prajjwal1 commited on
Commit
293ff2f
1 Parent(s): 5cb5ee1
Files changed (3) hide show
  1. config.json +1 -1
  2. pytorch_model.bin +2 -2
  3. trainer_state.json +103 -3187
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "ctrl",
3
  "architectures": [
4
  "CTRLLMHeadModel"
5
  ],
1
  {
2
+ "_name_or_path": "prajjwal1/ctrl_discovery_2",
3
  "architectures": [
4
  "CTRLLMHeadModel"
5
  ],
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9afa6354eb0df702a1b3ca6d22e4a7e8c1d152344c1c590d9037d1797f14da1f
3
- size 6552164800
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4ac8eabd147c74bf2489a369849673602db8aa91c3246688630e173cf113324e
3
+ size 6552164551
trainer_state.json CHANGED
@@ -2,3298 +2,214 @@
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
  "epoch": 3.0,
5
- "global_step": 273072,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
- "epoch": 0.01,
12
- "learning_rate": 4.990844905372942e-05,
13
- "loss": 1.2008,
14
  "step": 500
15
  },
16
  {
17
- "epoch": 0.01,
18
- "learning_rate": 4.9816898107458844e-05,
19
- "loss": 0.904,
20
  "step": 1000
21
  },
22
  {
23
- "epoch": 0.02,
24
- "learning_rate": 4.972534716118826e-05,
25
- "loss": 0.9083,
26
  "step": 1500
27
  },
28
  {
29
- "epoch": 0.02,
30
- "learning_rate": 4.963379621491768e-05,
31
- "loss": 0.9092,
32
  "step": 2000
33
  },
34
  {
35
- "epoch": 0.03,
36
- "learning_rate": 4.9542245268647095e-05,
37
- "loss": 0.8994,
38
  "step": 2500
39
  },
40
  {
41
- "epoch": 0.03,
42
- "learning_rate": 4.945069432237652e-05,
43
- "loss": 0.9007,
44
  "step": 3000
45
  },
46
  {
47
- "epoch": 0.04,
48
- "learning_rate": 4.9359143376105936e-05,
49
- "loss": 0.906,
50
  "step": 3500
51
  },
52
  {
53
- "epoch": 0.04,
54
- "learning_rate": 4.926759242983536e-05,
55
- "loss": 0.9026,
56
  "step": 4000
57
  },
58
  {
59
- "epoch": 0.05,
60
- "learning_rate": 4.917604148356478e-05,
61
- "loss": 0.8885,
62
  "step": 4500
63
  },
64
  {
65
- "epoch": 0.05,
66
- "learning_rate": 4.9084490537294195e-05,
67
- "loss": 0.8877,
68
  "step": 5000
69
  },
70
  {
71
- "epoch": 0.06,
72
- "learning_rate": 4.899293959102362e-05,
73
- "loss": 0.8886,
74
  "step": 5500
75
  },
76
  {
77
- "epoch": 0.07,
78
- "learning_rate": 4.890138864475303e-05,
79
- "loss": 0.9007,
80
  "step": 6000
81
  },
82
  {
83
- "epoch": 0.07,
84
- "learning_rate": 4.880983769848245e-05,
85
- "loss": 0.8963,
86
  "step": 6500
87
  },
88
  {
89
- "epoch": 0.08,
90
- "learning_rate": 4.871828675221187e-05,
91
- "loss": 0.8941,
92
  "step": 7000
93
  },
94
  {
95
- "epoch": 0.08,
96
- "learning_rate": 4.8626735805941294e-05,
97
- "loss": 0.8834,
98
  "step": 7500
99
  },
100
  {
101
- "epoch": 0.09,
102
- "learning_rate": 4.853518485967071e-05,
103
- "loss": 0.8852,
104
  "step": 8000
105
  },
106
  {
107
- "epoch": 0.09,
108
- "learning_rate": 4.8443633913400135e-05,
109
- "loss": 0.8888,
110
  "step": 8500
111
  },
112
  {
113
- "epoch": 0.1,
114
- "learning_rate": 4.835208296712955e-05,
115
- "loss": 0.8922,
116
  "step": 9000
117
  },
118
  {
119
- "epoch": 0.1,
120
- "learning_rate": 4.826053202085897e-05,
121
- "loss": 0.8927,
122
  "step": 9500
123
  },
124
  {
125
- "epoch": 0.11,
126
- "learning_rate": 4.816898107458839e-05,
127
- "loss": 0.8933,
128
  "step": 10000
129
  },
130
  {
131
- "epoch": 0.12,
132
- "learning_rate": 4.807743012831781e-05,
133
- "loss": 0.8863,
134
  "step": 10500
135
  },
136
  {
137
- "epoch": 0.12,
138
- "learning_rate": 4.798587918204723e-05,
139
- "loss": 0.8867,
140
  "step": 11000
141
  },
142
  {
143
- "epoch": 0.13,
144
- "learning_rate": 4.7894328235776645e-05,
145
- "loss": 0.8905,
146
  "step": 11500
147
  },
148
  {
149
- "epoch": 0.13,
150
- "learning_rate": 4.780277728950607e-05,
151
- "loss": 0.8957,
152
  "step": 12000
153
  },
154
  {
155
- "epoch": 0.14,
156
- "learning_rate": 4.7711226343235486e-05,
157
- "loss": 0.8859,
158
  "step": 12500
159
  },
160
  {
161
- "epoch": 0.14,
162
- "learning_rate": 4.761967539696491e-05,
163
- "loss": 0.8905,
164
  "step": 13000
165
  },
166
  {
167
- "epoch": 0.15,
168
- "learning_rate": 4.752812445069432e-05,
169
- "loss": 0.8861,
170
  "step": 13500
171
  },
172
  {
173
- "epoch": 0.15,
174
- "learning_rate": 4.7436573504423745e-05,
175
- "loss": 0.8919,
176
  "step": 14000
177
  },
178
  {
179
- "epoch": 0.16,
180
- "learning_rate": 4.734502255815316e-05,
181
- "loss": 0.8914,
182
  "step": 14500
183
  },
184
  {
185
- "epoch": 0.16,
186
- "learning_rate": 4.7253471611882586e-05,
187
- "loss": 0.8973,
188
  "step": 15000
189
  },
190
  {
191
- "epoch": 0.17,
192
- "learning_rate": 4.7161920665612e-05,
193
- "loss": 0.893,
194
  "step": 15500
195
  },
196
  {
197
- "epoch": 0.18,
198
- "learning_rate": 4.707036971934142e-05,
199
- "loss": 0.8924,
200
  "step": 16000
201
  },
202
- {
203
- "epoch": 0.18,
204
- "learning_rate": 4.6978818773070844e-05,
205
- "loss": 0.8805,
206
- "step": 16500
207
- },
208
- {
209
- "epoch": 0.19,
210
- "learning_rate": 4.6887267826800255e-05,
211
- "loss": 0.8978,
212
- "step": 17000
213
- },
214
- {
215
- "epoch": 0.19,
216
- "learning_rate": 4.679571688052968e-05,
217
- "loss": 0.89,
218
- "step": 17500
219
- },
220
- {
221
- "epoch": 0.2,
222
- "learning_rate": 4.6704165934259096e-05,
223
- "loss": 0.8874,
224
- "step": 18000
225
- },
226
- {
227
- "epoch": 0.2,
228
- "learning_rate": 4.661261498798852e-05,
229
- "loss": 0.8893,
230
- "step": 18500
231
- },
232
- {
233
- "epoch": 0.21,
234
- "learning_rate": 4.652106404171794e-05,
235
- "loss": 0.8872,
236
- "step": 19000
237
- },
238
- {
239
- "epoch": 0.21,
240
- "learning_rate": 4.642951309544736e-05,
241
- "loss": 0.8905,
242
- "step": 19500
243
- },
244
- {
245
- "epoch": 0.22,
246
- "learning_rate": 4.633796214917678e-05,
247
- "loss": 0.8919,
248
- "step": 20000
249
- },
250
- {
251
- "epoch": 0.23,
252
- "learning_rate": 4.6246411202906195e-05,
253
- "loss": 0.8846,
254
- "step": 20500
255
- },
256
- {
257
- "epoch": 0.23,
258
- "learning_rate": 4.615486025663561e-05,
259
- "loss": 0.8914,
260
- "step": 21000
261
- },
262
- {
263
- "epoch": 0.24,
264
- "learning_rate": 4.606330931036503e-05,
265
- "loss": 0.8863,
266
- "step": 21500
267
- },
268
- {
269
- "epoch": 0.24,
270
- "learning_rate": 4.5971758364094454e-05,
271
- "loss": 0.8921,
272
- "step": 22000
273
- },
274
- {
275
- "epoch": 0.25,
276
- "learning_rate": 4.588020741782387e-05,
277
- "loss": 0.8903,
278
- "step": 22500
279
- },
280
- {
281
- "epoch": 0.25,
282
- "learning_rate": 4.5788656471553295e-05,
283
- "loss": 0.8882,
284
- "step": 23000
285
- },
286
- {
287
- "epoch": 0.26,
288
- "learning_rate": 4.569710552528271e-05,
289
- "loss": 0.8945,
290
- "step": 23500
291
- },
292
- {
293
- "epoch": 0.26,
294
- "learning_rate": 4.5605554579012136e-05,
295
- "loss": 0.8944,
296
- "step": 24000
297
- },
298
- {
299
- "epoch": 0.27,
300
- "learning_rate": 4.5514003632741546e-05,
301
- "loss": 0.8784,
302
- "step": 24500
303
- },
304
- {
305
- "epoch": 0.27,
306
- "learning_rate": 4.542245268647097e-05,
307
- "loss": 0.8843,
308
- "step": 25000
309
- },
310
- {
311
- "epoch": 0.28,
312
- "learning_rate": 4.533090174020039e-05,
313
- "loss": 0.8882,
314
- "step": 25500
315
- },
316
- {
317
- "epoch": 0.29,
318
- "learning_rate": 4.5239350793929805e-05,
319
- "loss": 0.8891,
320
- "step": 26000
321
- },
322
- {
323
- "epoch": 0.29,
324
- "learning_rate": 4.514779984765923e-05,
325
- "loss": 0.8896,
326
- "step": 26500
327
- },
328
- {
329
- "epoch": 0.3,
330
- "learning_rate": 4.5056248901388646e-05,
331
- "loss": 0.8924,
332
- "step": 27000
333
- },
334
- {
335
- "epoch": 0.3,
336
- "learning_rate": 4.496469795511807e-05,
337
- "loss": 0.8812,
338
- "step": 27500
339
- },
340
- {
341
- "epoch": 0.31,
342
- "learning_rate": 4.487314700884749e-05,
343
- "loss": 0.8935,
344
- "step": 28000
345
- },
346
- {
347
- "epoch": 0.31,
348
- "learning_rate": 4.4781596062576904e-05,
349
- "loss": 0.8907,
350
- "step": 28500
351
- },
352
- {
353
- "epoch": 0.32,
354
- "learning_rate": 4.469004511630632e-05,
355
- "loss": 0.8887,
356
- "step": 29000
357
- },
358
- {
359
- "epoch": 0.32,
360
- "learning_rate": 4.4598494170035745e-05,
361
- "loss": 0.8898,
362
- "step": 29500
363
- },
364
- {
365
- "epoch": 0.33,
366
- "learning_rate": 4.450694322376516e-05,
367
- "loss": 0.8846,
368
- "step": 30000
369
- },
370
- {
371
- "epoch": 0.34,
372
- "learning_rate": 4.441539227749458e-05,
373
- "loss": 0.8858,
374
- "step": 30500
375
- },
376
- {
377
- "epoch": 0.34,
378
- "learning_rate": 4.4323841331224004e-05,
379
- "loss": 0.8933,
380
- "step": 31000
381
- },
382
- {
383
- "epoch": 0.35,
384
- "learning_rate": 4.423229038495342e-05,
385
- "loss": 0.8854,
386
- "step": 31500
387
- },
388
- {
389
- "epoch": 0.35,
390
- "learning_rate": 4.414073943868284e-05,
391
- "loss": 0.8931,
392
- "step": 32000
393
- },
394
- {
395
- "epoch": 0.36,
396
- "learning_rate": 4.4049188492412255e-05,
397
- "loss": 0.8904,
398
- "step": 32500
399
- },
400
- {
401
- "epoch": 0.36,
402
- "learning_rate": 4.395763754614168e-05,
403
- "loss": 0.884,
404
- "step": 33000
405
- },
406
- {
407
- "epoch": 0.37,
408
- "learning_rate": 4.3866086599871097e-05,
409
- "loss": 0.8963,
410
- "step": 33500
411
- },
412
- {
413
- "epoch": 0.37,
414
- "learning_rate": 4.377453565360052e-05,
415
- "loss": 0.89,
416
- "step": 34000
417
- },
418
- {
419
- "epoch": 0.38,
420
- "learning_rate": 4.368298470732994e-05,
421
- "loss": 0.8852,
422
