sanchit-gandhi HF staff commited on
Commit
728482e
1 Parent(s): 3dc454f

End of training

Browse files
all_results.json CHANGED
@@ -1,14 +1,14 @@
1
  {
2
  "epoch": 1.0,
3
- "eval_loss": 4.171594619750977,
4
- "eval_runtime": 915.7329,
5
  "eval_samples": 2642,
6
- "eval_samples_per_second": 2.885,
7
- "eval_steps_per_second": 0.241,
8
- "eval_wer": 2.334966283220944,
9
- "train_loss": 4.333276915630508,
10
- "train_runtime": 5595.6802,
11
  "train_samples": 28538,
12
- "train_samples_per_second": 5.1,
13
- "train_steps_per_second": 0.106
14
  }
 
1
  {
2
  "epoch": 1.0,
3
+ "eval_loss": 4.222872734069824,
4
+ "eval_runtime": 911.3606,
5
  "eval_samples": 2642,
6
+ "eval_samples_per_second": 2.899,
7
+ "eval_steps_per_second": 0.242,
8
+ "eval_wer": 2.421539071796906,
9
+ "train_loss": 4.444549621556343,
10
+ "train_runtime": 4661.3109,
11
  "train_samples": 28538,
12
+ "train_samples_per_second": 6.122,
13
+ "train_steps_per_second": 0.064
14
  }
eval_results.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
  "epoch": 1.0,
3
- "eval_loss": 4.171594619750977,
4
- "eval_runtime": 915.7329,
5
  "eval_samples": 2642,
6
- "eval_samples_per_second": 2.885,
7
- "eval_steps_per_second": 0.241,
8
- "eval_wer": 2.334966283220944
9
  }
 
1
  {
2
  "epoch": 1.0,
3
+ "eval_loss": 4.222872734069824,
4
+ "eval_runtime": 911.3606,
5
  "eval_samples": 2642,
6
+ "eval_samples_per_second": 2.899,
7
+ "eval_steps_per_second": 0.242,
8
+ "eval_wer": 2.421539071796906
9
  }
runs/Feb24_03-02-31_sanchit--v100/events.out.tfevents.1645677658.sanchit--v100.146177.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca0857be8b046f641f2ee90f04f4d06b36dccd818384ca43ff04a1381349c76f
3
+ size 358
train_results.json CHANGED
@@ -1,8 +1,8 @@
1
  {
2
  "epoch": 1.0,
3
- "train_loss": 4.333276915630508,
4
- "train_runtime": 5595.6802,
5
  "train_samples": 28538,
6
- "train_samples_per_second": 5.1,
7
- "train_steps_per_second": 0.106
8
  }
 
1
  {
2
  "epoch": 1.0,
3
+ "train_loss": 4.444549621556343,
4
+ "train_runtime": 4661.3109,
5
  "train_samples": 28538,
6
+ "train_samples_per_second": 6.122,
7
+ "train_steps_per_second": 0.064
8
  }
trainer_state.json CHANGED
@@ -2,3595 +2,1804 @@
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
  "epoch": 0.9987389659520807,
5
- "global_step": 594,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
  "epoch": 0.0,
12
- "learning_rate": 0.0,
13
- "loss": 4.8043,
14
  "step": 1
15
  },
16
  {
17
- "epoch": 0.0,
18
  "learning_rate": 2e-08,
19
- "loss": 4.9816,
20
  "step": 2
21
  },
22
  {
23
  "epoch": 0.01,
24
  "learning_rate": 4e-08,
25
- "loss": 4.8553,
26
  "step": 3
27
  },
28
  {
29
  "epoch": 0.01,
30
  "learning_rate": 6.000000000000001e-08,
31
- "loss": 4.8133,
32
  "step": 4
33
  },
34
  {
35
- "epoch": 0.01,
36
  "learning_rate": 8e-08,
37
- "loss": 4.7678,
38
  "step": 5
39
  },
40
  {
41
- "epoch": 0.01,
42
  "learning_rate": 1.0000000000000001e-07,
43
- "loss": 4.7823,
44
  "step": 6
45
  },
46
  {
47
- "epoch": 0.01,
48
  "learning_rate": 1.2000000000000002e-07,
49
- "loss": 4.8813,
50
  "step": 7
51
  },
52
  {
53
- "epoch": 0.01,
54
  "learning_rate": 1.4e-07,
55
- "loss": 4.7725,
56
  "step": 8
57
  },
58
  {
59
- "epoch": 0.02,
60
  "learning_rate": 1.6e-07,
61
- "loss": 4.8957,
62
  "step": 9
63
  },
64
  {
65
- "epoch": 0.02,
66
  "learning_rate": 1.8e-07,
67
- "loss": 4.8176,
68
  "step": 10
69
  },
70
  {
71
- "epoch": 0.02,
72
  "learning_rate": 2.0000000000000002e-07,
73
- "loss": 4.8163,
74
  "step": 11
75
  },
76
  {
77
- "epoch": 0.02,
78
  "learning_rate": 2.2e-07,
79
- "loss": 4.8674,
80
  "step": 12
81
  },
82
  {
83
- "epoch": 0.02,
84
  "learning_rate": 2.4000000000000003e-07,
85
- "loss": 4.8031,
86
  "step": 13
87
  },
88
  {
89
- "epoch": 0.02,
90
  "learning_rate": 2.6e-07,
91
- "loss": 4.7586,
92
  "step": 14
93
  },
94
  {
95
- "epoch": 0.03,
96
  "learning_rate": 2.8e-07,
97
- "loss": 4.6561,
98
  "step": 15
99
  },
100
  {
101
- "epoch": 0.03,
102
  "learning_rate": 3.0000000000000004e-07,
103
- "loss": 4.8397,
104
  "step": 16
105
  },
106
  {
107
- "epoch": 0.03,
108
  "learning_rate": 3.2e-07,
109
- "loss": 4.8695,
110
  "step": 17
111
  },
112
  {
113
- "epoch": 0.03,
114
  "learning_rate": 3.4000000000000003e-07,
115
- "loss": 4.9448,
116
  "step": 18
117
  },
118
  {
119
- "epoch": 0.03,
120
  "learning_rate": 3.6e-07,
121
- "loss": 4.7123,
122
  "step": 19
123
  },
124
  {
125
- "epoch": 0.03,
126
  "learning_rate": 3.8e-07,
127
- "loss": 4.9487,
128
  "step": 20
129
  },
130
  {
131
- "epoch": 0.04,
132
  "learning_rate": 4.0000000000000003e-07,
133
- "loss": 4.898,
134
  "step": 21
135
  },
136
  {
137
- "epoch": 0.04,
138
  "learning_rate": 4.2000000000000006e-07,
139
- "loss": 4.8324,
140
  "step": 22
141
  },
142
  {
143
- "epoch": 0.04,
144
  "learning_rate": 4.4e-07,
145
- "loss": 4.8531,
146
  "step": 23
147
  },
148
  {
149
- "epoch": 0.04,
150
  "learning_rate": 4.6000000000000004e-07,
151
- "loss": 4.9432,
152
  "step": 24
153
  },
154
  {
155
- "epoch": 0.04,
156
  "learning_rate": 4.800000000000001e-07,
157
- "loss": 4.7127,
158
  "step": 25
159
  },
160
  {
161
- "epoch": 0.04,
162
  "learning_rate": 5.000000000000001e-07,
163
- "loss": 5.1345,
164
  "step": 26
165
  },
166
  {
167
- "epoch": 0.05,
168
  "learning_rate": 5.2e-07,
169
- "loss": 4.7968,
170
  "step": 27
171
  },
172
  {
173
- "epoch": 0.05,
174
  "learning_rate": 5.4e-07,
175
- "loss": 4.6922,
176
  "step": 28
177
  },
178
  {
179
- "epoch": 0.05,
180
  "learning_rate": 5.6e-07,
181
- "loss": 4.7146,
182
  "step": 29
183
  },
184
  {
185
- "epoch": 0.05,
186
  "learning_rate": 5.800000000000001e-07,
187
- "loss": 4.8548,
188
  "step": 30
189
  },
190
  {
191
- "epoch": 0.05,
192
  "learning_rate": 6.000000000000001e-07,
193
- "loss": 4.8666,
194
  "step": 31
195
  },
196
  {
197
- "epoch": 0.05,
198
  "learning_rate": 6.200000000000001e-07,
199
- "loss": 4.7535,
200
  "step": 32
201
  },
202
  {
203
- "epoch": 0.06,
204
- "learning_rate": 6.200000000000001e-07,
205
- "loss": 4.7578,
206
  "step": 33
207
  },
208
  {
209
- "epoch": 0.06,
210
- "learning_rate": 6.4e-07,
211
- "loss": 4.6922,
212
  "step": 34
213
  },
214
  {
215
- "epoch": 0.06,
216
- "learning_rate": 6.6e-07,
217
- "loss": 4.7539,
218
  "step": 35
219
  },
220
  {
221
- "epoch": 0.06,
222
- "learning_rate": 6.800000000000001e-07,
223
- "loss": 4.9275,
224
  "step": 36
225
  },
226
  {
227
- "epoch": 0.06,
228
- "learning_rate": 7.000000000000001e-07,
229
- "loss": 4.7236,
230
  "step": 37
231
  },
232
  {
233
- "epoch": 0.06,
234
- "learning_rate": 7.2e-07,
235
- "loss": 4.8574,
236
  "step": 38
237
  },
238
  {
239
- "epoch": 0.07,
240
- "learning_rate": 7.4e-07,
241
- "loss": 5.0691,
242
  "step": 39
243
  },
244
  {
245
- "epoch": 0.07,
246
- "learning_rate": 7.6e-07,
247
- "loss": 4.6674,
248
  "step": 40
249
  },
250
  {
251
- "epoch": 0.07,
252
- "learning_rate": 7.8e-07,
253
- "loss": 4.7318,
254
  "step": 41
255
  },
256
  {
257
- "epoch": 0.07,
258
- "learning_rate": 8.000000000000001e-07,
259
- "loss": 4.9842,
260
  "step": 42
261
  },
262
  {
263
- "epoch": 0.07,
264
- "learning_rate": 8.200000000000001e-07,
265
- "loss": 4.7206,
266
  "step": 43
267
  },
268
  {
269
- "epoch": 0.07,
270
- "learning_rate": 8.400000000000001e-07,
271
- "loss": 4.8842,
272
  "step": 44
273
  },
274
  {
275
- "epoch": 0.08,
276
- "learning_rate": 8.6e-07,
277
- "loss": 5.0102,
278
  "step": 45
279
  },
280
  {
281
- "epoch": 0.08,
282
- "learning_rate": 8.8e-07,
283
- "loss": 5.1112,
284
  "step": 46
285
  },
286
  {
287
- "epoch": 0.08,
288
- "learning_rate": 9.000000000000001e-07,
289
- "loss": 5.0286,
290
  "step": 47
291
  },
292
  {
293
- "epoch": 0.08,
294
- "learning_rate": 9.200000000000001e-07,
295
- "loss": 5.4426,
296
  "step": 48
297
  },
298
  {
299
- "epoch": 0.08,
300
- "learning_rate": 9.400000000000001e-07,
301
- "loss": 5.4768,
302
  "step": 49
303
  },
304
  {
305
- "epoch": 0.08,
306
- "learning_rate": 9.400000000000001e-07,
307
- "loss": 5.3523,
308
  "step": 50
309
  },
310
  {
311
- "epoch": 0.09,
312
- "learning_rate": 9.600000000000001e-07,
313
- "loss": 4.5321,
314
  "step": 51
315
  },
316
  {
317
- "epoch": 0.09,
318
- "learning_rate": 9.800000000000001e-07,
319
- "loss": 4.5271,
320
  "step": 52
321
  },
322
  {
323
- "epoch": 0.09,
324
- "learning_rate": 1.0000000000000002e-06,
325
- "loss": 4.5945,
326
  "step": 53
327
  },
328
  {
329
- "epoch": 0.09,
330
- "learning_rate": 1.02e-06,
331
- "loss": 4.674,
332
  "step": 54
333
  },
334
  {
335
- "epoch": 0.09,
336
- "learning_rate": 1.04e-06,
337
- "loss": 4.5153,
338
  "step": 55
339
  },
340
  {
341
- "epoch": 0.09,
342
- "learning_rate": 1.06e-06,
343
- "loss": 4.6963,
344
  "step": 56
345
  },
346
  {
347
- "epoch": 0.1,
348
- "learning_rate": 1.08e-06,
349
- "loss": 4.56,
350
  "step": 57
351
  },
352
  {
353
- "epoch": 0.1,
354
- "learning_rate": 1.1e-06,
355
- "loss": 4.8043,
356
  "step": 58
357
  },
358
  {
359
- "epoch": 0.1,
360
- "learning_rate": 1.12e-06,
361
- "loss": 4.5976,
362
  "step": 59
363
  },
364
  {
365
- "epoch": 0.1,
366
- "learning_rate": 1.14e-06,
367
- "loss": 4.3466,
368
  "step": 60
369
  },
370
  {
371
- "epoch": 0.1,
372
- "learning_rate": 1.1600000000000001e-06,
373
- "loss": 4.6048,
374
  "step": 61
375
  },
376
  {
377
- "epoch": 0.1,
378
- "learning_rate": 1.1800000000000001e-06,
379
- "loss": 4.6691,
380
  "step": 62
381
  },
382
  {
383
- "epoch": 0.11,
384
- "learning_rate": 1.2000000000000002e-06,
385
- "loss": 4.5441,
386
  "step": 63
387
  },
388
  {
389
- "epoch": 0.11,
390
- "learning_rate": 1.2200000000000002e-06,
391
- "loss": 4.4973,
392
  "step": 64
393
  },
394
  {
395
- "epoch": 0.11,
396
- "learning_rate": 1.2400000000000002e-06,
397
- "loss": 4.5558,
398
  "step": 65
399
  },
400
  {
401
- "epoch": 0.11,
402
- "learning_rate": 1.26e-06,
403
- "loss": 4.4949,
404
  "step": 66
405
  },
406
  {
407
- "epoch": 0.11,
408
- "learning_rate": 1.28e-06,
409
- "loss": 4.4888,
410
  "step": 67
411
  },
412
  {
413
- "epoch": 0.11,
414
- "learning_rate": 1.3e-06,
415
- "loss": 4.4224,
416
  "step": 68
417
  },
418
  {
419
- "epoch": 0.12,
420
- "learning_rate": 1.32e-06,
421
- "loss": 4.544,
422
  "step": 69
423
  },
424
  {
425
- "epoch": 0.12,
426
- "learning_rate": 1.34e-06,
427
- "loss": 4.5557,
428
  "step": 70
429
  },
430
  {
431
- "epoch": 0.12,
432
- "learning_rate": 1.3600000000000001e-06,
433
- "loss": 4.6351,
434
  "step": 71
435
  },
436
  {
437
- "epoch": 0.12,
438
- "learning_rate": 1.3800000000000001e-06,
439
- "loss": 4.5891,
440
  "step": 72
441
  },
442
  {
443
- "epoch": 0.12,
444
- "learning_rate": 1.4000000000000001e-06,
445
- "loss": 4.385,
446
  "step": 73
447
  },
448
  {
449
- "epoch": 0.12,
450
- "learning_rate": 1.42e-06,
451
- "loss": 4.5821,
452
  "step": 74
453
  },
454
  {
455
- "epoch": 0.13,
456
- "learning_rate": 1.44e-06,
457
- "loss": 4.554,
458
  "step": 75
459
  },
460
  {
461
- "epoch": 0.13,
462
- "learning_rate": 1.46e-06,
463
- "loss": 4.5606,
464
  "step": 76
465
  },
466
  {
467
- "epoch": 0.13,
468
- "learning_rate": 1.48e-06,
469
- "loss": 4.5196,
470
  "step": 77
471
  },
472
  {
473
- "epoch": 0.13,
474
- "learning_rate": 1.5e-06,
475
- "loss": 4.5308,
476
  "step": 78
477
  },
478
  {
479
- "epoch": 0.13,
480
- "learning_rate": 1.52e-06,
481
- "loss": 4.4185,
482
  "step": 79
483
  },
484
  {
485
- "epoch": 0.13,
486
- "learning_rate": 1.54e-06,
487
- "loss": 4.555,
488
  "step": 80
489
  },
490
  {
491
- "epoch": 0.14,
492
- "learning_rate": 1.56e-06,
493
- "loss": 4.508,
494
  "step": 81
495
  },
496
  {
497
- "epoch": 0.14,
498
- "learning_rate": 1.5800000000000001e-06,
499
- "loss": 4.5176,
500
  "step": 82
501
  },
502
  {
503
- "epoch": 0.14,
504
- "learning_rate": 1.6000000000000001e-06,
505
- "loss": 4.5365,
506
  "step": 83
507
  },
508
  {
509
- "epoch": 0.14,
510
- "learning_rate": 1.6200000000000002e-06,
511
- "loss": 4.5456,
512
  "step": 84
513
  },
514
  {
515
- "epoch": 0.14,
516
- "learning_rate": 1.6400000000000002e-06,
517
- "loss": 4.4706,
518
  "step": 85
519
  },
520
  {
521
- "epoch": 0.14,
522
- "learning_rate": 1.6600000000000002e-06,
523
- "loss": 4.3598,
524
  "step": 86
525
  },
526
  {
527
- "epoch": 0.15,
528
- "learning_rate": 1.6800000000000002e-06,
529
- "loss": 4.6727,
530
  "step": 87
531
  },
532
  {
533
- "epoch": 0.15,
534
- "learning_rate": 1.7000000000000002e-06,
535
- "loss": 4.5677,
536
  "step": 88
537
  },
538
  {
539
- "epoch": 0.15,
540
- "learning_rate": 1.72e-06,
541
- "loss": 4.5185,
542
  "step": 89
543
  },
544
  {
545
- "epoch": 0.15,
546
- "learning_rate": 1.74e-06,
547
- "loss": 4.5671,
548
  "step": 90
549
  },
550
  {
551
- "epoch": 0.15,
552
- "learning_rate": 1.76e-06,
553
- "loss": 4.4984,
554
  "step": 91
555
  },
556
  {
557
- "epoch": 0.15,
558
- "learning_rate": 1.7800000000000001e-06,
559
- "loss": 4.6209,
560
  "step": 92
561
  },
562
  {
563
- "epoch": 0.16,
564
- "learning_rate": 1.8000000000000001e-06,
565
- "loss": 4.4814,
566
  "step": 93
567
  },
568
  {
569
- "epoch": 0.16,
570
- "learning_rate": 1.8200000000000002e-06,
571
- "loss": 4.7069,
572
  "step": 94
573
  },
574
  {
575
- "epoch": 0.16,
576
- "learning_rate": 1.8400000000000002e-06,
577
- "loss": 4.6537,
578
  "step": 95
579
  },
580
  {
581
- "epoch": 0.16,
582
- "learning_rate": 1.8600000000000002e-06,
583
- "loss": 4.6972,
584
  "step": 96
585
  },
586
  {
587
- "epoch": 0.16,
588
- "learning_rate": 1.8800000000000002e-06,
589
- "loss": 4.7654,
590
  "step": 97
591
  },
592
  {
593
- "epoch": 0.16,
594
- "learning_rate": 1.9000000000000002e-06,
595
- "loss": 4.5225,
596
  "step": 98
597
  },
598
  {
599
- "epoch": 0.17,
600
- "learning_rate": 1.9200000000000003e-06,
601
- "loss": 4.694,
602
  "step": 99
603
  },
604
  {
605
- "epoch": 0.17,
606
- "learning_rate": 1.94e-06,
607
- "loss": 4.487,
608
  "step": 100
609
  },
610
  {
611
- "epoch": 0.17,
612
- "learning_rate": 1.9600000000000003e-06,
613
- "loss": 4.539,
614
  "step": 101
615
  },
616
  {
617
- "epoch": 0.17,
