diff --git "a/checkpoint-4200/trainer_state.json" "b/checkpoint-4200/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-4200/trainer_state.json"
@@ -0,0 +1,5908 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6236543173212562,
+  "eval_steps": 500,
+  "global_step": 4200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.75,
+      "learning_rate": 2.9673590504451043e-07,
+      "loss": 1.433,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.483679525222552e-06,
+      "loss": 1.5006,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.967359050445104e-06,
+      "loss": 1.4919,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.4921875,
+      "learning_rate": 4.451038575667656e-06,
+      "loss": 1.4531,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.934718100890208e-06,
+      "loss": 1.4242,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.418397626112759e-06,
+      "loss": 1.3932,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8359375,
+      "learning_rate": 8.902077151335312e-06,
+      "loss": 1.3661,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.0385756676557864e-05,
+      "loss": 1.3677,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.1869436201780416e-05,
+      "loss": 1.3497,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.3353115727002968e-05,
+      "loss": 1.3049,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.4836795252225518e-05,
+      "loss": 1.2836,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.46484375,
+      "learning_rate": 1.6320474777448072e-05,
+      "loss": 1.3133,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.7804154302670624e-05,
+      "loss": 1.2596,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.408203125,
+      "learning_rate": 1.9287833827893176e-05,
+      "loss": 1.2612,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.0771513353115728e-05,
+      "loss": 1.2584,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.225519287833828e-05,
+      "loss": 1.2386,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.3738872403560832e-05,
+      "loss": 1.1833,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.5222551928783384e-05,
+      "loss": 1.2411,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.6706231454005936e-05,
+      "loss": 1.1956,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.818991097922849e-05,
+      "loss": 1.2327,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.9673590504451037e-05,
+      "loss": 1.1896,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.115727002967359e-05,
+      "loss": 1.1979,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.2640949554896144e-05,
+      "loss": 1.2004,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.41246290801187e-05,
+      "loss": 1.1624,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.400390625,
+      "learning_rate": 3.560830860534125e-05,
+      "loss": 1.1914,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.41796875,
+      "learning_rate": 3.70919881305638e-05,
+      "loss": 1.1952,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4140625,
+      "learning_rate": 3.857566765578635e-05,
+      "loss": 1.1538,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.005934718100891e-05,
+      "loss": 1.1702,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.44140625,
+      "learning_rate": 4.1543026706231456e-05,
+      "loss": 1.1683,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.3026706231454005e-05,
+      "loss": 1.1822,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.482421875,
+      "learning_rate": 4.451038575667656e-05,
+      "loss": 1.1649,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.46484375,
+      "learning_rate": 4.5994065281899116e-05,
+      "loss": 1.1868,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5,
+      "learning_rate": 4.7477744807121664e-05,
+      "loss": 1.161,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.451171875,
+      "learning_rate": 4.896142433234421e-05,
+      "loss": 1.1454,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.45703125,
+      "learning_rate": 5.044510385756677e-05,
+      "loss": 1.1388,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.490234375,
+      "learning_rate": 5.1928783382789324e-05,
+      "loss": 1.1253,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.51171875,
+      "learning_rate": 5.341246290801187e-05,
+      "loss": 1.1527,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.486328125,
+      "learning_rate": 5.489614243323442e-05,
+      "loss": 1.1234,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46875,
+      "learning_rate": 5.637982195845698e-05,
+      "loss": 1.113,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.458984375,
+      "learning_rate": 5.7863501483679525e-05,
+      "loss": 1.1368,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.484375,
+      "learning_rate": 5.9347181008902074e-05,
+      "loss": 1.137,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4765625,
+      "learning_rate": 6.0830860534124636e-05,
+      "loss": 1.1286,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.484375,
+      "learning_rate": 6.231454005934718e-05,
+      "loss": 1.1277,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.474609375,
+      "learning_rate": 6.379821958456974e-05,
+      "loss": 1.1268,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.47265625,
+      "learning_rate": 6.528189910979229e-05,
+      "loss": 1.0993,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.50390625,
+      "learning_rate": 6.676557863501484e-05,
+      "loss": 1.1017,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4609375,
+      "learning_rate": 6.82492581602374e-05,
+      "loss": 1.1345,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.48828125,
+      "learning_rate": 6.973293768545995e-05,
+      "loss": 1.1086,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45703125,
+      "learning_rate": 7.12166172106825e-05,
+      "loss": 1.0791,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.270029673590505e-05,
+      "loss": 1.1158,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46875,
+      "learning_rate": 7.41839762611276e-05,
+      "loss": 1.132,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4609375,
+      "learning_rate": 7.566765578635016e-05,
+      "loss": 1.1438,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.71513353115727e-05,
+      "loss": 1.1405,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.863501483679525e-05,
+      "loss": 1.1124,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.011869436201782e-05,
+      "loss": 1.0813,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.160237388724036e-05,
+      "loss": 1.1194,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.462890625,
+      "learning_rate": 8.308605341246291e-05,
+      "loss": 1.0927,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.456973293768546e-05,
+      "loss": 1.1277,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4453125,
+      "learning_rate": 8.605341246290801e-05,
+      "loss": 1.1271,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.435546875,
+      "learning_rate": 8.753709198813057e-05,
+      "loss": 1.1162,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.44921875,
+      "learning_rate": 8.902077151335312e-05,
+      "loss": 1.0857,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.050445103857568e-05,
+      "loss": 1.0869,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4453125,
+      "learning_rate": 9.198813056379823e-05,
+      "loss": 1.0655,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4296875,
+      "learning_rate": 9.347181008902077e-05,
+      "loss": 1.0585,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.41796875,
+      "learning_rate": 9.495548961424333e-05,
+      "loss": 1.1144,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.643916913946588e-05,
+      "loss": 1.0719,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.416015625,
+      "learning_rate": 9.792284866468843e-05,
+      "loss": 1.0919,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.423828125,
+      "learning_rate": 9.940652818991099e-05,
+      "loss": 1.1223,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00010089020771513354,
+      "loss": 1.0565,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00010237388724035609,
+      "loss": 1.0962,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00010385756676557865,
+      "loss": 1.0959,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.0001053412462908012,
+      "loss": 1.0628,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00010682492581602374,
+      "loss": 1.0975,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.0001083086053412463,
+      "loss": 1.0727,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00010979228486646884,
+      "loss": 1.0649,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00011127596439169139,
+      "loss": 1.0904,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00011275964391691397,
+      "loss": 1.079,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.0001142433234421365,
+      "loss": 1.0522,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00011572700296735905,
+      "loss": 1.0579,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.0001172106824925816,
+      "loss": 1.0871,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00011869436201780415,
+      "loss": 1.0936,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00012017804154302672,
+      "loss": 1.0734,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00012166172106824927,
+      "loss": 1.0657,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00012314540059347182,
+      "loss": 1.0884,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.00012462908011869436,
+      "loss": 1.0683,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00012611275964391692,
+      "loss": 1.1073,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00012759643916913948,
+      "loss": 1.0849,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00012908011869436204,
+      "loss": 1.0798,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00013056379821958458,
+      "loss": 1.1029,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.0001320474777448071,
+      "loss": 1.0664,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00013353115727002967,
+      "loss": 1.0998,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00013501483679525224,
+      "loss": 1.0834,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.0001364985163204748,
+      "loss": 1.062,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00013798219584569733,
+      "loss": 1.0825,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0001394658753709199,
+      "loss": 1.0689,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00014094955489614243,
+      "loss": 1.0557,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.000142433234421365,
+      "loss": 1.0582,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00014391691394658756,
+      "loss": 1.0921,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0001454005934718101,
+      "loss": 1.0544,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00014688427299703265,
+      "loss": 1.0333,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0001483679525222552,
+      "loss": 1.0454,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.37890625,
+      "learning_rate": 0.00014985163204747775,
+      "loss": 1.0434,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.390625,
+      "learning_rate": 0.0001513353115727003,
+      "loss": 1.0768,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00015281899109792285,
+      "loss": 1.0774,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001543026706231454,
+      "loss": 1.0438,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00015578635014836794,
+      "loss": 1.0573,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001572700296735905,
+      "loss": 1.0501,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.00015875370919881307,