- "step": 34500
423
- },
424
- {
425
- "epoch": 0.38,
426
- "learning_rate": 4.3591433761059355e-05,
427
- "loss": 0.8831,
428
- "step": 35000
429
- },
430
- {
431
- "epoch": 0.39,
432
- "learning_rate": 4.349988281478878e-05,
433
- "loss": 0.895,
434
- "step": 35500
435
- },
436
- {
437
- "epoch": 0.4,
438
- "learning_rate": 4.3408331868518196e-05,
439
- "loss": 0.8907,
440
- "step": 36000
441
- },
442
- {
443
- "epoch": 0.4,
444
- "learning_rate": 4.331678092224761e-05,
445
- "loss": 0.8956,
446
- "step": 36500
447
- },
448
- {
449
- "epoch": 0.41,
450
- "learning_rate": 4.322522997597703e-05,
451
- "loss": 0.8895,
452
- "step": 37000
453
- },
454
- {
455
- "epoch": 0.41,
456
- "learning_rate": 4.3133679029706454e-05,
457
- "loss": 0.8941,
458
- "step": 37500
459
- },
460
- {
461
- "epoch": 0.42,
462
- "learning_rate": 4.304212808343587e-05,
463
- "loss": 0.8919,
464
- "step": 38000
465
- },
466
- {
467
- "epoch": 0.42,
468
- "learning_rate": 4.2950577137165296e-05,
469
- "loss": 0.8986,
470
- "step": 38500
471
- },
472
- {
473
- "epoch": 0.43,
474
- "learning_rate": 4.285902619089471e-05,
475
- "loss": 0.8929,
476
- "step": 39000
477
- },
478
- {
479
- "epoch": 0.43,
480
- "learning_rate": 4.276747524462413e-05,
481
- "loss": 0.8906,
482
- "step": 39500
483
- },
484
- {
485
- "epoch": 0.44,
486
- "learning_rate": 4.267592429835355e-05,
487
- "loss": 0.8919,
488
- "step": 40000
489
- },
490
- {
491
- "epoch": 0.44,
492
- "learning_rate": 4.258437335208297e-05,
493
- "loss": 0.8983,
494
- "step": 40500
495
- },
496
- {
497
- "epoch": 0.45,
498
- "learning_rate": 4.249282240581239e-05,
499
- "loss": 0.8801,
500
- "step": 41000
501
- },
502
- {
503
- "epoch": 0.46,
504
- "learning_rate": 4.2401271459541805e-05,
505
- "loss": 0.8857,
506
- "step": 41500
507
- },
508
- {
509
- "epoch": 0.46,
510
- "learning_rate": 4.230972051327123e-05,
511
- "loss": 0.8778,
512
- "step": 42000
513
- },
514
- {
515
- "epoch": 0.47,
516
- "learning_rate": 4.221816956700065e-05,
517
- "loss": 0.8971,
518
- "step": 42500
519
- },
520
- {
521
- "epoch": 0.47,
522
- "learning_rate": 4.212661862073007e-05,
523
- "loss": 0.8845,
524
- "step": 43000
525
- },
526
- {
527
- "epoch": 0.48,
528
- "learning_rate": 4.203506767445948e-05,
529
- "loss": 0.8927,
530
- "step": 43500
531
- },
532
- {
533
- "epoch": 0.48,
534
- "learning_rate": 4.1943516728188905e-05,
535
- "loss": 0.8932,
536
- "step": 44000
537
- },
538
- {
539
- "epoch": 0.49,
540
- "learning_rate": 4.185196578191832e-05,
541
- "loss": 0.8867,
542
- "step": 44500
543
- },
544
- {
545
- "epoch": 0.49,
546
- "learning_rate": 4.1760414835647746e-05,
547
- "loss": 0.8911,
548
- "step": 45000
549
- },
550
- {
551
- "epoch": 0.5,
552
- "learning_rate": 4.166886388937716e-05,
553
- "loss": 0.8915,
554
- "step": 45500
555
- },
556
- {
557
- "epoch": 0.51,
558
- "learning_rate": 4.157731294310658e-05,
559
- "loss": 0.8925,
560
- "step": 46000
561
- },
562
- {
563
- "epoch": 0.51,
564
- "learning_rate": 4.1485761996836004e-05,
565
- "loss": 0.8828,
566
- "step": 46500
567
- },
568
- {
569
- "epoch": 0.52,
570
- "learning_rate": 4.1394211050565415e-05,
571
- "loss": 0.8966,
572
- "step": 47000
573
- },
574
- {
575
- "epoch": 0.52,
576
- "learning_rate": 4.130266010429484e-05,
577
- "loss": 0.8939,
578
- "step": 47500
579
- },
580
- {
581
- "epoch": 0.53,
582
- "learning_rate": 4.1211109158024256e-05,
583
- "loss": 0.8884,
584
- "step": 48000
585
- },
586
- {
587
- "epoch": 0.53,
588
- "learning_rate": 4.111955821175368e-05,
589
- "loss": 0.8925,
590
- "step": 48500
591
- },
592
- {
593
- "epoch": 0.54,
594
- "learning_rate": 4.10280072654831e-05,
595
- "loss": 0.8884,
596
- "step": 49000
597
- },
598
- {
599
- "epoch": 0.54,
600
- "learning_rate": 4.093645631921252e-05,
601
- "loss": 0.8912,
602
- "step": 49500
603
- },
604
- {
605
- "epoch": 0.55,
606
- "learning_rate": 4.084490537294194e-05,
607
- "loss": 0.8882,
608
- "step": 50000
609
- },
610
- {
611
- "epoch": 0.55,
612
- "learning_rate": 4.0753354426671356e-05,
613
- "loss": 0.8879,
614
- "step": 50500
615
- },
616
- {
617
- "epoch": 0.56,
618
- "learning_rate": 4.066180348040077e-05,
619
- "loss": 0.8872,
620
- "step": 51000
621
- },
622
- {
623
- "epoch": 0.57,
624
- "learning_rate": 4.057025253413019e-05,
625
- "loss": 0.9009,
626
- "step": 51500
627
- },
628
- {
629
- "epoch": 0.57,
630
- "learning_rate": 4.0478701587859614e-05,
631
- "loss": 0.8885,
632
- "step": 52000
633
- },
634
- {
635
- "epoch": 0.58,
636
- "learning_rate": 4.038715064158903e-05,
637
- "loss": 0.8896,
638
- "step": 52500
639
- },
640
- {
641
- "epoch": 0.58,
642
- "learning_rate": 4.0295599695318455e-05,
643
- "loss": 0.8895,
644
- "step": 53000
645
- },
646
- {
647
- "epoch": 0.59,
648
- "learning_rate": 4.020404874904787e-05,
649
- "loss": 0.8945,
650
- "step": 53500
651
- },
652
- {
653
- "epoch": 0.59,
654
- "learning_rate": 4.0112497802777296e-05,
655
- "loss": 0.8949,
656
- "step": 54000
657
- },
658
- {
659
- "epoch": 0.6,
660
- "learning_rate": 4.002094685650671e-05,
661
- "loss": 0.8928,
662
- "step": 54500
663
- },
664
- {
665
- "epoch": 0.6,
666
- "learning_rate": 3.992939591023613e-05,
667
- "loss": 0.9009,
668
- "step": 55000
669
- },
670
- {
671
- "epoch": 0.61,
672
- "learning_rate": 3.983784496396555e-05,
673
- "loss": 0.8918,
674
- "step": 55500
675
- },
676
- {
677
- "epoch": 0.62,
678
- "learning_rate": 3.9746294017694965e-05,
679
- "loss": 0.8809,
680
- "step": 56000
681
- },
682
- {
683
- "epoch": 0.62,
684
- "learning_rate": 3.965474307142439e-05,
685
- "loss": 0.8929,
686
- "step": 56500
687
- },
688
- {
689
- "epoch": 0.63,
690
- "learning_rate": 3.9563192125153806e-05,
691
- "loss": 0.8911,
692
- "step": 57000
693
- },
694
- {
695
- "epoch": 0.63,
696
- "learning_rate": 3.947164117888323e-05,
697
- "loss": 0.8933,
698
- "step": 57500
699
- },
700
- {
701
- "epoch": 0.64,
702
- "learning_rate": 3.938009023261265e-05,
703
- "loss": 0.8925,
704
- "step": 58000
705
- },
706
- {
707
- "epoch": 0.64,
708
- "learning_rate": 3.9288539286342065e-05,
709
- "loss": 0.8909,
710
- "step": 58500
711
- },
712
- {
713
- "epoch": 0.65,
714
- "learning_rate": 3.919698834007148e-05,
715
- "loss": 0.893,
716
- "step": 59000
717
- },
718
- {
719
- "epoch": 0.65,
720
- "learning_rate": 3.9105437393800906e-05,
721
- "loss": 0.8879,
722
- "step": 59500
723
- },
724
- {
725
- "epoch": 0.66,
726
- "learning_rate": 3.901388644753032e-05,
727
- "loss": 0.8947,
728
- "step": 60000
729
- },
730
- {
731
- "epoch": 0.66,
732
- "learning_rate": 3.892233550125975e-05,
733
- "loss": 0.8921,
734
- "step": 60500
735
- },
736
- {
737
- "epoch": 0.67,
738
- "learning_rate": 3.8830784554989164e-05,
739
- "loss": 0.8891,
740
- "step": 61000
741
- },
742
- {
743
- "epoch": 0.68,
744
- "learning_rate": 3.873923360871858e-05,
745
- "loss": 0.8894,
746
- "step": 61500
747
- },
748
- {
749
- "epoch": 0.68,
750
- "learning_rate": 3.8647682662448e-05,
751
- "loss": 0.8917,
752
- "step": 62000
753
- },
754
- {
755
- "epoch": 0.69,
756
- "learning_rate": 3.8556131716177416e-05,
757
- "loss": 0.8839,
758
- "step": 62500
759
- },
760
- {
761
- "epoch": 0.69,
762
- "learning_rate": 3.846458076990684e-05,
763
- "loss": 0.8944,
764
- "step": 63000
765
- },
766
- {
767
- "epoch": 0.7,
768
- "learning_rate": 3.837302982363626e-05,
769
- "loss": 0.8929,
770
- "step": 63500
771
- },
772
- {
773
- "epoch": 0.7,
774
- "learning_rate": 3.828147887736568e-05,
775
- "loss": 0.8811,
776
- "step": 64000
777
- },
778
- {
779
- "epoch": 0.71,
780
- "learning_rate": 3.81899279310951e-05,
781
- "loss": 0.89,
782
- "step": 64500
783
- },
784
- {
785
- "epoch": 0.71,
786
- "learning_rate": 3.809837698482452e-05,
787
- "loss": 0.8904,
788
- "step": 65000
789
- },
790
- {
791
- "epoch": 0.72,
792
- "learning_rate": 3.800682603855394e-05,
793
- "loss": 0.8901,
794
- "step": 65500
795
- },
796
- {
797
- "epoch": 0.73,
798
- "learning_rate": 3.7915275092283356e-05,
799
- "loss": 0.8976,
800
- "step": 66000
801
- },
802
- {
803
- "epoch": 0.73,
804
- "learning_rate": 3.7823724146012773e-05,
805
- "loss": 0.892,
806
- "step": 66500
807
- },
808
- {
809
- "epoch": 0.74,
810
- "learning_rate": 3.773217319974219e-05,
811
- "loss": 0.8906,
812
- "step": 67000
813
- },
814
- {
815
- "epoch": 0.74,
816
- "learning_rate": 3.7640622253471615e-05,
817
- "loss": 0.8846,
818
- "step": 67500
819
- },
820
- {
821
- "epoch": 0.75,
822
- "learning_rate": 3.754907130720103e-05,
823
- "loss": 0.8861,
824
- "step": 68000
825
- },
826
- {
827
- "epoch": 0.75,
828
- "learning_rate": 3.7457520360930456e-05,
829
- "loss": 0.8873,
830
- "step": 68500
831
- },
832
- {
833
- "epoch": 0.76,
834
- "learning_rate": 3.736596941465987e-05,
835
- "loss": 0.8897,
836
- "step": 69000
837
- },
838
- {
839
- "epoch": 0.76,
840
- "learning_rate": 3.727441846838929e-05,
841
- "loss": 0.8891,
842
- "step": 69500
843
- },
844
- {
845
- "epoch": 0.77,
846
- "learning_rate": 3.718286752211871e-05,
847
- "loss": 0.8875,
848
- "step": 70000
849
- },
850
- {
851
- "epoch": 0.77,
852
- "learning_rate": 3.709131657584813e-05,
853
- "loss": 0.903,
854
- "step": 70500
855
- },
856
- {
857
- "epoch": 0.78,
858
- "learning_rate": 3.699976562957755e-05,
859
- "loss": 0.9041,
860
- "step": 71000
861
- },
862
- {
863
- "epoch": 0.79,
864
- "learning_rate": 3.6908214683306966e-05,
865
- "loss": 0.9015,
866
- "step": 71500
867
- },
868
- {
869
- "epoch": 0.79,
870
- "learning_rate": 3.681666373703639e-05,