618
- "learning_rate": 1.98e-06,
619
- "loss": 4.3454,
620
  "step": 102
621
  },
622
  {
623
- "epoch": 0.17,
624
- "learning_rate": 2.0000000000000003e-06,
625
- "loss": 4.3138,
626
  "step": 103
627
  },
628
  {
629
- "epoch": 0.17,
630
- "learning_rate": 2.02e-06,
631
- "loss": 4.2958,
632
  "step": 104
633
  },
634
  {
635
- "epoch": 0.18,
636
- "learning_rate": 2.04e-06,
637
- "loss": 4.2833,
638
  "step": 105
639
  },
640
  {
641
- "epoch": 0.18,
642
- "learning_rate": 2.06e-06,
643
- "loss": 4.3276,
644
  "step": 106
645
  },
646
  {
647
- "epoch": 0.18,
648
- "learning_rate": 2.08e-06,
649
- "loss": 4.4692,
650
  "step": 107
651
  },
652
  {
653
- "epoch": 0.18,
654
- "learning_rate": 2.1000000000000002e-06,
655
- "loss": 4.4968,
656
  "step": 108
657
  },
658
  {
659
- "epoch": 0.18,
660
- "learning_rate": 2.12e-06,
661
- "loss": 4.5626,
662
  "step": 109
663
  },
664
  {
665
- "epoch": 0.18,
666
- "learning_rate": 2.1400000000000003e-06,
667
- "loss": 4.3518,
668
  "step": 110
669
  },
670
  {
671
- "epoch": 0.19,
672
- "learning_rate": 2.16e-06,
673
- "loss": 4.3489,
674
  "step": 111
675
  },
676
  {
677
- "epoch": 0.19,
678
- "learning_rate": 2.1800000000000003e-06,
679
- "loss": 4.4177,
680
  "step": 112
681
  },
682
  {
683
- "epoch": 0.19,
684
- "learning_rate": 2.2e-06,
685
- "loss": 4.4097,
686
  "step": 113
687
  },
688
  {
689
- "epoch": 0.19,
690
- "learning_rate": 2.2200000000000003e-06,
691
- "loss": 4.4427,
692
  "step": 114
693
  },
694
  {
695
- "epoch": 0.19,
696
- "learning_rate": 2.24e-06,
697
- "loss": 4.3799,
698
  "step": 115
699
  },
700
  {
701
- "epoch": 0.2,
702
- "learning_rate": 2.2600000000000004e-06,
703
- "loss": 4.3865,
704
  "step": 116
705
  },
706
  {
707
- "epoch": 0.2,
708
- "learning_rate": 2.28e-06,
709
- "loss": 4.4844,
710
  "step": 117
711
  },
712
  {
713
- "epoch": 0.2,
714
- "learning_rate": 2.3000000000000004e-06,
715
- "loss": 4.2464,
716
  "step": 118
717
  },
718
  {
719
- "epoch": 0.2,
720
- "learning_rate": 2.3200000000000002e-06,
721
- "loss": 4.2288,
722
  "step": 119
723
  },
724
  {
725
- "epoch": 0.2,
726
- "learning_rate": 2.3400000000000005e-06,
727
- "loss": 4.2251,
728
  "step": 120
729
  },
730
  {
731
- "epoch": 0.2,
732
- "learning_rate": 2.3600000000000003e-06,
733
- "loss": 4.4691,
734
  "step": 121
735
  },
736
  {
737
- "epoch": 0.21,
738
- "learning_rate": 2.38e-06,
739
- "loss": 4.5464,
740
  "step": 122
741
  },
742
  {
743
- "epoch": 0.21,
744
- "learning_rate": 2.4000000000000003e-06,
745
- "loss": 4.3148,
746
  "step": 123
747
  },
748
  {
749
- "epoch": 0.21,
750
- "learning_rate": 2.42e-06,
751
- "loss": 4.3815,
752
  "step": 124
753
  },
754
  {
755
- "epoch": 0.21,
756
- "learning_rate": 2.4400000000000004e-06,
757
- "loss": 4.3865,
758
  "step": 125
759
  },
760
  {
761
- "epoch": 0.21,
762
- "learning_rate": 2.46e-06,
763
- "loss": 4.2059,
764
  "step": 126
765
  },
766
  {
767
- "epoch": 0.21,
768
- "learning_rate": 2.4800000000000004e-06,
769
- "loss": 4.264,
770
  "step": 127
771
  },
772
  {
773
- "epoch": 0.22,
774
- "learning_rate": 2.5e-06,
775
- "loss": 4.2287,
776
  "step": 128
777
  },
778
  {
779
- "epoch": 0.22,
780
- "learning_rate": 2.52e-06,
781
- "loss": 4.3775,
782
  "step": 129
783
  },
784
  {
785
- "epoch": 0.22,
786
- "learning_rate": 2.5400000000000002e-06,
787
- "loss": 4.403,
788
  "step": 130
789
  },
790
  {
791
- "epoch": 0.22,
792
- "learning_rate": 2.56e-06,
793
- "loss": 4.3432,
794
  "step": 131
795
  },
796
  {
797
- "epoch": 0.22,
798
- "learning_rate": 2.5800000000000003e-06,
799
- "loss": 4.2978,
800
  "step": 132
801
  },
802
  {
803
- "epoch": 0.22,
804
- "learning_rate": 2.6e-06,
805
- "loss": 4.3968,
806
  "step": 133
807
  },
808
  {
809
- "epoch": 0.23,
810
- "learning_rate": 2.6200000000000003e-06,
811
- "loss": 4.4003,
812
  "step": 134
813
  },
814
  {
815
- "epoch": 0.23,
816
- "learning_rate": 2.64e-06,
817
- "loss": 4.449,
818
  "step": 135
819
  },
820
  {
821
- "epoch": 0.23,
822
- "learning_rate": 2.6600000000000004e-06,
823
- "loss": 4.3175,
824
  "step": 136
825
  },
826
  {
827
- "epoch": 0.23,
828
- "learning_rate": 2.68e-06,
829
- "loss": 4.3273,
830
  "step": 137
831
  },
832
  {
833
- "epoch": 0.23,
834
- "learning_rate": 2.7000000000000004e-06,
835
- "loss": 4.3344,
836
  "step": 138
837
  },
838
  {
839
- "epoch": 0.23,
840
- "learning_rate": 2.7200000000000002e-06,
841
- "loss": 4.32,
842
  "step": 139
843
  },
844
  {
845
- "epoch": 0.24,
846
- "learning_rate": 2.7400000000000004e-06,
847
- "loss": 4.5413,
848
  "step": 140
849
  },
850
  {
851
- "epoch": 0.24,
852
- "learning_rate": 2.7600000000000003e-06,
853
- "loss": 4.4599,
854
  "step": 141
855
  },
856
  {
857
- "epoch": 0.24,
858
- "learning_rate": 2.7800000000000005e-06,
859
- "loss": 4.495,
860
  "step": 142
861
  },
862
  {
863
- "epoch": 0.24,
864
- "learning_rate": 2.8000000000000003e-06,
865
- "loss": 4.538,
866
  "step": 143
867
  },
868
  {
869
- "epoch": 0.24,
870
- "learning_rate": 2.82e-06,
871
- "loss": 4.3266,
872
  "step": 144
873
  },
874
  {
875
- "epoch": 0.24,
876
- "learning_rate": 2.84e-06,
877
- "loss": 4.429,
878
  "step": 145
879
  },
880
  {
881
- "epoch": 0.25,
882
- "learning_rate": 2.86e-06,
883
- "loss": 4.7391,
884
  "step": 146
885
  },
886
  {
887
- "epoch": 0.25,
888
- "learning_rate": 2.88e-06,
889
- "loss": 4.7207,
890
  "step": 147
891
  },
892
  {
893
- "epoch": 0.25,
894
- "learning_rate": 2.9e-06,
895
- "loss": 4.7066,
896
  "step": 148
897
  },
898
  {
899
- "epoch": 0.25,
900
- "learning_rate": 2.92e-06,
901
- "loss": 4.6515,
902
  "step": 149
903
  },
904
  {
905
- "epoch": 0.25,
906
- "learning_rate": 2.9400000000000002e-06,
907
- "loss": 4.7034,
908
  "step": 150
909
  },
910
  {
911
- "epoch": 0.25,
912
- "learning_rate": 2.96e-06,
913
- "loss": 4.2933,
914
  "step": 151
915
  },
916
  {
917
- "epoch": 0.26,
918
- "learning_rate": 2.9800000000000003e-06,
919
- "loss": 4.2286,
920
  "step": 152
921
  },
922
  {
923
- "epoch": 0.26,
924
- "learning_rate": 3e-06,
925
- "loss": 4.22,
926
  "step": 153
927
  },
928
  {
929
- "epoch": 0.26,
930
- "learning_rate": 3.0200000000000003e-06,
931
- "loss": 4.1533,
932
  "step": 154
933
  },
934
  {
935
- "epoch": 0.26,
936
- "learning_rate": 3.04e-06,
937
- "loss": 4.196,
938
  "step": 155
939
  },
940
  {
941
- "epoch": 0.26,
942
- "learning_rate": 3.0600000000000003e-06,
943
- "loss": 4.184,
944
  "step": 156
945
  },
946
  {
947
- "epoch": 0.26,
948
- "learning_rate": 3.08e-06,
949
- "loss": 4.3548,
950
  "step": 157
951
  },
952
  {
953
- "epoch": 0.27,
954
- "learning_rate": 3.1000000000000004e-06,
955
- "loss": 4.1817,
956
  "step": 158
957
  },
958
  {
959
- "epoch": 0.27,
960
- "learning_rate": 3.12e-06,
961
- "loss": 4.1367,
962
  "step": 159
963
  },
964
  {
965
- "epoch": 0.27,
966
- "learning_rate": 3.1400000000000004e-06,
967
- "loss": 4.2649,
968
  "step": 160
969
  },
970
  {
971
- "epoch": 0.27,
972
- "learning_rate": 3.1600000000000002e-06,
973
- "loss": 4.3083,
974
  "step": 161
975
  },
976
  {
977
- "epoch": 0.27,
978
- "learning_rate": 3.1800000000000005e-06,
979
- "loss": 4.1601,
980
  "step": 162
981
  },
982
  {
983
- "epoch": 0.27,
984
- "learning_rate": 3.2000000000000003e-06,
985
- "loss": 4.2107,
986
  "step": 163
987
  },
988
  {
989
- "epoch": 0.28,
990
- "learning_rate": 3.2200000000000005e-06,
991
- "loss": 4.2489,
992
  "step": 164
993
  },
994
  {
995
- "epoch": 0.28,
996
- "learning_rate": 3.2400000000000003e-06,
997
- "loss": 4.2803,
998
  "step": 165
999
  },
1000
  {
1001
- "epoch": 0.28,
1002
- "learning_rate": 3.2600000000000006e-06,
1003
- "loss": 4.3913,
1004
  "step": 166
1005
  },
1006
  {
1007
- "epoch": 0.28,
1008
- "learning_rate": 3.2800000000000004e-06,
1009
- "loss": 4.228,
1010
  "step": 167
1011
  },
1012
  {
1013
- "epoch": 0.28,
1014
- "learning_rate": 3.3000000000000006e-06,
1015
- "loss": 4.4536,
1016
  "step": 168
1017
  },
1018
  {
1019
- "epoch": 0.28,
1020
- "learning_rate": 3.3200000000000004e-06,
1021
- "loss": 4.2729,
1022
  "step": 169
1023
  },
1024
  {
1025
- "epoch": 0.29,
1026
- "learning_rate": 3.3400000000000006e-06,
1027
- "loss": 4.3269,
1028
  "step": 170
1029
  },
1030
  {
1031
- "epoch": 0.29,
1032
- "learning_rate": 3.3600000000000004e-06,
1033
- "loss": 4.1987,
1034
  "step": 171
1035
  },
1036
  {
1037
- "epoch": 0.29,
1038
- "learning_rate": 3.3800000000000007e-06,
1039
- "loss": 4.1876,
1040
  "step": 172
1041
  },
1042
  {
1043
- "epoch": 0.29,
1044
- "learning_rate": 3.4000000000000005e-06,
1045
- "loss": 4.2146,
1046
  "step": 173
1047
  },
1048
  {
1049
- "epoch": 0.29,
1050
- "learning_rate": 3.4200000000000007e-06,
1051
- "loss": 4.1946,
1052
  "step": 174
1053
  },
1054
  {
1055
- "epoch": 0.29,
1056
- "learning_rate": 3.44e-06,
1057
- "loss": 4.3494,
1058
  "step": 175
1059
  },
1060
  {
1061
- "epoch": 0.3,
1062
- "learning_rate": 3.46e-06,
1063
- "loss": 4.3716,
1064
  "step": 176
1065
  },
1066
  {
1067
- "epoch": 0.3,
1068
- "learning_rate": 3.48e-06,
1069
- "loss": 4.3028,
1070
  "step": 177
1071
  },
1072
  {
1073
- "epoch": 0.3,
1074
- "learning_rate": 3.5e-06,
1075
- "loss": 4.1935,
1076
  "step": 178
1077
  },
1078
  {
1079
- "epoch": 0.3,
1080
- "learning_rate": 3.52e-06,
1081
- "loss": 4.331,
1082
  "step": 179
1083
  },
1084
  {
1085
- "epoch": 0.3,
1086
- "learning_rate": 3.54e-06,
1087
- "loss": 4.1889,
1088
  "step": 180
1089
  },
1090
  {
1091
- "epoch": 0.3,
1092
- "learning_rate": 3.5600000000000002e-06,
1093
- "loss": 4.2626,
1094
  "step": 181
1095
  },
1096
  {
1097
- "epoch": 0.31,
1098
- "learning_rate": 3.58e-06,
1099
- "loss": 4.261,
1100
  "step": 182
1101
  },
1102
  {
1103
- "epoch": 0.31,
1104
- "learning_rate": 3.6000000000000003e-06,
1105
- "loss": 4.239,
1106
  "step": 183
1107
  },
1108
  {
1109
- "epoch": 0.31,
1110
- "learning_rate": 3.62e-06,
1111
- "loss": 4.1259,
1112
  "step": 184
1113
  },
1114
  {
1115
- "epoch": 0.31,
1116
- "learning_rate": 3.6400000000000003e-06,
1117
- "loss": 4.4795,
1118
  "step": 185
1119
  },
1120
  {
1121
- "epoch": 0.31,
1122
- "learning_rate": 3.66e-06,
1123
- "loss": 4.2646,
1124
  "step": 186
1125
  },
1126
  {
1127
- "epoch": 0.31,
1128
- "learning_rate": 3.6800000000000003e-06,
1129
- "loss": 4.3469,
1130
  "step": 187
1131
  },
1132
  {
1133
- "epoch": 0.32,
1134
- "learning_rate": 3.7e-06,
1135
- "loss": 4.3586,
1136
  "step": 188
1137
  },
1138
  {
1139
- "epoch": 0.32,
1140
- "learning_rate": 3.7200000000000004e-06,
1141
- "loss": 4.5375,
1142
  "step": 189
1143
  },
1144
  {
1145
- "epoch": 0.32,
1146
- "learning_rate": 3.74e-06,
1147
- "loss": 4.2441,
1148
  "step": 190
1149
  },
1150
  {
1151
- "epoch": 0.32,
1152
- "learning_rate": 3.7600000000000004e-06,
1153
- "loss": 4.5268,
1154
  "step": 191
1155
  },
1156
  {
1157
- "epoch": 0.32,
1158
- "learning_rate": 3.7800000000000002e-06,
1159
- "loss": 4.4442,
1160
  "step": 192
1161
  },
1162
  {
1163
- "epoch": 0.32,
1164
- "learning_rate": 3.8000000000000005e-06,
1165
- "loss": 4.3399,
1166
  "step": 193
1167
  },
1168
  {
1169
- "epoch": 0.33,
1170
- "learning_rate": 3.820000000000001e-06,
1171
- "loss": 4.4032,
1172
  "step": 194
1173
  },
1174
  {
1175
- "epoch": 0.33,
1176
- "learning_rate": 3.8400000000000005e-06,
1177
- "loss": 4.3523,
1178
  "step": 195
1179
  },
1180
  {
1181
- "epoch": 0.33,
1182
- "learning_rate": 3.86e-06,
1183
- "loss": 4.259,
1184
  "step": 196
1185
  },
1186
  {
1187
- "epoch": 0.33,
1188
- "learning_rate": 3.88e-06,
1189
- "loss": 4.666,
1190
  "step": 197
1191
  },
1192
  {
1193
- "epoch": 0.33,
1194
- "learning_rate": 3.900000000000001e-06,
1195
- "loss": 4.6439,
1196
  "step": 198
1197
  },
1198
  {
1199
- "epoch": 0.33,
1200
- "learning_rate": 3.920000000000001e-06,
1201
- "loss": 4.5596,
1202
  "step": 199
1203
  },
1204
  {
1205
- "epoch": 0.34,
1206
- "learning_rate": 3.94e-06,
1207
- "loss": 4.7306,
1208
  "step": 200
1209
  },
1210
  {
1211
- "epoch": 0.34,
1212
- "learning_rate": 3.96e-06,
1213
- "loss": 4.1171,
1214
  "step": 201
1215
  },
1216
  {
1217
- "epoch": 0.34,
1218
- "learning_rate": 3.980000000000001e-06,
1219
- "loss": 4.2316,
1220
  "step": 202
1221
  },
1222
  {
1223
- "epoch": 0.34,
1224
- "learning_rate": 4.000000000000001e-06,
1225
- "loss": 4.2785,
1226
  "step": 203
1227
  },
1228
  {
1229
- "epoch": 0.34,
1230
- "learning_rate": 4.0200000000000005e-06,
1231
- "loss": 4.1903,
1232
  "step": 204
1233
  },
1234
  {
1235
- "epoch": 0.34,
1236
- "learning_rate": 4.04e-06,
1237
- "loss": 4.2281,
1238
  "step": 205
1239
  },
1240
  {
1241
- "epoch": 0.35,
1242
- "learning_rate": 4.060000000000001e-06,
1243
- "loss": 4.1607,
1244
  "step": 206
1245
  },
1246
  {
1247
- "epoch": 0.35,
1248
- "learning_rate": 4.08e-06,
1249
- "loss": 4.2343,
1250
  "step": 207
1251
  },
1252
  {
1253
- "epoch": 0.35,
1254
- "learning_rate": 4.1e-06,
1255
- "loss": 4.1985,
1256
  "step": 208
1257
  },
1258
  {
1259
- "epoch": 0.35,
1260
- "learning_rate": 4.12e-06,
1261
- "loss": 4.2208,
1262
  "step": 209
1263
  },
1264
  {
1265
- "epoch": 0.35,
1266
- "learning_rate": 4.14e-06,
1267
- "loss": 4.3262,
1268
  "step": 210
1269
  },
1270
  {
1271
- "epoch": 0.35,
1272
- "learning_rate": 4.16e-06,
1273
- "loss": 4.2993,
1274
  "step": 211
1275
  },
1276
  {
1277
- "epoch": 0.36,
1278
- "learning_rate": 4.18e-06,
1279
- "loss": 4.2977,
1280
  "step": 212
1281
  },
1282
  {
1283
- "epoch": 0.36,
1284
- "learning_rate": 4.2000000000000004e-06,
1285
- "loss": 4.2572,
1286
  "step": 213
1287
  },
1288
  {
1289
- "epoch": 0.36,
1290
- "learning_rate": 4.22e-06,
1291
- "loss": 4.3531,
1292
  "step": 214
1293
  },
1294
  {
1295
- "epoch": 0.36,
1296
- "learning_rate": 4.24e-06,
1297
- "loss": 4.2164,
1298
  "step": 215
1299
  },
1300
  {
1301
- "epoch": 0.36,
1302
- "learning_rate": 4.26e-06,
1303
- "loss": 4.2395,
1304
  "step": 216
1305
  },
1306
  {
1307
- "epoch": 0.36,
1308
- "learning_rate": 4.2800000000000005e-06,
1309
- "loss": 4.125,
1310
  "step": 217
1311
  },
1312
  {
1313
- "epoch": 0.37,
1314
- "learning_rate": 4.3e-06,
1315