+      "loss": 1.0413,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00016023738872403563,
+      "loss": 1.0234,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00016172106824925817,
+      "loss": 1.0538,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00016320474777448073,
+      "loss": 1.0493,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00016468842729970326,
+      "loss": 1.0686,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00016617210682492583,
+      "loss": 1.0794,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.0001676557863501484,
+      "loss": 1.0575,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00016913946587537092,
+      "loss": 1.0682,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00017062314540059348,
+      "loss": 1.0237,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00017210682492581602,
+      "loss": 1.0493,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00017359050445103858,
+      "loss": 1.0613,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.380859375,
+      "learning_rate": 0.00017507418397626114,
+      "loss": 1.0346,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00017655786350148368,
+      "loss": 1.0749,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00017804154302670624,
+      "loss": 1.0289,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00017952522255192878,
+      "loss": 1.0606,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.00018100890207715137,
+      "loss": 1.0166,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.375,
+      "learning_rate": 0.0001824925816023739,
+      "loss": 1.0347,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00018397626112759646,
+      "loss": 1.0388,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.000185459940652819,
+      "loss": 1.0545,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00018694362017804153,
+      "loss": 1.0281,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001884272997032641,
+      "loss": 1.0347,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00018991097922848666,
+      "loss": 1.0347,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019139465875370922,
+      "loss": 1.0594,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019287833827893175,
+      "loss": 1.0201,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00019436201780415432,
+      "loss": 1.0287,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019584569732937685,
+      "loss": 1.0777,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019732937685459941,
+      "loss": 1.0157,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019881305637982198,
+      "loss": 1.0474,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001999999865623139,
+      "loss": 1.0452,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.3828125,
+      "learning_rate": 0.00019999951624367985,
+      "loss": 1.02,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001999983740443526,
+      "loss": 1.041,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019999655997200635,
+      "loss": 1.0305,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001999940740388296,
+      "loss": 1.032,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019999091626152492,
+      "loss": 1.0027,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019998708666130893,
+      "loss": 1.0311,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.00019998258526391207,
+      "loss": 1.0165,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019997741209957853,
+      "loss": 1.013,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019997156720306597,
+      "loss": 0.9992,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019996505061364527,
+      "loss": 1.044,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001999578623751004,
+      "loss": 0.998,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019995000253572798,
+      "loss": 1.0354,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019994147114833698,
+      "loss": 1.0083,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019993226827024843,
+      "loss": 1.012,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019992239396329498,
+      "loss": 0.9953,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019991184829382057,
+      "loss": 1.0095,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019990063133267975,
+      "loss": 1.0518,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001998887431552376,
+      "loss": 1.0357,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019987618384136879,
+      "loss": 1.0542,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001998629534754574,
+      "loss": 1.0071,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019984905214639608,
+      "loss": 1.0414,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.00019983447994758563,
+      "loss": 1.003,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019981923697693437,
+      "loss": 1.0482,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00019980332333685729,
+      "loss": 1.0394,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019978673913427568,
+      "loss": 0.985,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019976948448061603,
+      "loss": 1.0149,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019975155949180967,
+      "loss": 1.0253,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019973296428829168,
+      "loss": 1.0067,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019971369899500024,
+      "loss": 1.013,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019969376374137578,
+      "loss": 1.0253,
+      "step": 825
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001996731586613601,
+      "loss": 1.0505,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019965188389339537,
+      "loss": 1.0199,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019962993958042336,
+      "loss": 1.0066,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019960732586988438,
+      "loss": 1.0226,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019958404291371635,
+      "loss": 0.9902,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.0001995600908683537,
+      "loss": 1.0338,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019953546989472633,
+      "loss": 1.0103,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019951018015825866,
+      "loss": 1.0237,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019948422182886833,
+      "loss": 0.972,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019945759508096527,
+      "loss": 1.0472,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019943030009345023,
+      "loss": 1.0156,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019940233704971388,
+      "loss": 1.0365,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019937370613763543,
+      "loss": 1.0098,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019934440754958136,
+      "loss": 1.0042,
+      "step": 895
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019931444148240423,
+      "loss": 1.0053,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019928380813744119,
+      "loss": 1.0311,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019925250772051276,
+      "loss": 0.9973,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019922054044192145,
+      "loss": 1.0023,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019918790651645023,
+      "loss": 1.0097,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019915460616336126,
+      "loss": 1.013,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001991206396063942,
+      "loss": 0.9989,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019908600707376495,
+      "loss": 0.9815,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001990507087981639,
+      "loss": 0.9753,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001990147450167545,
+      "loss": 1.0145,
+      "step": 945
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019897811597117168,
+      "loss": 1.0359,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019894082190751998,
+      "loss": 0.9902,
+      "step": 955
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019890286307637237,
+      "loss": 0.9997,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001988642397327681,
+      "loss": 1.0412,
+      "step": 965
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019882495213621116,
+      "loss": 0.9552,
+      "step": 970
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019878500055066866,
+      "loss": 0.9901,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019874438524456888,
+      "loss": 1.035,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001987031064907995,
+      "loss": 0.996,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001986611645667059,
+      "loss": 1.0248,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001986185597540891,
+      "loss": 0.9889,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019857529233920397,
+      "loss": 1.0151,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019853136261275737,
+      "loss": 0.9882,
+      "step": 1005
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019848677086990605,
+      "loss": 1.0051,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001984415174102548,
+      "loss": 0.9904,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001983956025378543,
+      "loss": 0.9899,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019834902656119924,
+      "loss": 0.999,
+      "step": 1025
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019830178979322614,
+      "loss": 0.9798,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019825389255131125,
+      "loss": 0.984,
+      "step": 1035
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001982053351572684,
+      "loss": 1.0014,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.000198156117937347,
+      "loss": 1.0148,
+      "step": 1045
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001981062412222296,
+      "loss": 1.0271,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019805570534702987,
+      "loss": 0.9902,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019800451065129018,
+      "loss": 0.9699,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019795265747897956,
+      "loss": 0.9911,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019790014617849106,
+      "loss": 0.9923,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019784697710263974,
+      "loss": 0.9976,
+      "step": 1075
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019779315060866007,
+      "loss": 0.9647,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019773866705820363,
+      "loss": 0.9589,
+      "step": 1085
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019768352681733662,
+      "loss": 1.005,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019762773025653747,
+      "loss": 1.0066,
+      "step": 1095
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001975712777506943,
+      "loss": 1.0118,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019751416967910248,
+      "loss": 1.0445,
+      "step": 1105
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019745640642546196,
+      "loss": 0.9766,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001973979883778747,
+      "loss": 0.9897,
+      "step": 1115
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019733891592884227,
+      "loss": 0.9826,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019727918947526292,
+      "loss": 0.988,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019721880941842913,
+      "loss": 1.0101,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019715777616402479,
+      "loss": 0.9744,
+      "step": 1135
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001970960901221225,
+      "loss": 0.9896,
+      "step": 1140
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019703375170718093,
+      "loss": 0.9786,
+      "step": 1145
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019697076133804185,
