14.5849,
+      "step": 4230
+    },
+    {
+      "epoch": 0.424,
+      "grad_norm": 1.2009106874465942,
+      "learning_rate": 3.4344191248551814e-05,
+      "loss": 14.6419,
+      "step": 4240
+    },
+    {
+      "epoch": 0.425,
+      "grad_norm": 1.336679458618164,
+      "learning_rate": 3.4303753001573164e-05,
+      "loss": 14.5884,
+      "step": 4250
+    },
+    {
+      "epoch": 0.425,
+      "eval_loss": 3.650291919708252,
+      "eval_runtime": 6.2357,
+      "eval_samples_per_second": 320.734,
+      "eval_steps_per_second": 20.046,
+      "step": 4250
+    },
+    {
+      "epoch": 0.426,
+      "grad_norm": 1.2796190977096558,
+      "learning_rate": 3.426345725995197e-05,
+      "loss": 13.906,
+      "step": 4260
+    },
+    {
+      "epoch": 0.427,
+      "grad_norm": 1.253543496131897,
+      "learning_rate": 3.422330318866262e-05,
+      "loss": 14.2191,
+      "step": 4270
+    },
+    {
+      "epoch": 0.428,
+      "grad_norm": 1.1889468431472778,
+      "learning_rate": 3.4183289959513575e-05,
+      "loss": 14.1989,
+      "step": 4280
+    },
+    {
+      "epoch": 0.429,
+      "grad_norm": 1.2647627592086792,
+      "learning_rate": 3.414341675107563e-05,
+      "loss": 14.2794,
+      "step": 4290
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3583110570907593,
+      "learning_rate": 3.4103682748611054e-05,
+      "loss": 14.8572,
+      "step": 4300
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 3.6479427814483643,
+      "eval_runtime": 6.2542,
+      "eval_samples_per_second": 319.787,
+      "eval_steps_per_second": 19.987,
+      "step": 4300
+    },
+    {
+      "epoch": 0.431,
+      "grad_norm": 1.3158838748931885,
+      "learning_rate": 3.4064087144003755e-05,
+      "loss": 14.4486,
+      "step": 4310
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 1.2332990169525146,
+      "learning_rate": 3.402462913569015e-05,
+      "loss": 14.2708,
+      "step": 4320
+    },
+    {
+      "epoch": 0.433,
+      "grad_norm": 1.1926401853561401,
+      "learning_rate": 3.3985307928591074e-05,
+      "loss": 14.385,
+      "step": 4330
+    },
+    {
+      "epoch": 0.434,
+      "grad_norm": 1.3099074363708496,
+      "learning_rate": 3.3946122734044455e-05,
+      "loss": 14.6973,
+      "step": 4340
+    },
+    {
+      "epoch": 0.435,
+      "grad_norm": 1.0637519359588623,
+      "learning_rate": 3.390707276973892e-05,
+      "loss": 14.905,
+      "step": 4350
+    },
+    {
+      "epoch": 0.435,
+      "eval_loss": 3.638732433319092,
+      "eval_runtime": 6.2436,
+      "eval_samples_per_second": 320.326,
+      "eval_steps_per_second": 20.02,
+      "step": 4350
+    },
+    {
+      "epoch": 0.436,
+      "grad_norm": 1.257131576538086,
+      "learning_rate": 3.3868157259648156e-05,
+      "loss": 14.6147,
+      "step": 4360
+    },
+    {
+      "epoch": 0.437,
+      "grad_norm": 1.0288323163986206,
+      "learning_rate": 3.382937543396614e-05,
+      "loss": 14.6657,
+      "step": 4370
+    },
+    {
+      "epoch": 0.438,
+      "grad_norm": 1.339852213859558,
+      "learning_rate": 3.379072652904321e-05,
+      "loss": 14.3125,
+      "step": 4380
+    },
+    {
+      "epoch": 0.439,
+      "grad_norm": 1.3522120714187622,
+      "learning_rate": 3.375220978732289e-05,
+      "loss": 13.5858,
+      "step": 4390
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.2488036155700684,
+      "learning_rate": 3.371382445727951e-05,
+      "loss": 14.0461,
+      "step": 4400
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 3.6404318809509277,
+      "eval_runtime": 6.2486,
+      "eval_samples_per_second": 320.074,
+      "eval_steps_per_second": 20.005,
+      "step": 4400
+    },
+    {
+      "epoch": 0.441,
+      "grad_norm": 1.4024121761322021,
+      "learning_rate": 3.3675569793356654e-05,
+      "loss": 14.3792,
+      "step": 4410
+    },
+    {
+      "epoch": 0.442,
+      "grad_norm": 1.574225664138794,
+      "learning_rate": 3.36374450559063e-05,
+      "loss": 14.2403,
+      "step": 4420
+    },
+    {
+      "epoch": 0.443,
+      "grad_norm": 1.3503550291061401,
+      "learning_rate": 3.359944951112878e-05,
+      "loss": 14.2765,
+      "step": 4430
+    },
+    {
+      "epoch": 0.444,
+      "grad_norm": 1.3308185338974,
+      "learning_rate": 3.356158243101345e-05,
+      "loss": 14.3184,
+      "step": 4440
+    },
+    {
+      "epoch": 0.445,
+      "grad_norm": 1.4911550283432007,
+      "learning_rate": 3.3523843093280096e-05,
+      "loss": 13.6826,
+      "step": 4450
+    },
+    {
+      "epoch": 0.445,
+      "eval_loss": 3.6506519317626953,
+      "eval_runtime": 6.2039,
+      "eval_samples_per_second": 322.377,
+      "eval_steps_per_second": 20.149,
+      "step": 4450
+    },
+    {
+      "epoch": 0.446,
+      "grad_norm": 1.5103703737258911,
+      "learning_rate": 3.348623078132111e-05,
+      "loss": 14.1831,
+      "step": 4460
+    },
+    {
+      "epoch": 0.447,
+      "grad_norm": 1.20027756690979,
+      "learning_rate": 3.344874478414426e-05,
+      "loss": 14.0568,
+      "step": 4470
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 1.173339605331421,
+      "learning_rate": 3.341138439631633e-05,
+      "loss": 14.4583,
+      "step": 4480
+    },
+    {
+      "epoch": 0.449,
+      "grad_norm": 1.2948070764541626,
+      "learning_rate": 3.337414891790731e-05,
+      "loss": 14.4602,
+      "step": 4490
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.2212433815002441,