- "loss": 4.2381,
1316
  "step": 218
1317
  },
1318
  {
1319
- "epoch": 0.37,
1320
- "learning_rate": 4.32e-06,
1321
- "loss": 4.2245,
1322
  "step": 219
1323
  },
1324
  {
1325
- "epoch": 0.37,
1326
- "learning_rate": 4.34e-06,
1327
- "loss": 4.3043,
1328
  "step": 220
1329
  },
1330
  {
1331
- "epoch": 0.37,
1332
- "learning_rate": 4.360000000000001e-06,
1333
- "loss": 4.3443,
1334
  "step": 221
1335
  },
1336
  {
1337
- "epoch": 0.37,
1338
- "learning_rate": 4.38e-06,
1339
- "loss": 4.3346,
1340
  "step": 222
1341
  },
1342
  {
1343
- "epoch": 0.37,
1344
- "learning_rate": 4.4e-06,
1345
- "loss": 4.1726,
1346
  "step": 223
1347
  },
1348
  {
1349
- "epoch": 0.38,
1350
- "learning_rate": 4.42e-06,
1351
- "loss": 4.2873,
1352
  "step": 224
1353
  },
1354
  {
1355
- "epoch": 0.38,
1356
- "learning_rate": 4.440000000000001e-06,
1357
- "loss": 4.4616,
1358
  "step": 225
1359
  },
1360
  {
1361
- "epoch": 0.38,
1362
- "learning_rate": 4.4600000000000005e-06,
1363
- "loss": 4.2406,
1364
  "step": 226
1365
  },
1366
  {
1367
- "epoch": 0.38,
1368
- "learning_rate": 4.48e-06,
1369
- "loss": 4.1887,
1370
  "step": 227
1371
  },
1372
  {
1373
- "epoch": 0.38,
1374
- "learning_rate": 4.5e-06,
1375
- "loss": 4.2441,
1376
  "step": 228
1377
  },
1378
  {
1379
- "epoch": 0.39,
1380
- "learning_rate": 4.520000000000001e-06,
1381
- "loss": 4.377,
1382
  "step": 229
1383
  },
1384
  {
1385
- "epoch": 0.39,
1386
- "learning_rate": 4.540000000000001e-06,
1387
- "loss": 4.3398,
1388
  "step": 230
1389
  },
1390
  {
1391
- "epoch": 0.39,
1392
- "learning_rate": 4.56e-06,
1393
- "loss": 4.2106,
1394
  "step": 231
1395
  },
1396
  {
1397
- "epoch": 0.39,
1398
- "learning_rate": 4.58e-06,
1399
- "loss": 4.2779,
1400
  "step": 232
1401
  },
1402
  {
1403
- "epoch": 0.39,
1404
- "learning_rate": 4.600000000000001e-06,
1405
- "loss": 4.3087,
1406
  "step": 233
1407
  },
1408
  {
1409
- "epoch": 0.39,
1410
- "learning_rate": 4.620000000000001e-06,
1411
- "loss": 4.2863,
1412
  "step": 234
1413
  },
1414
  {
1415
- "epoch": 0.4,
1416
- "learning_rate": 4.6400000000000005e-06,
1417
- "loss": 4.4418,
1418
  "step": 235
1419
  },
1420
  {
1421
- "epoch": 0.4,
1422
- "learning_rate": 4.66e-06,
1423
- "loss": 4.2034,
1424
  "step": 236
1425
  },
1426
  {
1427
- "epoch": 0.4,
1428
- "learning_rate": 4.680000000000001e-06,
1429
- "loss": 4.2032,
1430
  "step": 237
1431
  },
1432
  {
1433
- "epoch": 0.4,
1434
- "learning_rate": 4.7e-06,
1435
- "loss": 4.2322,
1436
  "step": 238
1437
  },
1438
  {
1439
- "epoch": 0.4,
1440
- "learning_rate": 4.7200000000000005e-06,
1441
- "loss": 4.1456,
1442
  "step": 239
1443
  },
1444
  {
1445
- "epoch": 0.4,
1446
- "learning_rate": 4.74e-06,
1447
- "loss": 4.3317,
1448
  "step": 240
1449
  },
1450
  {
1451
- "epoch": 0.41,
1452
- "learning_rate": 4.76e-06,
1453
- "loss": 4.2461,
1454
  "step": 241
1455
  },
1456
  {
1457
- "epoch": 0.41,
1458
- "learning_rate": 4.78e-06,
1459
- "loss": 4.2655,
1460
  "step": 242
1461
  },
1462
  {
1463
- "epoch": 0.41,
1464
- "learning_rate": 4.800000000000001e-06,
1465
- "loss": 4.2899,
1466
  "step": 243
1467
  },
1468
  {
1469
- "epoch": 0.41,
1470
- "learning_rate": 4.8200000000000004e-06,
1471
- "loss": 4.3705,
1472
  "step": 244
1473
  },
1474
  {
1475
- "epoch": 0.41,
1476
- "learning_rate": 4.84e-06,
1477
- "loss": 4.439,
1478
  "step": 245
1479
  },
1480
  {
1481
- "epoch": 0.41,
1482
- "learning_rate": 4.86e-06,
1483
- "loss": 4.4711,
1484
  "step": 246
1485
  },
1486
  {
1487
- "epoch": 0.42,
1488
- "learning_rate": 4.880000000000001e-06,
1489
- "loss": 4.4541,
1490
  "step": 247
1491
  },
1492
  {
1493
- "epoch": 0.42,
1494
- "learning_rate": 4.9000000000000005e-06,
1495
- "loss": 4.5902,
1496
  "step": 248
1497
  },
1498
  {
1499
- "epoch": 0.42,
1500
- "learning_rate": 4.92e-06,
1501
- "loss": 4.5564,
1502
  "step": 249
1503
  },
1504
  {
1505
- "epoch": 0.42,
1506
- "learning_rate": 4.94e-06,
1507
- "loss": 4.5735,
1508
  "step": 250
1509
  },
1510
  {
1511
- "epoch": 0.42,
1512
- "learning_rate": 4.960000000000001e-06,
1513
- "loss": 4.2696,
1514
  "step": 251
1515
  },
1516
  {
1517
- "epoch": 0.42,
1518
- "learning_rate": 4.980000000000001e-06,
1519
- "loss": 4.2687,
1520
  "step": 252
1521
  },
1522
  {
1523
- "epoch": 0.43,
1524
- "learning_rate": 5e-06,
1525
- "loss": 4.2286,
1526
  "step": 253
1527
  },
1528
  {
1529
- "epoch": 0.43,
1530
- "learning_rate": 5.02e-06,
1531
- "loss": 4.1151,
1532
  "step": 254
1533
  },
1534
  {
1535
- "epoch": 0.43,
1536
- "learning_rate": 5.04e-06,
1537
- "loss": 4.294,
1538
  "step": 255
1539
  },
1540
  {
1541
- "epoch": 0.43,
1542
- "learning_rate": 5.060000000000001e-06,
1543
- "loss": 4.244,
1544
  "step": 256
1545
  },
1546
  {
1547
- "epoch": 0.43,
1548
- "learning_rate": 5.0800000000000005e-06,
1549
- "loss": 4.1804,
1550
  "step": 257
1551
  },
1552
  {
1553
- "epoch": 0.43,
1554
- "learning_rate": 5.1e-06,
1555
- "loss": 4.2024,
1556
  "step": 258
1557
  },
1558
  {
1559
- "epoch": 0.44,
1560
- "learning_rate": 5.12e-06,
1561
- "loss": 4.2631,
1562
  "step": 259
1563
  },
1564
  {
1565
- "epoch": 0.44,
1566
- "learning_rate": 5.140000000000001e-06,
1567
- "loss": 4.269,
1568
  "step": 260
1569
  },
1570
  {
1571
- "epoch": 0.44,
1572
- "learning_rate": 5.1600000000000006e-06,
1573
- "loss": 4.2433,
1574
  "step": 261
1575
  },
1576
  {
1577
- "epoch": 0.44,
1578
- "learning_rate": 5.18e-06,
1579
- "loss": 4.1892,
1580
  "step": 262
1581
  },
1582
  {
1583
- "epoch": 0.44,
1584
- "learning_rate": 5.2e-06,
1585
- "loss": 4.2933,
1586
  "step": 263
1587
  },
1588
  {
1589
- "epoch": 0.44,
1590
- "learning_rate": 5.220000000000001e-06,
1591
- "loss": 4.0133,
1592
  "step": 264
1593
  },
1594
  {
1595
- "epoch": 0.45,
1596
- "learning_rate": 5.240000000000001e-06,
1597
- "loss": 4.1051,
1598
  "step": 265
1599
  },
1600
  {
1601
- "epoch": 0.45,
1602
- "learning_rate": 5.2600000000000005e-06,
1603
- "loss": 4.2157,
1604
  "step": 266
1605
  },
1606
  {
1607
- "epoch": 0.45,
1608
- "learning_rate": 5.28e-06,
1609
- "loss": 4.2397,
1610
  "step": 267
1611
  },
1612
- {
1613
- "epoch": 0.45,
1614
- "learning_rate": 5.300000000000001e-06,
1615
- "loss": 4.177,
1616
- "step": 268
1617
- },
1618
- {
1619
- "epoch": 0.45,
1620
- "learning_rate": 5.320000000000001e-06,
1621
- "loss": 4.292,
1622
- "step": 269
1623
- },
1624
- {
1625
- "epoch": 0.45,
1626
- "learning_rate": 5.3400000000000005e-06,
1627
- "loss": 4.2874,
1628
- "step": 270
1629
- },
1630
- {
1631
- "epoch": 0.46,
1632
- "learning_rate": 5.36e-06,
1633
- "loss": 4.2115,
1634
- "step": 271
1635
- },
1636
- {
1637
- "epoch": 0.46,
1638
- "learning_rate": 5.380000000000001e-06,
1639
- "loss": 4.3472,
1640
- "step": 272
1641
- },
1642
- {
1643
- "epoch": 0.46,
1644
- "learning_rate": 5.400000000000001e-06,
1645
- "loss": 4.2087,
1646
- "step": 273
1647
- },
1648
- {
1649
- "epoch": 0.46,
1650
- "learning_rate": 5.420000000000001e-06,
1651
- "loss": 4.2198,
1652
- "step": 274
1653
- },
1654
- {
1655
- "epoch": 0.46,
1656
- "learning_rate": 5.4400000000000004e-06,
1657
- "loss": 4.2855,
1658
- "step": 275
1659
- },
1660
- {
1661
- "epoch": 0.46,
1662
- "learning_rate": 5.460000000000001e-06,
1663
- "loss": 4.2137,
1664
- "step": 276
1665
- },
1666
- {
1667
- "epoch": 0.47,
1668
- "learning_rate": 5.480000000000001e-06,
1669
- "loss": 4.3496,
1670
- "step": 277
1671
- },
1672
- {
1673
- "epoch": 0.47,
1674
- "learning_rate": 5.500000000000001e-06,
1675
- "loss": 4.1356,
1676
- "step": 278
1677
- },
1678
- {
1679
- "epoch": 0.47,
1680
- "learning_rate": 5.5200000000000005e-06,
1681
- "loss": 4.317,
1682
- "step": 279
1683
- },
1684
- {
1685
- "epoch": 0.47,
1686
- "learning_rate": 5.540000000000001e-06,
1687
- "loss": 4.3791,
1688
- "step": 280
1689
- },
1690
- {
1691
- "epoch": 0.47,
1692
- "learning_rate": 5.560000000000001e-06,
1693
- "loss": 4.3842,
1694
- "step": 281
1695
- },
1696
- {
1697
- "epoch": 0.47,
1698
- "learning_rate": 5.580000000000001e-06,
1699
- "loss": 4.2575,
1700
- "step": 282
1701
- },
1702
- {
1703
- "epoch": 0.48,
1704
- "learning_rate": 5.600000000000001e-06,
1705
- "loss": 4.1577,
1706
- "step": 283
1707
- },
1708
- {
1709
- "epoch": 0.48,
1710
- "learning_rate": 5.620000000000001e-06,
1711
- "loss": 4.153,
1712
- "step": 284
1713
- },
1714
- {
1715
- "epoch": 0.48,
1716
- "learning_rate": 5.64e-06,
1717
- "loss": 4.4077,
1718
- "step": 285
1719
- },
1720
- {
1721
- "epoch": 0.48,
1722
- "learning_rate": 5.66e-06,
1723
- "loss": 4.1658,
1724
- "step": 286
1725
- },
1726
- {
1727
- "epoch": 0.48,
1728
- "learning_rate": 5.68e-06,
1729
- "loss": 4.5108,
1730
- "step": 287
1731
- },
1732
- {
1733
- "epoch": 0.48,
1734
- "learning_rate": 5.7e-06,
1735
- "loss": 4.0048,
1736
- "step": 288
1737
- },
1738
- {
1739
- "epoch": 0.49,
1740
- "learning_rate": 5.72e-06,
1741
- "loss": 4.2594,
1742
- "step": 289
1743
- },
1744
- {
1745
- "epoch": 0.49,
1746
- "learning_rate": 5.74e-06,
1747
- "loss": 4.2277,
1748
- "step": 290
1749
- },
1750
- {
1751
- "epoch": 0.49,
1752
- "learning_rate": 5.76e-06,
1753
- "loss": 4.4854,
1754
- "step": 291
1755
- },
1756
- {
1757
- "epoch": 0.49,
1758
- "learning_rate": 5.78e-06,
1759
- "loss": 4.1498,
1760
- "step": 292
1761
- },
1762
- {
1763
- "epoch": 0.49,
1764
- "learning_rate": 5.8e-06,
1765
- "loss": 4.3567,
1766
- "step": 293
1767
- },
1768
- {
1769
- "epoch": 0.49,
1770
- "learning_rate": 5.82e-06,
1771
- "loss": 4.3946,
1772
- "step": 294
1773
- },
1774
- {
1775
- "epoch": 0.5,
1776
- "learning_rate": 5.84e-06,
1777
- "loss": 4.3868,
1778
- "step": 295
1779
- },
1780
- {
1781
- "epoch": 0.5,
1782
- "learning_rate": 5.86e-06,
1783
- "loss": 4.2016,
1784
- "step": 296
1785
- },
1786
- {
1787
- "epoch": 0.5,
1788
- "learning_rate": 5.8800000000000005e-06,
1789
- "loss": 4.2791,
1790
- "step": 297
1791
- },
1792
- {
1793
- "epoch": 0.5,
1794
- "learning_rate": 5.9e-06,
1795
- "loss": 4.4595,
1796
- "step": 298
1797
- },
1798
- {
1799
- "epoch": 0.5,
1800
- "learning_rate": 5.92e-06,
1801
- "loss": 4.4781,
1802
- "step": 299
1803
- },
1804
- {
1805
- "epoch": 0.5,
1806
- "learning_rate": 5.94e-06,
1807
- "loss": 4.7221,
1808
- "step": 300
1809
- },
1810
- {
1811
- "epoch": 0.51,
1812
- "learning_rate": 5.9600000000000005e-06,
1813
- "loss": 4.3254,
1814
- "step": 301
1815
- },
1816
- {
1817
- "epoch": 0.51,
1818
- "learning_rate": 5.98e-06,
1819
- "loss": 4.266,
1820
- "step": 302
1821
- },
1822
- {
1823
- "epoch": 0.51,
1824
- "learning_rate": 6e-06,
1825
- "loss": 4.3527,
1826
- "step": 303
1827
- },
1828
- {
1829
- "epoch": 0.51,
1830
- "learning_rate": 6.02e-06,
1831
- "loss": 4.1289,
1832
- "step": 304
1833
- },
1834
- {
1835
- "epoch": 0.51,
1836
- "learning_rate": 6.040000000000001e-06,
1837
- "loss": 4.3227,
1838
- "step": 305
1839
- },
1840
- {
1841
- "epoch": 0.51,
1842
- "learning_rate": 6.0600000000000004e-06,
1843
- "loss": 4.3129,
1844
- "step": 306
1845
- },
1846
- {
1847
- "epoch": 0.52,
1848
- "learning_rate": 6.08e-06,
1849
- "loss": 4.1029,
1850
- "step": 307
1851
- },
1852
- {
1853
- "epoch": 0.52,
1854
- "learning_rate": 6.1e-06,
1855
- "loss": 4.3159,
1856
- "step": 308
1857
- },
1858
- {
1859
- "epoch": 0.52,
1860
- "learning_rate": 6.120000000000001e-06,
1861
- "loss": 4.2134,
1862
- "step": 309
1863
- },
1864
- {
1865
- "epoch": 0.52,
1866
- "learning_rate": 6.1400000000000005e-06,
1867
- "loss": 4.3056,
1868
- "step": 310
1869
- },
1870
- {
1871
- "epoch": 0.52,
1872
- "learning_rate": 6.16e-06,
1873
- "loss": 4.1592,
1874
- "step": 311
1875
- },
1876
- {
1877
- "epoch": 0.52,
1878
- "learning_rate": 6.18e-06,
1879
- "loss": 4.2038,
1880
- "step": 312
1881
- },
1882
- {
1883
- "epoch": 0.53,
1884
- "learning_rate": 6.200000000000001e-06,
1885
- "loss": 4.2221,
1886
- "step": 313
1887
- },
1888
- {
1889
- "epoch": 0.53,
1890
- "learning_rate": 6.220000000000001e-06,
1891
- "loss": 4.2282,
1892
- "step": 314
1893
- },
1894
- {
1895
- "epoch": 0.53,
1896
- "learning_rate": 6.24e-06,
1897
- "loss": 4.2493,
1898
- "step": 315
1899
- },
1900
- {
1901
- "epoch": 0.53,
1902
- "learning_rate": 6.26e-06,
1903
- "loss": 4.2448,
1904
- "step": 316
1905
- },
1906
- {
1907
- "epoch": 0.53,
1908
- "learning_rate": 6.280000000000001e-06,
1909
- "loss": 4.1976,
1910
- "step": 317
1911
- },
1912
- {
1913
- "epoch": 0.53,
1914
- "learning_rate": 6.300000000000001e-06,
1915
- "loss": 4.2818,
1916
- "step": 318
1917
- },
1918
- {
1919
- "epoch": 0.54,
1920
- "learning_rate": 6.3200000000000005e-06,
1921
- "loss": 4.1041,
1922
- "step": 319
1923
- },
1924
- {
1925
- "epoch": 0.54,
1926
- "learning_rate": 6.34e-06,
1927
- "loss": 4.2671,
1928
- "step": 320
1929
- },
1930
- {
1931
- "epoch": 0.54,
1932
- "learning_rate": 6.360000000000001e-06,
1933
- "loss": 4.1219,
1934
- "step": 321
1935
- },
1936
- {
1937
- "epoch": 0.54,
1938
- "learning_rate": 6.380000000000001e-06,
1939
- "loss": 4.1702,
1940
- "step": 322
1941
- },
1942
- {
1943
- "epoch": 0.54,
1944
- "learning_rate": 6.4000000000000006e-06,
1945
- "loss": 4.187,
1946
- "step": 323
1947
- },
1948
- {
1949
- "epoch": 0.54,
1950
- "learning_rate": 6.42e-06,
1951
- "loss": 4.2524,
1952
- "step": 324
1953
- },
1954
- {
1955
- "epoch": 0.55,
1956
- "learning_rate": 6.440000000000001e-06,
1957
- "loss": 4.2123,
1958
- "step": 325
1959
- },
1960
- {
1961
- "epoch": 0.55,
1962
- "learning_rate": 6.460000000000001e-06,
1963
- "loss": 4.1287,
1964
- "step": 326
1965
- },
1966
- {
1967
- "epoch": 0.55,
1968
- "learning_rate": 6.480000000000001e-06,
1969
- "loss": 4.4177,
1970
- "step": 327
1971
- },
1972
- {
1973
- "epoch": 0.55,
1974
- "learning_rate": 6.5000000000000004e-06,
1975
- "loss": 4.1621,
1976
- "step": 328
1977
- },
1978
- {
1979
- "epoch": 0.55,
1980
- "learning_rate": 6.520000000000001e-06,
1981
- "loss": 4.0732,
1982
- "step": 329
1983
- },
1984
- {
1985
- "epoch": 0.55,
1986
- "learning_rate": 6.540000000000001e-06,
1987
- "loss": 4.3004,
1988
- "step": 330
1989
- },
1990
- {
1991
- "epoch": 0.56,
1992
- "learning_rate": 6.560000000000001e-06,
1993
- "loss": 4.1879,
1994
- "step": 331
1995
- },
1996
- {
1997