+      "loss": 0.9958,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001969071194379275,
+      "loss": 0.9803,
+      "step": 1155
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019684282643443748,
+      "loss": 0.9894,
+      "step": 1160
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019677788275954624,
+      "loss": 1.0073,
+      "step": 1165
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019671228884959987,
+      "loss": 1.0145,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019664604514531332,
+      "loss": 1.0035,
+      "step": 1175
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001965791520917674,
+      "loss": 0.9856,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019651161013840583,
+      "loss": 0.9573,
+      "step": 1185
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00019644341973903208,
+      "loss": 0.9824,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019637458135180657,
+      "loss": 0.9873,
+      "step": 1195
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001963050954392433,
+      "loss": 0.9743,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019623496246820704,
+      "loss": 0.9987,
+      "step": 1205
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019616418290990993,
+      "loss": 1.0153,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0001960927572399084,
+      "loss": 0.9907,
+      "step": 1215
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019602068593810014,
+      "loss": 1.0148,
+      "step": 1220
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001959479694887206,
+      "loss": 1.0024,
+      "step": 1225
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019587460838033996,
+      "loss": 1.0298,
+      "step": 1230
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019580060310585973,
+      "loss": 0.9946,
+      "step": 1235
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019572595416250942,
+      "loss": 0.9961,
+      "step": 1240
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019565066205184332,
+      "loss": 1.0016,
+      "step": 1245
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019557472727973707,
+      "loss": 0.9829,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019549815035638414,
+      "loss": 0.9781,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00019542093179629268,
+      "loss": 1.0089,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001953430721182817,
+      "loss": 0.9942,
+      "step": 1265
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019526457184547793,
+      "loss": 0.9818,
+      "step": 1270
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019518543150531207,
+      "loss": 1.018,
+      "step": 1275
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019510565162951537,
+      "loss": 0.9697,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019502523275411599,
+      "loss": 1.0331,
+      "step": 1285
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019494417541943547,
+      "loss": 1.0071,
+      "step": 1290
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019486248017008503,
+      "loss": 0.9699,
+      "step": 1295
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019478014755496196,
+      "loss": 1.0168,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019469717812724575,
+      "loss": 0.9719,
+      "step": 1305
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019461357244439479,
+      "loss": 0.9974,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00019452933106814223,
+      "loss": 0.9897,
+      "step": 1315
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001944444545644923,
+      "loss": 0.9809,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001943589435037166,
+      "loss": 0.9601,
+      "step": 1325
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019427279846035025,
+      "loss": 0.9615,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019418602001318797,
+      "loss": 0.9888,
+      "step": 1335
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019409860874528017,
+      "loss": 1.0099,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00019401056524392916,
+      "loss": 0.9911,
+      "step": 1345
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019392189010068508,
+      "loss": 0.992,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.000193832583911342,
+      "loss": 0.9827,
+      "step": 1355
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001937426472759338,
+      "loss": 0.9473,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019365208079873036,
+      "loss": 0.9863,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001935608850882333,
+      "loss": 0.9662,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001934690607571719,
+      "loss": 0.9864,
+      "step": 1375
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019337660842249914,
+      "loss": 1.0151,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001932835287053874,
+      "loss": 0.9638,
+      "step": 1385
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019318982223122437,
+      "loss": 1.0149,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00019309548962960876,
+      "loss": 0.9827,
+      "step": 1395
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019300053153434622,
+      "loss": 0.9726,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019290494858344493,
+      "loss": 0.9742,
+      "step": 1405
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019280874141911137,
+      "loss": 0.9987,
+      "step": 1410
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019271191068774606,
+      "loss": 1.0067,
+      "step": 1415
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00019261445703993912,
+      "loss": 0.9833,
+      "step": 1420
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00019251638113046597,
+      "loss": 0.9599,
+      "step": 1425
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001924176836182829,
+      "loss": 0.9793,
+      "step": 1430
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019231836516652261,
+      "loss": 0.965,
+      "step": 1435
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001922184264424899,
+      "loss": 0.9598,
+      "step": 1440
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019211786811765692,
+      "loss": 0.9931,
+      "step": 1445
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019201669086765902,
+      "loss": 0.9571,
+      "step": 1450
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001919148953722898,
+      "loss": 0.9448,
+      "step": 1455
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001918124823154969,
+      "loss": 0.9523,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019170945238537718,
+      "loss": 0.9888,
+      "step": 1465
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019160580627417223,
+      "loss": 0.9735,
+      "step": 1470
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019150154467826357,
+      "loss": 0.9726,
+      "step": 1475
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00019139666829816817,
+      "loss": 0.9679,
+      "step": 1480
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.0001912911778385336,
+      "loss": 0.9913,
+      "step": 1485
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00019118507400813325,
+      "loss": 0.9736,
+      "step": 1490
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001910783575198618,
+      "loss": 0.9535,
+      "step": 1495
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001909710290907302,
+      "loss": 0.993,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019086308944186084,
+      "loss": 0.9645,
+      "step": 1505
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.000190754539298483,
+      "loss": 0.9694,
+      "step": 1510
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019064537938992757,
+      "loss": 0.9775,
+      "step": 1515
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001905356104496225,
+      "loss": 0.9388,
+      "step": 1520
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00019042523321508768,
+      "loss": 0.9755,
+      "step": 1525
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00019031424842793,
+      "loss": 1.0159,
+      "step": 1530
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00019020265683383842,
+      "loss": 0.9835,
+      "step": 1535
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.000190090459182579,
+      "loss": 0.9711,
+      "step": 1540
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018997765622798967,
+      "loss": 0.9742,
+      "step": 1545
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001898642487279754,
+      "loss": 0.9669,
+      "step": 1550
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.000189750237444503,
+      "loss": 0.9516,
+      "step": 1555
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018963562314359595,
+      "loss": 1.0002,
+      "step": 1560
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018952040659532936,
+      "loss": 0.937,
+      "step": 1565
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018940458857382467,
+      "loss": 0.9757,
+      "step": 1570
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00018928816985724458,
+      "loss": 0.9968,
+      "step": 1575
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001891711512277878,
+      "loss": 0.974,
+      "step": 1580
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018905353347168366,
+      "loss": 0.9641,
+      "step": 1585
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018893531737918702,
+      "loss": 0.9799,
+      "step": 1590
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001888165037445728,
+      "loss": 0.9385,
+      "step": 1595
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00018869709336613073,
+      "loss": 1.0199,
+      "step": 1600
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00018857708704615996,
+      "loss": 0.9743,
+      "step": 1605
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00018845648559096377,
+      "loss": 0.9641,
+      "step": 1610
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018833528981084384,
+      "loss": 0.9978,
+      "step": 1615
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018821350052009524,
+      "loss": 0.9661,
+      "step": 1620
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018809111853700068,
+      "loss": 0.9739,
+      "step": 1625
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018796814468382498,
+      "loss": 0.9467,
+      "step": 1630
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00018784457978680976,
+      "loss": 0.9578,
+      "step": 1635
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001877204246761677,
+      "loss": 0.9351,
+      "step": 1640
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00018759568018607707,
+      "loss": 0.969,
+      "step": 1645
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00018747034715467612,
+      "loss": 0.9618,
+      "step": 1650
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018734442642405724,
+      "loss": 0.9595,
+      "step": 1655
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001872179188402617,
+      "loss": 0.9795,
+      "step": 1660
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018709082525327362,
+      "loss": 0.9558,
+      "step": 1665
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018696314651701437,
+      "loss": 0.9677,
+      "step": 1670
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018683488348933693,
+      "loss": 0.9851,
+      "step": 1675
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018670603703202,
+      "loss": 0.9973,
+      "step": 1680
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00018657660801076214,
+      "loss": 0.9835,
+      "step": 1685
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018644659729517623,
+      "loss": 0.9342,
+      "step": 1690
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018631600575878343,