+      "learning_rate": 3.3337037654435325e-05,
+      "loss": 14.036,
+      "step": 4500
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 3.6365511417388916,
+      "eval_runtime": 6.2415,
+      "eval_samples_per_second": 320.433,
+      "eval_steps_per_second": 20.027,
+      "step": 4500
+    },
+    {
+      "epoch": 0.451,
+      "grad_norm": 1.335633397102356,
+      "learning_rate": 3.330004991681224e-05,
+      "loss": 13.7247,
+      "step": 4510
+    },
+    {
+      "epoch": 0.452,
+      "grad_norm": 1.226434588432312,
+      "learning_rate": 3.3263185021289925e-05,
+      "loss": 14.2707,
+      "step": 4520
+    },
+    {
+      "epoch": 0.453,
+      "grad_norm": 1.2056890726089478,
+      "learning_rate": 3.322644228940717e-05,
+      "loss": 14.7619,
+      "step": 4530
+    },
+    {
+      "epoch": 0.454,
+      "grad_norm": 1.4296637773513794,
+      "learning_rate": 3.318982104793721e-05,
+      "loss": 14.6893,
+      "step": 4540
+    },
+    {
+      "epoch": 0.455,
+      "grad_norm": 1.3162144422531128,
+      "learning_rate": 3.3153320628835953e-05,
+      "loss": 14.2959,
+      "step": 4550
+    },
+    {
+      "epoch": 0.455,
+      "eval_loss": 3.630129337310791,
+      "eval_runtime": 6.2099,
+      "eval_samples_per_second": 322.065,
+      "eval_steps_per_second": 20.129,
+      "step": 4550
+    },
+    {
+      "epoch": 0.456,
+      "grad_norm": 1.3302292823791504,
+      "learning_rate": 3.311694036919079e-05,
+      "loss": 14.435,
+      "step": 4560
+    },
+    {
+      "epoch": 0.457,
+      "grad_norm": 1.6379554271697998,
+      "learning_rate": 3.308067961117001e-05,
+      "loss": 14.5194,
+      "step": 4570
+    },
+    {
+      "epoch": 0.458,
+      "grad_norm": 1.1413871049880981,
+      "learning_rate": 3.3044537701972836e-05,
+      "loss": 14.5356,
+      "step": 4580
+    },
+    {
+      "epoch": 0.459,
+      "grad_norm": 1.333553671836853,
+      "learning_rate": 3.30085139937801e-05,
+      "loss": 13.9572,
+      "step": 4590
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.2290724515914917,
+      "learning_rate": 3.2972607843705445e-05,
+      "loss": 14.2627,
+      "step": 4600
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 3.626358985900879,
+      "eval_runtime": 6.1856,
+      "eval_samples_per_second": 323.334,
+      "eval_steps_per_second": 20.208,
+      "step": 4600
+    },
+    {
+      "epoch": 0.461,
+      "grad_norm": 1.2494893074035645,
+      "learning_rate": 3.293681861374713e-05,
+      "loss": 14.3398,
+      "step": 4610
+    },
+    {
+      "epoch": 0.462,
+      "grad_norm": 1.3624249696731567,
+      "learning_rate": 3.290114567074045e-05,
+      "loss": 14.3332,
+      "step": 4620
+    },
+    {
+      "epoch": 0.463,
+      "grad_norm": 1.1090985536575317,
+      "learning_rate": 3.2865588386310646e-05,
+      "loss": 13.7097,
+      "step": 4630
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 1.1444495916366577,
+      "learning_rate": 3.2830146136826445e-05,
+      "loss": 14.2853,
+      "step": 4640
+    },
+    {
+      "epoch": 0.465,
+      "grad_norm": 1.2577887773513794,
+      "learning_rate": 3.2794818303354126e-05,
+      "loss": 14.1538,
+      "step": 4650
+    },
+    {
+      "epoch": 0.465,
+      "eval_loss": 3.6183412075042725,
+      "eval_runtime": 6.2157,
+      "eval_samples_per_second": 321.764,
+      "eval_steps_per_second": 20.11,
+      "step": 4650
+    },
+    {
+      "epoch": 0.466,
+      "grad_norm": 1.1332851648330688,
+      "learning_rate": 3.2759604271612094e-05,
+      "loss": 13.2038,
+      "step": 4660
+    },
+    {
+      "epoch": 0.467,
+      "grad_norm": 1.2543771266937256,
+      "learning_rate": 3.272450343192603e-05,
+      "loss": 14.1238,
+      "step": 4670
+    },
+    {
+      "epoch": 0.468,
+      "grad_norm": 1.3030415773391724,
+      "learning_rate": 3.2689515179184576e-05,
+      "loss": 14.598,
+      "step": 4680
+    },
+    {
+      "epoch": 0.469,
+      "grad_norm": 1.4264740943908691,
+      "learning_rate": 3.265463891279551e-05,
+      "loss": 14.4331,
+      "step": 4690
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.3160494565963745,
+      "learning_rate": 3.2619874036642426e-05,
+      "loss": 14.5275,
+      "step": 4700
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 3.6218349933624268,
+      "eval_runtime": 6.246,
+      "eval_samples_per_second": 320.205,
+      "eval_steps_per_second": 20.013,
+      "step": 4700
+    },
+    {
+      "epoch": 0.471,
+      "grad_norm": 1.3615264892578125,
+      "learning_rate": 3.258521995904196e-05,
+      "loss": 14.6488,
+      "step": 4710
+    },
+    {
+      "epoch": 0.472,
+      "grad_norm": 1.199601173400879,
+      "learning_rate": 3.2550676092701496e-05,
+      "loss": 14.2667,
+      "step": 4720
+    },
+    {
+      "epoch": 0.473,
+      "grad_norm": 1.1377754211425781,
+      "learning_rate": 3.2516241854677356e-05,
+      "loss": 14.5754,
+      "step": 4730
+    },
+    {
+      "epoch": 0.474,
+      "grad_norm": 1.3825005292892456,
+      "learning_rate": 3.248191666633348e-05,
+      "loss": 14.6365,
+      "step": 4740
+    },
+    {
+      "epoch": 0.475,
+      "grad_norm": 1.1313741207122803,
+      "learning_rate": 3.244769995330059e-05,
+      "loss": 14.2485,
+      "step": 4750
+    },
+    {
+      "epoch": 0.475,
+      "eval_loss": 3.6164262294769287,
+      "eval_runtime": 6.2202,