- "epoch": 0.56,
1998
- "learning_rate": 6.5800000000000005e-06,
1999
- "loss": 4.1324,
2000
- "step": 332
2001
- },
2002
- {
2003
- "epoch": 0.56,
2004
- "learning_rate": 6.600000000000001e-06,
2005
- "loss": 4.2003,
2006
- "step": 333
2007
- },
2008
- {
2009
- "epoch": 0.56,
2010
- "learning_rate": 6.620000000000001e-06,
2011
- "loss": 4.2383,
2012
- "step": 334
2013
- },
2014
- {
2015
- "epoch": 0.56,
2016
- "learning_rate": 6.640000000000001e-06,
2017
- "loss": 4.2517,
2018
- "step": 335
2019
- },
2020
- {
2021
- "epoch": 0.56,
2022
- "learning_rate": 6.660000000000001e-06,
2023
- "loss": 4.1818,
2024
- "step": 336
2025
- },
2026
- {
2027
- "epoch": 0.57,
2028
- "learning_rate": 6.680000000000001e-06,
2029
- "loss": 4.148,
2030
- "step": 337
2031
- },
2032
- {
2033
- "epoch": 0.57,
2034
- "learning_rate": 6.700000000000001e-06,
2035
- "loss": 4.2455,
2036
- "step": 338
2037
- },
2038
- {
2039
- "epoch": 0.57,
2040
- "learning_rate": 6.720000000000001e-06,
2041
- "loss": 4.2638,
2042
- "step": 339
2043
- },
2044
- {
2045
- "epoch": 0.57,
2046
- "learning_rate": 6.740000000000001e-06,
2047
- "loss": 4.2869,
2048
- "step": 340
2049
- },
2050
- {
2051
- "epoch": 0.57,
2052
- "learning_rate": 6.760000000000001e-06,
2053
- "loss": 4.2918,
2054
- "step": 341
2055
- },
2056
- {
2057
- "epoch": 0.58,
2058
- "learning_rate": 6.780000000000001e-06,
2059
- "loss": 4.3944,
2060
- "step": 342
2061
- },
2062
- {
2063
- "epoch": 0.58,
2064
- "learning_rate": 6.800000000000001e-06,
2065
- "loss": 4.4212,
2066
- "step": 343
2067
- },
2068
- {
2069
- "epoch": 0.58,
2070
- "learning_rate": 6.820000000000001e-06,
2071
- "loss": 4.2959,
2072
- "step": 344
2073
- },
2074
- {
2075
- "epoch": 0.58,
2076
- "learning_rate": 6.8400000000000014e-06,
2077
- "loss": 4.4497,
2078
- "step": 345
2079
- },
2080
- {
2081
- "epoch": 0.58,
2082
- "learning_rate": 6.860000000000001e-06,
2083
- "loss": 4.1427,
2084
- "step": 346
2085
- },
2086
- {
2087
- "epoch": 0.58,
2088
- "learning_rate": 6.88e-06,
2089
- "loss": 4.2791,
2090
- "step": 347
2091
- },
2092
- {
2093
- "epoch": 0.59,
2094
- "learning_rate": 6.9e-06,
2095
- "loss": 4.6095,
2096
- "step": 348
2097
- },
2098
- {
2099
- "epoch": 0.59,
2100
- "learning_rate": 6.92e-06,
2101
- "loss": 4.4704,
2102
- "step": 349
2103
- },
2104
- {
2105
- "epoch": 0.59,
2106
- "learning_rate": 6.9400000000000005e-06,
2107
- "loss": 4.7726,
2108
- "step": 350
2109
- },
2110
- {
2111
- "epoch": 0.59,
2112
- "learning_rate": 6.96e-06,
2113
- "loss": 4.1317,
2114
- "step": 351
2115
- },
2116
- {
2117
- "epoch": 0.59,
2118
- "learning_rate": 6.98e-06,
2119
- "loss": 4.2371,
2120
- "step": 352
2121
- },
2122
- {
2123
- "epoch": 0.59,
2124
- "learning_rate": 7e-06,
2125
- "loss": 4.2641,
2126
- "step": 353
2127
- },
2128
- {
2129
- "epoch": 0.6,
2130
- "learning_rate": 7.0200000000000006e-06,
2131
- "loss": 4.1023,
2132
- "step": 354
2133
- },
2134
- {
2135
- "epoch": 0.6,
2136
- "learning_rate": 7.04e-06,
2137
- "loss": 4.1948,
2138
- "step": 355
2139
- },
2140
- {
2141
- "epoch": 0.6,
2142
- "learning_rate": 7.06e-06,
2143
- "loss": 4.1926,
2144
- "step": 356
2145
- },
2146
- {
2147
- "epoch": 0.6,
2148
- "learning_rate": 7.08e-06,
2149
- "loss": 4.215,
2150
- "step": 357
2151
- },
2152
- {
2153
- "epoch": 0.6,
2154
- "learning_rate": 7.100000000000001e-06,
2155
- "loss": 4.1484,
2156
- "step": 358
2157
- },
2158
- {
2159
- "epoch": 0.6,
2160
- "learning_rate": 7.1200000000000004e-06,
2161
- "loss": 4.1754,
2162
- "step": 359
2163
- },
2164
- {
2165
- "epoch": 0.61,
2166
- "learning_rate": 7.14e-06,
2167
- "loss": 4.1181,
2168
- "step": 360
2169
- },
2170
- {
2171
- "epoch": 0.61,
2172
- "learning_rate": 7.16e-06,
2173
- "loss": 4.2056,
2174
- "step": 361
2175
- },
2176
- {
2177
- "epoch": 0.61,
2178
- "learning_rate": 7.180000000000001e-06,
2179
- "loss": 4.2102,
2180
- "step": 362
2181
- },
2182
- {
2183
- "epoch": 0.61,
2184
- "learning_rate": 7.2000000000000005e-06,
2185
- "loss": 4.2621,
2186
- "step": 363
2187
- },
2188
- {
2189
- "epoch": 0.61,
2190
- "learning_rate": 7.22e-06,
2191
- "loss": 4.1432,
2192
- "step": 364
2193
- },
2194
- {
2195
- "epoch": 0.61,
2196
- "learning_rate": 7.24e-06,
2197
- "loss": 4.2162,
2198
- "step": 365
2199
- },
2200
- {
2201
- "epoch": 0.62,
2202
- "learning_rate": 7.260000000000001e-06,
2203
- "loss": 4.1775,
2204
- "step": 366
2205
- },
2206
- {
2207
- "epoch": 0.62,
2208
- "learning_rate": 7.280000000000001e-06,
2209
- "loss": 4.1788,
2210
- "step": 367
2211
- },
2212
- {
2213
- "epoch": 0.62,
2214
- "learning_rate": 7.3e-06,
2215
- "loss": 4.1636,
2216
- "step": 368
2217
- },
2218
- {
2219
- "epoch": 0.62,
2220
- "learning_rate": 7.32e-06,
2221
- "loss": 4.2602,
2222
- "step": 369
2223
- },
2224
- {
2225
- "epoch": 0.62,
2226
- "learning_rate": 7.340000000000001e-06,
2227
- "loss": 4.1437,
2228
- "step": 370
2229
- },
2230
- {
2231
- "epoch": 0.62,
2232
- "learning_rate": 7.360000000000001e-06,
2233
- "loss": 4.0553,
2234
- "step": 371
2235
- },
2236
- {
2237
- "epoch": 0.63,
2238
- "learning_rate": 7.3800000000000005e-06,
2239
- "loss": 4.219,
2240
- "step": 372
2241
- },
2242
- {
2243
- "epoch": 0.63,
2244
- "learning_rate": 7.4e-06,
2245
- "loss": 4.0653,
2246
- "step": 373
2247
- },
2248
- {
2249
- "epoch": 0.63,
2250
- "learning_rate": 7.420000000000001e-06,
2251
- "loss": 4.0878,
2252
- "step": 374
2253
- },
2254
- {
2255
- "epoch": 0.63,
2256
- "learning_rate": 7.440000000000001e-06,
2257
- "loss": 4.2545,
2258
- "step": 375
2259
- },
2260
- {
2261
- "epoch": 0.63,
2262
- "learning_rate": 7.4600000000000006e-06,
2263
- "loss": 4.1381,
2264
- "step": 376
2265
- },
2266
- {
2267
- "epoch": 0.63,
2268
- "learning_rate": 7.48e-06,
2269
- "loss": 4.2293,
2270
- "step": 377
2271
- },
2272
- {
2273
- "epoch": 0.64,
2274
- "learning_rate": 7.500000000000001e-06,
2275
- "loss": 4.2858,
2276
- "step": 378
2277
- },
2278
- {
2279
- "epoch": 0.64,
2280
- "learning_rate": 7.520000000000001e-06,
2281
- "loss": 4.1157,
2282
- "step": 379
2283
- },
2284
- {
2285
- "epoch": 0.64,
2286
- "learning_rate": 7.540000000000001e-06,
2287
- "loss": 4.3363,
2288
- "step": 380
2289
- },
2290
- {
2291
- "epoch": 0.64,
2292
- "learning_rate": 7.5600000000000005e-06,
2293
- "loss": 4.2621,
2294
- "step": 381
2295
- },
2296
- {
2297
- "epoch": 0.64,
2298
- "learning_rate": 7.58e-06,
2299
- "loss": 4.1278,
2300
- "step": 382
2301
- },
2302
- {
2303
- "epoch": 0.64,
2304
- "learning_rate": 7.600000000000001e-06,
2305
- "loss": 4.0316,
2306
- "step": 383
2307
- },
2308
- {
2309
- "epoch": 0.65,
2310
- "learning_rate": 7.620000000000001e-06,
2311
- "loss": 4.2188,
2312
- "step": 384
2313
- },
2314
- {
2315
- "epoch": 0.65,
2316
- "learning_rate": 7.640000000000001e-06,
2317
- "loss": 4.4046,
2318
- "step": 385
2319
- },
2320
- {
2321
- "epoch": 0.65,
2322
- "learning_rate": 7.660000000000001e-06,
2323
- "loss": 4.1662,
2324
- "step": 386
2325
- },
2326
- {
2327
- "epoch": 0.65,
2328
- "learning_rate": 7.680000000000001e-06,
2329
- "loss": 4.3235,
2330
- "step": 387
2331
- },
2332
- {
2333
- "epoch": 0.65,
2334
- "learning_rate": 7.7e-06,
2335
- "loss": 4.2542,
2336
- "step": 388
2337
- },
2338
- {
2339
- "epoch": 0.65,
2340
- "learning_rate": 7.72e-06,
2341
- "loss": 4.142,
2342
- "step": 389
2343
- },
2344
- {
2345
- "epoch": 0.66,
2346
- "learning_rate": 7.74e-06,
2347
- "loss": 4.1514,
2348
- "step": 390
2349
- },
2350
- {
2351
- "epoch": 0.66,
2352
- "learning_rate": 7.76e-06,
2353
- "loss": 4.366,
2354
- "step": 391
2355
- },
2356
- {
2357
- "epoch": 0.66,
2358
- "learning_rate": 7.78e-06,
2359
- "loss": 4.2973,
2360
- "step": 392
2361
- },
2362
- {
2363
- "epoch": 0.66,
2364
- "learning_rate": 7.800000000000002e-06,
2365
- "loss": 4.5381,
2366
- "step": 393
2367
- },
2368
- {
2369
- "epoch": 0.66,
2370
- "learning_rate": 7.820000000000001e-06,
2371
- "loss": 4.3709,
2372
- "step": 394
2373
- },
2374
- {
2375
- "epoch": 0.66,
2376
- "learning_rate": 7.840000000000001e-06,
2377
- "loss": 4.2935,
2378
- "step": 395
2379
- },
2380
- {
2381
- "epoch": 0.67,
2382
- "learning_rate": 7.860000000000001e-06,
2383
- "loss": 4.2228,
2384
- "step": 396
2385
- },
2386
- {
2387
- "epoch": 0.67,
2388
- "learning_rate": 7.88e-06,
2389
- "loss": 4.4299,
2390
- "step": 397
2391
- },
2392
- {
2393
- "epoch": 0.67,
2394
- "learning_rate": 7.9e-06,
2395
- "loss": 4.6182,
2396
- "step": 398
2397
- },
2398
- {
2399
- "epoch": 0.67,
2400
- "learning_rate": 7.92e-06,
2401
- "loss": 4.43,
2402
- "step": 399
2403
- },
2404
- {
2405
- "epoch": 0.67,
2406
- "learning_rate": 7.94e-06,
2407
- "loss": 4.5297,
2408
- "step": 400
2409
- },
2410
- {
2411
- "epoch": 0.67,
2412
- "learning_rate": 7.960000000000002e-06,
2413
- "loss": 4.2478,
2414
- "step": 401
2415
- },
2416
- {
2417
- "epoch": 0.68,
2418
- "learning_rate": 7.980000000000002e-06,
2419
- "loss": 4.15,
2420
- "step": 402
2421
- },
2422
- {
2423
- "epoch": 0.68,
2424
- "learning_rate": 8.000000000000001e-06,
2425
- "loss": 4.2147,
2426
- "step": 403
2427
- },
2428
- {
2429
- "epoch": 0.68,
2430
- "learning_rate": 8.020000000000001e-06,
2431
- "loss": 4.1922,
2432
- "step": 404
2433
- },
2434
- {
2435
- "epoch": 0.68,
2436
- "learning_rate": 8.040000000000001e-06,
2437
- "loss": 4.1304,
2438
- "step": 405
2439
- },
2440
- {
2441
- "epoch": 0.68,
2442
- "learning_rate": 8.06e-06,
2443
- "loss": 4.1817,
2444
- "step": 406
2445
- },
2446
- {
2447
- "epoch": 0.68,
2448
- "learning_rate": 8.08e-06,
2449
- "loss": 4.1808,
2450
- "step": 407
2451
- },
2452
- {
2453
- "epoch": 0.69,
2454
- "learning_rate": 8.1e-06,
2455
- "loss": 4.0998,
2456
- "step": 408
2457
- },
2458
- {
2459
- "epoch": 0.69,
2460
- "learning_rate": 8.120000000000002e-06,
2461
- "loss": 4.1643,
2462
- "step": 409
2463
- },
2464
- {
2465
- "epoch": 0.69,
2466
- "learning_rate": 8.14e-06,
2467
- "loss": 4.1098,
2468
- "step": 410
2469
- },
2470
- {
2471
- "epoch": 0.69,
2472
- "learning_rate": 8.16e-06,
2473
- "loss": 4.1225,
2474
- "step": 411
2475
- },
2476
- {
2477
- "epoch": 0.69,
2478
- "learning_rate": 8.18e-06,
2479
- "loss": 4.2878,
2480
- "step": 412
2481
- },
2482
- {
2483
- "epoch": 0.69,
2484
- "learning_rate": 8.2e-06,
2485
- "loss": 4.1765,
2486
- "step": 413
2487
- },
2488
- {
2489
- "epoch": 0.7,
2490
- "learning_rate": 8.220000000000001e-06,
2491
- "loss": 4.2272,
2492
- "step": 414
2493
- },
2494
- {
2495
- "epoch": 0.7,
2496
- "learning_rate": 8.24e-06,
2497
- "loss": 4.2027,
2498
- "step": 415
2499
- },
2500
- {
2501
- "epoch": 0.7,
2502
- "learning_rate": 8.26e-06,
2503
- "loss": 4.1183,
2504
- "step": 416
2505
- },
2506
- {
2507
- "epoch": 0.7,
2508
- "learning_rate": 8.28e-06,
2509
- "loss": 4.1938,
2510
- "step": 417
2511
- },
2512
- {
2513
- "epoch": 0.7,
2514
- "learning_rate": 8.3e-06,
2515
- "loss": 4.0316,
2516
- "step": 418
2517
- },
2518
- {
2519
- "epoch": 0.7,
2520
- "learning_rate": 8.32e-06,
2521
- "loss": 4.2936,
2522
- "step": 419
2523
- },
2524
- {
2525
- "epoch": 0.71,
2526
- "learning_rate": 8.34e-06,
2527
- "loss": 4.1993,
2528
- "step": 420
2529
- },
2530
- {
2531
- "epoch": 0.71,
2532
- "learning_rate": 8.36e-06,
2533
- "loss": 3.973,
2534
- "step": 421
2535
- },
2536
- {
2537
- "epoch": 0.71,
2538
- "learning_rate": 8.380000000000001e-06,
2539
- "loss": 4.1127,
2540
- "step": 422
2541
- },
2542
- {
2543
- "epoch": 0.71,
2544
- "learning_rate": 8.400000000000001e-06,
2545
- "loss": 4.2403,
2546
- "step": 423
2547
- },
2548
- {
2549
- "epoch": 0.71,
2550
- "learning_rate": 8.42e-06,
2551
- "loss": 4.1748,
2552
- "step": 424
2553
- },
2554
- {
2555
- "epoch": 0.71,
2556
- "learning_rate": 8.44e-06,
2557
- "loss": 4.1619,
2558
- "step": 425
2559
- },
2560
- {
2561
- "epoch": 0.72,
2562
- "learning_rate": 8.46e-06,
2563
- "loss": 4.2359,
2564
- "step": 426
2565
- },
2566
- {
2567
- "epoch": 0.72,
2568
- "learning_rate": 8.48e-06,
2569
- "loss": 4.1836,
2570
- "step": 427
2571
- },
2572
- {
2573
- "epoch": 0.72,
2574
- "learning_rate": 8.5e-06,
2575
- "loss": 4.1234,
2576
- "step": 428
2577
- },
2578
- {
2579
- "epoch": 0.72,
2580
- "learning_rate": 8.52e-06,
2581
- "loss": 4.2324,
2582
- "step": 429
2583
- },
2584
- {
2585
- "epoch": 0.72,
2586
- "learning_rate": 8.540000000000001e-06,
2587
- "loss": 4.1858,
2588
- "step": 430
2589
- },
2590
- {
2591
- "epoch": 0.72,
2592
- "learning_rate": 8.560000000000001e-06,
2593
- "loss": 4.2511,
2594
- "step": 431
2595
- },
2596
- {
2597
- "epoch": 0.73,
2598
- "learning_rate": 8.580000000000001e-06,
2599
- "loss": 4.0518,
2600
- "step": 432
2601
- },
2602
- {
2603
- "epoch": 0.73,
2604
- "learning_rate": 8.6e-06,
2605
- "loss": 4.2866,
2606
- "step": 433
2607
- },
2608
- {
2609
- "epoch": 0.73,
2610
- "learning_rate": 8.62e-06,
2611
- "loss": 4.2093,
2612
- "step": 434
2613
- },
2614
- {
2615
- "epoch": 0.73,
2616
- "learning_rate": 8.64e-06,
2617
- "loss": 4.2793,
2618
- "step": 435
2619
- },
2620
- {
2621
- "epoch": 0.73,
2622
- "learning_rate": 8.66e-06,
2623
- "loss": 4.1262,
2624
- "step": 436
2625
- },
2626
- {
2627
- "epoch": 0.73,
2628
- "learning_rate": 8.68e-06,
2629
- "loss": 4.0688,
2630
- "step": 437
2631
- },
2632
- {
2633
- "epoch": 0.74,
2634
- "learning_rate": 8.700000000000001e-06,
2635
- "loss": 4.2754,
2636
- "step": 438
2637
- },
2638
- {
2639
- "epoch": 0.74,
2640
- "learning_rate": 8.720000000000001e-06,
2641
- "loss": 4.1836,
2642
- "step": 439
2643
- },
2644
- {
2645
- "epoch": 0.74,
2646
- "learning_rate": 8.740000000000001e-06,
2647
- "loss": 4.3704,
2648
- "step": 440
2649
- },
2650
- {
2651
- "epoch": 0.74,
2652
- "learning_rate": 8.76e-06,
2653
- "loss": 4.2427,
2654
- "step": 441
2655
- },
2656
- {
2657
- "epoch": 0.74,
2658
- "learning_rate": 8.78e-06,
2659
- "loss": 4.422,
2660
- "step": 442
2661
- },
2662
- {
2663
- "epoch": 0.74,
2664
- "learning_rate": 8.8e-06,
2665
- "loss": 4.3321,
2666
- "step": 443
2667
- },
2668
- {
2669
- "epoch": 0.75,
2670
- "learning_rate": 8.82e-06,
2671
- "loss": 4.2648,
2672
- "step": 444
2673
- },
2674
- {
2675
- "epoch": 0.75,
2676
- "learning_rate": 8.84e-06,
2677
- "loss": 4.089,
2678
- "step": 445
2679
- },
2680
- {