+      "loss": 1.0092,
+      "step": 1695
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00018618483427900722,
+      "loss": 0.9792,
+      "step": 1700
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018605308373716782,
+      "loss": 0.9774,
+      "step": 1705
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018592075501847584,
+      "loss": 0.9629,
+      "step": 1710
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018578784901202675,
+      "loss": 0.9647,
+      "step": 1715
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001856543666107947,
+      "loss": 0.9881,
+      "step": 1720
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001855203087116265,
+      "loss": 0.9917,
+      "step": 1725
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001853856762152356,
+      "loss": 0.9852,
+      "step": 1730
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018525047002619616,
+      "loss": 0.9541,
+      "step": 1735
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018511469105293683,
+      "loss": 0.971,
+      "step": 1740
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001849783402077348,
+      "loss": 0.9864,
+      "step": 1745
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001848414184067094,
+      "loss": 0.9857,
+      "step": 1750
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00018470392656981633,
+      "loss": 0.9607,
+      "step": 1755
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001845658656208411,
+      "loss": 0.9854,
+      "step": 1760
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001844272364873931,
+      "loss": 0.9878,
+      "step": 1765
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001842880401008992,
+      "loss": 0.9802,
+      "step": 1770
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018414827739659755,
+      "loss": 0.9668,
+      "step": 1775
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018400794931353132,
+      "loss": 0.9624,
+      "step": 1780
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00018386705679454242,
+      "loss": 1.0008,
+      "step": 1785
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.000183725600786265,
+      "loss": 0.9585,
+      "step": 1790
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018358358223911932,
+      "loss": 0.9947,
+      "step": 1795
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018344100210730514,
+      "loss": 0.9581,
+      "step": 1800
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018329786134879553,
+      "loss": 0.9528,
+      "step": 1805
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018315416092533023,
+      "loss": 0.9702,
+      "step": 1810
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018300990180240931,
+      "loss": 0.9341,
+      "step": 1815
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00018286508494928666,
+      "loss": 0.9587,
+      "step": 1820
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00018271971133896345,
+      "loss": 0.9632,
+      "step": 1825
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001825737819481817,
+      "loss": 0.949,
+      "step": 1830
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00018242729775741744,
+      "loss": 0.9854,
+      "step": 1835
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001822802597508745,
+      "loss": 0.9745,
+      "step": 1840
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00018213266891647765,
+      "loss": 0.9367,
+      "step": 1845
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018198452624586592,
+      "loss": 0.9518,
+      "step": 1850
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018183583273438623,
+      "loss": 0.9493,
+      "step": 1855
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018168658938108636,
+      "loss": 0.9703,
+      "step": 1860
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018153679718870843,
+      "loss": 1.0251,
+      "step": 1865
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00018138645716368212,
+      "loss": 0.9541,
+      "step": 1870
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00018123557031611798,
+      "loss": 0.9702,
+      "step": 1875
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018108413765980045,
+      "loss": 0.9776,
+      "step": 1880
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018093216021218127,
+      "loss": 0.9974,
+      "step": 1885
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001807796389943725,
+      "loss": 0.9357,
+      "step": 1890
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00018062657503113968,
+      "loss": 0.97,
+      "step": 1895
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018047296935089503,
+      "loss": 0.9495,
+      "step": 1900
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00018031882298569048,
+      "loss": 0.977,
+      "step": 1905
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00018016413697121065,
+      "loss": 0.9441,
+      "step": 1910
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001800089123467661,
+      "loss": 0.9695,
+      "step": 1915
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00017985315015528607,
+      "loss": 0.9556,
+      "step": 1920
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00017969685144331182,
+      "loss": 0.9801,
+      "step": 1925
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00017954001726098917,
+      "loss": 1.0291,
+      "step": 1930
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00017938264866206193,
+      "loss": 0.9283,
+      "step": 1935
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001792247467038644,
+      "loss": 0.9569,
+      "step": 1940
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00017906631244731446,
+      "loss": 0.9667,
+      "step": 1945
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00017890734695690652,
+      "loss": 0.953,
+      "step": 1950
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001787478513007041,
+      "loss": 0.9536,
+      "step": 1955
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017858782655033291,
+      "loss": 0.9896,
+      "step": 1960
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001784272737809736,
+      "loss": 0.9257,
+      "step": 1965
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017826619407135445,
+      "loss": 0.9844,
+      "step": 1970
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017810458850374414,
+      "loss": 0.9824,
+      "step": 1975
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001779424581639445,
+      "loss": 0.9524,
+      "step": 1980
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001777798041412833,
+      "loss": 0.9813,
+      "step": 1985
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00017761662752860678,
+      "loss": 0.9497,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00017745292942227244,
+      "loss": 0.921,
+      "step": 1995
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001772887109221415,
+      "loss": 0.9375,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00017712397313157175,
+      "loss": 0.9381,
+      "step": 2005
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017695871715740994,
+      "loss": 0.954,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00017679294410998442,
+      "loss": 0.9281,
+      "step": 2015
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017662665510309768,
+      "loss": 0.9522,
+      "step": 2020
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00017645985125401884,
+      "loss": 0.9619,
+      "step": 2025
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00017629253368347622,
+      "loss": 0.9485,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001761247035156497,
+      "loss": 0.9487,
+      "step": 2035
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00017595636187816318,
+      "loss": 0.9368,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00017578750990207716,
+      "loss": 0.923,
+      "step": 2045
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00017561814872188092,
+      "loss": 0.9878,
+      "step": 2050
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00017544827947548502,
+      "loss": 1.0012,
+      "step": 2055
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001752779033042137,
+      "loss": 1.0007,
+      "step": 2060
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00017510702135279708,
+      "loss": 0.9638,
+      "step": 2065
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00017493563476936351,
+      "loss": 0.968,
+      "step": 2070
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00017476374470543199,
+      "loss": 0.9327,
+      "step": 2075
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00017459135231590424,
+      "loss": 0.958,
+      "step": 2080
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00017441845875905702,
+      "loss": 0.9515,
+      "step": 2085
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00017424506519653438,
+      "loss": 0.9442,
+      "step": 2090
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00017407117279333987,
+      "loss": 0.942,
+      "step": 2095
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001738967827178286,
+      "loss": 0.9547,
+      "step": 2100
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00017372189614169947,
+      "loss": 0.966,
+      "step": 2105
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017354651423998733,
+      "loss": 0.9754,
+      "step": 2110
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00017337063819105496,
+      "loss": 0.928,
+      "step": 2115
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00017319426917658537,
+      "loss": 0.9721,
+      "step": 2120
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00017301740838157362,
+      "loss": 0.9716,
+      "step": 2125
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00017284005699431896,
+      "loss": 0.9606,
+      "step": 2130
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.000172662216206417,
+      "loss": 0.9619,
+      "step": 2135
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00017248388721275129,
+      "loss": 0.9236,
+      "step": 2140
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00017230507121148575,
+      "loss": 0.9592,
+      "step": 2145
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00017212576940405647,
+      "loss": 0.9507,
+      "step": 2150
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00017194598299516338,
+      "loss": 0.9622,
+      "step": 2155
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017176571319276257,
+      "loss": 0.9527,
+      "step": 2160
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00017158496120805788,
+      "loss": 0.9426,
+      "step": 2165
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00017140372825549284,
+      "loss": 0.9517,
+      "step": 2170
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00017122201555274261,
+      "loss": 0.9285,
+      "step": 2175
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00017103982432070563,
+      "loss": 0.9671,
+      "step": 2180
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00017085715578349557,
+      "loss": 0.9375,
+      "step": 2185
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00017067401116843296,
+      "loss": 0.954,
+      "step": 2190
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001704903917060371,
+      "loss": 0.9486,
+      "step": 2195
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00017030629863001764,
+      "loss": 0.9823,
+      "step": 2200
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001701217331772664,
+      "loss": 0.9401,
+      "step": 2205
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00016993669658784904,
+      "loss": 0.9571,
+      "step": 2210
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001697511901049967,
+      "loss": 0.9801,
+      "step": 2215
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00016956521497509764,