871
- "loss": 0.8953,
872
- "step": 72000
873
- },
874
- {
875
- "epoch": 0.8,
876
- "learning_rate": 3.672511279076581e-05,
877
- "loss": 0.8998,
878
- "step": 72500
879
- },
880
- {
881
- "epoch": 0.8,
882
- "learning_rate": 3.663356184449523e-05,
883
- "loss": 0.8869,
884
- "step": 73000
885
- },
886
- {
887
- "epoch": 0.81,
888
- "learning_rate": 3.654201089822464e-05,
889
- "loss": 0.8871,
890
- "step": 73500
891
- },
892
- {
893
- "epoch": 0.81,
894
- "learning_rate": 3.6450459951954065e-05,
895
- "loss": 0.898,
896
- "step": 74000
897
- },
898
- {
899
- "epoch": 0.82,
900
- "learning_rate": 3.635890900568348e-05,
901
- "loss": 0.8971,
902
- "step": 74500
903
- },
904
- {
905
- "epoch": 0.82,
906
- "learning_rate": 3.6267358059412906e-05,
907
- "loss": 0.8986,
908
- "step": 75000
909
- },
910
- {
911
- "epoch": 0.83,
912
- "learning_rate": 3.6175807113142324e-05,
913
- "loss": 0.8942,
914
- "step": 75500
915
- },
916
- {
917
- "epoch": 0.83,
918
- "learning_rate": 3.608425616687174e-05,
919
- "loss": 0.8951,
920
- "step": 76000
921
- },
922
- {
923
- "epoch": 0.84,
924
- "learning_rate": 3.5992705220601165e-05,
925
- "loss": 0.8935,
926
- "step": 76500
927
- },
928
- {
929
- "epoch": 0.85,
930
- "learning_rate": 3.5901154274330575e-05,
931
- "loss": 0.8812,
932
- "step": 77000
933
- },
934
- {
935
- "epoch": 0.85,
936
- "learning_rate": 3.580960332806e-05,
937
- "loss": 0.8954,
938
- "step": 77500
939
- },
940
- {
941
- "epoch": 0.86,
942
- "learning_rate": 3.5718052381789416e-05,
943
- "loss": 0.8942,
944
- "step": 78000
945
- },
946
- {
947
- "epoch": 0.86,
948
- "learning_rate": 3.562650143551884e-05,
949
- "loss": 0.8932,
950
- "step": 78500
951
- },
952
- {
953
- "epoch": 0.87,
954
- "learning_rate": 3.553495048924826e-05,
955
- "loss": 0.8915,
956
- "step": 79000
957
- },
958
- {
959
- "epoch": 0.87,
960
- "learning_rate": 3.544339954297768e-05,
961
- "loss": 0.8794,
962
- "step": 79500
963
- },
964
- {
965
- "epoch": 0.88,
966
- "learning_rate": 3.53518485967071e-05,
967
- "loss": 0.8998,
968
- "step": 80000
969
- },
970
- {
971
- "epoch": 0.88,
972
- "learning_rate": 3.5260297650436516e-05,
973
- "loss": 0.8926,
974
- "step": 80500
975
- },
976
- {
977
- "epoch": 0.89,
978
- "learning_rate": 3.516874670416593e-05,
979
- "loss": 0.8897,
980
- "step": 81000
981
- },
982
- {
983
- "epoch": 0.9,
984
- "learning_rate": 3.507719575789536e-05,
985
- "loss": 0.8856,
986
- "step": 81500
987
- },
988
- {
989
- "epoch": 0.9,
990
- "learning_rate": 3.4985644811624774e-05,
991
- "loss": 0.8958,
992
- "step": 82000
993
- },
994
- {
995
- "epoch": 0.91,
996
- "learning_rate": 3.489409386535419e-05,
997
- "loss": 0.8901,
998
- "step": 82500
999
- },
1000
- {
1001
- "epoch": 0.91,
1002
- "learning_rate": 3.4802542919083615e-05,
1003
- "loss": 0.8843,
1004
- "step": 83000
1005
- },
1006
- {
1007
- "epoch": 0.92,
1008
- "learning_rate": 3.471099197281303e-05,
1009
- "loss": 0.8991,
1010
- "step": 83500
1011
- },
1012
- {
1013
- "epoch": 0.92,
1014
- "learning_rate": 3.4619441026542456e-05,
1015
- "loss": 0.8977,
1016
- "step": 84000
1017
- },
1018
- {
1019
- "epoch": 0.93,
1020
- "learning_rate": 3.4527890080271874e-05,
1021
- "loss": 0.8917,
1022
- "step": 84500
1023
- },
1024
- {
1025
- "epoch": 0.93,
1026
- "learning_rate": 3.443633913400129e-05,
1027
- "loss": 0.9006,
1028
- "step": 85000
1029
- },
1030
- {
1031
- "epoch": 0.94,
1032
- "learning_rate": 3.434478818773071e-05,
1033
- "loss": 0.8889,
1034
- "step": 85500
1035
- },
1036
- {
1037
- "epoch": 0.94,
1038
- "learning_rate": 3.425323724146013e-05,
1039
- "loss": 0.8955,
1040
- "step": 86000
1041
- },
1042
- {
1043
- "epoch": 0.95,
1044
- "learning_rate": 3.416168629518955e-05,
1045
- "loss": 0.898,
1046
- "step": 86500
1047
- },
1048
- {
1049
- "epoch": 0.96,
1050
- "learning_rate": 3.4070135348918966e-05,
1051
- "loss": 0.8845,
1052
- "step": 87000
1053
- },
1054
- {
1055
- "epoch": 0.96,
1056
- "learning_rate": 3.397858440264839e-05,
1057
- "loss": 0.8898,
1058
- "step": 87500
1059
- },
1060
- {
1061
- "epoch": 0.97,
1062
- "learning_rate": 3.388703345637781e-05,
1063
- "loss": 0.8933,
1064
- "step": 88000
1065
- },
1066
- {
1067
- "epoch": 0.97,
1068
- "learning_rate": 3.3795482510107225e-05,
1069
- "loss": 0.8903,
1070
- "step": 88500
1071
- },
1072
- {
1073
- "epoch": 0.98,
1074
- "learning_rate": 3.370393156383664e-05,
1075
- "loss": 0.8931,
1076
- "step": 89000
1077
- },
1078
- {
1079
- "epoch": 0.98,
1080
- "learning_rate": 3.3612380617566066e-05,
1081
- "loss": 0.9005,
1082
- "step": 89500
1083
- },
1084
- {
1085
- "epoch": 0.99,
1086
- "learning_rate": 3.352082967129548e-05,
1087
- "loss": 0.8839,
1088
- "step": 90000
1089
- },
1090
- {
1091
- "epoch": 0.99,
1092
- "learning_rate": 3.342927872502491e-05,
1093
- "loss": 0.898,
1094
- "step": 90500
1095
- },
1096
- {
1097
- "epoch": 1.0,
1098
- "learning_rate": 3.3337727778754324e-05,
1099
- "loss": 0.8996,
1100
- "step": 91000
1101
- },
1102
- {
1103
- "epoch": 1.01,
1104
- "learning_rate": 3.324617683248374e-05,
1105
- "loss": 0.7676,
1106
- "step": 91500
1107
- },
1108
- {
1109
- "epoch": 1.01,
1110
- "learning_rate": 3.3154625886213165e-05,
1111
- "loss": 0.7695,
1112
- "step": 92000
1113
- },
1114
- {
1115
- "epoch": 1.02,
1116
- "learning_rate": 3.3063074939942576e-05,
1117
- "loss": 0.7604,
1118
- "step": 92500
1119
- },
1120
- {
1121
- "epoch": 1.02,
1122
- "learning_rate": 3.2971523993672e-05,
1123
- "loss": 0.761,
1124
- "step": 93000
1125
- },
1126
- {
1127
- "epoch": 1.03,
1128
- "learning_rate": 3.287997304740142e-05,
1129
- "loss": 0.7576,
1130
- "step": 93500
1131
- },
1132
- {
1133
- "epoch": 1.03,
1134
- "learning_rate": 3.278842210113084e-05,
1135
- "loss": 0.7618,
1136
- "step": 94000
1137
- },
1138
- {
1139
- "epoch": 1.04,
1140
- "learning_rate": 3.269687115486026e-05,
1141
- "loss": 0.7585,
1142
- "step": 94500
1143
- },
1144
- {
1145
- "epoch": 1.04,
1146
- "learning_rate": 3.260532020858968e-05,
1147
- "loss": 0.7636,
1148
- "step": 95000
1149
- },
1150
- {
1151
- "epoch": 1.05,
1152
- "learning_rate": 3.25137692623191e-05,
1153
- "loss": 0.7609,
1154
- "step": 95500
1155
- },
1156
- {
1157
- "epoch": 1.05,
1158
- "learning_rate": 3.2422218316048517e-05,
1159
- "loss": 0.7605,
1160
- "step": 96000
1161
- },
1162
- {
1163
- "epoch": 1.06,
1164
- "learning_rate": 3.2330667369777934e-05,
1165
- "loss": 0.7686,
1166
- "step": 96500
1167
- },
1168
- {
1169
- "epoch": 1.07,
1170
- "learning_rate": 3.223911642350735e-05,
1171
- "loss": 0.7633,
1172
- "step": 97000
1173
- },
1174
- {
1175
- "epoch": 1.07,
1176
- "learning_rate": 3.2147565477236775e-05,
1177
- "loss": 0.7644,
1178
- "step": 97500
1179
- },
1180
- {
1181
- "epoch": 1.08,
1182
- "learning_rate": 3.205601453096619e-05,
1183
- "loss": 0.7687,
1184
- "step": 98000
1185
- },
1186
- {
1187
- "epoch": 1.08,
1188
- "learning_rate": 3.1964463584695616e-05,
1189
- "loss": 0.767,
1190
- "step": 98500
1191
- },
1192
- {
1193
- "epoch": 1.09,
1194
- "learning_rate": 3.187291263842503e-05,
1195
- "loss": 0.7663,
1196
- "step": 99000
1197
- },
1198
- {
1199
- "epoch": 1.09,
1200
- "learning_rate": 3.178136169215446e-05,
1201
- "loss": 0.7657,
1202
- "step": 99500
1203
- },
1204
- {
1205
- "epoch": 1.1,
1206
- "learning_rate": 3.168981074588387e-05,
1207
- "loss": 0.7682,
1208
- "step": 100000
1209
- },
1210
- {
1211
- "epoch": 1.1,
1212
- "learning_rate": 3.159825979961329e-05,
1213
- "loss": 0.7721,
1214
- "step": 100500
1215
- },
1216
- {
1217
- "epoch": 1.11,
1218
- "learning_rate": 3.150670885334271e-05,
1219
- "loss": 0.7673,
1220
- "step": 101000
1221
- },
1222
- {
1223
- "epoch": 1.12,
1224
- "learning_rate": 3.1415157907072126e-05,
1225
- "loss": 0.7649,
1226
- "step": 101500
1227
- },
1228
- {
1229
- "epoch": 1.12,
1230
- "learning_rate": 3.132360696080155e-05,
1231
- "loss": 0.7737,
1232
- "step": 102000
1233
- },
1234
- {
1235
- "epoch": 1.13,
1236
- "learning_rate": 3.123205601453097e-05,
1237
- "loss": 0.7688,
1238
- "step": 102500
1239
- },
1240
- {
1241
- "epoch": 1.13,
1242
- "learning_rate": 3.114050506826039e-05,
1243
- "loss": 0.7705,
1244
- "step": 103000
1245
- },
1246
- {
1247
- "epoch": 1.14,
1248
- "learning_rate": 3.10489541219898e-05,
1249
- "loss": 0.764,
1250
- "step": 103500
1251
- },
1252
- {
1253
- "epoch": 1.14,
1254
- "learning_rate": 3.0957403175719225e-05,
1255
- "loss": 0.7654,
1256
- "step": 104000
1257
- },
1258
- {
1259
- "epoch": 1.15,
1260
- "learning_rate": 3.086585222944864e-05,
1261
- "loss": 0.7762,
1262
- "step": 104500
1263
- },
1264
- {
1265
- "epoch": 1.15,
1266
- "learning_rate": 3.0774301283178067e-05,
1267
- "loss": 0.7708,
1268
- "step": 105000
1269
- },
1270
- {
1271
- "epoch": 1.16,
1272
- "learning_rate": 3.0682750336907484e-05,
1273
- "loss": 0.778,
1274
- "step": 105500
1275
- },
1276
- {
1277
- "epoch": 1.16,
1278
- "learning_rate": 3.05911993906369e-05,
1279
- "loss": 0.7741,
1280
- "step": 106000
1281
- },
1282
- {
1283
- "epoch": 1.17,
1284
- "learning_rate": 3.0499648444366325e-05,
1285
- "loss": 0.7714,
1286
- "step": 106500
1287
- },
1288
- {
1289
- "epoch": 1.18,
1290
- "learning_rate": 3.0408097498095746e-05,
1291
- "loss": 0.774,
1292
- "step": 107000
1293
- },
1294
- {
1295
- "epoch": 1.18,
1296
- "learning_rate": 3.031654655182516e-05,
1297
- "loss": 0.7704,
1298
- "step": 107500
1299
- },
1300
- {
1301
- "epoch": 1.19,
1302
- "learning_rate": 3.022499560555458e-05,
1303
- "loss": 0.7797,
1304