2681
- "epoch": 0.75,
2682
- "learning_rate": 8.860000000000002e-06,
2683
- "loss": 4.3901,
2684
- "step": 446
2685
- },
2686
- {
2687
- "epoch": 0.75,
2688
- "learning_rate": 8.880000000000001e-06,
2689
- "loss": 4.3074,
2690
- "step": 447
2691
- },
2692
- {
2693
- "epoch": 0.75,
2694
- "learning_rate": 8.900000000000001e-06,
2695
- "loss": 4.2438,
2696
- "step": 448
2697
- },
2698
- {
2699
- "epoch": 0.75,
2700
- "learning_rate": 8.920000000000001e-06,
2701
- "loss": 4.4442,
2702
- "step": 449
2703
- },
2704
- {
2705
- "epoch": 0.76,
2706
- "learning_rate": 8.94e-06,
2707
- "loss": 4.5537,
2708
- "step": 450
2709
- },
2710
- {
2711
- "epoch": 0.76,
2712
- "learning_rate": 8.96e-06,
2713
- "loss": 4.2158,
2714
- "step": 451
2715
- },
2716
- {
2717
- "epoch": 0.76,
2718
- "learning_rate": 8.98e-06,
2719
- "loss": 4.2222,
2720
- "step": 452
2721
- },
2722
- {
2723
- "epoch": 0.76,
2724
- "learning_rate": 9e-06,
2725
- "loss": 4.1596,
2726
- "step": 453
2727
- },
2728
- {
2729
- "epoch": 0.76,
2730
- "learning_rate": 9.020000000000002e-06,
2731
- "loss": 4.0322,
2732
- "step": 454
2733
- },
2734
- {
2735
- "epoch": 0.77,
2736
- "learning_rate": 9.040000000000002e-06,
2737
- "loss": 4.2099,
2738
- "step": 455
2739
- },
2740
- {
2741
- "epoch": 0.77,
2742
- "learning_rate": 9.060000000000001e-06,
2743
- "loss": 4.1521,
2744
- "step": 456
2745
- },
2746
- {
2747
- "epoch": 0.77,
2748
- "learning_rate": 9.080000000000001e-06,
2749
- "loss": 4.2381,
2750
- "step": 457
2751
- },
2752
- {
2753
- "epoch": 0.77,
2754
- "learning_rate": 9.100000000000001e-06,
2755
- "loss": 4.2634,
2756
- "step": 458
2757
- },
2758
- {
2759
- "epoch": 0.77,
2760
- "learning_rate": 9.12e-06,
2761
- "loss": 4.1887,
2762
- "step": 459
2763
- },
2764
- {
2765
- "epoch": 0.77,
2766
- "learning_rate": 9.14e-06,
2767
- "loss": 4.1982,
2768
- "step": 460
2769
- },
2770
- {
2771
- "epoch": 0.78,
2772
- "learning_rate": 9.16e-06,
2773
- "loss": 4.1105,
2774
- "step": 461
2775
- },
2776
- {
2777
- "epoch": 0.78,
2778
- "learning_rate": 9.180000000000002e-06,
2779
- "loss": 4.1386,
2780
- "step": 462
2781
- },
2782
- {
2783
- "epoch": 0.78,
2784
- "learning_rate": 9.200000000000002e-06,
2785
- "loss": 4.1434,
2786
- "step": 463
2787
- },
2788
- {
2789
- "epoch": 0.78,
2790
- "learning_rate": 9.220000000000002e-06,
2791
- "loss": 4.2041,
2792
- "step": 464
2793
- },
2794
- {
2795
- "epoch": 0.78,
2796
- "learning_rate": 9.240000000000001e-06,
2797
- "loss": 4.2266,
2798
- "step": 465
2799
- },
2800
- {
2801
- "epoch": 0.78,
2802
- "learning_rate": 9.260000000000001e-06,
2803
- "loss": 4.1666,
2804
- "step": 466
2805
- },
2806
- {
2807
- "epoch": 0.79,
2808
- "learning_rate": 9.280000000000001e-06,
2809
- "loss": 4.1755,
2810
- "step": 467
2811
- },
2812
- {
2813
- "epoch": 0.79,
2814
- "learning_rate": 9.3e-06,
2815
- "loss": 4.1936,
2816
- "step": 468
2817
- },
2818
- {
2819
- "epoch": 0.79,
2820
- "learning_rate": 9.32e-06,
2821
- "loss": 4.1813,
2822
- "step": 469
2823
- },
2824
- {
2825
- "epoch": 0.79,
2826
- "learning_rate": 9.340000000000002e-06,
2827
- "loss": 4.1513,
2828
- "step": 470
2829
- },
2830
- {
2831
- "epoch": 0.79,
2832
- "learning_rate": 9.360000000000002e-06,
2833
- "loss": 4.0757,
2834
- "step": 471
2835
- },
2836
- {
2837
- "epoch": 0.79,
2838
- "learning_rate": 9.38e-06,
2839
- "loss": 4.1831,
2840
- "step": 472
2841
- },
2842
- {
2843
- "epoch": 0.8,
2844
- "learning_rate": 9.4e-06,
2845
- "loss": 4.1809,
2846
- "step": 473
2847
- },
2848
- {
2849
- "epoch": 0.8,
2850
- "learning_rate": 9.42e-06,
2851
- "loss": 4.0661,
2852
- "step": 474
2853
- },
2854
- {
2855
- "epoch": 0.8,
2856
- "learning_rate": 9.440000000000001e-06,
2857
- "loss": 4.1123,
2858
- "step": 475
2859
- },
2860
- {
2861
- "epoch": 0.8,
2862
- "learning_rate": 9.460000000000001e-06,
2863
- "loss": 4.1363,
2864
- "step": 476
2865
- },
2866
- {
2867
- "epoch": 0.8,
2868
- "learning_rate": 9.48e-06,
2869
- "loss": 4.1625,
2870
- "step": 477
2871
- },
2872
- {
2873
- "epoch": 0.8,
2874
- "learning_rate": 9.5e-06,
2875
- "loss": 4.1857,
2876
- "step": 478
2877
- },
2878
- {
2879
- "epoch": 0.81,
2880
- "learning_rate": 9.52e-06,
2881
- "loss": 4.0617,
2882
- "step": 479
2883
- },
2884
- {
2885
- "epoch": 0.81,
2886
- "learning_rate": 9.54e-06,
2887
- "loss": 4.0652,
2888
- "step": 480
2889
- },
2890
- {
2891
- "epoch": 0.81,
2892
- "learning_rate": 9.56e-06,
2893
- "loss": 4.3014,
2894
- "step": 481
2895
- },
2896
- {
2897
- "epoch": 0.81,
2898
- "learning_rate": 9.58e-06,
2899
- "loss": 4.1331,
2900
- "step": 482
2901
- },
2902
- {
2903
- "epoch": 0.81,
2904
- "learning_rate": 9.600000000000001e-06,
2905
- "loss": 4.2987,
2906
- "step": 483
2907
- },
2908
- {
2909
- "epoch": 0.81,
2910
- "learning_rate": 9.620000000000001e-06,
2911
- "loss": 4.2466,
2912
- "step": 484
2913
- },
2914
- {
2915
- "epoch": 0.82,
2916
- "learning_rate": 9.640000000000001e-06,
2917
- "loss": 4.2273,
2918
- "step": 485
2919
- },
2920
- {
2921
- "epoch": 0.82,
2922
- "learning_rate": 9.66e-06,
2923
- "loss": 4.1424,
2924
- "step": 486
2925
- },
2926
- {
2927
- "epoch": 0.82,
2928
- "learning_rate": 9.68e-06,
2929
- "loss": 4.2659,
2930
- "step": 487
2931
- },
2932
- {
2933
- "epoch": 0.82,
2934
- "learning_rate": 9.7e-06,
2935
- "loss": 4.2862,
2936
- "step": 488
2937
- },
2938
- {
2939
- "epoch": 0.82,
2940
- "learning_rate": 9.72e-06,
2941
- "loss": 4.2092,
2942
- "step": 489
2943
- },
2944
- {
2945
- "epoch": 0.82,
2946
- "learning_rate": 9.74e-06,
2947
- "loss": 4.1234,
2948
- "step": 490
2949
- },
2950
- {
2951
- "epoch": 0.83,
2952
- "learning_rate": 9.760000000000001e-06,
2953
- "loss": 4.3266,
2954
- "step": 491
2955
- },
2956
- {
2957
- "epoch": 0.83,
2958
- "learning_rate": 9.780000000000001e-06,
2959
- "loss": 4.1458,
2960
- "step": 492
2961
- },
2962
- {
2963
- "epoch": 0.83,
2964
- "learning_rate": 9.800000000000001e-06,
2965
- "loss": 4.027,
2966
- "step": 493
2967
- },
2968
- {
2969
- "epoch": 0.83,
2970
- "learning_rate": 9.820000000000001e-06,
2971
- "loss": 4.2423,
2972
- "step": 494
2973
- },
2974
- {
2975
- "epoch": 0.83,
2976
- "learning_rate": 9.84e-06,
2977
- "loss": 4.3469,
2978
- "step": 495
2979
- },
2980
- {
2981
- "epoch": 0.83,
2982
- "learning_rate": 9.86e-06,
2983
- "loss": 4.391,
2984
- "step": 496
2985
- },
2986
- {
2987
- "epoch": 0.84,
2988
- "learning_rate": 9.88e-06,
2989
- "loss": 4.267,
2990
- "step": 497
2991
- },
2992
- {
2993
- "epoch": 0.84,
2994
- "learning_rate": 9.9e-06,
2995
- "loss": 4.4238,
2996
- "step": 498
2997
- },
2998
- {
2999
- "epoch": 0.84,
3000
- "learning_rate": 9.920000000000002e-06,
3001
- "loss": 4.4835,
3002
- "step": 499
3003
- },
3004
- {
3005
- "epoch": 0.84,
3006
- "learning_rate": 9.940000000000001e-06,
3007
- "loss": 4.4132,
3008
- "step": 500
3009
- },
3010
- {
3011
- "epoch": 0.84,
3012
- "eval_loss": 4.189728260040283,
3013
- "eval_runtime": 775.4835,
3014
- "eval_samples_per_second": 3.407,
3015
- "eval_steps_per_second": 0.285,
3016
- "eval_wer": 1.9666600555335185,
3017
- "step": 500
3018
- },
3019
- {
3020
- "epoch": 0.84,
3021
- "learning_rate": 9.960000000000001e-06,
3022
- "loss": 4.2325,
3023
- "step": 501
3024
- },
3025
- {
3026
- "epoch": 0.84,
3027
- "learning_rate": 9.980000000000001e-06,
3028
- "loss": 4.1381,
3029
- "step": 502
3030
- },
3031
- {
3032
- "epoch": 0.85,
3033
- "learning_rate": 1e-05,
3034
- "loss": 4.1025,
3035
- "step": 503
3036
- },
3037
- {
3038
- "epoch": 0.85,
3039
- "learning_rate": 9.893617021276596e-06,
3040
- "loss": 4.1918,
3041
- "step": 504
3042
- },
3043
- {
3044
- "epoch": 0.85,
3045
- "learning_rate": 9.787234042553192e-06,
3046
- "loss": 4.1578,
3047
- "step": 505
3048
- },
3049
- {
3050
- "epoch": 0.85,
3051
- "learning_rate": 9.680851063829787e-06,
3052
- "loss": 4.2885,
3053
- "step": 506
3054
- },
3055
- {
3056
- "epoch": 0.85,
3057
- "learning_rate": 9.574468085106385e-06,
3058
- "loss": 4.1569,
3059
- "step": 507
3060
- },
3061
- {
3062
- "epoch": 0.85,
3063
- "learning_rate": 9.46808510638298e-06,
3064
- "loss": 4.1487,
3065
- "step": 508
3066
- },
3067
- {
3068
- "epoch": 0.86,
3069
- "learning_rate": 9.361702127659576e-06,
3070
- "loss": 4.0881,
3071
- "step": 509
3072
- },
3073
- {
3074
- "epoch": 0.86,
3075
- "learning_rate": 9.255319148936171e-06,
3076
- "loss": 4.1581,
3077
- "step": 510
3078
- },
3079
- {
3080
- "epoch": 0.86,
3081
- "learning_rate": 9.148936170212767e-06,
3082
- "loss": 4.2344,
3083
- "step": 511
3084
- },
3085
- {
3086
- "epoch": 0.86,
3087
- "learning_rate": 9.042553191489362e-06,
3088
- "loss": 4.1285,
3089
- "step": 512
3090
- },
3091
- {
3092
- "epoch": 0.86,
3093
- "learning_rate": 8.936170212765958e-06,
3094
- "loss": 4.1244,
3095
- "step": 513
3096
- },
3097
- {
3098
- "epoch": 0.86,
3099
- "learning_rate": 8.829787234042555e-06,
3100
- "loss": 4.0797,
3101
- "step": 514
3102
- },
3103
- {
3104
- "epoch": 0.87,
3105
- "learning_rate": 8.72340425531915e-06,
3106
- "loss": 4.2064,
3107
- "step": 515
3108
- },
3109
- {
3110
- "epoch": 0.87,
3111
- "learning_rate": 8.617021276595746e-06,
3112
- "loss": 4.1293,
3113
- "step": 516
3114
- },
3115
- {
3116
- "epoch": 0.87,
3117
- "learning_rate": 8.510638297872341e-06,
3118
- "loss": 4.1691,
3119
- "step": 517
3120
- },
3121
- {
3122
- "epoch": 0.87,
3123
- "learning_rate": 8.404255319148937e-06,
3124
- "loss": 4.23,
3125
- "step": 518
3126
- },
3127
- {
3128
- "epoch": 0.87,
3129
- "learning_rate": 8.297872340425532e-06,
3130
- "loss": 4.045,
3131
- "step": 519
3132
- },
3133
- {
3134
- "epoch": 0.87,
3135
- "learning_rate": 8.191489361702128e-06,
3136
- "loss": 4.1642,
3137
- "step": 520
3138
- },
3139
- {
3140
- "epoch": 0.88,
3141
- "learning_rate": 8.085106382978723e-06,
3142
- "loss": 4.1907,
3143
- "step": 521
3144
- },
3145
- {
3146
- "epoch": 0.88,
3147
- "learning_rate": 7.97872340425532e-06,
3148
- "loss": 4.2534,
3149
- "step": 522
3150
- },
3151
- {
3152
- "epoch": 0.88,
3153
- "learning_rate": 7.872340425531916e-06,
3154
- "loss": 4.0643,
3155
- "step": 523
3156
- },
3157
- {
3158
- "epoch": 0.88,
3159
- "learning_rate": 7.765957446808511e-06,
3160
- "loss": 3.9933,
3161
- "step": 524
3162
- },
3163
- {
3164
- "epoch": 0.88,
3165
- "learning_rate": 7.659574468085107e-06,
3166
- "loss": 4.0862,
3167
- "step": 525
3168
- },
3169
- {
3170
- "epoch": 0.88,
3171
- "learning_rate": 7.553191489361703e-06,
3172
- "loss": 4.0739,
3173
- "step": 526
3174
- },
3175
- {
3176
- "epoch": 0.89,
3177
- "learning_rate": 7.446808510638298e-06,
3178
- "loss": 4.2288,
3179
- "step": 527
3180
- },
3181
- {
3182
- "epoch": 0.89,
3183
- "learning_rate": 7.340425531914894e-06,
3184
- "loss": 4.1022,
3185
- "step": 528
3186
- },
3187
- {
3188
- "epoch": 0.89,
3189
- "learning_rate": 7.234042553191491e-06,
3190
- "loss": 4.0489,
3191
- "step": 529
3192
- },
3193
- {
3194
- "epoch": 0.89,
3195
- "learning_rate": 7.127659574468085e-06,
3196
- "loss": 4.0736,
3197
- "step": 530
3198
- },
3199
- {
3200
- "epoch": 0.89,
3201
- "learning_rate": 7.021276595744682e-06,
3202
- "loss": 4.0327,
3203
- "step": 531
3204
- },
3205
- {
3206
- "epoch": 0.89,
3207
- "learning_rate": 6.914893617021278e-06,
3208
- "loss": 4.2801,
3209
- "step": 532
3210
- },
3211
  {
3212
  "epoch": 0.9,
3213
- "learning_rate": 6.808510638297873e-06,
3214
- "loss": 4.0947,
3215
- "step": 533
3216
- },
3217
- {
3218
- "epoch": 0.9,
3219
- "learning_rate": 6.702127659574469e-06,
3220
- "loss": 4.1191,
3221
- "step": 534
3222
- },
3223
- {
3224
- "epoch": 0.9,
3225
- "learning_rate": 6.595744680851064e-06,
3226
- "loss": 4.1445,
3227
- "step": 535
3228
- },
3229
- {
3230
- "epoch": 0.9,
3231
- "learning_rate": 6.48936170212766e-06,
3232
- "loss": 3.9945,
3233
- "step": 536
3234
- },
3235
- {
3236
- "epoch": 0.9,
3237
- "learning_rate": 6.382978723404256e-06,
3238
- "loss": 4.2223,
3239
- "step": 537
3240
  },
3241
  {
3242
  "epoch": 0.9,
3243
- "learning_rate": 6.276595744680851e-06,
3244
- "loss": 4.2336,
3245
- "step": 538
3246
- },
3247
- {
3248
- "epoch": 0.91,
3249
- "learning_rate": 6.170212765957447e-06,
3250
- "loss": 4.0644,
3251
- "step": 539
3252
- },
3253
- {
3254
- "epoch": 0.91,
3255
- "learning_rate": 6.063829787234044e-06,
3256
- "loss": 4.1494,
3257
- "step": 540
3258
- },
3259
- {
3260
- "epoch": 0.91,
3261
- "learning_rate": 5.957446808510638e-06,
3262
  "loss": 4.1364,
3263
- "step": 541
3264
  },
3265
  {
3266
  "epoch": 0.91,
3267
- "learning_rate": 5.851063829787235e-06,
3268
- "loss": 4.1308,
3269
- "step": 542
3270
  },
3271
  {
3272
  "epoch": 0.91,
3273
- "learning_rate": 5.744680851063831e-06,
3274
- "loss": 4.3163,
3275
- "step": 543
3276
  },
3277
  {
3278
  "epoch": 0.91,
3279
- "learning_rate": 5.638297872340426e-06,
3280
- "loss": 4.156,
3281
- "step": 544
3282
- },
3283
- {
3284
- "epoch": 0.92,
3285
- "learning_rate": 5.531914893617022e-06,
3286
- "loss": 4.2632,
3287
- "step": 545
3288
- },
3289
- {
3290
- "epoch": 0.92,
3291
- "learning_rate": 5.425531914893617e-06,
3292
- "loss": 4.4659,
3293
- "step": 546
3294
- },
3295
- {
3296
- "epoch": 0.92,
3297
- "learning_rate": 5.319148936170213e-06,
3298
- "loss": 4.2235,
3299
- "step": 547
3300
  },
3301
  {
3302
  "epoch": 0.92,
3303
- "learning_rate": 5.212765957446809e-06,
3304
- "loss": 4.4768,
3305
- "step": 548
3306
  },
3307
  {
3308
  "epoch": 0.92,
3309
- "learning_rate": 5.106382978723404e-06,
3310
- "loss": 4.6745,
3311
- "step": 549
3312
  },
3313
  {
3314
  "epoch": 0.92,
3315
- "learning_rate": 5e-06,
3316
- "loss": 4.3527,
3317
- "step": 550
3318
- },
3319
- {
3320
- "epoch": 0.93,
3321
- "learning_rate": 4.893617021276596e-06,
3322
- "loss": 4.1747,
3323
- "step": 551
3324
- },
3325
- {
3326
- "epoch": 0.93,
3327
- "learning_rate": 4.787234042553192e-06,
3328
- "loss": 4.0934,
3329
- "step": 552
3330
- },
3331
- {
3332
- "epoch": 0.93,
3333
- "learning_rate": 4.680851063829788e-06,
3334
- "loss": 4.082,
3335
- "step": 553
3336
  },
3337
  {
3338
  "epoch": 0.93,
3339
- "learning_rate": 4.574468085106383e-06,
3340
- "loss": 4.0284,
3341
- "step": 554
3342
  },
3343
  {
3344
  "epoch": 0.93,
3345