+      "eval_samples_per_second": 321.531,
+      "eval_steps_per_second": 20.096,
+      "step": 4750
+    },
+    {
+      "epoch": 0.476,
+      "grad_norm": 1.176073431968689,
+      "learning_rate": 3.241359114543582e-05,
+      "loss": 14.2109,
+      "step": 4760
+    },
+    {
+      "epoch": 0.477,
+      "grad_norm": 1.5828211307525635,
+      "learning_rate": 3.237958967678283e-05,
+      "loss": 14.7457,
+      "step": 4770
+    },
+    {
+      "epoch": 0.478,
+      "grad_norm": 1.4503830671310425,
+      "learning_rate": 3.234569498553233e-05,
+      "loss": 14.3239,
+      "step": 4780
+    },
+    {
+      "epoch": 0.479,
+      "grad_norm": 1.7589170932769775,
+      "learning_rate": 3.231190651398314e-05,
+      "loss": 14.0929,
+      "step": 4790
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.3697690963745117,
+      "learning_rate": 3.227822370850359e-05,
+      "loss": 14.5645,
+      "step": 4800
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 3.614612579345703,
+      "eval_runtime": 6.2212,
+      "eval_samples_per_second": 321.483,
+      "eval_steps_per_second": 20.093,
+      "step": 4800
+    },
+    {
+      "epoch": 0.481,
+      "grad_norm": 1.6647683382034302,
+      "learning_rate": 3.224464601949349e-05,
+      "loss": 14.3627,
+      "step": 4810
+    },
+    {
+      "epoch": 0.482,
+      "grad_norm": 1.6885759830474854,
+      "learning_rate": 3.2211172901346385e-05,
+      "loss": 14.3836,
+      "step": 4820
+    },
+    {
+      "epoch": 0.483,
+      "grad_norm": 1.4781463146209717,
+      "learning_rate": 3.21778038124124e-05,
+      "loss": 14.2925,
+      "step": 4830
+    },
+    {
+      "epoch": 0.484,
+      "grad_norm": 1.5786219835281372,
+      "learning_rate": 3.2144538214961344e-05,
+      "loss": 13.8596,
+      "step": 4840
+    },
+    {
+      "epoch": 0.485,
+      "grad_norm": 1.4004464149475098,
+      "learning_rate": 3.21113755751464e-05,
+      "loss": 13.7206,
+      "step": 4850
+    },
+    {
+      "epoch": 0.485,
+      "eval_loss": 3.6186439990997314,
+      "eval_runtime": 6.2035,
+      "eval_samples_per_second": 322.396,
+      "eval_steps_per_second": 20.15,
+      "step": 4850
+    },
+    {
+      "epoch": 0.486,
+      "grad_norm": 1.272031545639038,
+      "learning_rate": 3.207831536296808e-05,
+      "loss": 14.5275,
+      "step": 4860
+    },
+    {
+      "epoch": 0.487,
+      "grad_norm": 1.3145509958267212,
+      "learning_rate": 3.2045357052238676e-05,
+      "loss": 14.3756,
+      "step": 4870
+    },
+    {
+      "epoch": 0.488,
+      "grad_norm": 1.108602523803711,
+      "learning_rate": 3.201250012054707e-05,
+      "loss": 13.865,
+      "step": 4880
+    },
+    {
+      "epoch": 0.489,
+      "grad_norm": 1.4979068040847778,
+      "learning_rate": 3.197974404922397e-05,
+      "loss": 14.5053,
+      "step": 4890
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.2225371599197388,
+      "learning_rate": 3.194708832330752e-05,
+      "loss": 14.6196,
+      "step": 4900
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 3.6117711067199707,
+      "eval_runtime": 6.2829,
+      "eval_samples_per_second": 318.327,
+      "eval_steps_per_second": 19.895,
+      "step": 4900
+    },
+    {
+      "epoch": 0.491,
+      "grad_norm": 1.3729444742202759,
+      "learning_rate": 3.191453243150929e-05,
+      "loss": 14.44,
+      "step": 4910
+    },
+    {
+      "epoch": 0.492,
+      "grad_norm": 1.4063236713409424,
+      "learning_rate": 3.188207586618064e-05,
+      "loss": 14.2631,
+      "step": 4920
+    },
+    {
+      "epoch": 0.493,
+      "grad_norm": 1.1830472946166992,
+      "learning_rate": 3.1849718123279517e-05,
+      "loss": 13.5341,
+      "step": 4930
+    },
+    {
+      "epoch": 0.494,
+      "grad_norm": 1.3388463258743286,
+      "learning_rate": 3.181745870233753e-05,
+      "loss": 13.9621,
+      "step": 4940
+    },
+    {
+      "epoch": 0.495,
+      "grad_norm": 1.2672926187515259,
+      "learning_rate": 3.178529710642749e-05,
+      "loss": 14.2384,
+      "step": 4950
+    },
+    {
+      "epoch": 0.495,
+      "eval_loss": 3.6163549423217773,
+      "eval_runtime": 6.2068,
+      "eval_samples_per_second": 322.225,
+      "eval_steps_per_second": 20.139,
+      "step": 4950
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 1.2444629669189453,
+      "learning_rate": 3.175323284213118e-05,
+      "loss": 14.3422,
+      "step": 4960
+    },
+    {
+      "epoch": 0.497,
+      "grad_norm": 1.3806087970733643,
+      "learning_rate": 3.172126541950766e-05,
+      "loss": 14.1642,
+      "step": 4970
+    },
+    {
+      "epoch": 0.498,
+      "grad_norm": 1.3681821823120117,
+      "learning_rate": 3.1689394352061735e-05,
+      "loss": 14.0664,
+      "step": 4980
+    },
+    {
+      "epoch": 0.499,
+      "grad_norm": 1.6592974662780762,
+      "learning_rate": 3.165761915671293e-05,
+      "loss": 14.3502,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1652058362960815,
+      "learning_rate": 3.162593935376469e-05,
+      "loss": 14.4776,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 3.6070852279663086,
+      "eval_runtime": 6.2187,
+      "eval_samples_per_second": 321.609,
+      "eval_steps_per_second": 20.101,
+      "step": 5000
+    },
+    {
+      "epoch": 0.501,
+      "grad_norm": 1.1766422986984253,
+      "learning_rate": 3.159435446687396e-05,
+      "loss": 14.1349,
+      "step": 5010
+    },