- "step": 108000
1305
- },
1306
- {
1307
- "epoch": 1.19,
1308
- "learning_rate": 3.0133444659284e-05,
1309
- "loss": 0.7647,
1310
- "step": 108500
1311
- },
1312
- {
1313
- "epoch": 1.2,
1314
- "learning_rate": 3.0041893713013418e-05,
1315
- "loss": 0.774,
1316
- "step": 109000
1317
- },
1318
- {
1319
- "epoch": 1.2,
1320
- "learning_rate": 2.9950342766742838e-05,
1321
- "loss": 0.7714,
1322
- "step": 109500
1323
- },
1324
- {
1325
- "epoch": 1.21,
1326
- "learning_rate": 2.985879182047226e-05,
1327
- "loss": 0.7638,
1328
- "step": 110000
1329
- },
1330
- {
1331
- "epoch": 1.21,
1332
- "learning_rate": 2.976724087420168e-05,
1333
- "loss": 0.7724,
1334
- "step": 110500
1335
- },
1336
- {
1337
- "epoch": 1.22,
1338
- "learning_rate": 2.9675689927931093e-05,
1339
- "loss": 0.7723,
1340
- "step": 111000
1341
- },
1342
- {
1343
- "epoch": 1.22,
1344
- "learning_rate": 2.9584138981660514e-05,
1345
- "loss": 0.767,
1346
- "step": 111500
1347
- },
1348
- {
1349
- "epoch": 1.23,
1350
- "learning_rate": 2.9492588035389934e-05,
1351
- "loss": 0.7745,
1352
- "step": 112000
1353
- },
1354
- {
1355
- "epoch": 1.24,
1356
- "learning_rate": 2.9401037089119355e-05,
1357
- "loss": 0.7657,
1358
- "step": 112500
1359
- },
1360
- {
1361
- "epoch": 1.24,
1362
- "learning_rate": 2.9309486142848776e-05,
1363
- "loss": 0.7706,
1364
- "step": 113000
1365
- },
1366
- {
1367
- "epoch": 1.25,
1368
- "learning_rate": 2.9217935196578196e-05,
1369
- "loss": 0.7767,
1370
- "step": 113500
1371
- },
1372
- {
1373
- "epoch": 1.25,
1374
- "learning_rate": 2.9126384250307613e-05,
1375
- "loss": 0.7777,
1376
- "step": 114000
1377
- },
1378
- {
1379
- "epoch": 1.26,
1380
- "learning_rate": 2.9034833304037034e-05,
1381
- "loss": 0.7705,
1382
- "step": 114500
1383
- },
1384
- {
1385
- "epoch": 1.26,
1386
- "learning_rate": 2.8943282357766448e-05,
1387
- "loss": 0.7829,
1388
- "step": 115000
1389
- },
1390
- {
1391
- "epoch": 1.27,
1392
- "learning_rate": 2.8851731411495868e-05,
1393
- "loss": 0.775,
1394
- "step": 115500
1395
- },
1396
- {
1397
- "epoch": 1.27,
1398
- "learning_rate": 2.876018046522529e-05,
1399
- "loss": 0.7815,
1400
- "step": 116000
1401
- },
1402
- {
1403
- "epoch": 1.28,
1404
- "learning_rate": 2.866862951895471e-05,
1405
- "loss": 0.7703,
1406
- "step": 116500
1407
- },
1408
- {
1409
- "epoch": 1.29,
1410
- "learning_rate": 2.857707857268413e-05,
1411
- "loss": 0.7736,
1412
- "step": 117000
1413
- },
1414
- {
1415
- "epoch": 1.29,
1416
- "learning_rate": 2.848552762641355e-05,
1417
- "loss": 0.7747,
1418
- "step": 117500
1419
- },
1420
- {
1421
- "epoch": 1.3,
1422
- "learning_rate": 2.839397668014297e-05,
1423
- "loss": 0.7707,
1424
- "step": 118000
1425
- },
1426
- {
1427
- "epoch": 1.3,
1428
- "learning_rate": 2.8302425733872385e-05,
1429
- "loss": 0.7666,
1430
- "step": 118500
1431
- },
1432
- {
1433
- "epoch": 1.31,
1434
- "learning_rate": 2.8210874787601806e-05,
1435
- "loss": 0.7867,
1436
- "step": 119000
1437
- },
1438
- {
1439
- "epoch": 1.31,
1440
- "learning_rate": 2.8119323841331223e-05,
1441
- "loss": 0.7759,
1442
- "step": 119500
1443
- },
1444
- {
1445
- "epoch": 1.32,
1446
- "learning_rate": 2.8027772895060643e-05,
1447
- "loss": 0.7743,
1448
- "step": 120000
1449
- },
1450
- {
1451
- "epoch": 1.32,
1452
- "learning_rate": 2.7936221948790064e-05,
1453
- "loss": 0.7708,
1454
- "step": 120500
1455
- },
1456
- {
1457
- "epoch": 1.33,
1458
- "learning_rate": 2.7844671002519484e-05,
1459
- "loss": 0.7725,
1460
- "step": 121000
1461
- },
1462
- {
1463
- "epoch": 1.33,
1464
- "learning_rate": 2.7753120056248905e-05,
1465
- "loss": 0.7687,
1466
- "step": 121500
1467
- },
1468
- {
1469
- "epoch": 1.34,
1470
- "learning_rate": 2.7661569109978326e-05,
1471
- "loss": 0.7779,
1472
- "step": 122000
1473
- },
1474
- {
1475
- "epoch": 1.35,
1476
- "learning_rate": 2.757001816370774e-05,
1477
- "loss": 0.774,
1478
- "step": 122500
1479
- },
1480
- {
1481
- "epoch": 1.35,
1482
- "learning_rate": 2.747846721743716e-05,
1483
- "loss": 0.7671,
1484
- "step": 123000
1485
- },
1486
- {
1487
- "epoch": 1.36,
1488
- "learning_rate": 2.738691627116658e-05,
1489
- "loss": 0.7721,
1490
- "step": 123500
1491
- },
1492
- {
1493
- "epoch": 1.36,
1494
- "learning_rate": 2.7295365324896e-05,
1495
- "loss": 0.7758,
1496
- "step": 124000
1497
- },
1498
- {
1499
- "epoch": 1.37,
1500
- "learning_rate": 2.720381437862542e-05,
1501
- "loss": 0.7689,
1502
- "step": 124500
1503
- },
1504
- {
1505
- "epoch": 1.37,
1506
- "learning_rate": 2.711226343235484e-05,
1507
- "loss": 0.773,
1508
- "step": 125000
1509
- },
1510
- {
1511
- "epoch": 1.38,
1512
- "learning_rate": 2.702071248608426e-05,
1513
- "loss": 0.7776,
1514
- "step": 125500
1515
- },
1516
- {
1517
- "epoch": 1.38,
1518
- "learning_rate": 2.6929161539813673e-05,
1519
- "loss": 0.7724,
1520
- "step": 126000
1521
- },
1522
- {
1523
- "epoch": 1.39,
1524
- "learning_rate": 2.6837610593543094e-05,
1525
- "loss": 0.7885,
1526
- "step": 126500
1527
- },
1528
- {
1529
- "epoch": 1.4,
1530
- "learning_rate": 2.6746059647272515e-05,
1531
- "loss": 0.7737,
1532
- "step": 127000
1533
- },
1534
- {
1535
- "epoch": 1.4,
1536
- "learning_rate": 2.6654508701001935e-05,
1537
- "loss": 0.7715,
1538
- "step": 127500
1539
- },
1540
- {
1541
- "epoch": 1.41,
1542
- "learning_rate": 2.6562957754731356e-05,
1543
- "loss": 0.7709,
1544
- "step": 128000
1545
- },
1546
- {
1547
- "epoch": 1.41,
1548
- "learning_rate": 2.6471406808460776e-05,
1549
- "loss": 0.7732,
1550
- "step": 128500
1551
- },
1552
- {
1553
- "epoch": 1.42,
1554
- "learning_rate": 2.6379855862190193e-05,
1555
- "loss": 0.7802,
1556
- "step": 129000
1557
- },
1558
- {
1559
- "epoch": 1.42,
1560
- "learning_rate": 2.6288304915919614e-05,
1561
- "loss": 0.7736,
1562
- "step": 129500
1563
- },
1564
- {
1565
- "epoch": 1.43,
1566
- "learning_rate": 2.6196753969649028e-05,
1567
- "loss": 0.782,
1568
- "step": 130000
1569
- },
1570
- {
1571
- "epoch": 1.43,
1572
- "learning_rate": 2.610520302337845e-05,
1573
- "loss": 0.7728,
1574
- "step": 130500
1575
- },
1576
- {
1577
- "epoch": 1.44,
1578
- "learning_rate": 2.601365207710787e-05,
1579
- "loss": 0.7773,
1580
- "step": 131000
1581
- },
1582
- {
1583
- "epoch": 1.44,
1584
- "learning_rate": 2.592210113083729e-05,
1585
- "loss": 0.7658,
1586
- "step": 131500
1587
- },
1588
- {
1589
- "epoch": 1.45,
1590
- "learning_rate": 2.583055018456671e-05,
1591
- "loss": 0.7782,
1592
- "step": 132000
1593
- },
1594
- {
1595
- "epoch": 1.46,
1596
- "learning_rate": 2.573899923829613e-05,
1597
- "loss": 0.7794,
1598
- "step": 132500
1599
- },
1600
- {
1601
- "epoch": 1.46,
1602
- "learning_rate": 2.564744829202555e-05,
1603
- "loss": 0.7806,
1604
- "step": 133000
1605
- },
1606
- {
1607
- "epoch": 1.47,
1608
- "learning_rate": 2.5555897345754965e-05,
1609
- "loss": 0.7708,
1610
- "step": 133500
1611
- },
1612
- {
1613
- "epoch": 1.47,
1614
- "learning_rate": 2.5464346399484386e-05,
1615
- "loss": 0.7712,
1616
- "step": 134000
1617
- },
1618
- {
1619
- "epoch": 1.48,
1620
- "learning_rate": 2.5372795453213806e-05,
1621
- "loss": 0.781,
1622
- "step": 134500
1623
- },
1624
- {
1625
- "epoch": 1.48,
1626
- "learning_rate": 2.5281244506943223e-05,
1627
- "loss": 0.7807,
1628
- "step": 135000
1629
- },
1630
- {
1631
- "epoch": 1.49,
1632
- "learning_rate": 2.5189693560672644e-05,
1633
- "loss": 0.7862,
1634
- "step": 135500
1635
- },
1636
- {
1637
- "epoch": 1.49,
1638
- "learning_rate": 2.5098142614402065e-05,
1639
- "loss": 0.7796,
1640
- "step": 136000
1641
- },
1642
- {
1643
- "epoch": 1.5,
1644
- "learning_rate": 2.5006591668131485e-05,
1645
- "loss": 0.7842,
1646
- "step": 136500
1647
- },
1648
- {
1649
- "epoch": 1.51,
1650
- "learning_rate": 2.4915040721860902e-05,
1651
- "loss": 0.782,
1652
- "step": 137000
1653
- },
1654
- {
1655
- "epoch": 1.51,
1656
- "learning_rate": 2.4823489775590323e-05,
1657
- "loss": 0.7808,
1658
- "step": 137500
1659
- },
1660
- {
1661
- "epoch": 1.52,
1662
- "learning_rate": 2.4731938829319744e-05,
1663
- "loss": 0.7748,
1664
- "step": 138000
1665
- },
1666
- {
1667
- "epoch": 1.52,
1668
- "learning_rate": 2.464038788304916e-05,
1669
- "loss": 0.7783,
1670
- "step": 138500
1671
- },
1672
- {
1673
- "epoch": 1.53,
1674
- "learning_rate": 2.454883693677858e-05,
1675
- "loss": 0.7832,
1676
- "step": 139000
1677
- },
1678
- {
1679
- "epoch": 1.53,
1680
- "learning_rate": 2.4457285990508e-05,
1681
- "loss": 0.7795,
1682
- "step": 139500
1683
- },
1684
- {
1685
- "epoch": 1.54,
1686
- "learning_rate": 2.4365735044237416e-05,
1687
- "loss": 0.7836,
1688
- "step": 140000
1689
- },
1690
- {
1691
- "epoch": 1.54,
1692
- "learning_rate": 2.4274184097966836e-05,
1693
- "loss": 0.7779,
1694
- "step": 140500
1695
- },
1696
- {
1697
- "epoch": 1.55,
1698
- "learning_rate": 2.4182633151696257e-05,
1699
- "loss": 0.7813,
1700
- "step": 141000
1701
- },
1702
- {
1703
- "epoch": 1.55,
1704
- "learning_rate": 2.4091082205425677e-05,
1705
- "loss": 0.7818,
1706
- "step": 141500
1707
- },
1708
- {
1709
- "epoch": 1.56,
1710
- "learning_rate": 2.3999531259155095e-05,
1711
- "loss": 0.7813,
1712
- "step": 142000
1713
- },
1714
- {
1715
- "epoch": 1.57,
1716
- "learning_rate": 2.3907980312884515e-05,
1717
- "loss": 0.7787,
1718
- "step": 142500
1719
- },
1720
- {
1721
- "epoch": 1.57,
1722
- "learning_rate": 2.3816429366613936e-05,
1723
- "loss": 0.784,
1724
- "step": 143000
1725
- },
1726
- {
1727
- "epoch": 1.58,
1728
- "learning_rate": 2.3724878420343356e-05,