- "learning_rate": 4.468085106382979e-06,
3346
- "loss": 4.0693,
3347
- "step": 555
3348
  },
3349
  {
3350
  "epoch": 0.93,
3351
- "learning_rate": 4.361702127659575e-06,
3352
- "loss": 4.1033,
3353
- "step": 556
3354
- },
3355
- {
3356
- "epoch": 0.94,
3357
- "learning_rate": 4.255319148936171e-06,
3358
- "loss": 4.1894,
3359
- "step": 557
3360
- },
3361
- {
3362
- "epoch": 0.94,
3363
- "learning_rate": 4.148936170212766e-06,
3364
- "loss": 4.2196,
3365
- "step": 558
3366
- },
3367
- {
3368
- "epoch": 0.94,
3369
- "learning_rate": 4.042553191489362e-06,
3370
- "loss": 4.0573,
3371
- "step": 559
3372
  },
3373
  {
3374
  "epoch": 0.94,
3375
- "learning_rate": 3.936170212765958e-06,
3376
- "loss": 4.098,
3377
- "step": 560
3378
  },
3379
  {
3380
  "epoch": 0.94,
3381
- "learning_rate": 3.8297872340425535e-06,
3382
- "loss": 4.4263,
3383
- "step": 561
3384
  },
3385
  {
3386
  "epoch": 0.94,
3387
- "learning_rate": 3.723404255319149e-06,
3388
- "loss": 4.0279,
3389
- "step": 562
3390
- },
3391
- {
3392
- "epoch": 0.95,
3393
- "learning_rate": 3.6170212765957453e-06,
3394
- "loss": 4.0855,
3395
- "step": 563
3396
- },
3397
- {
3398
- "epoch": 0.95,
3399
- "learning_rate": 3.510638297872341e-06,
3400
- "loss": 4.1792,
3401
- "step": 564
3402
- },
3403
- {
3404
- "epoch": 0.95,
3405
- "learning_rate": 3.4042553191489363e-06,
3406
- "loss": 4.1505,
3407
- "step": 565
3408
  },
3409
  {
3410
  "epoch": 0.95,
3411
- "learning_rate": 3.297872340425532e-06,
3412
- "loss": 4.0332,
3413
- "step": 566
3414
  },
3415
  {
3416
  "epoch": 0.95,
3417
- "learning_rate": 3.191489361702128e-06,
3418
- "loss": 4.1336,
3419
- "step": 567
3420
- },
3421
- {
3422
- "epoch": 0.96,
3423
- "learning_rate": 3.0851063829787237e-06,
3424
- "loss": 4.0822,
3425
- "step": 568
3426
- },
3427
- {
3428
- "epoch": 0.96,
3429
- "learning_rate": 2.978723404255319e-06,
3430
- "loss": 3.9525,
3431
- "step": 569
3432
- },
3433
- {
3434
- "epoch": 0.96,
3435
- "learning_rate": 2.8723404255319155e-06,
3436
- "loss": 4.1345,
3437
- "step": 570
3438
  },
3439
  {
3440
  "epoch": 0.96,
3441
- "learning_rate": 2.765957446808511e-06,
3442
- "loss": 4.1153,
3443
- "step": 571
3444
  },
3445
  {
3446
  "epoch": 0.96,
3447
- "learning_rate": 2.6595744680851065e-06,
3448
- "loss": 4.1605,
3449
- "step": 572
3450
  },
3451
  {
3452
  "epoch": 0.96,
3453
- "learning_rate": 2.553191489361702e-06,
3454
- "loss": 4.2209,
3455
- "step": 573
3456
- },
3457
- {
3458
- "epoch": 0.97,
3459
- "learning_rate": 2.446808510638298e-06,
3460
- "loss": 4.1386,
3461
- "step": 574
3462
- },
3463
- {
3464
- "epoch": 0.97,
3465
- "learning_rate": 2.340425531914894e-06,
3466
- "loss": 4.2482,
3467
- "step": 575
3468
- },
3469
- {
3470
- "epoch": 0.97,
3471
- "learning_rate": 2.2340425531914894e-06,
3472
- "loss": 4.1276,
3473
- "step": 576
3474
  },
3475
  {
3476
  "epoch": 0.97,
3477
- "learning_rate": 2.1276595744680853e-06,
3478
- "loss": 4.1114,
3479
- "step": 577
3480
  },
3481
  {
3482
  "epoch": 0.97,
3483
- "learning_rate": 2.021276595744681e-06,
3484
- "loss": 4.0383,
3485
- "step": 578
3486
  },
3487
  {
3488
  "epoch": 0.97,
3489
- "learning_rate": 1.9148936170212767e-06,
3490
- "loss": 3.9893,
3491
- "step": 579
3492
- },
3493
- {
3494
- "epoch": 0.98,
3495
- "learning_rate": 1.8085106382978727e-06,
3496
- "loss": 4.2624,
3497
- "step": 580
3498
- },
3499
- {
3500
- "epoch": 0.98,
3501
- "learning_rate": 1.7021276595744682e-06,
3502
- "loss": 4.3654,
3503
- "step": 581
3504
- },
3505
- {
3506
- "epoch": 0.98,
3507
- "learning_rate": 1.595744680851064e-06,
3508
- "loss": 4.2641,
3509
- "step": 582
3510
  },
3511
  {
3512
  "epoch": 0.98,
3513
- "learning_rate": 1.4893617021276596e-06,
3514
- "loss": 4.0602,
3515
- "step": 583
3516
  },
3517
  {
3518
  "epoch": 0.98,
3519
- "learning_rate": 1.3829787234042555e-06,
3520
- "loss": 4.1982,
3521
- "step": 584
3522
  },
3523
  {
3524
  "epoch": 0.98,
3525
- "learning_rate": 1.276595744680851e-06,
3526
- "loss": 4.1675,
3527
- "step": 585
3528
- },
3529
- {
3530
- "epoch": 0.99,
3531
- "learning_rate": 1.170212765957447e-06,
3532
- "loss": 4.063,
3533
- "step": 586
3534
- },
3535
- {
3536
- "epoch": 0.99,
3537
- "learning_rate": 1.0638297872340427e-06,
3538
- "loss": 4.2433,
3539
- "step": 587
3540
- },
3541
- {
3542
- "epoch": 0.99,
3543
- "learning_rate": 9.574468085106384e-07,
3544
- "loss": 4.2181,
3545
- "step": 588
3546
  },
3547
  {
3548
  "epoch": 0.99,
3549
- "learning_rate": 8.510638297872341e-07,
3550
- "loss": 4.3395,
3551
- "step": 589
3552
  },
3553
  {
3554
  "epoch": 0.99,
3555
- "learning_rate": 7.446808510638298e-07,
3556
- "loss": 4.1872,
3557
- "step": 590
3558
  },
3559
  {
3560
  "epoch": 0.99,
3561
- "learning_rate": 6.382978723404255e-07,
3562
- "loss": 4.399,
3563
- "step": 591
3564
- },
3565
- {
3566
- "epoch": 1.0,
3567
- "learning_rate": 5.319148936170213e-07,
3568
- "loss": 4.2918,
3569
- "step": 592
3570
  },
3571
  {
3572
  "epoch": 1.0,
3573
- "learning_rate": 4.2553191489361704e-07,
3574
- "loss": 4.4872,
3575
- "step": 593
3576
  },
3577
  {
3578
  "epoch": 1.0,
3579
- "learning_rate": 3.1914893617021275e-07,
3580
- "loss": 4.4735,
3581
- "step": 594
3582
  },
3583
  {
3584
  "epoch": 1.0,
3585
- "step": 594,
3586
  "total_flos": 0.0,
3587
- "train_loss": 4.333276915630508,
3588
- "train_runtime": 5595.6802,
3589
- "train_samples_per_second": 5.1,
3590
- "train_steps_per_second": 0.106
3591
  }
3592
  ],
3593
- "max_steps": 594,
3594
  "num_train_epochs": 1,
3595
  "total_flos": 0.0,
3596
  "trial_name": null,
 
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
  "epoch": 0.9987389659520807,
5
+ "global_step": 297,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
  "epoch": 0.0,
12
+ "learning_rate": 2e-08,
13
+ "loss": 4.7862,
14
  "step": 1
15
  },
16
  {
17
+ "epoch": 0.01,
18
  "learning_rate": 2e-08,
19
+ "loss": 4.8712,
20
  "step": 2
21
  },
22
  {
23
  "epoch": 0.01,
24
  "learning_rate": 4e-08,
25
+ "loss": 4.9393,
26
  "step": 3
27
  },
28
  {
29
  "epoch": 0.01,
30
  "learning_rate": 6.000000000000001e-08,
31
+ "loss": 4.8916,
32
  "step": 4
33
  },
34
  {
35
+ "epoch": 0.02,
36
  "learning_rate": 8e-08,
37
+ "loss": 4.7806,
38
  "step": 5
39
  },
40
  {
41
+ "epoch": 0.02,
42
  "learning_rate": 1.0000000000000001e-07,
43
+ "loss": 4.9123,
44
  "step": 6
45
  },
46
  {
47
+ "epoch": 0.02,
48
  "learning_rate": 1.2000000000000002e-07,
49
+ "loss": 4.8899,
50
  "step": 7
51
  },
52
  {
53
+ "epoch": 0.03,
54
  "learning_rate": 1.4e-07,
55
+ "loss": 4.9651,
56
  "step": 8
57
  },
58
  {
59
+ "epoch": 0.03,
60
  "learning_rate": 1.6e-07,
61
+ "loss": 4.8742,
62
  "step": 9
63
  },
64
  {
65
+ "epoch": 0.03,
66
  "learning_rate": 1.8e-07,
67
+ "loss": 4.749,
68
  "step": 10
69
  },
70
  {
71
+ "epoch": 0.04,
72
  "learning_rate": 2.0000000000000002e-07,
73
+ "loss": 4.8902,
74
  "step": 11
75
  },
76
  {
77
+ "epoch": 0.04,
78
  "learning_rate": 2.2e-07,
79
+ "loss": 4.9384,
80
  "step": 12
81
  },
82
  {
83
+ "epoch": 0.04,
84
  "learning_rate": 2.4000000000000003e-07,
85
+ "loss": 4.8481,
86
  "step": 13
87
  },
88
  {
89
+ "epoch": 0.05,
90
  "learning_rate": 2.6e-07,
91
+ "loss": 4.8025,
92
  "step": 14
93
  },
94
  {
95
+ "epoch": 0.05,
96
  "learning_rate": 2.8e-07,
97
+ "loss": 4.8327,
98
  "step": 15
99
  },
100
  {
101
+ "epoch": 0.05,
102
  "learning_rate": 3.0000000000000004e-07,
103
+ "loss": 4.7626,
104
  "step": 16
105
  },
106
  {
107
+ "epoch": 0.06,
108
  "learning_rate": 3.2e-07,
109
+ "loss": 4.8658,
110
  "step": 17
111
  },
112
  {
113
+ "epoch": 0.06,
114
  "learning_rate": 3.4000000000000003e-07,
115
+ "loss": 4.8883,
116
  "step": 18
117
  },
118
  {
119
+ "epoch": 0.06,
120
  "learning_rate": 3.6e-07,
121
+ "loss": 4.8363,
122
  "step": 19
123
  },
124
  {
125
+ "epoch": 0.07,
126
  "learning_rate": 3.8e-07,
127
+ "loss": 4.9294,
128
  "step": 20
129
  },
130
  {
131
+ "epoch": 0.07,
132
  "learning_rate": 4.0000000000000003e-07,
133
+ "loss": 4.9403,
134
  "step": 21
135
  },
136
  {
137
+ "epoch": 0.07,
138
  "learning_rate": 4.2000000000000006e-07,
139
+ "loss": 4.8762,
140
  "step": 22
141
  },
142
  {
143
+ "epoch": 0.08,
144
  "learning_rate": 4.4e-07,
145
+ "loss": 4.8012,
146
  "step": 23
147
  },
148
  {
149
+ "epoch": 0.08,
150
  "learning_rate": 4.6000000000000004e-07,
151
+ "loss": 4.9476,
152
  "step": 24
153
  },
154
  {
155
+ "epoch": 0.08,
156
  "learning_rate": 4.800000000000001e-07,
157
+ "loss": 4.7731,
158
  "step": 25
159
  },
160
  {
161
+ "epoch": 0.09,
162
  "learning_rate": 5.000000000000001e-07,
163
+ "loss": 5.086,
164
  "step": 26
165
  },
166
  {
167
+ "epoch": 0.09,
168
  "learning_rate": 5.2e-07,
169
+ "loss": 4.8115,
170
  "step": 27
171
  },
172
  {
173
+ "epoch": 0.09,
174
  "learning_rate": 5.4e-07,
175
+ "loss": 4.8335,
176
  "step": 28
177
  },
178
  {
179
+ "epoch": 0.1,
180
  "learning_rate": 5.6e-07,
181
+ "loss": 4.7309,
182
  "step": 29
183
  },
184
  {
185
+ "epoch": 0.1,
186
  "learning_rate": 5.800000000000001e-07,
187
+ "loss": 4.8824,
188
  "step": 30
189
  },
190
  {
191
+ "epoch": 0.1,
192
  "learning_rate": 6.000000000000001e-07,
193
+ "loss": 4.875,
194
  "step": 31
195
  },
196
  {
197
+ "epoch": 0.11,
198
  "learning_rate": 6.200000000000001e-07,
199
+ "loss": 4.8,
200
  "step": 32
201
  },
202
  {
203
+ "epoch": 0.11,
204
+ "learning_rate": 6.4e-07,
205
+ "loss": 4.8361,
206
  "step": 33
207
  },
208
  {
209
+ "epoch": 0.11,
210
+ "learning_rate": 6.6e-07,
211
+ "loss": 4.7767,
212
  "step": 34
213
  },
214
  {
215
+ "epoch": 0.12,
216
+ "learning_rate": 6.800000000000001e-07,
217
+ "loss": 4.8105,
218
  "step": 35
219
  },
220
  {
221
+ "epoch": 0.12,
222
+ "learning_rate": 7.000000000000001e-07,
223
+ "loss": 4.7742,
224
  "step": 36
225
  },
226
  {
227
+ "epoch": 0.12,
228
+ "learning_rate": 7.2e-07,
229
+ "loss": 4.9031,
230
  "step": 37
231
  },
232
  {
233
+ "epoch": 0.13,
234
+ "learning_rate": 7.4e-07,
235
+ "loss": 4.8678,
236
  "step": 38
237
  },
238
  {
239
+ "epoch": 0.13,
240
+ "learning_rate": 7.6e-07,
241
+ "loss": 4.9272,
242
  "step": 39
243
  },
244
  {
245
+ "epoch": 0.13,
246
+ "learning_rate": 7.8e-07,
247
+ "loss": 4.764,
248
  "step": 40
249
  },
250
  {
251
+ "epoch": 0.14,
252
+ "learning_rate": 8.000000000000001e-07,
253
+ "loss": 4.8599,
254
  "step": 41
255
  },
256
  {
257
+ "epoch": 0.14,
258
+ "learning_rate": 8.200000000000001e-07,
259
+ "loss": 4.9507,
260
  "step": 42
261
  },
262
  {
263
+ "epoch": 0.14,
264
+ "learning_rate": 8.400000000000001e-07,
265
+ "loss": 4.7209,
266
  "step": 43
267
  },
268
  {
269
+ "epoch": 0.15,
270
+ "learning_rate": 8.6e-07,
271
+ "loss": 5.0393,
272
  "step": 44
273
  },
274
  {
275
+ "epoch": 0.15,
276
+ "learning_rate": 8.8e-07,
277
+ "loss": 5.0139,
278
  "step": 45
279
  },
280
  {
281
+ "epoch": 0.15,
282
+ "learning_rate": 9.000000000000001e-07,
283
+ "loss": 5.0851,
284
  "step": 46
285
  },
286
  {
287
+ "epoch": 0.16,
288
+ "learning_rate": 9.200000000000001e-07,
289
+ "loss": 5.1161,
290
  "step": 47
291
  },
292
  {
293
+ "epoch": 0.16,
294
+ "learning_rate": 9.400000000000001e-07,
295
+ "loss": 5.2155,
296
  "step": 48
297
  },
298
  {
299
+ "epoch": 0.16,
300
+ "learning_rate": 9.600000000000001e-07,
301
+ "loss": 5.2669,
302
  "step": 49
303
  },
304
  {
305
+ "epoch": 0.17,
306
+ "learning_rate": 9.800000000000001e-07,
307
+ "loss": 5.1875,
308
  "step": 50
309
  },
310
  {
311
+ "epoch": 0.17,
312
+ "learning_rate": 1.0000000000000002e-06,
313
+ "loss": 4.673,
314
  "step": 51
315
  },
316
  {
317
+ "epoch": 0.17,
318
+ "learning_rate": 1.02e-06,
319
+ "loss": 4.5594,
320
  "step": 52
321
  },
322
  {
323
+ "epoch": 0.18,
324
+ "learning_rate": 1.04e-06,
325
+ "loss": 4.5074,
326
  "step": 53
327
  },
328
  {
329
+ "epoch": 0.18,
330
+ "learning_rate": 1.06e-06,
331
+ "loss": 4.5297,
332
  "step": 54
333
  },
334
  {
335
+ "epoch": 0.18,
336
+ "learning_rate": 1.08e-06,
337
+ "loss": 4.4645,
338
  "step": 55
339
  },
340
  {
341
+ "epoch": 0.19,
342
+ "learning_rate": 1.1e-06,
343
+ "loss": 4.4743,
344
  "step": 56
345
  },
346
  {
347
+ "epoch": 0.19,
348
+ "learning_rate": 1.12e-06,
349
+ "loss": 4.6112,
350
  "step": 57
351
  },
352
  {
353
+ "epoch": 0.2,
354
+ "learning_rate": 1.14e-06,
355
+ "loss": 4.6401,
356
  "step": 58
357
  },
358
  {
359
+ "epoch": 0.2,
360
+ "learning_rate": 1.1600000000000001e-06,
361
+ "loss": 4.6055,
362
  "step": 59
363
  },
364
  {
365
+ "epoch": 0.2,
366
+ "learning_rate": 1.1800000000000001e-06,
367
+ "loss": 4.5181,
368
  "step": 60
369
  },
370
  {
371
+ "epoch": 0.21,
372
+ "learning_rate": 1.2000000000000002e-06,
373
+ "loss": 4.5917,
374
  "step": 61
375
  },
376
  {
377
+ "epoch": 0.21,
378
+ "learning_rate": 1.2200000000000002e-06,
379
+ "loss": 4.4533,
380
  "step": 62
381
  },
382
  {
383
+ "epoch": 0.21,
384
+ "learning_rate": 1.2400000000000002e-06,
385
+ "loss": 4.5889,
386
  "step": 63
387
  },
388
  {
389
+ "epoch": 0.22,
390
+ "learning_rate": 1.26e-06,
391
+ "loss": 4.5351,
392
  "step": 64
393
  },
394
  {
395
+ "epoch": 0.22,
396
+ "learning_rate": 1.28e-06,
397
+ "loss": 4.5995,
398
  "step": 65
399
  },
400
  {
401
+ "epoch": 0.22,
402
+ "learning_rate": 1.3e-06,
403
+ "loss": 4.5842,
404
  "step": 66
405
  },
406
  {
407
+ "epoch": 0.23,
408
+ "learning_rate": 1.32e-06,
409
+ "loss": 4.5496,
410
  "step": 67
411
  },
412
  {
413
+ "epoch": 0.23,
414
+ "learning_rate": 1.34e-06,
415
+ "loss": 4.5796,
416
  "step": 68
417
  },
418
  {
419
+ "epoch": 0.23,
420
+ "learning_rate": 1.3600000000000001e-06,
421
+ "loss": 4.5204,
422
  "step": 69
423
  },
424
  {
425
+ "epoch": 0.24,
426
+ "learning_rate": 1.3800000000000001e-06,
427
+ "loss": 4.4595,
428
  "step": 70
429
  },
430
  {
431
+ "epoch": 0.24,
432
+ "learning_rate": 1.4000000000000001e-06,
433
+ "loss": 4.4492,
434
  "step": 71
435
  },
436
  {
437
+ "epoch": 0.24,
438
+ "learning_rate": 1.42e-06,
439
+ "loss": 4.5821,
440
  "step": 72
441
  },
442
  {
443
+ "epoch": 0.25,
444
+ "learning_rate": 1.44e-06,
445
+ "loss": 4.4998,
446
  "step": 73
447
  },
448
  {
449
+ "epoch": 0.25,