+    {
+      "epoch": 0.502,
+      "grad_norm": 1.2750104665756226,
+      "learning_rate": 3.1562864023021174e-05,
+      "loss": 14.2648,
+      "step": 5020
+    },
+    {
+      "epoch": 0.503,
+      "grad_norm": 1.1030397415161133,
+      "learning_rate": 3.1531467552480395e-05,
+      "loss": 14.0325,
+      "step": 5030
+    },
+    {
+      "epoch": 0.504,
+      "grad_norm": 1.3132902383804321,
+      "learning_rate": 3.1500164588789964e-05,
+      "loss": 14.5553,
+      "step": 5040
+    },
+    {
+      "epoch": 0.505,
+      "grad_norm": 1.1227859258651733,
+      "learning_rate": 3.146895466872338e-05,
+      "loss": 14.3816,
+      "step": 5050
+    },
+    {
+      "epoch": 0.505,
+      "eval_loss": 3.6058602333068848,
+      "eval_runtime": 6.2308,
+      "eval_samples_per_second": 320.985,
+      "eval_steps_per_second": 20.062,
+      "step": 5050
+    },
+    {
+      "epoch": 0.506,
+      "grad_norm": 1.3989307880401611,
+      "learning_rate": 3.1437837332260465e-05,
+      "loss": 14.3869,
+      "step": 5060
+    },
+    {
+      "epoch": 0.507,
+      "grad_norm": 1.2606079578399658,
+      "learning_rate": 3.140681212255896e-05,
+      "loss": 14.8284,
+      "step": 5070
+    },
+    {
+      "epoch": 0.508,
+      "grad_norm": 1.5170297622680664,
+      "learning_rate": 3.137587858592628e-05,
+      "loss": 14.2912,
+      "step": 5080
+    },
+    {
+      "epoch": 0.509,
+      "grad_norm": 1.1421058177947998,
+      "learning_rate": 3.13450362717917e-05,
+      "loss": 14.3119,
+      "step": 5090
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.2525449991226196,
+      "learning_rate": 3.131428473267876e-05,
+      "loss": 14.4662,
+      "step": 5100
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 3.6014516353607178,
+      "eval_runtime": 6.2101,
+      "eval_samples_per_second": 322.058,
+      "eval_steps_per_second": 20.129,
+      "step": 5100
+    },
+    {
+      "epoch": 0.511,
+      "grad_norm": 1.1750966310501099,
+      "learning_rate": 3.1283623524178034e-05,
+      "loss": 14.0454,
+      "step": 5110
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 1.2625198364257812,
+      "learning_rate": 3.125305220492011e-05,
+      "loss": 14.3145,
+      "step": 5120
+    },
+    {
+      "epoch": 0.513,
+      "grad_norm": 1.3217575550079346,
+      "learning_rate": 3.122257033654893e-05,
+      "loss": 14.3861,
+      "step": 5130
+    },
+    {
+      "epoch": 0.514,
+      "grad_norm": 1.1578750610351562,
+      "learning_rate": 3.119217748369538e-05,
+      "loss": 14.7037,
+      "step": 5140
+    },
+    {
+      "epoch": 0.515,
+      "grad_norm": 1.0977482795715332,
+      "learning_rate": 3.11618732139512e-05,
+      "loss": 14.5877,
+      "step": 5150
+    },
+    {
+      "epoch": 0.515,
+      "eval_loss": 3.6061079502105713,
+      "eval_runtime": 6.2037,
+      "eval_samples_per_second": 322.391,
+      "eval_steps_per_second": 20.149,
+      "step": 5150
+    },
+    {
+      "epoch": 0.516,
+      "grad_norm": 1.5148530006408691,
+      "learning_rate": 3.113165709784312e-05,
+      "loss": 14.0579,
+      "step": 5160
+    },
+    {
+      "epoch": 0.517,
+      "grad_norm": 1.2597321271896362,
+      "learning_rate": 3.110152870880728e-05,
+      "loss": 14.3179,
+      "step": 5170
+    },
+    {
+      "epoch": 0.518,
+      "grad_norm": 1.0222440958023071,
+      "learning_rate": 3.107148762316397e-05,
+      "loss": 14.5666,
+      "step": 5180
+    },
+    {
+      "epoch": 0.519,
+      "grad_norm": 1.2599363327026367,
+      "learning_rate": 3.104153342009261e-05,
+      "loss": 14.1644,
+      "step": 5190
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.148889422416687,
+      "learning_rate": 3.101166568160696e-05,
+      "loss": 13.9043,
+      "step": 5200
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 3.596761703491211,
+      "eval_runtime": 6.2279,
+      "eval_samples_per_second": 321.133,
+      "eval_steps_per_second": 20.071,
+      "step": 5200
+    },
+    {
+      "epoch": 0.521,
+      "grad_norm": 1.254384160041809,
+      "learning_rate": 3.098188399253065e-05,
+      "loss": 14.405,
+      "step": 5210
+    },
+    {
+      "epoch": 0.522,
+      "grad_norm": 1.6923539638519287,
+      "learning_rate": 3.0952187940472915e-05,
+      "loss": 14.4278,
+      "step": 5220
+    },
+    {
+      "epoch": 0.523,
+      "grad_norm": 1.6045749187469482,
+      "learning_rate": 3.092257711580463e-05,
+      "loss": 14.2743,
+      "step": 5230
+    },
+    {
+      "epoch": 0.524,
+      "grad_norm": 1.2014061212539673,
+      "learning_rate": 3.089305111163457e-05,
+      "loss": 14.4292,
+      "step": 5240
+    },
+    {
+      "epoch": 0.525,
+      "grad_norm": 1.3017672300338745,
+      "learning_rate": 3.0863609523785955e-05,
+      "loss": 14.5071,
+      "step": 5250
+    },
+    {
+      "epoch": 0.525,
+      "eval_loss": 3.590036153793335,
+      "eval_runtime": 6.1954,
+      "eval_samples_per_second": 322.822,
+      "eval_steps_per_second": 20.176,
+      "step": 5250
+    },
+    {
+      "epoch": 0.526,
+      "grad_norm": 1.3010127544403076,
+      "learning_rate": 3.083425195077315e-05,
+      "loss": 14.6025,
+      "step": 5260
+    },
+    {
+      "epoch": 0.527,
+      "grad_norm": 1.4094475507736206,
+      "learning_rate": 3.080497799377872e-05,
+      "loss": 14.0731,
+      "step": 5270
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 1.1461422443389893,