+      "loss": 0.943,
+      "step": 2220
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001693787724476889,
+      "loss": 0.973,
+      "step": 2225
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00016919186377544788,
+      "loss": 0.9662,
+      "step": 2230
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00016900449021418394,
+      "loss": 0.9618,
+      "step": 2235
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00016881665302282995,
+      "loss": 0.9454,
+      "step": 2240
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00016862835346343385,
+      "loss": 0.9414,
+      "step": 2245
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00016843959280115015,
+      "loss": 0.9437,
+      "step": 2250
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00016825037230423139,
+      "loss": 0.9761,
+      "step": 2255
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00016806069324401977,
+      "loss": 0.9458,
+      "step": 2260
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00016787055689493837,
+      "loss": 0.9676,
+      "step": 2265
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00016767996453448283,
+      "loss": 0.9729,
+      "step": 2270
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00016748891744321263,
+      "loss": 0.9613,
+      "step": 2275
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001672974169047425,
+      "loss": 0.9453,
+      "step": 2280
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00016710546420573377,
+      "loss": 0.9538,
+      "step": 2285
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00016691306063588583,
+      "loss": 0.9484,
+      "step": 2290
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.0001667202074879274,
+      "loss": 0.9501,
+      "step": 2295
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00016652690605760775,
+      "loss": 0.9739,
+      "step": 2300
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016633315764368818,
+      "loss": 0.9428,
+      "step": 2305
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001661389635479332,
+      "loss": 0.938,
+      "step": 2310
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00016594432507510175,
+      "loss": 0.9647,
+      "step": 2315
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016574924353293845,
+      "loss": 0.9413,
+      "step": 2320
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001655537202321649,
+      "loss": 0.9443,
+      "step": 2325
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00016535775648647075,
+      "loss": 0.9407,
+      "step": 2330
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001651613536125049,
+      "loss": 0.9803,
+      "step": 2335
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001649645129298668,
+      "loss": 0.9349,
+      "step": 2340
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.0001647672357610973,
+      "loss": 0.9581,
+      "step": 2345
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00016456952343167007,
+      "loss": 0.9243,
+      "step": 2350
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00016437137726998255,
+      "loss": 0.9591,
+      "step": 2355
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00016417279860734692,
+      "loss": 0.9286,
+      "step": 2360
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00016397378877798134,
+      "loss": 0.9695,
+      "step": 2365
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001637743491190009,
+      "loss": 0.9696,
+      "step": 2370
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00016357448097040867,
+      "loss": 0.9465,
+      "step": 2375
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00016337418567508665,
+      "loss": 0.9405,
+      "step": 2380
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00016317346457878675,
+      "loss": 0.9151,
+      "step": 2385
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.0001629723190301218,
+      "loss": 0.9319,
+      "step": 2390
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00016277075038055634,
+      "loss": 0.956,
+      "step": 2395
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00016256875998439796,
+      "loss": 0.9475,
+      "step": 2400
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00016236634919878756,
+      "loss": 0.9747,
+      "step": 2405
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00016216351938369073,
+      "loss": 0.9405,
+      "step": 2410
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00016196027190188848,
+      "loss": 0.9888,
+      "step": 2415
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00016175660811896796,
+      "loss": 0.9063,
+      "step": 2420
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00016155252940331342,
+      "loss": 0.9454,
+      "step": 2425
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.000161348037126097,
+      "loss": 0.9783,
+      "step": 2430
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001611431326612695,
+      "loss": 0.9951,
+      "step": 2435
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00016093781738555106,
+      "loss": 0.9194,
+      "step": 2440
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001607320926784221,
+      "loss": 0.9692,
+      "step": 2445
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00016052595992211387,
+      "loss": 0.8929,
+      "step": 2450
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001603194205015993,
+      "loss": 0.9456,
+      "step": 2455
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00016011247580458355,
+      "loss": 0.956,
+      "step": 2460
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00015990512722149482,
+      "loss": 0.9113,
+      "step": 2465
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00015969737614547494,
+      "loss": 0.9445,
+      "step": 2470
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00015948922397237007,
+      "loss": 0.957,
+      "step": 2475
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00015928067210072122,
+      "loss": 0.924,
+      "step": 2480
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001590717219317549,
+      "loss": 0.9558,
+      "step": 2485
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00015886237486937378,
+      "loss": 0.9632,
+      "step": 2490
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.375,
+      "learning_rate": 0.00015865263232014715,
+      "loss": 0.9477,
+      "step": 2495
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001584424956933015,
+      "loss": 0.9443,
+      "step": 2500
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001582319664007111,
+      "loss": 0.9316,
+      "step": 2505
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00015802104585688851,
+      "loss": 0.9205,
+      "step": 2510
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00015780973547897494,
+      "loss": 0.924,
+      "step": 2515
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001575980366867309,
+      "loss": 0.9498,
+      "step": 2520
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00015738595090252657,
+      "loss": 0.9677,
+      "step": 2525
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00015717347955133233,
+      "loss": 0.9786,
+      "step": 2530
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00015696062406070902,
+      "loss": 0.9523,
+      "step": 2535
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00015674738586079852,
+      "loss": 0.9496,
+      "step": 2540
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001565337663843141,
+      "loss": 0.9439,
+      "step": 2545
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00015631976706653063,
+      "loss": 0.9496,
+      "step": 2550
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00015610538934527526,
+      "loss": 0.9315,
+      "step": 2555
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00015589063466091743,
+      "loss": 0.9595,
+      "step": 2560
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001556755044563594,
+      "loss": 0.9524,
+      "step": 2565
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00015546000017702648,
+      "loss": 0.9442,
+      "step": 2570
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00015524412327085725,
+      "loss": 0.9807,
+      "step": 2575
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00015502787518829406,
+      "loss": 0.9385,
+      "step": 2580
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.328125,
+      "learning_rate": 0.00015481125738227305,
+      "loss": 0.9618,
+      "step": 2585
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00015459427130821442,
+      "loss": 0.9443,
+      "step": 2590
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001543769184240128,
+      "loss": 0.9428,
+      "step": 2595
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00015415920019002736,
+      "loss": 0.9349,
+      "step": 2600
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00015394111806907188,
+      "loss": 0.9325,
+      "step": 2605
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00015372267352640513,
+      "loss": 0.9622,
+      "step": 2610
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00015350386802972097,
+      "loss": 0.934,
+      "step": 2615
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00015328470304913833,
+      "loss": 0.9729,
+      "step": 2620
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00015306518005719157,
+      "loss": 0.9473,
+      "step": 2625
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00015284530052882045,
+      "loss": 0.9268,
+      "step": 2630
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00015262506594136016,
+      "loss": 0.9737,
+      "step": 2635
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00015240447777453153,
+      "loss": 0.9231,
+      "step": 2640
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00015218353751043107,
+      "loss": 0.9359,
+      "step": 2645
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00015196224663352093,
+      "loss": 0.9394,
+      "step": 2650
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00015174060663061898,
+      "loss": 0.9556,
+      "step": 2655
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00015151861899088877,
+      "loss": 0.9345,
+      "step": 2660
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001512962852058297,
+      "loss": 0.9363,
+      "step": 2665
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00015107360676926666,
+      "loss": 0.9575,
+      "step": 2670
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00015085058517734043,
+      "loss": 0.9824,
+      "step": 2675
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001506272219284972,
+      "loss": 0.95,
+      "step": 2680
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00015040351852347878,
+      "loss": 0.9348,
+      "step": 2685
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001501794764653124,
+      "loss": 0.9585,
+      "step": 2690
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00014995509725930078,
+      "loss": 0.9342,
+      "step": 2695
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001497303824130117,
+      "loss": 0.9266,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00014950533343626812,
+      "loss": 0.9552,
+      "step": 2705
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.000149279951841138,
+      "loss": 0.9514,
+      "step": 2710
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00014905423914192412,
+      "loss": 0.9697,
+      "step": 2715
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001488281968551538,
+      "loss": 0.9494,
+      "step": 2720
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00014860182649956892,
+      "loss": 0.9667,
+      "step": 2725
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001483751295961156,
+      "loss": 0.9504,
+      "step": 2730
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001481481076679338,
+      "loss": 0.953,
+      "step": 2735