1729
- "loss": 0.7923,
1730
- "step": 143500
1731
- },
1732
- {
1733
- "epoch": 1.58,
1734
- "learning_rate": 2.3633327474072774e-05,
1735
- "loss": 0.7752,
1736
- "step": 144000
1737
- },
1738
- {
1739
- "epoch": 1.59,
1740
- "learning_rate": 2.354177652780219e-05,
1741
- "loss": 0.7755,
1742
- "step": 144500
1743
- },
1744
- {
1745
- "epoch": 1.59,
1746
- "learning_rate": 2.345022558153161e-05,
1747
- "loss": 0.7859,
1748
- "step": 145000
1749
- },
1750
- {
1751
- "epoch": 1.6,
1752
- "learning_rate": 2.3358674635261032e-05,
1753
- "loss": 0.7763,
1754
- "step": 145500
1755
- },
1756
- {
1757
- "epoch": 1.6,
1758
- "learning_rate": 2.326712368899045e-05,
1759
- "loss": 0.7812,
1760
- "step": 146000
1761
- },
1762
- {
1763
- "epoch": 1.61,
1764
- "learning_rate": 2.317557274271987e-05,
1765
- "loss": 0.7764,
1766
- "step": 146500
1767
- },
1768
- {
1769
- "epoch": 1.61,
1770
- "learning_rate": 2.308402179644929e-05,
1771
- "loss": 0.7743,
1772
- "step": 147000
1773
- },
1774
- {
1775
- "epoch": 1.62,
1776
- "learning_rate": 2.2992470850178707e-05,
1777
- "loss": 0.7861,
1778
- "step": 147500
1779
- },
1780
- {
1781
- "epoch": 1.63,
1782
- "learning_rate": 2.2900919903908128e-05,
1783
- "loss": 0.7746,
1784
- "step": 148000
1785
- },
1786
- {
1787
- "epoch": 1.63,
1788
- "learning_rate": 2.280936895763755e-05,
1789
- "loss": 0.7802,
1790
- "step": 148500
1791
- },
1792
- {
1793
- "epoch": 1.64,
1794
- "learning_rate": 2.271781801136697e-05,
1795
- "loss": 0.7859,
1796
- "step": 149000
1797
- },
1798
- {
1799
- "epoch": 1.64,
1800
- "learning_rate": 2.2626267065096386e-05,
1801
- "loss": 0.7774,
1802
- "step": 149500
1803
- },
1804
- {
1805
- "epoch": 1.65,
1806
- "learning_rate": 2.2534716118825804e-05,
1807
- "loss": 0.779,
1808
- "step": 150000
1809
- },
1810
- {
1811
- "epoch": 1.65,
1812
- "learning_rate": 2.2443165172555224e-05,
1813
- "loss": 0.7815,
1814
- "step": 150500
1815
- },
1816
- {
1817
- "epoch": 1.66,
1818
- "learning_rate": 2.2351614226284645e-05,
1819
- "loss": 0.7847,
1820
- "step": 151000
1821
- },
1822
- {
1823
- "epoch": 1.66,
1824
- "learning_rate": 2.2260063280014062e-05,
1825
- "loss": 0.7717,
1826
- "step": 151500
1827
- },
1828
- {
1829
- "epoch": 1.67,
1830
- "learning_rate": 2.2168512333743483e-05,
1831
- "loss": 0.7701,
1832
- "step": 152000
1833
- },
1834
- {
1835
- "epoch": 1.68,
1836
- "learning_rate": 2.2076961387472903e-05,
1837
- "loss": 0.7807,
1838
- "step": 152500
1839
- },
1840
- {
1841
- "epoch": 1.68,
1842
- "learning_rate": 2.1985410441202324e-05,
1843
- "loss": 0.7786,
1844
- "step": 153000
1845
- },
1846
- {
1847
- "epoch": 1.69,
1848
- "learning_rate": 2.189385949493174e-05,
1849
- "loss": 0.7894,
1850
- "step": 153500
1851
- },
1852
- {
1853
- "epoch": 1.69,
1854
- "learning_rate": 2.180230854866116e-05,
1855
- "loss": 0.7766,
1856
- "step": 154000
1857
- },
1858
- {
1859
- "epoch": 1.7,
1860
- "learning_rate": 2.171075760239058e-05,
1861
- "loss": 0.7792,
1862
- "step": 154500
1863
- },
1864
- {
1865
- "epoch": 1.7,
1866
- "learning_rate": 2.1619206656119996e-05,
1867
- "loss": 0.7785,
1868
- "step": 155000
1869
- },
1870
- {
1871
- "epoch": 1.71,
1872
- "learning_rate": 2.1527655709849416e-05,
1873
- "loss": 0.7795,
1874
- "step": 155500
1875
- },
1876
- {
1877
- "epoch": 1.71,
1878
- "learning_rate": 2.1436104763578837e-05,
1879
- "loss": 0.7835,
1880
- "step": 156000
1881
- },
1882
- {
1883
- "epoch": 1.72,
1884
- "learning_rate": 2.1344553817308258e-05,
1885
- "loss": 0.7737,
1886
- "step": 156500
1887
- },
1888
- {
1889
- "epoch": 1.72,
1890
- "learning_rate": 2.1253002871037675e-05,
1891
- "loss": 0.785,
1892
- "step": 157000
1893
- },
1894
- {
1895
- "epoch": 1.73,
1896
- "learning_rate": 2.1161451924767095e-05,
1897
- "loss": 0.7859,
1898
- "step": 157500
1899
- },
1900
- {
1901
- "epoch": 1.74,
1902
- "learning_rate": 2.1069900978496516e-05,
1903
- "loss": 0.7842,
1904
- "step": 158000
1905
- },
1906
- {
1907
- "epoch": 1.74,
1908
- "learning_rate": 2.0978350032225936e-05,
1909
- "loss": 0.7808,
1910
- "step": 158500
1911
- },
1912
- {
1913
- "epoch": 1.75,
1914
- "learning_rate": 2.0886799085955354e-05,
1915
- "loss": 0.7732,
1916
- "step": 159000
1917
- },
1918
- {
1919
- "epoch": 1.75,
1920
- "learning_rate": 2.0795248139684774e-05,
1921
- "loss": 0.7785,
1922
- "step": 159500
1923
- },
1924
- {
1925
- "epoch": 1.76,
1926
- "learning_rate": 2.070369719341419e-05,
1927
- "loss": 0.7794,
1928
- "step": 160000
1929
- },
1930
- {
1931
- "epoch": 1.76,
1932
- "learning_rate": 2.0612146247143612e-05,
1933
- "loss": 0.7893,
1934
- "step": 160500
1935
- },
1936
- {
1937
- "epoch": 1.77,
1938
- "learning_rate": 2.052059530087303e-05,
1939
- "loss": 0.779,
1940
- "step": 161000
1941
- },
1942
- {
1943
- "epoch": 1.77,
1944
- "learning_rate": 2.042904435460245e-05,
1945
- "loss": 0.7792,
1946
- "step": 161500
1947
- },
1948
- {
1949
- "epoch": 1.78,
1950
- "learning_rate": 2.033749340833187e-05,
1951
- "loss": 0.7902,
1952
- "step": 162000
1953
- },
1954
- {
1955
- "epoch": 1.79,
1956
- "learning_rate": 2.0245942462061288e-05,
1957
- "loss": 0.7835,
1958
- "step": 162500
1959
- },
1960
- {
1961
- "epoch": 1.79,
1962
- "learning_rate": 2.0154391515790708e-05,
1963
- "loss": 0.7831,
1964
- "step": 163000
1965
- },
1966
- {
1967
- "epoch": 1.8,
1968
- "learning_rate": 2.006284056952013e-05,
1969
- "loss": 0.7852,
1970
- "step": 163500
1971
- },
1972
- {
1973
- "epoch": 1.8,
1974
- "learning_rate": 1.997128962324955e-05,
1975
- "loss": 0.7855,
1976
- "step": 164000
1977
- },
1978
- {
1979
- "epoch": 1.81,
1980
- "learning_rate": 1.9879738676978966e-05,
1981
- "loss": 0.7821,
1982
- "step": 164500
1983
- },
1984
- {
1985
- "epoch": 1.81,
1986
- "learning_rate": 1.9788187730708384e-05,
1987
- "loss": 0.7762,
1988
- "step": 165000
1989
- },
1990
- {
1991
- "epoch": 1.82,
1992
- "learning_rate": 1.9696636784437804e-05,
1993
- "loss": 0.7852,
1994
- "step": 165500
1995
- },
1996
- {
1997
- "epoch": 1.82,
1998
- "learning_rate": 1.9605085838167225e-05,
1999
- "loss": 0.7857,
2000
- "step": 166000
2001
- },
2002
- {
2003
- "epoch": 1.83,
2004
- "learning_rate": 1.9513534891896642e-05,
2005
- "loss": 0.7872,
2006
- "step": 166500
2007
- },
2008
- {
2009
- "epoch": 1.83,
2010
- "learning_rate": 1.9421983945626063e-05,
2011
- "loss": 0.7749,
2012
- "step": 167000
2013
- },
2014
- {
2015
- "epoch": 1.84,
2016
- "learning_rate": 1.9330432999355483e-05,
2017
- "loss": 0.7867,
2018
- "step": 167500
2019
- },
2020
- {
2021
- "epoch": 1.85,
2022
- "learning_rate": 1.9238882053084904e-05,
2023
- "loss": 0.7801,
2024
- "step": 168000
2025
- },
2026
- {
2027
- "epoch": 1.85,
2028
- "learning_rate": 1.914733110681432e-05,
2029
- "loss": 0.7734,
2030
- "step": 168500
2031
- },
2032
- {
2033
- "epoch": 1.86,
2034
- "learning_rate": 1.905578016054374e-05,
2035
- "loss": 0.7798,
2036
- "step": 169000
2037
- },
2038
- {
2039
- "epoch": 1.86,
2040
- "learning_rate": 1.896422921427316e-05,
2041
- "loss": 0.7862,
2042
- "step": 169500
2043
- },
2044
- {
2045
- "epoch": 1.87,
2046
- "learning_rate": 1.887267826800258e-05,
2047
- "loss": 0.788,
2048
- "step": 170000
2049
- },
2050
- {
2051
- "epoch": 1.87,
2052
- "learning_rate": 1.8781127321731997e-05,
2053
- "loss": 0.7863,
2054
- "step": 170500
2055
- },
2056
- {
2057
- "epoch": 1.88,
2058
- "learning_rate": 1.8689576375461417e-05,
2059
- "loss": 0.7865,
2060
- "step": 171000
2061
- },
2062
- {
2063
- "epoch": 1.88,
2064
- "learning_rate": 1.8598025429190838e-05,
2065
- "loss": 0.79,
2066
- "step": 171500
2067
- },
2068
- {
2069
- "epoch": 1.89,
2070
- "learning_rate": 1.8506474482920255e-05,
2071
- "loss": 0.7871,
2072
- "step": 172000
2073
- },
2074
- {
2075
- "epoch": 1.9,
2076
- "learning_rate": 1.8414923536649675e-05,
2077
- "loss": 0.7882,
2078
- "step": 172500
2079
- },
2080
- {
2081
- "epoch": 1.9,
2082
- "learning_rate": 1.8323372590379096e-05,
2083
- "loss": 0.7811,
2084
- "step": 173000
2085
- },
2086
- {
2087
- "epoch": 1.91,
2088
- "learning_rate": 1.8231821644108517e-05,
2089
- "loss": 0.7827,
2090
- "step": 173500
2091
- },
2092
- {
2093
- "epoch": 1.91,
2094
- "learning_rate": 1.8140270697837934e-05,
2095
- "loss": 0.7741,
2096
- "step": 174000
2097
- },
2098
- {
2099
- "epoch": 1.92,
2100
- "learning_rate": 1.8048719751567354e-05,
2101
- "loss": 0.7743,
2102
- "step": 174500
2103
- },
2104
- {
2105
- "epoch": 1.92,
2106
- "learning_rate": 1.795716880529677e-05,
2107
- "loss": 0.785,
2108
- "step": 175000
2109
- },
2110
- {
2111
- "epoch": 1.93,
2112
- "learning_rate": 1.7865617859026192e-05,
2113
- "loss": 0.7774,
2114
- "step": 175500
2115
- },
2116
- {
2117
- "epoch": 1.93,
2118
- "learning_rate": 1.777406691275561e-05,
2119
- "loss": 0.7921,
2120
- "step": 176000
2121
- },
2122
- {
2123
- "epoch": 1.94,
2124
- "learning_rate": 1.768251596648503e-05,
2125
- "loss": 0.7827,
2126
- "step": 176500
2127
- },
2128
- {
2129
- "epoch": 1.94,
2130
- "learning_rate": 1.759096502021445e-05,
2131
- "loss": 0.7891,
2132
- "step": 177000
2133
- },
2134
- {
2135
- "epoch": 1.95,
2136
- "learning_rate": 1.7499414073943868e-05,
2137
- "loss": 0.7873,
2138
- "step": 177500
2139
- },
2140
- {
2141
- "epoch": 1.96,
2142
- "learning_rate": 1.7407863127673288e-05,
2143
- "loss": 0.7817,
2144
- "step": 178000
2145
- },
2146
- {
2147
- "epoch": 1.96,
2148
- "learning_rate": 1.731631218140271e-05,
2149
- "loss": 0.7802,
2150
- "step": 178500
2151
- },
2152
- {
2153