+      "learning_rate": 3.077578725663068e-05,
+      "loss": 13.7408,
+      "step": 5280
+    },
+    {
+      "epoch": 0.529,
+      "grad_norm": 1.1306099891662598,
+      "learning_rate": 3.0746679345779934e-05,
+      "loss": 13.4037,
+      "step": 5290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.536644697189331,
+      "learning_rate": 3.071765387027802e-05,
+      "loss": 13.4816,
+      "step": 5300
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 3.6046133041381836,
+      "eval_runtime": 6.2359,
+      "eval_samples_per_second": 320.723,
+      "eval_steps_per_second": 20.045,
+      "step": 5300
+    },
+    {
+      "epoch": 0.531,
+      "grad_norm": 1.5308995246887207,
+      "learning_rate": 3.068871044175507e-05,
+      "loss": 14.0583,
+      "step": 5310
+    },
+    {
+      "epoch": 0.532,
+      "grad_norm": 1.1854599714279175,
+      "learning_rate": 3.0659848674397934e-05,
+      "loss": 13.6901,
+      "step": 5320
+    },
+    {
+      "epoch": 0.533,
+      "grad_norm": 1.1849660873413086,
+      "learning_rate": 3.063106818492863e-05,
+      "loss": 13.6696,
+      "step": 5330
+    },
+    {
+      "epoch": 0.534,
+      "grad_norm": 1.3578708171844482,
+      "learning_rate": 3.060236859258293e-05,
+      "loss": 13.4662,
+      "step": 5340
+    },
+    {
+      "epoch": 0.535,
+      "grad_norm": 1.1757259368896484,
+      "learning_rate": 3.057374951908926e-05,
+      "loss": 13.7274,
+      "step": 5350
+    },
+    {
+      "epoch": 0.535,
+      "eval_loss": 3.6120777130126953,
+      "eval_runtime": 6.1981,
+      "eval_samples_per_second": 322.682,
+      "eval_steps_per_second": 20.168,
+      "step": 5350
+    },
+    {
+      "epoch": 0.536,
+      "grad_norm": 1.3536651134490967,
+      "learning_rate": 3.054521058864766e-05,
+      "loss": 14.1579,
+      "step": 5360
+    },
+    {
+      "epoch": 0.537,
+      "grad_norm": 1.3435899019241333,
+      "learning_rate": 3.051675142790916e-05,
+      "loss": 14.4204,
+      "step": 5370
+    },
+    {
+      "epoch": 0.538,
+      "grad_norm": 1.1941726207733154,
+      "learning_rate": 3.048837166595521e-05,
+      "loss": 14.1706,
+      "step": 5380
+    },
+    {
+      "epoch": 0.539,
+      "grad_norm": 1.123294472694397,
+      "learning_rate": 3.0460070934277386e-05,
+      "loss": 13.9552,
+      "step": 5390
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.2226771116256714,
+      "learning_rate": 3.0431848866757295e-05,
+      "loss": 14.0163,
+      "step": 5400
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 3.5943922996520996,
+      "eval_runtime": 6.2301,
+      "eval_samples_per_second": 321.024,
+      "eval_steps_per_second": 20.064,
+      "step": 5400
+    },
+    {
+      "epoch": 0.541,
+      "grad_norm": 1.2817046642303467,
+      "learning_rate": 3.040370509964669e-05,
+      "loss": 13.9803,
+      "step": 5410
+    },
+    {
+      "epoch": 0.542,
+      "grad_norm": 1.2297040224075317,
+      "learning_rate": 3.037563927154777e-05,
+      "loss": 14.6411,
+      "step": 5420
+    },
+    {
+      "epoch": 0.543,
+      "grad_norm": 1.2498054504394531,
+      "learning_rate": 3.0347651023393664e-05,
+      "loss": 14.4703,
+      "step": 5430
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 1.189957857131958,
+      "learning_rate": 3.0319739998429204e-05,
+      "loss": 13.9314,
+      "step": 5440
+    },
+    {
+      "epoch": 0.545,
+      "grad_norm": 1.176761507987976,
+      "learning_rate": 3.0291905842191765e-05,
+      "loss": 13.9819,
+      "step": 5450
+    },
+    {
+      "epoch": 0.545,
+      "eval_loss": 3.5868208408355713,
+      "eval_runtime": 6.2396,
+      "eval_samples_per_second": 320.535,
+      "eval_steps_per_second": 20.033,
+      "step": 5450
+    },
+    {
+      "epoch": 0.546,
+      "grad_norm": 1.2379381656646729,
+      "learning_rate": 3.0264148202492366e-05,
+      "loss": 14.4884,
+      "step": 5460
+    },
+    {
+      "epoch": 0.547,
+      "grad_norm": 1.1885240077972412,
+      "learning_rate": 3.0236466729396972e-05,
+      "loss": 13.8977,
+      "step": 5470
+    },
+    {
+      "epoch": 0.548,
+      "grad_norm": 1.4787578582763672,
+      "learning_rate": 3.020886107520798e-05,
+      "loss": 14.1968,
+      "step": 5480
+    },
+    {
+      "epoch": 0.549,
+      "grad_norm": 1.121922254562378,
+      "learning_rate": 3.0181330894445807e-05,
+      "loss": 13.8813,
+      "step": 5490
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.4813346862792969,
+      "learning_rate": 3.0153875843830832e-05,
+      "loss": 14.2594,
+      "step": 5500
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 3.5890376567840576,
+      "eval_runtime": 6.2112,
+      "eval_samples_per_second": 322.001,
+      "eval_steps_per_second": 20.125,
+      "step": 5500
+    },
+    {
+      "epoch": 0.551,
+      "grad_norm": 1.2030844688415527,
+      "learning_rate": 3.0126495582265325e-05,
+      "loss": 14.2157,
+      "step": 5510
+    },
+    {
+      "epoch": 0.552,
+      "grad_norm": 1.1894972324371338,
+      "learning_rate": 3.0099189770815694e-05,
+      "loss": 13.9861,
+      "step": 5520
+    },
+    {
+      "epoch": 0.553,
+      "grad_norm": 1.2028629779815674,
+      "learning_rate": 3.0071958072694866e-05,
+      "loss": 13.931,
+      "step": 5530
+    },
+    {
+      "epoch": 0.554,
+      "grad_norm": 1.2379300594329834,
+      "learning_rate": 3.004480015324483e-05,