450
+ "learning_rate": 1.46e-06,
451
+ "loss": 4.4812,
452
  "step": 74
453
  },
454
  {
455
+ "epoch": 0.25,
456
+ "learning_rate": 1.48e-06,
457
+ "loss": 4.5778,
458
  "step": 75
459
  },
460
  {
461
+ "epoch": 0.26,
462
+ "learning_rate": 1.5e-06,
463
+ "loss": 4.4696,
464
  "step": 76
465
  },
466
  {
467
+ "epoch": 0.26,
468
+ "learning_rate": 1.52e-06,
469
+ "loss": 4.4308,
470
  "step": 77
471
  },
472
  {
473
+ "epoch": 0.26,
474
+ "learning_rate": 1.54e-06,
475
+ "loss": 4.4694,
476
  "step": 78
477
  },
478
  {
479
+ "epoch": 0.27,
480
+ "learning_rate": 1.56e-06,
481
+ "loss": 4.3769,
482
  "step": 79
483
  },
484
  {
485
+ "epoch": 0.27,
486
+ "learning_rate": 1.5800000000000001e-06,
487
+ "loss": 4.5486,
488
  "step": 80
489
  },
490
  {
491
+ "epoch": 0.27,
492
+ "learning_rate": 1.6000000000000001e-06,
493
+ "loss": 4.4755,
494
  "step": 81
495
  },
496
  {
497
+ "epoch": 0.28,
498
+ "learning_rate": 1.6200000000000002e-06,
499
+ "loss": 4.4662,
500
  "step": 82
501
  },
502
  {
503
+ "epoch": 0.28,
504
+ "learning_rate": 1.6400000000000002e-06,
505
+ "loss": 4.4412,
506
  "step": 83
507
  },
508
  {
509
+ "epoch": 0.28,
510
+ "learning_rate": 1.6600000000000002e-06,
511
+ "loss": 4.4749,
512
  "step": 84
513
  },
514
  {
515
+ "epoch": 0.29,
516
+ "learning_rate": 1.6800000000000002e-06,
517
+ "loss": 4.5525,
518
  "step": 85
519
  },
520
  {
521
+ "epoch": 0.29,
522
+ "learning_rate": 1.7000000000000002e-06,
523
+ "loss": 4.4525,
524
  "step": 86
525
  },
526
  {
527
+ "epoch": 0.29,
528
+ "learning_rate": 1.72e-06,
529
+ "loss": 4.5006,
530
  "step": 87
531
  },
532
  {
533
+ "epoch": 0.3,
534
+ "learning_rate": 1.74e-06,
535
+ "loss": 4.5137,
536
  "step": 88
537
  },
538
  {
539
+ "epoch": 0.3,
540
+ "learning_rate": 1.76e-06,
541
+ "loss": 4.549,
542
  "step": 89
543
  },
544
  {
545
+ "epoch": 0.3,
546
+ "learning_rate": 1.7800000000000001e-06,
547
+ "loss": 4.5675,
548
  "step": 90
549
  },
550
  {
551
+ "epoch": 0.31,
552
+ "learning_rate": 1.8000000000000001e-06,
553
+ "loss": 4.595,
554
  "step": 91
555
  },
556
  {
557
+ "epoch": 0.31,
558
+ "learning_rate": 1.8200000000000002e-06,
559
+ "loss": 4.6441,
560
  "step": 92
561
  },
562
  {
563
+ "epoch": 0.31,
564
+ "learning_rate": 1.8400000000000002e-06,
565
+ "loss": 4.5773,
566
  "step": 93
567
  },
568
  {
569
+ "epoch": 0.32,
570
+ "learning_rate": 1.8600000000000002e-06,
571
+ "loss": 4.522,
572
  "step": 94
573
  },
574
  {
575
+ "epoch": 0.32,
576
+ "learning_rate": 1.8800000000000002e-06,
577
+ "loss": 4.4908,
578
  "step": 95
579
  },
580
  {
581
+ "epoch": 0.32,
582
+ "learning_rate": 1.9000000000000002e-06,
583
+ "loss": 4.6946,
584
  "step": 96
585
  },
586
  {
587
+ "epoch": 0.33,
588
+ "learning_rate": 1.9200000000000003e-06,
589
+ "loss": 4.8747,
590
  "step": 97
591
  },
592
  {
593
+ "epoch": 0.33,
594
+ "learning_rate": 1.94e-06,
595
+ "loss": 4.8247,
596
  "step": 98
597
  },
598
  {
599
+ "epoch": 0.33,
600
+ "learning_rate": 1.9600000000000003e-06,
601
+ "loss": 4.786,
602
  "step": 99
603
  },
604
  {
605
+ "epoch": 0.34,
606
+ "learning_rate": 1.98e-06,
607
+ "loss": 4.8753,
608
  "step": 100
609
  },
610
  {
611
+ "epoch": 0.34,
612
+ "learning_rate": 2.0000000000000003e-06,
613
+ "loss": 4.332,
614
  "step": 101
615
  },
616
  {
617
+ "epoch": 0.34,
618
+ "learning_rate": 2.02e-06,
619
+ "loss": 4.3795,
620
  "step": 102
621
  },
622
  {
623
+ "epoch": 0.35,
624
+ "learning_rate": 2.04e-06,
625
+ "loss": 4.3807,
626
  "step": 103
627
  },
628
  {
629
+ "epoch": 0.35,
630
+ "learning_rate": 2.06e-06,
631
+ "loss": 4.2868,
632
  "step": 104
633
  },
634
  {
635
+ "epoch": 0.35,
636
+ "learning_rate": 2.08e-06,
637
+ "loss": 4.3938,
638
  "step": 105
639
  },
640
  {
641
+ "epoch": 0.36,
642
+ "learning_rate": 2.1000000000000002e-06,
643
+ "loss": 4.3391,
644
  "step": 106
645
  },
646
  {
647
+ "epoch": 0.36,
648
+ "learning_rate": 2.12e-06,
649
+ "loss": 4.2794,
650
  "step": 107
651
  },
652
  {
653
+ "epoch": 0.36,
654
+ "learning_rate": 2.1400000000000003e-06,
655
+ "loss": 4.3783,
656
  "step": 108
657
  },
658
  {
659
+ "epoch": 0.37,
660
+ "learning_rate": 2.16e-06,
661
+ "loss": 4.382,
662
  "step": 109
663
  },
664
  {
665
+ "epoch": 0.37,
666
+ "learning_rate": 2.1800000000000003e-06,
667
+ "loss": 4.3473,
668
  "step": 110
669
  },
670
  {
671
+ "epoch": 0.37,
672
+ "learning_rate": 2.2e-06,
673
+ "loss": 4.4273,
674
  "step": 111
675
  },
676
  {
677
+ "epoch": 0.38,
678
+ "learning_rate": 2.2200000000000003e-06,
679
+ "loss": 4.3599,
680
  "step": 112
681
  },
682
  {
683
+ "epoch": 0.38,
684
+ "learning_rate": 2.24e-06,
685
+ "loss": 4.2758,
686
  "step": 113
687
  },
688
  {
689
+ "epoch": 0.38,
690
+ "learning_rate": 2.2600000000000004e-06,
691
+ "loss": 4.3514,
692
  "step": 114
693
  },
694
  {
695
+ "epoch": 0.39,
696
+ "learning_rate": 2.28e-06,
697
+ "loss": 4.3079,
698
  "step": 115
699
  },
700
  {
701
+ "epoch": 0.39,
702
+ "learning_rate": 2.3000000000000004e-06,
703
+ "loss": 4.3189,
704
  "step": 116
705
  },
706
  {
707
+ "epoch": 0.39,
708
+ "learning_rate": 2.3200000000000002e-06,
709
+ "loss": 4.2784,
710
  "step": 117
711
  },
712
  {
713
+ "epoch": 0.4,
714
+ "learning_rate": 2.3400000000000005e-06,
715
+ "loss": 4.3314,
716
  "step": 118
717
  },
718
  {
719
+ "epoch": 0.4,
720
+ "learning_rate": 2.3600000000000003e-06,
721
+ "loss": 4.3653,
722
  "step": 119
723
  },
724
  {
725
+ "epoch": 0.4,
726
+ "learning_rate": 2.38e-06,
727
+ "loss": 4.36,
728
  "step": 120
729
  },
730
  {
731
+ "epoch": 0.41,
732
+ "learning_rate": 2.4000000000000003e-06,
733
+ "loss": 4.3902,
734
  "step": 121
735
  },
736
  {
737
+ "epoch": 0.41,
738
+ "learning_rate": 2.42e-06,
739
+ "loss": 4.4351,
740
  "step": 122
741
  },
742
  {
743
+ "epoch": 0.41,
744
+ "learning_rate": 2.4400000000000004e-06,
745
+ "loss": 4.2534,
746
  "step": 123
747
  },
748
  {
749
+ "epoch": 0.42,
750
+ "learning_rate": 2.46e-06,
751
+ "loss": 4.3887,
752
  "step": 124
753
  },
754
  {
755
+ "epoch": 0.42,
756
+ "learning_rate": 2.4800000000000004e-06,
757
+ "loss": 4.4345,
758
  "step": 125
759
  },
760
  {
761
+ "epoch": 0.42,
762
+ "learning_rate": 2.5e-06,
763
+ "loss": 4.3609,
764
  "step": 126
765
  },
766
  {
767
+ "epoch": 0.43,
768
+ "learning_rate": 2.52e-06,
769
+ "loss": 4.4055,
770
  "step": 127
771
  },
772
  {
773
+ "epoch": 0.43,
774
+ "learning_rate": 2.5400000000000002e-06,
775
+ "loss": 4.2582,
776
  "step": 128
777
  },
778
  {
779
+ "epoch": 0.43,
780
+ "learning_rate": 2.56e-06,
781
+ "loss": 4.4031,
782
  "step": 129
783
  },
784
  {
785
+ "epoch": 0.44,
786
+ "learning_rate": 2.5800000000000003e-06,
787
+ "loss": 4.4855,
788
  "step": 130
789
  },
790
  {
791
+ "epoch": 0.44,
792
+ "learning_rate": 2.6e-06,
793
+ "loss": 4.4102,
794
  "step": 131
795
  },
796
  {
797
+ "epoch": 0.44,
798
+ "learning_rate": 2.6200000000000003e-06,
799
+ "loss": 4.2256,
800
  "step": 132
801
  },
802
  {
803
+ "epoch": 0.45,
804
+ "learning_rate": 2.64e-06,
805
+ "loss": 4.4007,
806
  "step": 133
807
  },
808
  {
809
+ "epoch": 0.45,
810
+ "learning_rate": 2.6600000000000004e-06,
811
+ "loss": 4.2806,
812
  "step": 134
813
  },
814
  {
815
+ "epoch": 0.45,
816
+ "learning_rate": 2.68e-06,
817
+ "loss": 4.4708,
818
  "step": 135
819
  },
820
  {
821
+ "epoch": 0.46,
822
+ "learning_rate": 2.7000000000000004e-06,
823
+ "loss": 4.3876,
824
  "step": 136
825
  },
826
  {
827
+ "epoch": 0.46,
828
+ "learning_rate": 2.7200000000000002e-06,
829
+ "loss": 4.3317,
830
  "step": 137
831
  },
832
  {
833
+ "epoch": 0.46,
834
+ "learning_rate": 2.7400000000000004e-06,
835
+ "loss": 4.2371,
836
  "step": 138
837
  },
838
  {
839
+ "epoch": 0.47,
840
+ "learning_rate": 2.7600000000000003e-06,
841
+ "loss": 4.2725,
842
  "step": 139
843
  },
844
  {
845
+ "epoch": 0.47,
846
+ "learning_rate": 2.7800000000000005e-06,
847
+ "loss": 4.3226,
848
  "step": 140
849
  },
850
  {
851
+ "epoch": 0.47,
852
+ "learning_rate": 2.8000000000000003e-06,
853
+ "loss": 4.5303,
854
  "step": 141
855
  },
856
  {
857
+ "epoch": 0.48,
858
+ "learning_rate": 2.82e-06,
859
+ "loss": 4.2812,
860
  "step": 142
861
  },
862
  {
863
+ "epoch": 0.48,
864
+ "learning_rate": 2.84e-06,
865
+ "loss": 4.3971,
866
  "step": 143
867
  },
868
  {
869
+ "epoch": 0.48,
870
+ "learning_rate": 2.86e-06,
871
+ "loss": 4.426,
872
  "step": 144
873
  },
874
  {
875
+ "epoch": 0.49,
876
+ "learning_rate": 2.88e-06,
877
+ "loss": 4.4879,
878
  "step": 145
879
  },
880
  {
881
+ "epoch": 0.49,
882
+ "learning_rate": 2.9e-06,
883
+ "loss": 4.3671,
884
  "step": 146
885
  },
886
  {
887
+ "epoch": 0.49,
888
+ "learning_rate": 2.92e-06,
889
+ "loss": 4.4797,
890
  "step": 147
891
  },
892
  {
893
+ "epoch": 0.5,
894
+ "learning_rate": 2.9400000000000002e-06,
895
+ "loss": 4.6502,
896
  "step": 148
897
  },
898
  {
899
+ "epoch": 0.5,
900
+ "learning_rate": 2.96e-06,
901
+ "loss": 4.6188,
902
  "step": 149
903
  },
904
  {
905
+ "epoch": 0.5,
906
+ "learning_rate": 2.9800000000000003e-06,
907
+ "loss": 4.7733,
908
  "step": 150
909
  },
910
  {
911
+ "epoch": 0.51,
912
+ "learning_rate": 3e-06,
913
+ "loss": 4.3361,
914
  "step": 151
915
  },
916
  {
917
+ "epoch": 0.51,
918
+ "learning_rate": 3.0200000000000003e-06,
919
+ "loss": 4.3216,
920
  "step": 152
921
  },
922
  {
923
+ "epoch": 0.51,
924
+ "learning_rate": 3.04e-06,
925
+ "loss": 4.3243,
926
  "step": 153
927
  },
928
  {
929
+ "epoch": 0.52,
930
+ "learning_rate": 3.0600000000000003e-06,
931
+ "loss": 4.2681,
932
  "step": 154
933
  },
934
  {
935
+ "epoch": 0.52,
936
+ "learning_rate": 3.08e-06,
937
+ "loss": 4.2981,
938
  "step": 155
939
  },
940
  {
941
+ "epoch": 0.52,
942
+ "learning_rate": 3.1000000000000004e-06,
943
+ "loss": 4.3336,
944
  "step": 156
945
  },
946
  {
947
+ "epoch": 0.53,
948
+ "learning_rate": 3.12e-06,
949
+ "loss": 4.2307,
950
  "step": 157
951
  },
952
  {
953
+ "epoch": 0.53,
954
+ "learning_rate": 3.1400000000000004e-06,
955
+ "loss": 4.3043,
956
  "step": 158
957
  },
958
  {
959
+ "epoch": 0.53,
960
+ "learning_rate": 3.1600000000000002e-06,
961
+ "loss": 4.28,
962
  "step": 159
963
  },
964
  {
965
+ "epoch": 0.54,
966
+ "learning_rate": 3.1800000000000005e-06,
967
+ "loss": 4.2736,
968
  "step": 160
969
  },
970
  {
971
+ "epoch": 0.54,
972
+ "learning_rate": 3.2000000000000003e-06,
973
+ "loss": 4.2756,
974
  "step": 161
975
  },
976
  {
977
+ "epoch": 0.54,
978
+ "learning_rate": 3.2200000000000005e-06,
979
+ "loss": 4.2625,
980
  "step": 162
981
  },
982
  {
983
+ "epoch": 0.55,
984
+ "learning_rate": 3.2400000000000003e-06,
985
+ "loss": 4.3129,
986
  "step": 163
987
  },
988
  {
989
+ "epoch": 0.55,
990
+ "learning_rate": 3.2600000000000006e-06,
991
+ "loss": 4.2453,
992
  "step": 164
993
  },
994
  {
995
+ "epoch": 0.55,
996
+ "learning_rate": 3.2800000000000004e-06,
997
+ "loss": 4.3079,
998
  "step": 165
999
  },
1000
  {
1001
+ "epoch": 0.56,
1002
+ "learning_rate": 3.3000000000000006e-06,
1003
+ "loss": 4.2754,
1004
  "step": 166
1005
  },
1006
  {
1007
+ "epoch": 0.56,
1008
+ "learning_rate": 3.3200000000000004e-06,
1009
+ "loss": 4.3026,
1010
  "step": 167
1011
  },
1012
  {
1013
+ "epoch": 0.56,
1014
+ "learning_rate": 3.3400000000000006e-06,
1015
+ "loss": 4.2798,
1016
  "step": 168
1017
  },
1018
  {
1019
+ "epoch": 0.57,
1020
+ "learning_rate": 3.3600000000000004e-06,
1021
+ "loss": 4.2466,
1022
  "step": 169
1023
  },
1024
  {
1025
+ "epoch": 0.57,
1026
+ "learning_rate": 3.3800000000000007e-06,
1027
+ "loss": 4.1995,
1028
  "step": 170
1029
  },
1030
  {
1031
+ "epoch": 0.58,
1032
+ "learning_rate": 3.4000000000000005e-06,
1033
+ "loss": 4.1824,
1034
  "step": 171
1035
  },
1036
  {
1037
+ "epoch": 0.58,
1038
+ "learning_rate": 3.4200000000000007e-06,
1039
+ "loss": 4.2936,
1040
  "step": 172
1041
  },
1042
  {
1043
+ "epoch": 0.58,
1044
+ "learning_rate": 3.44e-06,
1045
+ "loss": 4.1523,
1046
  "step": 173
1047
  },
1048
  {
1049
+ "epoch": 0.59,
1050
+ "learning_rate": 3.46e-06,
1051
+ "loss": 4.2063,
1052
  "step": 174
1053
  },
1054
  {
1055
+ "epoch": 0.59,
1056
+ "learning_rate": 3.48e-06,
1057
+ "loss": 4.3818,
1058
  "step": 175
1059
  },
1060
  {
1061
+ "epoch": 0.59,
1062
+ "learning_rate": 3.5e-06,
1063
+ "loss": 4.2121,
1064
  "step": 176
1065
  },
1066
  {
1067
+ "epoch": 0.6,
1068
+ "learning_rate": 3.52e-06,
1069
+ "loss": 4.35,
1070
  "step": 177
1071
  },
1072
  {
1073
+ "epoch": 0.6,
1074
+ "learning_rate": 3.54e-06,
1075
+ "loss": 4.2109,
1076
  "step": 178
1077
  },
1078
  {
1079
+ "epoch": 0.6,
1080
+ "learning_rate": 3.5600000000000002e-06,
1081
+ "loss": 4.2574,
1082
  "step": 179
1083
  },
1084
  {
1085
+ "epoch": 0.61,
1086
+ "learning_rate": 3.58e-06,
1087
+ "loss": 4.3989,
1088
  "step": 180
1089
  },
1090
  {
1091
+ "epoch": 0.61,
1092
+ "learning_rate": 3.6000000000000003e-06,
1093
+ "loss": 4.2638,
1094
  "step": 181
1095
  },
1096
  {
1097
+ "epoch": 0.61,
1098
+ "learning_rate": 3.62e-06,
1099
+ "loss": 4.2118,
1100
  "step": 182
1101
  },
1102
  {
1103
+ "epoch": 0.62,
1104
+ "learning_rate": 3.6400000000000003e-06,
1105
+ "loss": 4.1813,
1106
  "step": 183
1107
  },
1108
  {
1109
+ "epoch": 0.62,
1110
+ "learning_rate": 3.66e-06,
1111
+ "loss": 4.2778,
1112
  "step": 184
1113
  },
1114
  {
1115
+ "epoch": 0.62,
1116
+ "learning_rate": 3.6800000000000003e-06,
1117
+ "loss": 4.3297,
1118
  "step": 185
1119
  },
1120
  {
1121
+ "epoch": 0.63,
1122
+ "learning_rate": 3.7e-06,
1123
+ "loss": 4.272,
1124
  "step": 186
1125
  },
1126
  {
1127
+ "epoch": 0.63,
1128
+ "learning_rate": 3.7200000000000004e-06,
1129
+ "loss": 4.3077,
1130
  "step": 187
1131
  },
1132
  {
1133
+ "epoch": 0.63,
1134
+ "learning_rate": 3.74e-06,
1135
+ "loss": 4.2812,
1136
  "step": 188
1137
  },
1138
  {
1139
+ "epoch": 0.64,
1140
+ "learning_rate": 3.7600000000000004e-06,
1141
+ "loss": 4.2457,
1142
  "step": 189
1143
  },
1144
  {
1145
+ "epoch": 0.64,
1146
+ "learning_rate": 3.7800000000000002e-06,
1147
+ "loss": 4.3328,
1148
  "step": 190
1149
  },
1150
  {
1151
+ "epoch": 0.64,
1152
+ "learning_rate": 3.8000000000000005e-06,