- "epoch": 1.97,
2154
- "learning_rate": 1.722476123513213e-05,
2155
- "loss": 0.7864,
2156
- "step": 179000
2157
- },
2158
- {
2159
- "epoch": 1.97,
2160
- "learning_rate": 1.7133210288861547e-05,
2161
- "loss": 0.7872,
2162
- "step": 179500
2163
- },
2164
- {
2165
- "epoch": 1.98,
2166
- "learning_rate": 1.7041659342590964e-05,
2167
- "loss": 0.7952,
2168
- "step": 180000
2169
- },
2170
- {
2171
- "epoch": 1.98,
2172
- "learning_rate": 1.6950108396320384e-05,
2173
- "loss": 0.7861,
2174
- "step": 180500
2175
- },
2176
- {
2177
- "epoch": 1.99,
2178
- "learning_rate": 1.6858557450049805e-05,
2179
- "loss": 0.7835,
2180
- "step": 181000
2181
- },
2182
- {
2183
- "epoch": 1.99,
2184
- "learning_rate": 1.6767006503779222e-05,
2185
- "loss": 0.7922,
2186
- "step": 181500
2187
- },
2188
- {
2189
- "epoch": 2.0,
2190
- "learning_rate": 1.6675455557508643e-05,
2191
- "loss": 0.7884,
2192
- "step": 182000
2193
- },
2194
- {
2195
- "epoch": 2.0,
2196
- "learning_rate": 1.6583904611238063e-05,
2197
- "loss": 0.6872,
2198
- "step": 182500
2199
- },
2200
- {
2201
- "epoch": 2.01,
2202
- "learning_rate": 1.6492353664967484e-05,
2203
- "loss": 0.6674,
2204
- "step": 183000
2205
- },
2206
- {
2207
- "epoch": 2.02,
2208
- "learning_rate": 1.64008027186969e-05,
2209
- "loss": 0.6726,
2210
- "step": 183500
2211
- },
2212
- {
2213
- "epoch": 2.02,
2214
- "learning_rate": 1.630925177242632e-05,
2215
- "loss": 0.6642,
2216
- "step": 184000
2217
- },
2218
- {
2219
- "epoch": 2.03,
2220
- "learning_rate": 1.6217700826155742e-05,
2221
- "loss": 0.6699,
2222
- "step": 184500
2223
- },
2224
- {
2225
- "epoch": 2.03,
2226
- "learning_rate": 1.612614987988516e-05,
2227
- "loss": 0.674,
2228
- "step": 185000
2229
- },
2230
- {
2231
- "epoch": 2.04,
2232
- "learning_rate": 1.6034598933614577e-05,
2233
- "loss": 0.6681,
2234
- "step": 185500
2235
- },
2236
- {
2237
- "epoch": 2.04,
2238
- "learning_rate": 1.5943047987343997e-05,
2239
- "loss": 0.6729,
2240
- "step": 186000
2241
- },
2242
- {
2243
- "epoch": 2.05,
2244
- "learning_rate": 1.5851497041073418e-05,
2245
- "loss": 0.671,
2246
- "step": 186500
2247
- },
2248
- {
2249
- "epoch": 2.05,
2250
- "learning_rate": 1.5759946094802835e-05,
2251
- "loss": 0.6731,
2252
- "step": 187000
2253
- },
2254
- {
2255
- "epoch": 2.06,
2256
- "learning_rate": 1.5668395148532256e-05,
2257
- "loss": 0.6765,
2258
- "step": 187500
2259
- },
2260
- {
2261
- "epoch": 2.07,
2262
- "learning_rate": 1.5576844202261676e-05,
2263
- "loss": 0.6757,
2264
- "step": 188000
2265
- },
2266
- {
2267
- "epoch": 2.07,
2268
- "learning_rate": 1.5485293255991097e-05,
2269
- "loss": 0.6647,
2270
- "step": 188500
2271
- },
2272
- {
2273
- "epoch": 2.08,
2274
- "learning_rate": 1.5393742309720514e-05,
2275
- "loss": 0.6795,
2276
- "step": 189000
2277
- },
2278
- {
2279
- "epoch": 2.08,
2280
- "learning_rate": 1.5302191363449934e-05,
2281
- "loss": 0.6673,
2282
- "step": 189500
2283
- },
2284
- {
2285
- "epoch": 2.09,
2286
- "learning_rate": 1.5210640417179353e-05,
2287
- "loss": 0.6701,
2288
- "step": 190000
2289
- },
2290
- {
2291
- "epoch": 2.09,
2292
- "learning_rate": 1.5119089470908774e-05,
2293
- "loss": 0.6726,
2294
- "step": 190500
2295
- },
2296
- {
2297
- "epoch": 2.1,
2298
- "learning_rate": 1.5027538524638191e-05,
2299
- "loss": 0.6646,
2300
- "step": 191000
2301
- },
2302
- {
2303
- "epoch": 2.1,
2304
- "learning_rate": 1.493598757836761e-05,
2305
- "loss": 0.6707,
2306
- "step": 191500
2307
- },
2308
- {
2309
- "epoch": 2.11,
2310
- "learning_rate": 1.484443663209703e-05,
2311
- "loss": 0.6727,
2312
- "step": 192000
2313
- },
2314
- {
2315
- "epoch": 2.11,
2316
- "learning_rate": 1.4752885685826448e-05,
2317
- "loss": 0.6799,
2318
- "step": 192500
2319
- },
2320
- {
2321
- "epoch": 2.12,
2322
- "learning_rate": 1.4661334739555868e-05,
2323
- "loss": 0.6755,
2324
- "step": 193000
2325
- },
2326
- {
2327
- "epoch": 2.13,
2328
- "learning_rate": 1.4569783793285289e-05,
2329
- "loss": 0.6706,
2330
- "step": 193500
2331
- },
2332
- {
2333
- "epoch": 2.13,
2334
- "learning_rate": 1.4478232847014708e-05,
2335
- "loss": 0.6699,
2336
- "step": 194000
2337
- },
2338
- {
2339
- "epoch": 2.14,
2340
- "learning_rate": 1.4386681900744125e-05,
2341
- "loss": 0.6771,
2342
- "step": 194500
2343
- },
2344
- {
2345
- "epoch": 2.14,
2346
- "learning_rate": 1.4295130954473546e-05,
2347
- "loss": 0.6832,
2348
- "step": 195000
2349
- },
2350
- {
2351
- "epoch": 2.15,
2352
- "learning_rate": 1.4203580008202966e-05,
2353
- "loss": 0.6803,
2354
- "step": 195500
2355
- },
2356
- {
2357
- "epoch": 2.15,
2358
- "learning_rate": 1.4112029061932385e-05,
2359
- "loss": 0.6646,
2360
- "step": 196000
2361
- },
2362
- {
2363
- "epoch": 2.16,
2364
- "learning_rate": 1.4020478115661804e-05,
2365
- "loss": 0.6792,
2366
- "step": 196500
2367
- },
2368
- {
2369
- "epoch": 2.16,
2370
- "learning_rate": 1.3928927169391223e-05,
2371
- "loss": 0.6656,
2372
- "step": 197000
2373
- },
2374
- {
2375
- "epoch": 2.17,
2376
- "learning_rate": 1.3837376223120643e-05,
2377
- "loss": 0.6719,
2378
- "step": 197500
2379
- },
2380
- {
2381
- "epoch": 2.18,
2382
- "learning_rate": 1.3745825276850064e-05,
2383
- "loss": 0.6716,
2384
- "step": 198000
2385
- },
2386
- {
2387
- "epoch": 2.18,
2388
- "learning_rate": 1.3654274330579481e-05,
2389
- "loss": 0.6728,
2390
- "step": 198500
2391
- },
2392
- {
2393
- "epoch": 2.19,
2394
- "learning_rate": 1.35627233843089e-05,
2395
- "loss": 0.6708,
2396
- "step": 199000
2397
- },
2398
- {
2399
- "epoch": 2.19,
2400
- "learning_rate": 1.347117243803832e-05,
2401
- "loss": 0.6755,
2402
- "step": 199500
2403
- },
2404
- {
2405
- "epoch": 2.2,
2406
- "learning_rate": 1.3379621491767741e-05,
2407
- "loss": 0.6762,
2408
- "step": 200000
2409
- },
2410
- {
2411
- "epoch": 2.2,
2412
- "learning_rate": 1.3288070545497158e-05,
2413
- "loss": 0.6811,
2414
- "step": 200500
2415
- },
2416
- {
2417
- "epoch": 2.21,
2418
- "learning_rate": 1.3196519599226579e-05,
2419
- "loss": 0.676,
2420
- "step": 201000
2421
- },
2422
- {
2423
- "epoch": 2.21,
2424
- "learning_rate": 1.3104968652955998e-05,
2425
- "loss": 0.6763,
2426
- "step": 201500
2427
- },
2428
- {
2429
- "epoch": 2.22,
2430
- "learning_rate": 1.3013417706685415e-05,
2431
- "loss": 0.6701,
2432
- "step": 202000
2433
- },
2434
- {
2435
- "epoch": 2.22,
2436
- "learning_rate": 1.2921866760414836e-05,
2437
- "loss": 0.6742,
2438
- "step": 202500
2439
- },
2440
- {
2441
- "epoch": 2.23,
2442
- "learning_rate": 1.2830315814144256e-05,
2443
- "loss": 0.6745,
2444
- "step": 203000
2445
- },
2446
- {
2447
- "epoch": 2.24,
2448
- "learning_rate": 1.2738764867873675e-05,
2449
- "loss": 0.6716,
2450
- "step": 203500
2451
- },
2452
- {
2453
- "epoch": 2.24,
2454
- "learning_rate": 1.2647213921603094e-05,
2455
- "loss": 0.6737,
2456
- "step": 204000
2457
- },
2458
- {
2459
- "epoch": 2.25,
2460
- "learning_rate": 1.2555662975332513e-05,
2461
- "loss": 0.6726,
2462
- "step": 204500
2463
- },
2464
- {
2465
- "epoch": 2.25,
2466
- "learning_rate": 1.2464112029061933e-05,
2467
- "loss": 0.6713,
2468
- "step": 205000
2469
- },
2470
- {
2471
- "epoch": 2.26,
2472
- "learning_rate": 1.2372561082791352e-05,
2473
- "loss": 0.6828,
2474
- "step": 205500
2475
- },
2476
- {
2477
- "epoch": 2.26,
2478
- "learning_rate": 1.2281010136520773e-05,
2479
- "loss": 0.6741,
2480
- "step": 206000
2481
- },
2482
- {
2483
- "epoch": 2.27,
2484
- "learning_rate": 1.218945919025019e-05,
2485
- "loss": 0.6757,
2486
- "step": 206500
2487
- },
2488
- {
2489
- "epoch": 2.27,
2490
- "learning_rate": 1.209790824397961e-05,
2491
- "loss": 0.6788,
2492
- "step": 207000
2493
- },
2494
- {
2495
- "epoch": 2.28,
2496
- "learning_rate": 1.200635729770903e-05,
2497
- "loss": 0.6765,
2498
- "step": 207500
2499
- },
2500
- {
2501
- "epoch": 2.29,
2502
- "learning_rate": 1.1914806351438449e-05,
2503
- "loss": 0.6758,
2504
- "step": 208000
2505
- },
2506
- {
2507
- "epoch": 2.29,
2508
- "learning_rate": 1.1823255405167869e-05,
2509
- "loss": 0.6792,
2510
- "step": 208500
2511
- },
2512
- {
2513
- "epoch": 2.3,
2514
- "learning_rate": 1.1731704458897288e-05,
2515
- "loss": 0.6772,
2516
- "step": 209000
2517
- },
2518
- {
2519
- "epoch": 2.3,
2520
- "learning_rate": 1.1640153512626707e-05,
2521
- "loss": 0.6787,
2522
- "step": 209500
2523
- },
2524
- {
2525
- "epoch": 2.31,
2526
- "learning_rate": 1.1548602566356126e-05,
2527
- "loss": 0.6822,
2528
- "step": 210000
2529
- },
2530
- {
2531
- "epoch": 2.31,
2532
- "learning_rate": 1.1457051620085546e-05,
2533
- "loss": 0.6723,
2534
- "step": 210500
2535
- },
2536
- {
2537
- "epoch": 2.32,
2538
- "learning_rate": 1.1365500673814965e-05,
2539
- "loss": 0.6687,
2540
- "step": 211000
2541
- },
2542
- {
2543
- "epoch": 2.32,
2544
- "learning_rate": 1.1273949727544384e-05,
2545
- "loss": 0.6849,
2546
- "step": 211500
2547
- },
2548
- {
2549
- "epoch": 2.33,
2550
- "learning_rate": 1.1182398781273803e-05,
2551
- "loss": 0.6772,
2552
- "step": 212000
2553
- },
2554
- {
2555
- "epoch": 2.33,
2556
- "learning_rate": 1.1090847835003224e-05,
2557
- "loss": 0.6738,
2558
- "step": 212500
2559
- },
2560
- {
2561
- "epoch": 2.34,
2562
- "learning_rate": 1.0999296888732642e-05,
2563
- "loss": 0.6732,
2564
- "step": 213000
2565
- },
2566
- {
2567
- "epoch": 2.35,
2568
- "learning_rate": 1.0907745942462063e-05,
2569
- "loss": 0.6776,
2570
- "step": 213500
2571
- },
2572
- {
2573
- "epoch": 2.35,
2574
- "learning_rate": 1.081619499619148e-05,
2575