+      "loss": 14.1274,
+      "step": 5540
+    },
+    {
+      "epoch": 0.555,
+      "grad_norm": 1.0493221282958984,
+      "learning_rate": 3.0017715679919344e-05,
+      "loss": 14.3283,
+      "step": 5550
+    },
+    {
+      "epoch": 0.555,
+      "eval_loss": 3.582146644592285,
+      "eval_runtime": 6.224,
+      "eval_samples_per_second": 321.336,
+      "eval_steps_per_second": 20.084,
+      "step": 5550
+    },
+    {
+      "epoch": 0.556,
+      "grad_norm": 1.2582511901855469,
+      "learning_rate": 2.9990704322266887e-05,
+      "loss": 14.4304,
+      "step": 5560
+    },
+    {
+      "epoch": 0.557,
+      "grad_norm": 1.212406873703003,
+      "learning_rate": 2.9963765751913658e-05,
+      "loss": 14.5878,
+      "step": 5570
+    },
+    {
+      "epoch": 0.558,
+      "grad_norm": 1.2840884923934937,
+      "learning_rate": 2.9936899642546866e-05,
+      "loss": 14.2102,
+      "step": 5580
+    },
+    {
+      "epoch": 0.559,
+      "grad_norm": 1.3281713724136353,
+      "learning_rate": 2.991010566989808e-05,
+      "loss": 13.6666,
+      "step": 5590
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5215142965316772,
+      "learning_rate": 2.9883383511726847e-05,
+      "loss": 13.9324,
+      "step": 5600
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 3.5781474113464355,
+      "eval_runtime": 6.226,
+      "eval_samples_per_second": 321.232,
+      "eval_steps_per_second": 20.077,
+      "step": 5600
+    },
+    {
+      "epoch": 0.561,
+      "grad_norm": 1.1610621213912964,
+      "learning_rate": 2.985673284780436e-05,
+      "loss": 14.2228,
+      "step": 5610
+    },
+    {
+      "epoch": 0.562,
+      "grad_norm": 1.1308223009109497,
+      "learning_rate": 2.9830153359897363e-05,
+      "loss": 14.2885,
+      "step": 5620
+    },
+    {
+      "epoch": 0.563,
+      "grad_norm": 1.1470520496368408,
+      "learning_rate": 2.9803644731752194e-05,
+      "loss": 14.5404,
+      "step": 5630
+    },
+    {
+      "epoch": 0.564,
+      "grad_norm": 1.3501935005187988,
+      "learning_rate": 2.9777206649078993e-05,
+      "loss": 14.237,
+      "step": 5640
+    },
+    {
+      "epoch": 0.565,
+      "grad_norm": 1.115102767944336,
+      "learning_rate": 2.9750838799536022e-05,
+      "loss": 13.7697,
+      "step": 5650
+    },
+    {
+      "epoch": 0.565,
+      "eval_loss": 3.5766561031341553,
+      "eval_runtime": 6.2493,
+      "eval_samples_per_second": 320.035,
+      "eval_steps_per_second": 20.002,
+      "step": 5650
+    },
+    {
+      "epoch": 0.566,
+      "grad_norm": 0.9969695210456848,
+      "learning_rate": 2.9724540872714186e-05,
+      "loss": 14.0174,
+      "step": 5660
+    },
+    {
+      "epoch": 0.567,
+      "grad_norm": 1.4996564388275146,
+      "learning_rate": 2.969831256012167e-05,
+      "loss": 14.4387,
+      "step": 5670
+    },
+    {
+      "epoch": 0.568,
+      "grad_norm": 1.0316441059112549,
+      "learning_rate": 2.9672153555168763e-05,
+      "loss": 14.0463,
+      "step": 5680
+    },
+    {
+      "epoch": 0.569,
+      "grad_norm": 1.5262383222579956,
+      "learning_rate": 2.9646063553152792e-05,
+      "loss": 14.1116,
+      "step": 5690
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.442413568496704,
+      "learning_rate": 2.962004225124321e-05,
+      "loss": 13.7311,
+      "step": 5700
+    },
+    {
+      "epoch": 0.57,
+      "eval_loss": 3.5711216926574707,
+      "eval_runtime": 6.2251,
+      "eval_samples_per_second": 321.28,
+      "eval_steps_per_second": 20.08,
+      "step": 5700
+    },
+    {
+      "epoch": 0.571,
+      "grad_norm": 1.0696613788604736,
+      "learning_rate": 2.9594089348466825e-05,
+      "loss": 13.9055,
+      "step": 5710
+    },
+    {
+      "epoch": 0.572,
+      "grad_norm": 1.1490660905838013,
+      "learning_rate": 2.9568204545693235e-05,
+      "loss": 13.5724,
+      "step": 5720
+    },
+    {
+      "epoch": 0.573,
+      "grad_norm": 1.0024884939193726,
+      "learning_rate": 2.9542387545620252e-05,
+      "loss": 13.8718,
+      "step": 5730
+    },
+    {
+      "epoch": 0.574,
+      "grad_norm": 1.1411281824111938,
+      "learning_rate": 2.9516638052759665e-05,
+      "loss": 13.9561,
+      "step": 5740
+    },
+    {
+      "epoch": 0.575,
+      "grad_norm": 1.3843379020690918,
+      "learning_rate": 2.9490955773422944e-05,
+      "loss": 13.747,
+      "step": 5750
+    },
+    {
+      "epoch": 0.575,
+      "eval_loss": 3.592867136001587,
+      "eval_runtime": 6.2333,
+      "eval_samples_per_second": 320.858,
+      "eval_steps_per_second": 20.054,
+      "step": 5750
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 1.0970510244369507,
+      "learning_rate": 2.946534041570722e-05,
+      "loss": 14.139,
+      "step": 5760
+    },
+    {
+      "epoch": 0.577,
+      "grad_norm": 1.0204675197601318,
+      "learning_rate": 2.9439791689481354e-05,
+      "loss": 13.7387,
+      "step": 5770
+    },
+    {
+      "epoch": 0.578,
+      "grad_norm": 1.233353853225708,
+      "learning_rate": 2.9414309306372117e-05,
+      "loss": 14.0551,
+      "step": 5780
+    },
+    {
+      "epoch": 0.579,
+      "grad_norm": 1.2701174020767212,
+      "learning_rate": 2.938889297975051e-05,
+      "loss": 14.2149,
+      "step": 5790
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.036139726638794,
+      "learning_rate": 2.936354242471826e-05,
+      "loss": 13.816,
+      "step": 5800