1153
+ "loss": 4.3663,
1154
  "step": 191
1155
  },
1156
  {
1157
+ "epoch": 0.65,
1158
+ "learning_rate": 3.820000000000001e-06,
1159
+ "loss": 4.4388,
1160
  "step": 192
1161
  },
1162
  {
1163
+ "epoch": 0.65,
1164
+ "learning_rate": 3.8400000000000005e-06,
1165
+ "loss": 4.5001,
1166
  "step": 193
1167
  },
1168
  {
1169
+ "epoch": 0.65,
1170
+ "learning_rate": 3.86e-06,
1171
+ "loss": 4.4342,
1172
  "step": 194
1173
  },
1174
  {
1175
+ "epoch": 0.66,
1176
+ "learning_rate": 3.88e-06,
1177
+ "loss": 4.4113,
1178
  "step": 195
1179
  },
1180
  {
1181
+ "epoch": 0.66,
1182
+ "learning_rate": 3.900000000000001e-06,
1183
+ "loss": 4.3612,
1184
  "step": 196
1185
  },
1186
  {
1187
+ "epoch": 0.66,
1188
+ "learning_rate": 3.920000000000001e-06,
1189
+ "loss": 4.2716,
1190
  "step": 197
1191
  },
1192
  {
1193
+ "epoch": 0.67,
1194
+ "learning_rate": 3.94e-06,
1195
+ "loss": 4.7501,
1196
  "step": 198
1197
  },
1198
  {
1199
+ "epoch": 0.67,
1200
+ "learning_rate": 3.96e-06,
1201
+ "loss": 4.5016,
1202
  "step": 199
1203
  },
1204
  {
1205
+ "epoch": 0.67,
1206
+ "learning_rate": 3.980000000000001e-06,
1207
+ "loss": 4.7116,
1208
  "step": 200
1209
  },
1210
  {
1211
+ "epoch": 0.68,
1212
+ "learning_rate": 4.000000000000001e-06,
1213
+ "loss": 4.3106,
1214
  "step": 201
1215
  },
1216
  {
1217
+ "epoch": 0.68,
1218
+ "learning_rate": 4.0200000000000005e-06,
1219
+ "loss": 4.2625,
1220
  "step": 202
1221
  },
1222
  {
1223
+ "epoch": 0.68,
1224
+ "learning_rate": 4.04e-06,
1225
+ "loss": 4.2393,
1226
  "step": 203
1227
  },
1228
  {
1229
+ "epoch": 0.69,
1230
+ "learning_rate": 4.060000000000001e-06,
1231
+ "loss": 4.1657,
1232
  "step": 204
1233
  },
1234
  {
1235
+ "epoch": 0.69,
1236
+ "learning_rate": 4.08e-06,
1237
+ "loss": 4.2338,
1238
  "step": 205
1239
  },
1240
  {
1241
+ "epoch": 0.69,
1242
+ "learning_rate": 4.1e-06,
1243
+ "loss": 4.229,
1244
  "step": 206
1245
  },
1246
  {
1247
+ "epoch": 0.7,
1248
+ "learning_rate": 4.12e-06,
1249
+ "loss": 4.2768,
1250
  "step": 207
1251
  },
1252
  {
1253
+ "epoch": 0.7,
1254
+ "learning_rate": 4.14e-06,
1255
+ "loss": 4.245,
1256
  "step": 208
1257
  },
1258
  {
1259
+ "epoch": 0.7,
1260
+ "learning_rate": 4.16e-06,
1261
+ "loss": 4.2213,
1262
  "step": 209
1263
  },
1264
  {
1265
+ "epoch": 0.71,
1266
+ "learning_rate": 4.18e-06,
1267
+ "loss": 4.3171,
1268
  "step": 210
1269
  },
1270
  {
1271
+ "epoch": 0.71,
1272
+ "learning_rate": 4.2000000000000004e-06,
1273
+ "loss": 4.1127,
1274
  "step": 211
1275
  },
1276
  {
1277
+ "epoch": 0.71,
1278
+ "learning_rate": 4.22e-06,
1279
+ "loss": 4.2402,
1280
  "step": 212
1281
  },
1282
  {
1283
+ "epoch": 0.72,
1284
+ "learning_rate": 4.24e-06,
1285
+ "loss": 4.265,
1286
  "step": 213
1287
  },
1288
  {
1289
+ "epoch": 0.72,
1290
+ "learning_rate": 4.26e-06,
1291
+ "loss": 4.2684,
1292
  "step": 214
1293
  },
1294
  {
1295
+ "epoch": 0.72,
1296
+ "learning_rate": 4.2800000000000005e-06,
1297
+ "loss": 4.2735,
1298
  "step": 215
1299
  },
1300
  {
1301
+ "epoch": 0.73,
1302
+ "learning_rate": 4.3e-06,
1303
+ "loss": 4.2555,
1304
  "step": 216
1305
  },
1306
  {
1307
+ "epoch": 0.73,
1308
+ "learning_rate": 4.32e-06,
1309
+ "loss": 4.2211,
1310
  "step": 217
1311
  },
1312
  {
1313
+ "epoch": 0.73,
1314
+ "learning_rate": 4.34e-06,
1315
+ "loss": 4.1925,
1316
  "step": 218
1317
  },
1318
  {
1319
+ "epoch": 0.74,
1320
+ "learning_rate": 4.360000000000001e-06,
1321
+ "loss": 4.1934,
1322
  "step": 219
1323
  },
1324
  {
1325
+ "epoch": 0.74,
1326
+ "learning_rate": 4.38e-06,
1327
+ "loss": 4.2549,
1328
  "step": 220
1329
  },
1330
  {
1331
+ "epoch": 0.74,
1332
+ "learning_rate": 4.4e-06,
1333
+ "loss": 4.1733,
1334
  "step": 221
1335
  },
1336
  {
1337
+ "epoch": 0.75,
1338
+ "learning_rate": 4.42e-06,
1339
+ "loss": 4.1752,
1340
  "step": 222
1341
  },
1342
  {
1343
+ "epoch": 0.75,
1344
+ "learning_rate": 4.440000000000001e-06,
1345
+ "loss": 4.1934,
1346
  "step": 223
1347
  },
1348
  {
1349
+ "epoch": 0.75,
1350
+ "learning_rate": 4.4600000000000005e-06,
1351
+ "loss": 4.2419,
1352
  "step": 224
1353
  },
1354
  {
1355
+ "epoch": 0.76,
1356
+ "learning_rate": 4.48e-06,
1357
+ "loss": 4.2159,
1358
  "step": 225
1359
  },
1360
  {
1361
+ "epoch": 0.76,
1362
+ "learning_rate": 4.5e-06,
1363
+ "loss": 4.226,
1364
  "step": 226
1365
  },
1366
  {
1367
+ "epoch": 0.76,
1368
+ "learning_rate": 4.520000000000001e-06,
1369
+ "loss": 4.2576,
1370
  "step": 227
1371
  },
1372
  {
1373
+ "epoch": 0.77,
1374
+ "learning_rate": 4.540000000000001e-06,
1375
+ "loss": 4.1938,
1376
  "step": 228
1377
  },
1378
  {
1379
+ "epoch": 0.77,
1380
+ "learning_rate": 4.56e-06,
1381
+ "loss": 4.2063,
1382
  "step": 229
1383
  },
1384
  {
1385
+ "epoch": 0.77,
1386
+ "learning_rate": 4.58e-06,
1387
+ "loss": 4.2194,
1388
  "step": 230
1389
  },
1390
  {
1391
+ "epoch": 0.78,
1392
+ "learning_rate": 4.600000000000001e-06,
1393
+ "loss": 4.3171,
1394
  "step": 231
1395
  },
1396
  {
1397
+ "epoch": 0.78,
1398
+ "learning_rate": 4.620000000000001e-06,
1399
+ "loss": 4.1888,
1400
  "step": 232
1401
  },
1402
  {
1403
+ "epoch": 0.78,
1404
+ "learning_rate": 4.6400000000000005e-06,
1405
+ "loss": 4.2994,
1406
  "step": 233
1407
  },
1408
  {
1409
+ "epoch": 0.79,
1410
+ "learning_rate": 4.66e-06,
1411
+ "loss": 4.2743,
1412
  "step": 234
1413
  },
1414
  {
1415
+ "epoch": 0.79,
1416
+ "learning_rate": 4.680000000000001e-06,
1417
+ "loss": 4.3265,
1418
  "step": 235
1419
  },
1420
  {
1421
+ "epoch": 0.79,
1422
+ "learning_rate": 4.7e-06,
1423
+ "loss": 4.146,
1424
  "step": 236
1425
  },
1426
  {
1427
+ "epoch": 0.8,
1428
+ "learning_rate": 4.7200000000000005e-06,
1429
+ "loss": 4.2531,
1430
  "step": 237
1431
  },
1432
  {
1433
+ "epoch": 0.8,
1434
+ "learning_rate": 4.74e-06,
1435
+ "loss": 4.3796,
1436
  "step": 238
1437
  },
1438
  {
1439
+ "epoch": 0.8,
1440
+ "learning_rate": 4.76e-06,
1441
+ "loss": 4.2271,
1442
  "step": 239
1443
  },
1444
  {
1445
+ "epoch": 0.81,
1446
+ "learning_rate": 4.78e-06,
1447
+ "loss": 4.2738,
1448
  "step": 240
1449
  },
1450
  {
1451
+ "epoch": 0.81,
1452
+ "learning_rate": 4.800000000000001e-06,
1453
+ "loss": 4.3837,
1454
  "step": 241
1455
  },
1456
  {
1457
+ "epoch": 0.81,
1458
+ "learning_rate": 4.8200000000000004e-06,
1459
+ "loss": 4.3178,
1460
  "step": 242
1461
  },
1462
  {
1463
+ "epoch": 0.82,
1464
+ "learning_rate": 4.84e-06,
1465
+ "loss": 4.2609,
1466
  "step": 243
1467
  },
1468
  {
1469
+ "epoch": 0.82,
1470
+ "learning_rate": 4.86e-06,
1471
+ "loss": 4.2858,
1472
  "step": 244
1473
  },
1474
  {
1475
+ "epoch": 0.82,
1476
+ "learning_rate": 4.880000000000001e-06,
1477
+ "loss": 4.2618,
1478
  "step": 245
1479
  },
1480
  {
1481
+ "epoch": 0.83,
1482
+ "learning_rate": 4.9000000000000005e-06,
1483
+ "loss": 4.4445,
1484
  "step": 246
1485
  },
1486
  {
1487
+ "epoch": 0.83,
1488
+ "learning_rate": 4.92e-06,
1489
+ "loss": 4.3901,
1490
  "step": 247
1491
  },
1492
  {
1493
+ "epoch": 0.83,
1494
+ "learning_rate": 4.94e-06,
1495
+ "loss": 4.3951,
1496
  "step": 248
1497
  },
1498
  {
1499
+ "epoch": 0.84,
1500
+ "learning_rate": 4.960000000000001e-06,
1501
+ "loss": 4.5076,
1502
  "step": 249
1503
  },
1504
  {
1505
+ "epoch": 0.84,
1506
+ "learning_rate": 4.980000000000001e-06,
1507
+ "loss": 4.5527,
1508
  "step": 250
1509
  },
1510
  {
1511
+ "epoch": 0.84,
1512
+ "learning_rate": 5e-06,
1513
+ "loss": 4.2833,
1514
  "step": 251
1515
  },
1516
  {
1517
+ "epoch": 0.85,
1518
+ "learning_rate": 5.02e-06,
1519
+ "loss": 4.206,
1520
  "step": 252
1521
  },
1522
  {
1523
+ "epoch": 0.85,
1524
+ "learning_rate": 5.04e-06,
1525
+ "loss": 4.1542,
1526
  "step": 253
1527
  },
1528
  {
1529
+ "epoch": 0.85,
1530
+ "learning_rate": 5.060000000000001e-06,
1531
+ "loss": 4.1541,
1532
  "step": 254
1533
  },
1534
  {
1535
+ "epoch": 0.86,
1536
+ "learning_rate": 5.0800000000000005e-06,
1537
+ "loss": 4.1914,
1538
  "step": 255
1539
  },
1540
  {
1541
+ "epoch": 0.86,
1542
+ "learning_rate": 5.1e-06,
1543
+ "loss": 4.1824,
1544
  "step": 256
1545
  },
1546
  {
1547
+ "epoch": 0.86,
1548
+ "learning_rate": 5.12e-06,
1549
+ "loss": 4.3117,
1550
  "step": 257
1551
  },
1552
  {
1553
+ "epoch": 0.87,
1554
+ "learning_rate": 5.140000000000001e-06,
1555
+ "loss": 4.192,
1556
  "step": 258
1557
  },
1558
  {
1559
+ "epoch": 0.87,
1560
+ "learning_rate": 5.1600000000000006e-06,
1561
+ "loss": 4.2298,
1562
  "step": 259
1563
  },
1564
  {
1565
+ "epoch": 0.87,
1566
+ "learning_rate": 5.18e-06,
1567
+ "loss": 4.1561,
1568
  "step": 260
1569
  },
1570
  {
1571
+ "epoch": 0.88,
1572
+ "learning_rate": 5.2e-06,
1573
+ "loss": 4.2479,
1574
  "step": 261
1575
  },
1576
  {
1577
+ "epoch": 0.88,
1578
+ "learning_rate": 5.220000000000001e-06,
1579
+ "loss": 4.2713,
1580
  "step": 262
1581
  },
1582
  {
1583
+ "epoch": 0.88,
1584
+ "learning_rate": 5.240000000000001e-06,
1585
+ "loss": 4.1502,
1586
  "step": 263
1587
  },
1588
  {
1589
+ "epoch": 0.89,
1590
+ "learning_rate": 5.2600000000000005e-06,
1591
+ "loss": 4.1789,
1592
  "step": 264
1593
  },
1594
  {
1595
+ "epoch": 0.89,
1596
+ "learning_rate": 5.28e-06,
1597
+ "loss": 4.2032,
1598
  "step": 265
1599
  },
1600
  {
1601
+ "epoch": 0.89,
1602
+ "learning_rate": 5.300000000000001e-06,
1603
+ "loss": 4.2824,
1604
  "step": 266
1605
  },
1606
  {
1607
+ "epoch": 0.9,
1608
+ "learning_rate": 5.320000000000001e-06,
1609
+ "loss": 4.2033,
1610
  "step": 267
1611
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1612
  {
1613
  "epoch": 0.9,
1614
+ "learning_rate": 5.3400000000000005e-06,
1615
+ "loss": 4.1714,
1616
+ "step": 268
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1617
  },
1618
  {
1619
  "epoch": 0.9,
1620
+ "learning_rate": 5.36e-06,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1621
  "loss": 4.1364,
1622
+ "step": 269
1623
  },
1624
  {
1625
  "epoch": 0.91,
1626
+ "learning_rate": 5.380000000000001e-06,
1627
+ "loss": 4.261,
1628
+ "step": 270
1629
  },
1630
  {
1631
  "epoch": 0.91,
1632
+ "learning_rate": 5.400000000000001e-06,
1633
+ "loss": 4.1305,
1634
+ "step": 271
1635
  },
1636
  {
1637
  "epoch": 0.91,
1638
+ "learning_rate": 5.420000000000001e-06,
1639
+ "loss": 4.1106,
1640
+ "step": 272
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1641
  },
1642
  {
1643
  "epoch": 0.92,
1644
+ "learning_rate": 5.4400000000000004e-06,
1645
+ "loss": 4.1723,
1646
+ "step": 273
1647
  },
1648
  {
1649
  "epoch": 0.92,
1650
+ "learning_rate": 5.460000000000001e-06,
1651
+ "loss": 4.1495,
1652
+ "step": 274
1653
  },
1654
  {
1655
  "epoch": 0.92,
1656
+ "learning_rate": 5.480000000000001e-06,
1657
+ "loss": 4.2855,
1658
+ "step": 275
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1659
  },
1660
  {
1661
  "epoch": 0.93,
1662
+ "learning_rate": 5.500000000000001e-06,
1663
+ "loss": 4.199,
1664
+ "step": 276
1665
  },
1666
  {
1667
  "epoch": 0.93,
1668
+ "learning_rate": 5.5200000000000005e-06,
1669
+ "loss": 4.2175,
1670
+ "step": 277
1671
  },
1672
  {
1673
  "epoch": 0.93,
1674
+ "learning_rate": 5.540000000000001e-06,
1675
+ "loss": 4.152,
1676
+ "step": 278
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1677
  },
1678
  {
1679
  "epoch": 0.94,
1680
+ "learning_rate": 5.560000000000001e-06,
1681
+ "loss": 4.1857,
1682
+ "step": 279
1683
  },
1684
  {
1685
  "epoch": 0.94,
1686
+ "learning_rate": 5.580000000000001e-06,
1687
+ "loss": 4.1679,
1688
+ "step": 280
1689
  },
1690
  {
1691
  "epoch": 0.94,
1692
+ "learning_rate": 5.600000000000001e-06,
1693
+ "loss": 4.2036,
1694
+ "step": 281
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1695
  },
1696
  {
1697
  "epoch": 0.95,
1698
+ "learning_rate": 5.620000000000001e-06,
1699
+ "loss": 4.2049,
1700
+ "step": 282
1701
  },
1702
  {
1703
  "epoch": 0.95,
1704
+ "learning_rate": 5.64e-06,
1705
+ "loss": 4.3554,
1706
+ "step": 283
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1707
  },
1708
  {
1709
  "epoch": 0.96,
1710
+ "learning_rate": 5.66e-06,
1711
+ "loss": 4.1291,
1712
+ "step": 284
1713
  },
1714
  {
1715
  "epoch": 0.96,
1716
+ "learning_rate": 5.68e-06,
1717
+ "loss": 4.2472,
1718
+ "step": 285
1719
  },
1720
  {
1721
  "epoch": 0.96,
1722
+ "learning_rate": 5.7e-06,
1723
+ "loss": 4.2736,
1724
+ "step": 286
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1725
  },
1726
  {
1727
  "epoch": 0.97,
1728
+ "learning_rate": 5.72e-06,
1729
+ "loss": 4.1614,
1730
+ "step": 287
1731
  },
1732
  {
1733
  "epoch": 0.97,
1734
+ "learning_rate": 5.74e-06,
1735
+ "loss": 4.2049,
1736
+ "step": 288
1737
  },
1738
  {
1739
  "epoch": 0.97,
1740
+ "learning_rate": 5.76e-06,
1741
+ "loss": 4.206,
1742
+ "step": 289
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1743
  },
1744
  {
1745
  "epoch": 0.98,
1746
+ "learning_rate": 5.78e-06,
1747
+ "loss": 4.2975,
1748
+ "step": 290
1749
  },
1750
  {
1751
  "epoch": 0.98,
1752
+ "learning_rate": 5.8e-06,
1753
+ "loss": 4.2788,
1754
+ "step": 291
1755
  },
1756
  {
1757
  "epoch": 0.98,
1758
+ "learning_rate": 5.82e-06,
1759
+ "loss": 4.3599,
1760
+ "step": 292
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1761
  },
1762
  {
1763
  "epoch": 0.99,
1764
+ "learning_rate": 5.84e-06,
1765
+ "loss": 4.3441,
1766
+ "step": 293
1767
  },
1768
  {
1769
  "epoch": 0.99,
1770
+ "learning_rate": 5.86e-06,
1771
+ "loss": 4.5265,
1772
+ "step": 294
1773
  },
1774
  {
1775
  "epoch": 0.99,
1776
+ "learning_rate": 5.8800000000000005e-06,
1777
+ "loss": 4.3277,
1778
+ "step": 295
 
 
 
 
 
 
1779
  },
1780
  {
1781
  "epoch": 1.0,
1782
+ "learning_rate": 5.9e-06,
1783
+ "loss": 4.6971,
1784
+ "step": 296
1785
  },
1786
  {
1787
  "epoch": 1.0,
1788
+ "learning_rate": 5.92e-06,
1789
+ "loss": 4.5581,
1790
+ "step": 297
1791
  },
1792
  {
1793
  "epoch": 1.0,
1794
+ "step": 297,
1795
  "total_flos": 0.0,
1796
+ "train_loss": 4.444549621556343,
1797
+ "train_runtime": 4661.3109,
1798
+ "train_samples_per_second": 6.122,
1799
+ "train_steps_per_second": 0.064
1800
  }
1801
  ],
1802
+ "max_steps": 297,
1803
  "num_train_epochs": 1,
1804
  "total_flos": 0.0,
1805
  "trial_name": null,