- "loss": 0.6744,
2576
- "step": 214000
2577
- },
2578
- {
2579
- "epoch": 2.36,
2580
- "learning_rate": 1.07246440499209e-05,
2581
- "loss": 0.6792,
2582
- "step": 214500
2583
- },
2584
- {
2585
- "epoch": 2.36,
2586
- "learning_rate": 1.063309310365032e-05,
2587
- "loss": 0.6808,
2588
- "step": 215000
2589
- },
2590
- {
2591
- "epoch": 2.37,
2592
- "learning_rate": 1.0541542157379739e-05,
2593
- "loss": 0.6777,
2594
- "step": 215500
2595
- },
2596
- {
2597
- "epoch": 2.37,
2598
- "learning_rate": 1.0449991211109159e-05,
2599
- "loss": 0.683,
2600
- "step": 216000
2601
- },
2602
- {
2603
- "epoch": 2.38,
2604
- "learning_rate": 1.0358440264838578e-05,
2605
- "loss": 0.6767,
2606
- "step": 216500
2607
- },
2608
- {
2609
- "epoch": 2.38,
2610
- "learning_rate": 1.0266889318567997e-05,
2611
- "loss": 0.6744,
2612
- "step": 217000
2613
- },
2614
- {
2615
- "epoch": 2.39,
2616
- "learning_rate": 1.0175338372297416e-05,
2617
- "loss": 0.6777,
2618
- "step": 217500
2619
- },
2620
- {
2621
- "epoch": 2.39,
2622
- "learning_rate": 1.0083787426026836e-05,
2623
- "loss": 0.6794,
2624
- "step": 218000
2625
- },
2626
- {
2627
- "epoch": 2.4,
2628
- "learning_rate": 9.992236479756255e-06,
2629
- "loss": 0.6705,
2630
- "step": 218500
2631
- },
2632
- {
2633
- "epoch": 2.41,
2634
- "learning_rate": 9.900685533485674e-06,
2635
- "loss": 0.6696,
2636
- "step": 219000
2637
- },
2638
- {
2639
- "epoch": 2.41,
2640
- "learning_rate": 9.809134587215093e-06,
2641
- "loss": 0.6829,
2642
- "step": 219500
2643
- },
2644
- {
2645
- "epoch": 2.42,
2646
- "learning_rate": 9.717583640944514e-06,
2647
- "loss": 0.6702,
2648
- "step": 220000
2649
- },
2650
- {
2651
- "epoch": 2.42,
2652
- "learning_rate": 9.626032694673932e-06,
2653
- "loss": 0.6759,
2654
- "step": 220500
2655
- },
2656
- {
2657
- "epoch": 2.43,
2658
- "learning_rate": 9.534481748403353e-06,
2659
- "loss": 0.6822,
2660
- "step": 221000
2661
- },
2662
- {
2663
- "epoch": 2.43,
2664
- "learning_rate": 9.442930802132772e-06,
2665
- "loss": 0.6771,
2666
- "step": 221500
2667
- },
2668
- {
2669
- "epoch": 2.44,
2670
- "learning_rate": 9.35137985586219e-06,
2671
- "loss": 0.6752,
2672
- "step": 222000
2673
- },
2674
- {
2675
- "epoch": 2.44,
2676
- "learning_rate": 9.25982890959161e-06,
2677
- "loss": 0.6752,
2678
- "step": 222500
2679
- },
2680
- {
2681
- "epoch": 2.45,
2682
- "learning_rate": 9.16827796332103e-06,
2683
- "loss": 0.6752,
2684
- "step": 223000
2685
- },
2686
- {
2687
- "epoch": 2.46,
2688
- "learning_rate": 9.07672701705045e-06,
2689
- "loss": 0.6829,
2690
- "step": 223500
2691
- },
2692
- {
2693
- "epoch": 2.46,
2694
- "learning_rate": 8.985176070779868e-06,
2695
- "loss": 0.6732,
2696
- "step": 224000
2697
- },
2698
- {
2699
- "epoch": 2.47,
2700
- "learning_rate": 8.893625124509287e-06,
2701
- "loss": 0.683,
2702
- "step": 224500
2703
- },
2704
- {
2705
- "epoch": 2.47,
2706
- "learning_rate": 8.802074178238706e-06,
2707
- "loss": 0.6717,
2708
- "step": 225000
2709
- },
2710
- {
2711
- "epoch": 2.48,
2712
- "learning_rate": 8.710523231968126e-06,
2713
- "loss": 0.678,
2714
- "step": 225500
2715
- },
2716
- {
2717
- "epoch": 2.48,
2718
- "learning_rate": 8.618972285697545e-06,
2719
- "loss": 0.6752,
2720
- "step": 226000
2721
- },
2722
- {
2723
- "epoch": 2.49,
2724
- "learning_rate": 8.527421339426964e-06,
2725
- "loss": 0.6805,
2726
- "step": 226500
2727
- },
2728
- {
2729
- "epoch": 2.49,
2730
- "learning_rate": 8.435870393156383e-06,
2731
- "loss": 0.6821,
2732
- "step": 227000
2733
- },
2734
- {
2735
- "epoch": 2.5,
2736
- "learning_rate": 8.344319446885804e-06,
2737
- "loss": 0.6802,
2738
- "step": 227500
2739
- },
2740
- {
2741
- "epoch": 2.5,
2742
- "learning_rate": 8.252768500615223e-06,
2743
- "loss": 0.6703,
2744
- "step": 228000
2745
- },
2746
- {
2747
- "epoch": 2.51,
2748
- "learning_rate": 8.161217554344643e-06,
2749
- "loss": 0.6724,
2750
- "step": 228500
2751
- },
2752
- {
2753
- "epoch": 2.52,
2754
- "learning_rate": 8.069666608074062e-06,
2755
- "loss": 0.6758,
2756
- "step": 229000
2757
- },
2758
- {
2759
- "epoch": 2.52,
2760
- "learning_rate": 7.978115661803481e-06,
2761
- "loss": 0.6798,
2762
- "step": 229500
2763
- },
2764
- {
2765
- "epoch": 2.53,
2766
- "learning_rate": 7.8865647155329e-06,
2767
- "loss": 0.6832,
2768
- "step": 230000
2769
- },
2770
- {
2771
- "epoch": 2.53,
2772
- "learning_rate": 7.79501376926232e-06,
2773
- "loss": 0.6691,
2774
- "step": 230500
2775
- },
2776
- {
2777
- "epoch": 2.54,
2778
- "learning_rate": 7.70346282299174e-06,
2779
- "loss": 0.6848,
2780
- "step": 231000
2781
- },
2782
- {
2783
- "epoch": 2.54,
2784
- "learning_rate": 7.611911876721157e-06,
2785
- "loss": 0.6699,
2786
- "step": 231500
2787
- },
2788
- {
2789
- "epoch": 2.55,
2790
- "learning_rate": 7.520360930450578e-06,
2791
- "loss": 0.682,
2792
- "step": 232000
2793
- },
2794
- {
2795
- "epoch": 2.55,
2796
- "learning_rate": 7.428809984179996e-06,
2797
- "loss": 0.6789,
2798
- "step": 232500
2799
- },
2800
- {
2801
- "epoch": 2.56,
2802
- "learning_rate": 7.3372590379094165e-06,
2803
- "loss": 0.6749,
2804
- "step": 233000
2805
- },
2806
- {
2807
- "epoch": 2.57,
2808
- "learning_rate": 7.245708091638835e-06,
2809
- "loss": 0.6749,
2810
- "step": 233500
2811
- },
2812
- {
2813
- "epoch": 2.57,
2814
- "learning_rate": 7.154157145368255e-06,
2815
- "loss": 0.6771,
2816
- "step": 234000
2817
- },
2818
- {
2819
- "epoch": 2.58,
2820
- "learning_rate": 7.062606199097674e-06,
2821
- "loss": 0.6848,
2822
- "step": 234500
2823
- },
2824
- {
2825
- "epoch": 2.58,
2826
- "learning_rate": 6.971055252827094e-06,
2827
- "loss": 0.677,
2828
- "step": 235000
2829
- },
2830
- {
2831
- "epoch": 2.59,
2832
- "learning_rate": 6.879504306556513e-06,
2833
- "loss": 0.6817,
2834
- "step": 235500
2835
- },
2836
- {
2837
- "epoch": 2.59,
2838
- "learning_rate": 6.787953360285932e-06,
2839
- "loss": 0.6776,
2840
- "step": 236000
2841
- },
2842
- {
2843
- "epoch": 2.6,
2844
- "learning_rate": 6.696402414015351e-06,
2845
- "loss": 0.6794,
2846
- "step": 236500
2847
- },
2848
- {
2849
- "epoch": 2.6,
2850
- "learning_rate": 6.604851467744772e-06,
2851
- "loss": 0.6825,
2852
- "step": 237000
2853
- },
2854
- {
2855
- "epoch": 2.61,
2856
- "learning_rate": 6.51330052147419e-06,
2857
- "loss": 0.6717,
2858
- "step": 237500
2859
- },
2860
- {
2861
- "epoch": 2.61,
2862
- "learning_rate": 6.4217495752036104e-06,
2863
- "loss": 0.6819,
2864
- "step": 238000
2865
- },
2866
- {
2867
- "epoch": 2.62,
2868
- "learning_rate": 6.330198628933029e-06,
2869
- "loss": 0.674,
2870
- "step": 238500
2871
- },
2872
- {
2873
- "epoch": 2.63,
2874
- "learning_rate": 6.238647682662448e-06,
2875
- "loss": 0.6757,
2876
- "step": 239000
2877
- },
2878
- {
2879
- "epoch": 2.63,
2880
- "learning_rate": 6.147096736391868e-06,
2881
- "loss": 0.6771,
2882
- "step": 239500
2883
- },
2884
- {
2885
- "epoch": 2.64,
2886
- "learning_rate": 6.055545790121287e-06,
2887
- "loss": 0.6848,
2888
- "step": 240000
2889
- },
2890
- {
2891
- "epoch": 2.64,
2892
- "learning_rate": 5.9639948438507066e-06,
2893
- "loss": 0.6724,
2894
- "step": 240500
2895
- },
2896
- {
2897
- "epoch": 2.65,
2898
- "learning_rate": 5.872443897580126e-06,
2899
- "loss": 0.6797,
2900
- "step": 241000
2901
- },
2902
- {
2903
- "epoch": 2.65,
2904
- "learning_rate": 5.780892951309545e-06,
2905
- "loss": 0.676,
2906
- "step": 241500
2907
- },
2908
- {
2909
- "epoch": 2.66,
2910
- "learning_rate": 5.689342005038965e-06,
2911
- "loss": 0.6777,
2912
- "step": 242000
2913
- },
2914
- {
2915
- "epoch": 2.66,
2916
- "learning_rate": 5.597791058768383e-06,
2917
- "loss": 0.6707,
2918
- "step": 242500
2919
- },
2920
- {
2921
- "epoch": 2.67,
2922
- "learning_rate": 5.506240112497803e-06,
2923
- "loss": 0.6837,
2924
- "step": 243000
2925
- },
2926
- {
2927
- "epoch": 2.68,
2928
- "learning_rate": 5.414689166227222e-06,
2929
- "loss": 0.6732,
2930
- "step": 243500
2931
- },
2932
- {
2933
- "epoch": 2.68,
2934
- "learning_rate": 5.323138219956641e-06,
2935
- "loss": 0.6745,
2936
- "step": 244000
2937
- },
2938
- {
2939
- "epoch": 2.69,
2940
- "learning_rate": 5.231587273686061e-06,
2941
- "loss": 0.6766,
2942
- "step": 244500
2943
- },
2944
- {
2945
- "epoch": 2.69,
2946
- "learning_rate": 5.14003632741548e-06,
2947
- "loss": 0.6726,
2948
- "step": 245000
2949
- },
2950
- {
2951
- "epoch": 2.7,
2952
- "learning_rate": 5.0484853811449e-06,
2953
- "loss": 0.6821,
2954
- "step": 245500
2955
- },
2956
- {
2957
- "epoch": 2.7,
2958
- "learning_rate": 4.956934434874319e-06,
2959
- "loss": 0.6708,
2960
- "step": 246000
2961
- },
2962
- {
2963
- "epoch": 2.71,
2964
- "learning_rate": 4.865383488603738e-06,
2965
- "loss": 0.6837,
2966
- "step": 246500
2967
- },
2968
- {
2969
- "epoch": 2.71,
2970
- "learning_rate": 4.773832542333158e-06,
2971
- "loss": 0.6752,
2972
- "step": 247000
2973
- },
2974
- {
2975
- "epoch": 2.72,
2976
- "learning_rate": 4.682281596062577e-06,
2977
- "loss": 0.6718,
2978
- "step": 247500
2979
- },
2980
- {
2981
- "epoch": 2.72,
2982
- "learning_rate": 4.590730649791997e-06,
2983
- "loss": 0.6812,
2984
- "step": 248000
2985
- },
2986
- {
2987
- "epoch": 2.73,
2988
- "learning_rate": 4.499179703521416e-06,
2989
- "loss": 0.6803,
2990
- "step": 248500
2991
- },
2992
- {
2993
- "epoch": 2.74,
2994
- "learning_rate": 4.407628757250835e-06,
2995
- "loss": 0.6803,
2996
- "step": 249000
2997
- },
2998
- {
2999
- "epoch": 2.74,
3000
- "learning_rate": 4.316077810980255e-06,
3001
- "loss": 0.6784,