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 3.5758726596832275,
+      "eval_runtime": 6.2078,
+      "eval_samples_per_second": 322.177,
+      "eval_steps_per_second": 20.136,
+      "step": 5800
+    },
+    {
+      "epoch": 0.581,
+      "grad_norm": 1.3539680242538452,
+      "learning_rate": 2.933825735809436e-05,
+      "loss": 13.9908,
+      "step": 5810
+    },
+    {
+      "epoch": 0.582,
+      "grad_norm": 1.2056362628936768,
+      "learning_rate": 2.931303749840183e-05,
+      "loss": 14.2202,
+      "step": 5820
+    },
+    {
+      "epoch": 0.583,
+      "grad_norm": 1.1992346048355103,
+      "learning_rate": 2.9287882565854518e-05,
+      "loss": 14.2938,
+      "step": 5830
+    },
+    {
+      "epoch": 0.584,
+      "grad_norm": 1.378831148147583,
+      "learning_rate": 2.9262792282344083e-05,
+      "loss": 13.963,
+      "step": 5840
+    },
+    {
+      "epoch": 0.585,
+      "grad_norm": 1.3450469970703125,
+      "learning_rate": 2.9237766371427078e-05,
+      "loss": 14.0479,
+      "step": 5850
+    },
+    {
+      "epoch": 0.585,
+      "eval_loss": 3.5703623294830322,
+      "eval_runtime": 6.195,
+      "eval_samples_per_second": 322.841,
+      "eval_steps_per_second": 20.178,
+      "step": 5850
+    },
+    {
+      "epoch": 0.586,
+      "grad_norm": 1.115608811378479,
+      "learning_rate": 2.9212804558312162e-05,
+      "loss": 14.1431,
+      "step": 5860
+    },
+    {
+      "epoch": 0.587,
+      "grad_norm": 1.0845974683761597,
+      "learning_rate": 2.918790656984741e-05,
+      "loss": 14.2136,
+      "step": 5870
+    },
+    {
+      "epoch": 0.588,
+      "grad_norm": 1.293483018875122,
+      "learning_rate": 2.9163072134507774e-05,
+      "loss": 14.0732,
+      "step": 5880
+    },
+    {
+      "epoch": 0.589,
+      "grad_norm": 1.5905728340148926,
+      "learning_rate": 2.9138300982382654e-05,
+      "loss": 14.1266,
+      "step": 5890
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.4333081245422363,
+      "learning_rate": 2.9113592845163552e-05,
+      "loss": 14.3786,
+      "step": 5900
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 3.5610904693603516,
+      "eval_runtime": 6.2518,
+      "eval_samples_per_second": 319.906,
+      "eval_steps_per_second": 19.994,
+      "step": 5900
+    },
+    {
+      "epoch": 0.591,
+      "grad_norm": 1.3458027839660645,
+      "learning_rate": 2.9088947456131888e-05,
+      "loss": 14.4405,
+      "step": 5910
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 1.0692472457885742,
+      "learning_rate": 2.906436455014689e-05,
+      "loss": 14.2587,
+      "step": 5920
+    },
+    {
+      "epoch": 0.593,
+      "grad_norm": 1.1816102266311646,
+      "learning_rate": 2.903984386363363e-05,
+      "loss": 14.0342,
+      "step": 5930
+    },
+    {
+      "epoch": 0.594,
+      "grad_norm": 1.4598188400268555,
+      "learning_rate": 2.901538513457116e-05,
+      "loss": 14.1905,
+      "step": 5940
+    },
+    {
+      "epoch": 0.595,
+      "grad_norm": 0.9857292771339417,
+      "learning_rate": 2.8990988102480705e-05,
+      "loss": 13.8953,
+      "step": 5950
+    },
+    {
+      "epoch": 0.595,
+      "eval_loss": 3.569197654724121,
+      "eval_runtime": 6.2134,
+      "eval_samples_per_second": 321.884,
+      "eval_steps_per_second": 20.118,
+      "step": 5950
+    },
+    {
+      "epoch": 0.596,
+      "grad_norm": 1.5652998685836792,
+      "learning_rate": 2.8966652508414088e-05,
+      "loss": 14.1159,
+      "step": 5960
+    },
+    {
+      "epoch": 0.597,
+      "grad_norm": 1.135145664215088,
+      "learning_rate": 2.8942378094942106e-05,
+      "loss": 14.1481,
+      "step": 5970
+    },
+    {
+      "epoch": 0.598,
+      "grad_norm": 1.1402225494384766,
+      "learning_rate": 2.8918164606143166e-05,
+      "loss": 14.3713,
+      "step": 5980
+    },
+    {
+      "epoch": 0.599,
+      "grad_norm": 1.2710378170013428,
+      "learning_rate": 2.8894011787591897e-05,
+      "loss": 14.1149,
+      "step": 5990
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.2332837581634521,
+      "learning_rate": 2.8869919386347948e-05,
+      "loss": 14.2201,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 3.5695314407348633,
+      "eval_runtime": 6.2251,
+      "eval_samples_per_second": 321.282,
+      "eval_steps_per_second": 20.08,
+      "step": 6000
+    },
+    {
+      "epoch": 0.601,
+      "grad_norm": 1.117622971534729,
+      "learning_rate": 2.884588715094486e-05,
+      "loss": 14.3036,
+      "step": 6010
+    },
+    {
+      "epoch": 0.602,
+      "grad_norm": 1.1040174961090088,
+      "learning_rate": 2.8821914831379037e-05,
+      "loss": 13.8687,
+      "step": 6020
+    },
+    {
+      "epoch": 0.603,
+      "grad_norm": 1.7144644260406494,
+      "learning_rate": 2.879800217909883e-05,
+      "loss": 13.7791,
+      "step": 6030
+    },
+    {
+      "epoch": 0.604,
+      "grad_norm": 1.907306432723999,
+      "learning_rate": 2.87741489469937e-05,
+      "loss": 13.8325,
+      "step": 6040
+    },
+    {
+      "epoch": 0.605,
+      "grad_norm": 1.427022099494934,
+      "learning_rate": 2.8750354889383513e-05,
+      "loss": 14.1623,
+      "step": 6050
+    },
+    {
+      "epoch": 0.605,
+      "eval_loss": 3.5647103786468506,
+      "eval_runtime": 6.1899,
+      "eval_samples_per_second": 323.108,
+      "eval_steps_per_second": 20.194,
+      "step": 6050
+    },
+    {
+      "epoch": 0.606,