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00014792076224034753,
+      "loss": 0.9548,
+      "step": 2740
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00014769309484085412,
+      "loss": 0.9632,
+      "step": 2745
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00014746510699911432,
+      "loss": 0.9619,
+      "step": 2750
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00014723680024694184,
+      "loss": 0.9155,
+      "step": 2755
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00014700817611829308,
+      "loss": 0.9234,
+      "step": 2760
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00014677923614925685,
+      "loss": 0.934,
+      "step": 2765
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.000146549981878044,
+      "loss": 0.9354,
+      "step": 2770
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00014632041484497727,
+      "loss": 0.9554,
+      "step": 2775
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00014609053659248058,
+      "loss": 0.9443,
+      "step": 2780
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00014586034866506906,
+      "loss": 0.9584,
+      "step": 2785
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00014562985260933845,
+      "loss": 0.9035,
+      "step": 2790
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00014539904997395468,
+      "loss": 0.9215,
+      "step": 2795
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00014516794230964365,
+      "loss": 0.9279,
+      "step": 2800
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00014493653116918066,
+      "loss": 0.9395,
+      "step": 2805
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001447048181073799,
+      "loss": 0.9314,
+      "step": 2810
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00014447280468108436,
+      "loss": 0.9497,
+      "step": 2815
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00014424049244915493,
+      "loss": 0.9324,
+      "step": 2820
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00014400788297246024,
+      "loss": 0.9351,
+      "step": 2825
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001437749778138659,
+      "loss": 0.9482,
+      "step": 2830
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00014354177853822443,
+      "loss": 0.938,
+      "step": 2835
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00014330828671236425,
+      "loss": 0.9478,
+      "step": 2840
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001430745039050794,
+      "loss": 0.9596,
+      "step": 2845
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00014284043168711906,
+      "loss": 0.9186,
+      "step": 2850
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00014260607163117694,
+      "loss": 0.9631,
+      "step": 2855
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00014237142531188055,
+      "loss": 0.9411,
+      "step": 2860
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00014213649430578083,
+      "loss": 0.9558,
+      "step": 2865
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00014190128019134153,
+      "loss": 0.9684,
+      "step": 2870
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00014166578454892853,
+      "loss": 0.9226,
+      "step": 2875
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00014143000896079918,
+      "loss": 0.9103,
+      "step": 2880
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00014119395501109182,
+      "loss": 0.9262,
+      "step": 2885
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00014095762428581506,
+      "loss": 0.9444,
+      "step": 2890
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001407210183728371,
+      "loss": 0.9438,
+      "step": 2895
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00014048413886187503,
+      "loss": 0.9699,
+      "step": 2900
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00014024698734448431,
+      "loss": 0.9255,
+      "step": 2905
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00014000956541404785,
+      "loss": 0.9733,
+      "step": 2910
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001397718746657655,
+      "loss": 0.9554,
+      "step": 2915
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001395339166966433,
+      "loss": 0.9273,
+      "step": 2920
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001392956931054825,
+      "loss": 0.9364,
+      "step": 2925
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00013905720549286932,
+      "loss": 0.9433,
+      "step": 2930
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001388184554611636,
+      "loss": 0.9288,
+      "step": 2935
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001385794446144885,
+      "loss": 0.9338,
+      "step": 2940
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001383401745587196,
+      "loss": 1.0093,
+      "step": 2945
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00013810064690147387,
+      "loss": 0.924,
+      "step": 2950
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001378608632520993,
+      "loss": 0.9731,
+      "step": 2955
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00013762082522166363,
+      "loss": 0.9236,
+      "step": 2960
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001373805344229439,
+      "loss": 0.9378,
+      "step": 2965
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00013713999247041533,
+      "loss": 0.929,
+      "step": 2970
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00013689920098024078,
+      "loss": 0.9338,
+      "step": 2975
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001366581615702596,
+      "loss": 0.9437,
+      "step": 2980
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00013641687585997677,
+      "loss": 0.9331,
+      "step": 2985
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00013617534547055236,
+      "loss": 0.9655,
+      "step": 2990
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001359335720247902,
+      "loss": 0.9384,
+      "step": 2995
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001356915571471273,
+      "loss": 0.9497,
+      "step": 3000
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001354493024636227,
+      "loss": 0.934,
+      "step": 3005
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001352068096019468,
+      "loss": 0.9272,
+      "step": 3010
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00013496408019137018,
+      "loss": 0.9428,
+      "step": 3015
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00013472111586275274,
+      "loss": 0.9461,
+      "step": 3020
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001344779182485328,
+      "loss": 0.9403,
+      "step": 3025
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001342344889827161,
+      "loss": 0.9324,
+      "step": 3030
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001339908297008648,
+      "loss": 0.9427,
+      "step": 3035
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00013374694204008647,
+      "loss": 0.9369,
+      "step": 3040
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00013350282763902315,
+      "loss": 0.9251,
+      "step": 3045
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001332584881378403,
+      "loss": 0.9365,
+      "step": 3050
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00013301392517821577,
+      "loss": 0.9184,
+      "step": 3055
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00013276914040332889,
+      "loss": 0.9231,
+      "step": 3060
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001325241354578492,
+      "loss": 0.9217,
+      "step": 3065
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001322789119879256,
+      "loss": 0.9712,
+      "step": 3070
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00013203347164117524,
+      "loss": 0.9239,
+      "step": 3075
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00013178781606667234,
+      "loss": 0.9567,
+      "step": 3080
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00013154194691493732,
+      "loss": 0.9389,
+      "step": 3085
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001312958658379255,
+      "loss": 0.9429,
+      "step": 3090
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00013104957448901614,
+      "loss": 0.9226,
+      "step": 3095
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00013080307452300127,
+      "loss": 0.9155,
+      "step": 3100
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00013055636759607458,
+      "loss": 0.9516,
+      "step": 3105
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00013030945536582025,
+      "loss": 0.9284,
+      "step": 3110
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00013006233949120199,
+      "loss": 0.9113,
+      "step": 3115
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00012981502163255166,
+      "loss": 0.9471,
+      "step": 3120
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001295675034515582,
+      "loss": 0.9416,
+      "step": 3125
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00012931978661125655,
+      "loss": 0.9395,
+      "step": 3130
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00012907187277601641,
+      "loss": 0.9367,
+      "step": 3135
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012882376361153102,
+      "loss": 0.923,
+      "step": 3140
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00012857546078480598,
+      "loss": 0.9115,
+      "step": 3145
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00012832696596414817,
+      "loss": 0.9182,
+      "step": 3150
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00012807828081915436,
+      "loss": 0.913,
+      "step": 3155
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001278294070207001,
+      "loss": 0.9438,
+      "step": 3160
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.0001275803462409285,
+      "loss": 0.9487,
+      "step": 3165
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00012733110015323898,
+      "loss": 0.9521,
+      "step": 3170
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.0001270816704322759,
+      "loss": 0.9146,
+      "step": 3175
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00012683205875391754,
+      "loss": 0.9211,
+      "step": 3180
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00012658226679526476,
+      "loss": 0.9297,
+      "step": 3185
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00012633229623462951,
+      "loss": 0.9186,
+      "step": 3190
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012608214875152392,
+      "loss": 0.9443,
+      "step": 3195
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00012583182602664877,
+      "loss": 0.9399,
+      "step": 3200
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00012558132974188223,
+      "loss": 0.9342,
+      "step": 3205
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00012533066158026862,
+      "loss": 0.9088,
+      "step": 3210
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00012507982322600703,
+      "loss": 0.9777,
+      "step": 3215
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00012482881636444014,
+      "loss": 0.944,
+      "step": 3220
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00012457764268204277,
+      "loss": 0.9307,
+      "step": 3225
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001243263038664105,
+      "loss": 0.9396,
+      "step": 3230
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00012407480160624848,
+      "loss": 0.9138,
+      "step": 3235
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001238231375913601,
+      "loss": 0.9583,
+      "step": 3240
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00012357131351263537,
+      "loss": 0.9799,
+      "step": 3245
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012331933106203986,
+      "loss": 0.9619,
+      "step": 3250
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00012306719193260323,
+      "loss": 0.9188,
+      "step": 3255