+      "grad_norm": 1.3600339889526367,
+      "learning_rate": 2.872661976200789e-05,
+      "loss": 13.9789,
+      "step": 6060
+    },
+    {
+      "epoch": 0.607,
+      "grad_norm": 1.3188380002975464,
+      "learning_rate": 2.8702943322015686e-05,
+      "loss": 13.7975,
+      "step": 6070
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 1.5352420806884766,
+      "learning_rate": 2.867932532795456e-05,
+      "loss": 14.3163,
+      "step": 6080
+    },
+    {
+      "epoch": 0.609,
+      "grad_norm": 1.3321303129196167,
+      "learning_rate": 2.865576553976062e-05,
+      "loss": 14.1487,
+      "step": 6090
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.168168067932129,
+      "learning_rate": 2.8632263718748197e-05,
+      "loss": 14.033,
+      "step": 6100
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 3.5599112510681152,
+      "eval_runtime": 6.2025,
+      "eval_samples_per_second": 322.449,
+      "eval_steps_per_second": 20.153,
+      "step": 6100
+    },
+    {
+      "epoch": 0.611,
+      "grad_norm": 1.336086630821228,
+      "learning_rate": 2.8608819627599655e-05,
+      "loss": 13.561,
+      "step": 6110
+    },
+    {
+      "epoch": 0.612,
+      "grad_norm": 1.248259425163269,
+      "learning_rate": 2.8585433030355346e-05,
+      "loss": 13.796,
+      "step": 6120
+    },
+    {
+      "epoch": 0.613,
+      "grad_norm": 1.2325026988983154,
+      "learning_rate": 2.856210369240364e-05,
+      "loss": 13.5651,
+      "step": 6130
+    },
+    {
+      "epoch": 0.614,
+      "grad_norm": 1.2659862041473389,
+      "learning_rate": 2.8538831380471032e-05,
+      "loss": 14.1842,
+      "step": 6140
+    },
+    {
+      "epoch": 0.615,
+      "grad_norm": 1.4123190641403198,
+      "learning_rate": 2.851561586261235e-05,
+      "loss": 14.1946,
+      "step": 6150
+    },
+    {
+      "epoch": 0.615,
+      "eval_loss": 3.5558834075927734,
+      "eval_runtime": 6.1869,
+      "eval_samples_per_second": 323.262,
+      "eval_steps_per_second": 20.204,
+      "step": 6150
+    },
+    {
+      "epoch": 0.616,
+      "grad_norm": 1.2844250202178955,
+      "learning_rate": 2.8492456908201033e-05,
+      "loss": 13.8504,
+      "step": 6160
+    },
+    {
+      "epoch": 0.617,
+      "grad_norm": 1.3036625385284424,
+      "learning_rate": 2.8469354287919516e-05,
+      "loss": 13.455,
+      "step": 6170
+    },
+    {
+      "epoch": 0.618,
+      "grad_norm": 1.658996343612671,
+      "learning_rate": 2.8446307773749708e-05,
+      "loss": 13.9047,
+      "step": 6180
+    },
+    {
+      "epoch": 0.619,
+      "grad_norm": 1.3913613557815552,
+      "learning_rate": 2.8423317138963523e-05,
+      "loss": 14.359,
+      "step": 6190
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.2245821952819824,
+      "learning_rate": 2.8400382158113496e-05,
+      "loss": 13.9187,
+      "step": 6200
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 3.5561468601226807,
+      "eval_runtime": 6.1814,
+      "eval_samples_per_second": 323.553,
+      "eval_steps_per_second": 20.222,
+      "step": 6200
+    },
+    {
+      "epoch": 0.621,
+      "grad_norm": 1.4613145589828491,
+      "learning_rate": 2.8377502607023556e-05,
+      "loss": 13.7003,
+      "step": 6210
+    },
+    {
+      "epoch": 0.622,
+      "grad_norm": 1.1367855072021484,
+      "learning_rate": 2.8354678262779732e-05,
+      "loss": 14.1309,
+      "step": 6220
+    },
+    {
+      "epoch": 0.623,
+      "grad_norm": 1.282486081123352,
+      "learning_rate": 2.8331908903721126e-05,
+      "loss": 14.1894,
+      "step": 6230
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 1.3200207948684692,
+      "learning_rate": 2.830919430943082e-05,
+      "loss": 13.954,
+      "step": 6240
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 1.4164808988571167,
+      "learning_rate": 2.828653426072691e-05,
+      "loss": 14.2861,
+      "step": 6250
+    },
+    {
+      "epoch": 0.625,
+      "eval_loss": 3.5531787872314453,
+      "eval_runtime": 6.1823,
+      "eval_samples_per_second": 323.506,
+      "eval_steps_per_second": 20.219,
+      "step": 6250
+    },
+    {
+      "epoch": 0.626,
+      "grad_norm": 1.3889133930206299,
+      "learning_rate": 2.8263928539653644e-05,
+      "loss": 14.2126,
+      "step": 6260
+    },
+    {
+      "epoch": 0.627,
+      "grad_norm": 1.80168879032135,
+      "learning_rate": 2.824137692947261e-05,
+      "loss": 14.0486,
+      "step": 6270
+    },
+    {
+      "epoch": 0.628,
+      "grad_norm": 1.1456998586654663,
+      "learning_rate": 2.8218879214653993e-05,
+      "loss": 13.9792,
+      "step": 6280
+    },
+    {
+      "epoch": 0.629,
+      "grad_norm": 1.2505043745040894,
+      "learning_rate": 2.8196435180867964e-05,
+      "loss": 13.6514,
+      "step": 6290
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.216320276260376,
+      "learning_rate": 2.817404461497606e-05,
+      "loss": 14.249,
+      "step": 6300
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 3.5436816215515137,
+      "eval_runtime": 6.218,
+      "eval_samples_per_second": 321.648,
+      "eval_steps_per_second": 20.103,
+      "step": 6300
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 50,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 5,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 0
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.3989320935079936e+17,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}