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00012281489781840781,
+      "loss": 0.9593,
+      "step": 3260
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.0001225624504145772,
+      "loss": 0.938,
+      "step": 3265
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00012230985141726498,
+      "loss": 0.9237,
+      "step": 3270
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00012205710252364329,
+      "loss": 0.9314,
+      "step": 3275
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00012180420543189131,
+      "loss": 0.9093,
+      "step": 3280
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00012155116184118402,
+      "loss": 0.9335,
+      "step": 3285
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00012129797345168073,
+      "loss": 0.9643,
+      "step": 3290
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00012104464196451353,
+      "loss": 0.9507,
+      "step": 3295
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00012079116908177593,
+      "loss": 0.9201,
+      "step": 3300
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00012053755650651166,
+      "loss": 0.9375,
+      "step": 3305
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00012028380594270283,
+      "loss": 0.9681,
+      "step": 3310
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00012002991909525873,
+      "loss": 0.923,
+      "step": 3315
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011977589767000433,
+      "loss": 0.9143,
+      "step": 3320
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011952174337366881,
+      "loss": 0.9205,
+      "step": 3325
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00011926745791387406,
+      "loss": 0.9113,
+      "step": 3330
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.00011901304299912326,
+      "loss": 0.9357,
+      "step": 3335
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00011875850033878939,
+      "loss": 0.9552,
+      "step": 3340
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00011850383164310371,
+      "loss": 0.9234,
+      "step": 3345
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00011824903862314427,
+      "loss": 0.9085,
+      "step": 3350
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00011799412299082448,
+      "loss": 0.9226,
+      "step": 3355
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011773908645888152,
+      "loss": 0.9107,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00011748393074086497,
+      "loss": 0.9402,
+      "step": 3365
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011722865755112504,
+      "loss": 0.9419,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00011697326860480133,
+      "loss": 0.9146,
+      "step": 3375
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00011671776561781123,
+      "loss": 0.8996,
+      "step": 3380
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011646215030683818,
+      "loss": 0.9089,
+      "step": 3385
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.0001162064243893205,
+      "loss": 0.9257,
+      "step": 3390
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.333984375,
+      "learning_rate": 0.00011595058958343952,
+      "loss": 0.924,
+      "step": 3395
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00011569464760810825,
+      "loss": 0.9558,
+      "step": 3400
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00011543860018295966,
+      "loss": 0.9002,
+      "step": 3405
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00011518244902833537,
+      "loss": 0.9107,
+      "step": 3410
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00011492619586527385,
+      "loss": 0.9712,
+      "step": 3415
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.357421875,
+      "learning_rate": 0.0001146698424154989,
+      "loss": 0.9171,
+      "step": 3420
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00011441339040140824,
+      "loss": 0.9107,
+      "step": 3425
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00011415684154606177,
+      "loss": 0.9396,
+      "step": 3430
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011390019757317003,
+      "loss": 0.9101,
+      "step": 3435
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00011364346020708266,
+      "loss": 0.9324,
+      "step": 3440
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011338663117277686,
+      "loss": 0.933,
+      "step": 3445
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00011312971219584563,
+      "loss": 0.9153,
+      "step": 3450
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00011287270500248631,
+      "loss": 0.9401,
+      "step": 3455
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00011261561131948897,
+      "loss": 0.9264,
+      "step": 3460
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00011235843287422482,
+      "loss": 0.92,
+      "step": 3465
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00011210117139463452,
+      "loss": 0.9187,
+      "step": 3470
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00011184382860921663,
+      "loss": 0.949,
+      "step": 3475
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011158640624701603,
+      "loss": 0.9201,
+      "step": 3480
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00011132890603761221,
+      "loss": 0.9096,
+      "step": 3485
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00011107132971110779,
+      "loss": 0.9248,
+      "step": 3490
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00011081367899811668,
+      "loss": 0.9189,
+      "step": 3495
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00011055595562975267,
+      "loss": 0.9182,
+      "step": 3500
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00011029816133761772,
+      "loss": 0.9549,
+      "step": 3505
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00011004029785379024,
+      "loss": 0.9158,
+      "step": 3510
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00010978236691081365,
+      "loss": 0.9104,
+      "step": 3515
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00010952437024168444,
+      "loss": 0.9305,
+      "step": 3520
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00010926630957984087,
+      "loss": 0.9383,
+      "step": 3525
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00010900818665915109,
+      "loss": 0.9124,
+      "step": 3530
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.00010875000321390154,
+      "loss": 0.9169,
+      "step": 3535
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00010849176097878535,
+      "loss": 0.9044,
+      "step": 3540
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00010823346168889062,
+      "loss": 0.9234,
+      "step": 3545
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00010797510707968878,
+      "loss": 0.9431,
+      "step": 3550
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00010771669888702303,
+      "loss": 0.9263,
+      "step": 3555
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00010745823884709647,
+      "loss": 0.9423,
+      "step": 3560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00010719972869646062,
+      "loss": 0.9232,
+      "step": 3565
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00010694117017200372,
+      "loss": 0.962,
+      "step": 3570
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00010668256501093892,
+      "loss": 0.935,
+      "step": 3575
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00010642391495079278,
+      "loss": 0.9212,
+      "step": 3580
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00010616522172939356,
+      "loss": 0.9269,
+      "step": 3585
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00010590648708485946,
+      "loss": 0.9182,
+      "step": 3590
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.000105647712755587,
+      "loss": 0.9437,
+      "step": 3595
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00010538890048023937,
+      "loss": 0.9449,
+      "step": 3600
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001051300519977347,
+      "loss": 0.9023,
+      "step": 3605
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00010487116904723433,
+      "loss": 0.9136,
+      "step": 3610
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00010461225336813128,
+      "loss": 0.9317,
+      "step": 3615
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00010435330670003842,
+      "loss": 0.8979,
+      "step": 3620
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00010409433078277684,
+      "loss": 0.9319,
+      "step": 3625
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00010383532735636411,
+      "loss": 0.9344,
+      "step": 3630
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00010357629816100272,
+      "loss": 0.907,
+      "step": 3635
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001033172449370682,
+      "loss": 0.9222,
+      "step": 3640
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00010305816942509761,
+      "loss": 0.9384,
+      "step": 3645
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00010279907336577765,
+      "loss": 0.9195,
+      "step": 3650
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00010253995849993321,
+      "loss": 0.9177,
+      "step": 3655
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001022808265685154,
+      "loss": 0.9058,
+      "step": 3660
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.0001020216793125901,
+      "loss": 0.9293,
+      "step": 3665
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.34765625,
+      "learning_rate": 0.00010176251847332614,
+      "loss": 0.8824,
+      "step": 3670
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.00010150334579198353,
+      "loss": 0.9316,
+      "step": 3675
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.00010124416300990196,
+      "loss": 0.9351,
+      "step": 3680
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00010098497186848888,
+      "loss": 0.9187,
+      "step": 3685
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00010072577410920794,
+      "loss": 0.9019,
+      "step": 3690
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00010046657147356733,
+      "loss": 0.9152,
+      "step": 3695
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00010020736570310789,
+      "loss": 0.904,
+      "step": 3700
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.99481585393916e-05,
+      "loss": 0.9462,
+      "step": 3705
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.968895172398974e-05,
+      "loss": 0.9428,
+      "step": 3710
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.94297469984713e-05,
+      "loss": 0.9453,
+      "step": 3715
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.917054610439124e-05,
+      "loss": 0.9176,
+      "step": 3720
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.89113507832787e-05,
+      "loss": 0.951,
+      "step": 3725
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.865216277662545e-05,
+      "loss": 0.8904,
+      "step": 3730
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.83929838258741e-05,
+      "loss": 0.9197,
+      "step": 3735
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.813381567240639e-05,
+      "loss": 0.9342,
+      "step": 3740
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.787466005753152e-05,
+      "loss": 0.9713,
+      "step": 3745
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.761551872247449e-05,
+      "loss": 0.9556,
+      "step": 3750
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.735639340836428e-05,
+      "loss": 0.9125,
+      "step": 3755
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.709728585622229e-05,
+      "loss": 0.9716,
+      "step": 3760
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.68381978069506e-05,
+      "loss": 0.9105,
+      "step": 3765
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.657913100132011e-05,
+      "loss": 0.8839,
+      "step": 3770
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.632008717995916e-05,
+      "loss": 0.9204,
+      "step": 3775
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.606106808334165e-05,
+      "loss": 0.8863,
+      "step": 3780
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.580207545177516e-05,
+      "loss": 0.905,
+      "step": 3785
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.554311102538966e-05,
+      "loss": 0.9207,
+      "step": 3790
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.528417654412564e-05,
+      "loss": 0.9598,
+      "step": 3795
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.502527374772217e-05,
+      "loss": 0.943,
+      "step": 3800
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.476640437570562e-05,
+      "loss": 0.9201,
+      "step": 3805
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.450757016737776e-05,
+      "loss": 0.9366,
+      "step": 3810
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.424877286180404e-05,
+      "loss": 0.9357,
+      "step": 3815
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.3990014197802e-05,
+      "loss": 0.9431,
+      "step": 3820
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.37312959139296e-05,
+      "loss": 0.9362,
+      "step": 3825
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.347261974847341e-05,
+      "loss": 0.9157,
+      "step": 3830
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.321398743943706e-05,
+      "loss": 0.9213,
+      "step": 3835
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.295540072452951e-05,
+      "loss": 0.9502,
+      "step": 3840
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.269686134115336e-05,
+      "loss": 0.9224,
+      "step": 3845
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.243837102639328e-05,
+      "loss": 0.8954,
+      "step": 3850
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.217993151700408e-05,
+      "loss": 0.9021,
+      "step": 3855
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.19215445493994e-05,
+      "loss": 0.9367,
+      "step": 3860
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.166321185963984e-05,
+      "loss": 0.9301,
+      "step": 3865
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.140493518342113e-05,
+      "loss": 0.9468,
+      "step": 3870
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.114671625606285e-05,
+      "loss": 0.891,
+      "step": 3875
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.088855681249658e-05,
+      "loss": 0.938,
+      "step": 3880
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.063045858725406e-05,
+      "loss": 0.9287,
+      "step": 3885
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.037242331445588e-05,
+      "loss": 0.8992,
+      "step": 3890
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.011445272779962e-05,
+      "loss": 0.9444,
+      "step": 3895
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.985654856054818e-05,
+      "loss": 0.9119,
+      "step": 3900
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.34375,
+      "learning_rate": 8.95987125455183e-05,
+      "loss": 0.904,
+      "step": 3905
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.33984375,
+      "learning_rate": 8.934094641506873e-05,
+      "loss": 0.9223,
+      "step": 3910
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.349609375,
+      "learning_rate": 8.908325190108873e-05,
+      "loss": 0.9288,
+      "step": 3915
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.349609375,
+      "learning_rate": 8.882563073498635e-05,
+      "loss": 0.9177,
+      "step": 3920
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.359375,
+      "learning_rate": 8.856808464767689e-05,
+      "loss": 0.888,
+      "step": 3925
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.345703125,
+      "learning_rate": 8.831061536957107e-05,
+      "loss": 0.9174,
+      "step": 3930
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.359375,
+      "learning_rate": 8.80532246305637e-05,
+      "loss": 0.9345,
+      "step": 3935
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.333984375,
+      "learning_rate": 8.779591416002179e-05,
+      "loss": 0.926,
+      "step": 3940
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.349609375,
+      "learning_rate": 8.753868568677311e-05,
+      "loss": 0.9409,
+      "step": 3945
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.34375,
+      "learning_rate": 8.728154093909441e-05,
+      "loss": 0.918,
+      "step": 3950
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.36328125,
+      "learning_rate": 8.702448164470007e-05,
+      "loss": 0.9695,
+      "step": 3955
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.349609375,
+      "learning_rate": 8.676750953073011e-05,
+      "loss": 0.9221,
+      "step": 3960
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.341796875,
+      "learning_rate": 8.65106263237389e-05,
+      "loss": 0.9461,
+      "step": 3965
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3515625,
+      "learning_rate": 8.625383374968357e-05,
+      "loss": 0.9107,
+      "step": 3970
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.359375,
+      "learning_rate": 8.599713353391207e-05,
+      "loss": 0.9238,
+      "step": 3975
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.34765625,
+      "learning_rate": 8.574052740115201e-05,
+      "loss": 0.9063,
+      "step": 3980
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.34765625,
+      "learning_rate": 8.548401707549878e-05,
+      "loss": 0.9457,
+      "step": 3985
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.353515625,
+      "learning_rate": 8.522760428040402e-05,
+      "loss": 0.9385,
+      "step": 3990
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.341796875,
+      "learning_rate": 8.49712907386642e-05,
+      "loss": 0.9101,
+      "step": 3995
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.34375,
+      "learning_rate": 8.471507817240882e-05,
+      "loss": 0.9426,
+      "step": 4000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.345703125,
+      "learning_rate": 8.445896830308898e-05,
+      "loss": 0.9284,
+      "step": 4005
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.34375,
+      "learning_rate": 8.420296285146574e-05,
+      "loss": 0.8853,
+      "step": 4010
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.33984375,
+      "learning_rate": 8.394706353759869e-05,
+      "loss": 0.9316,
+      "step": 4015
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3515625,
+      "learning_rate": 8.369127208083418e-05,
+      "loss": 0.9209,
+      "step": 4020
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3515625,
+      "learning_rate": 8.343559019979392e-05,
+      "loss": 0.903,
+      "step": 4025
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35546875,
+      "learning_rate": 8.318001961236349e-05,
+      "loss": 0.9319,
+      "step": 4030
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.333984375,
+      "learning_rate": 8.292456203568055e-05,
+      "loss": 0.8953,
+      "step": 4035
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.34765625,
+      "learning_rate": 8.266921918612359e-05,
+      "loss": 0.9299,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3515625,
+      "learning_rate": 8.241399277930021e-05,
+      "loss": 0.9355,
+      "step": 4045
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.330078125,
+      "learning_rate": 8.215888453003562e-05,
+      "loss": 0.8886,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.337890625,
+      "learning_rate": 8.190389615236123e-05,
+      "loss": 0.9026,
+      "step": 4055
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.341796875,
+      "learning_rate": 8.164902935950303e-05,
+      "loss": 0.8896,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.345703125,
+      "learning_rate": 8.139428586386998e-05,
+      "loss": 0.9258,
+      "step": 4065
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3515625,
+      "learning_rate": 8.11396673770428e-05,
+      "loss": 0.9139,
+      "step": 4070
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.34375,
+      "learning_rate": 8.088517560976226e-05,
+      "loss": 0.9078,
+      "step": 4075
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.353515625,
+      "learning_rate": 8.063081227191755e-05,
+      "loss": 0.9161,
+      "step": 4080
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.34765625,
+      "learning_rate": 8.037657907253523e-05,
+      "loss": 0.9098,
+      "step": 4085
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.337890625,
+      "learning_rate": 8.012247771976726e-05,
+      "loss": 0.9088,
+      "step": 4090
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.349609375,
+      "learning_rate": 7.986850992087984e-05,
+      "loss": 0.9156,
+      "step": 4095
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3515625,
+      "learning_rate": 7.961467738224189e-05,
+      "loss": 0.9148,
+      "step": 4100
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.936098180931341e-05,
+      "loss": 0.9172,
+      "step": 4105
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.910742490663425e-05,
+      "loss": 0.9105,
+      "step": 4110
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.337890625,
+      "learning_rate": 7.885400837781255e-05,
+      "loss": 0.9309,
+      "step": 4115
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.337890625,
+      "learning_rate": 7.860073392551323e-05,
+      "loss": 0.8877,
+      "step": 4120
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.349609375,
+      "learning_rate": 7.83476032514467e-05,
+      "loss": 0.9208,
+      "step": 4125
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.34375,
+      "learning_rate": 7.809461805635734e-05,
+      "loss": 0.8915,
+      "step": 4130
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.357421875,
+      "learning_rate": 7.784178004001197e-05,
+      "loss": 0.9121,
+      "step": 4135
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.758909090118868e-05,
+      "loss": 0.9053,
+      "step": 4140
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.733655233766528e-05,
+      "loss": 0.9288,
+      "step": 4145
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.708416604620772e-05,
+      "loss": 0.9522,
+      "step": 4150
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.349609375,
+      "learning_rate": 7.683193372255898e-05,
+      "loss": 0.8911,
+      "step": 4155
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.34375,
+      "learning_rate": 7.657985706142767e-05,
+      "loss": 0.917,
+      "step": 4160
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.33984375,
+      "learning_rate": 7.632793775647625e-05,
+      "loss": 0.8925,
+      "step": 4165
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.349609375,
+      "learning_rate": 7.607617750031014e-05,
+      "loss": 0.9262,
+      "step": 4170
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.345703125,
+      "learning_rate": 7.58245779844661e-05,
+      "loss": 0.954,
+      "step": 4175
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.35546875,
+      "learning_rate": 7.557314089940085e-05,
+      "loss": 0.9071,
+      "step": 4180
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3515625,
+      "learning_rate": 7.532186793447977e-05,
+      "loss": 0.9207,
+      "step": 4185
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.507076077796565e-05,
+      "loss": 0.9352,
+      "step": 4190
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.34375,
+      "learning_rate": 7.481982111700705e-05,
+      "loss": 0.9581,
+      "step": 4195
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.357421875,
+      "learning_rate": 7.456905063762731e-05,
+      "loss": 0.9343,
+      "step": 4200
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 6734,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 5.906288014140637e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}