3002
- "step": 249500
3003
- },
3004
- {
3005
- "epoch": 2.75,
3006
- "learning_rate": 4.224526864709674e-06,
3007
- "loss": 0.6765,
3008
- "step": 250000
3009
- },
3010
- {
3011
- "epoch": 2.75,
3012
- "learning_rate": 4.132975918439093e-06,
3013
- "loss": 0.6792,
3014
- "step": 250500
3015
- },
3016
- {
3017
- "epoch": 2.76,
3018
- "learning_rate": 4.0414249721685125e-06,
3019
- "loss": 0.6786,
3020
- "step": 251000
3021
- },
3022
- {
3023
- "epoch": 2.76,
3024
- "learning_rate": 3.949874025897931e-06,
3025
- "loss": 0.6765,
3026
- "step": 251500
3027
- },
3028
- {
3029
- "epoch": 2.77,
3030
- "learning_rate": 3.858323079627351e-06,
3031
- "loss": 0.673,
3032
- "step": 252000
3033
- },
3034
- {
3035
- "epoch": 2.77,
3036
- "learning_rate": 3.7667721333567704e-06,
3037
- "loss": 0.6727,
3038
- "step": 252500
3039
- },
3040
- {
3041
- "epoch": 2.78,
3042
- "learning_rate": 3.6752211870861897e-06,
3043
- "loss": 0.6751,
3044
- "step": 253000
3045
- },
3046
- {
3047
- "epoch": 2.78,
3048
- "learning_rate": 3.583670240815609e-06,
3049
- "loss": 0.6832,
3050
- "step": 253500
3051
- },
3052
- {
3053
- "epoch": 2.79,
3054
- "learning_rate": 3.4921192945450287e-06,
3055
- "loss": 0.6742,
3056
- "step": 254000
3057
- },
3058
- {
3059
- "epoch": 2.8,
3060
- "learning_rate": 3.400568348274448e-06,
3061
- "loss": 0.6738,
3062
- "step": 254500
3063
- },
3064
- {
3065
- "epoch": 2.8,
3066
- "learning_rate": 3.3090174020038674e-06,
3067
- "loss": 0.681,
3068
- "step": 255000
3069
- },
3070
- {
3071
- "epoch": 2.81,
3072
- "learning_rate": 3.2174664557332867e-06,
3073
- "loss": 0.6724,
3074
- "step": 255500
3075
- },
3076
- {
3077
- "epoch": 2.81,
3078
- "learning_rate": 3.125915509462706e-06,
3079
- "loss": 0.6734,
3080
- "step": 256000
3081
- },
3082
- {
3083
- "epoch": 2.82,
3084
- "learning_rate": 3.0343645631921253e-06,
3085
- "loss": 0.6821,
3086
- "step": 256500
3087
- },
3088
- {
3089
- "epoch": 2.82,
3090
- "learning_rate": 2.9428136169215446e-06,
3091
- "loss": 0.6754,
3092
- "step": 257000
3093
- },
3094
- {
3095
- "epoch": 2.83,
3096
- "learning_rate": 2.851262670650964e-06,
3097
- "loss": 0.6805,
3098
- "step": 257500
3099
- },
3100
- {
3101
- "epoch": 2.83,
3102
- "learning_rate": 2.7597117243803832e-06,
3103
- "loss": 0.6777,
3104
- "step": 258000
3105
- },
3106
- {
3107
- "epoch": 2.84,
3108
- "learning_rate": 2.668160778109803e-06,
3109
- "loss": 0.6756,
3110
- "step": 258500
3111
- },
3112
- {
3113
- "epoch": 2.85,
3114
- "learning_rate": 2.5766098318392223e-06,
3115
- "loss": 0.6835,
3116
- "step": 259000
3117
- },
3118
- {
3119
- "epoch": 2.85,
3120
- "learning_rate": 2.4850588855686416e-06,
3121
- "loss": 0.675,
3122
- "step": 259500
3123
- },
3124
- {
3125
- "epoch": 2.86,
3126
- "learning_rate": 2.3935079392980604e-06,
3127
- "loss": 0.6733,
3128
- "step": 260000
3129
- },
3130
- {
3131
- "epoch": 2.86,
3132
- "learning_rate": 2.3019569930274798e-06,
3133
- "loss": 0.6761,
3134
- "step": 260500
3135
- },
3136
- {
3137
- "epoch": 2.87,
3138
- "learning_rate": 2.2104060467568995e-06,
3139
- "loss": 0.6753,
3140
- "step": 261000
3141
- },
3142
- {
3143
- "epoch": 2.87,
3144
- "learning_rate": 2.118855100486319e-06,
3145
- "loss": 0.679,
3146
- "step": 261500
3147
- },
3148
- {
3149
- "epoch": 2.88,
3150
- "learning_rate": 2.027304154215738e-06,
3151
- "loss": 0.6707,
3152
- "step": 262000
3153
- },
3154
- {
3155
- "epoch": 2.88,
3156
- "learning_rate": 1.9357532079451574e-06,
3157
- "loss": 0.6744,
3158
- "step": 262500
3159
- },
3160
- {
3161
- "epoch": 2.89,
3162
- "learning_rate": 1.844202261674577e-06,
3163
- "loss": 0.678,
3164
- "step": 263000
3165
- },
3166
- {
3167
- "epoch": 2.89,
3168
- "learning_rate": 1.7526513154039962e-06,
3169
- "loss": 0.6655,
3170
- "step": 263500
3171
- },
3172
- {
3173
- "epoch": 2.9,
3174
- "learning_rate": 1.6611003691334153e-06,
3175
- "loss": 0.6812,
3176
- "step": 264000
3177
- },
3178
- {
3179
- "epoch": 2.91,
3180
- "learning_rate": 1.5695494228628347e-06,
3181
- "loss": 0.6797,
3182
- "step": 264500
3183
- },
3184
- {
3185
- "epoch": 2.91,
3186
- "learning_rate": 1.4779984765922542e-06,
3187
- "loss": 0.6727,
3188
- "step": 265000
3189
- },
3190
- {
3191
- "epoch": 2.92,
3192
- "learning_rate": 1.3864475303216735e-06,
3193
- "loss": 0.6738,
3194
- "step": 265500
3195
- },
3196
- {
3197
- "epoch": 2.92,
3198
- "learning_rate": 1.2948965840510928e-06,
3199
- "loss": 0.6799,
3200
- "step": 266000
3201
- },
3202
- {
3203
- "epoch": 2.93,
3204
- "learning_rate": 1.203345637780512e-06,
3205
- "loss": 0.6779,
3206
- "step": 266500
3207
- },
3208
- {
3209
- "epoch": 2.93,
3210
- "learning_rate": 1.1117946915099314e-06,
3211
- "loss": 0.68,
3212
- "step": 267000
3213
- },
3214
- {
3215
- "epoch": 2.94,
3216
- "learning_rate": 1.020243745239351e-06,
3217
- "loss": 0.6783,
3218
- "step": 267500
3219
- },
3220
- {
3221
- "epoch": 2.94,
3222
- "learning_rate": 9.286927989687702e-07,
3223
- "loss": 0.6851,
3224
- "step": 268000
3225
- },
3226
- {
3227
- "epoch": 2.95,
3228
- "learning_rate": 8.371418526981894e-07,
3229
- "loss": 0.6734,
3230
- "step": 268500
3231
- },
3232
- {
3233
- "epoch": 2.96,
3234
- "learning_rate": 7.455909064276089e-07,
3235
- "loss": 0.6796,
3236
- "step": 269000
3237
- },
3238
- {
3239
- "epoch": 2.96,
3240
- "learning_rate": 6.540399601570283e-07,
3241
- "loss": 0.6754,
3242
- "step": 269500
3243
- },
3244
- {
3245
- "epoch": 2.97,
3246
- "learning_rate": 5.624890138864476e-07,
3247
- "loss": 0.6813,
3248
- "step": 270000
3249
- },
3250
- {
3251
- "epoch": 2.97,
3252
- "learning_rate": 4.7093806761586695e-07,
3253
- "loss": 0.6758,
3254
- "step": 270500
3255
- },
3256
- {
3257
- "epoch": 2.98,
3258
- "learning_rate": 3.7938712134528625e-07,
3259
- "loss": 0.6788,
3260
- "step": 271000
3261
- },
3262
- {
3263
- "epoch": 2.98,
3264
- "learning_rate": 2.8783617507470556e-07,
3265
- "loss": 0.6753,
3266
- "step": 271500
3267
- },
3268
- {
3269
- "epoch": 2.99,
3270
- "learning_rate": 1.962852288041249e-07,
3271
- "loss": 0.6819,
3272
- "step": 272000
3273
- },
3274
- {
3275
- "epoch": 2.99,
3276
- "learning_rate": 1.0473428253354427e-07,
3277
- "loss": 0.675,
3278
- "step": 272500
3279
- },
3280
- {
3281
- "epoch": 3.0,
3282
- "learning_rate": 1.3183336262963613e-08,
3283
- "loss": 0.6794,
3284
- "step": 273000
3285
- },
3286
  {
3287
  "epoch": 3.0,
3288
- "step": 273072,
3289
- "total_flos": 4122145169772675840,
3290
- "train_runtime": 310227.8743,
3291
- "train_samples_per_second": 0.88
3292
  }
3293
  ],
3294
- "max_steps": 273072,
3295
  "num_train_epochs": 3,
3296
- "total_flos": 4122145169772675840,
3297
  "trial_name": null,
3298
  "trial_params": null
3299
  }
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
  "epoch": 3.0,
5
+ "global_step": 16314,
6
  "is_hyper_param_search": false,
7
  "is_local_process_zero": true,
8
  "is_world_process_zero": true,
9
  "log_history": [
10
  {
11
+ "epoch": 0.09,
12
+ "learning_rate": 4.846757386293981e-05,
13
+ "loss": 0.9182,
14
  "step": 500
15
  },
16
  {
17
+ "epoch": 0.18,
18
+ "learning_rate": 4.6935147725879615e-05,
19
+ "loss": 0.9324,
20
  "step": 1000
21
  },
22
  {
23
+ "epoch": 0.28,
24
+ "learning_rate": 4.540272158881942e-05,
25
+ "loss": 0.9328,
26
  "step": 1500
27
  },
28
  {
29
+ "epoch": 0.37,
30
+ "learning_rate": 4.387029545175923e-05,
31
+ "loss": 0.9331,
32
  "step": 2000
33
  },
34
  {
35
+ "epoch": 0.46,
36
+ "learning_rate": 4.2337869314699034e-05,
37
+ "loss": 0.9462,
38
  "step": 2500
39
  },
40
  {
41
+ "epoch": 0.55,
42
+ "learning_rate": 4.080544317763884e-05,
43
+ "loss": 0.9241,
44
  "step": 3000
45
  },
46
  {
47
+ "epoch": 0.64,
48
+ "learning_rate": 3.9273017040578646e-05,
49
+ "loss": 0.9323,
50
  "step": 3500
51
  },
52
  {
53
+ "epoch": 0.74,
54
+ "learning_rate": 3.774059090351845e-05,
55
+ "loss": 0.9296,
56
  "step": 4000
57
  },
58
  {
59
+ "epoch": 0.83,
60
+ "learning_rate": 3.620816476645826e-05,
61
+ "loss": 0.9296,
62
  "step": 4500
63
  },
64
  {
65
+ "epoch": 0.92,
66
+ "learning_rate": 3.4675738629398065e-05,
67
+ "loss": 0.9232,
68
  "step": 5000
69
  },
70
  {
71
+ "epoch": 1.01,
72
+ "learning_rate": 3.314331249233787e-05,
73
+ "loss": 0.9,
74
  "step": 5500
75
  },
76
  {
77
+ "epoch": 1.1,
78
+ "learning_rate": 3.161088635527768e-05,
79
+ "loss": 0.7056,
80
  "step": 6000
81
  },
82
  {
83
+ "epoch": 1.2,
84
+ "learning_rate": 3.0078460218217487e-05,
85
+ "loss": 0.7021,
86
  "step": 6500
87
  },
88
  {
89
+ "epoch": 1.29,
90
+ "learning_rate": 2.854603408115729e-05,
91
+ "loss": 0.6979,
92
  "step": 7000
93
  },
94
  {
95
+ "epoch": 1.38,
96
+ "learning_rate": 2.7013607944097096e-05,
97
+ "loss": 0.7094,
98
  "step": 7500
99
  },
100
  {
101
+ "epoch": 1.47,
102
+ "learning_rate": 2.5481181807036903e-05,
103
+ "loss": 0.7099,
104
  "step": 8000
105
  },
106
  {
107
+ "epoch": 1.56,
108
+ "learning_rate": 2.394875566997671e-05,
109
+ "loss": 0.7178,
110
  "step": 8500
111
  },
112
  {
113
+ "epoch": 1.66,
114
+ "learning_rate": 2.2416329532916515e-05,
115
+ "loss": 0.7136,
116
  "step": 9000
117
  },
118
  {
119
+ "epoch": 1.75,
120
+ "learning_rate": 2.088390339585632e-05,
121
+ "loss": 0.71,
122
  "step": 9500
123
  },
124
  {
125
+ "epoch": 1.84,
126
+ "learning_rate": 1.9351477258796128e-05,
127
+ "loss": 0.7183,
128
  "step": 10000
129
  },
130
  {
131
+ "epoch": 1.93,
132
+ "learning_rate": 1.7819051121735934e-05,
133
+ "loss": 0.7188,
134
  "step": 10500
135
  },
136
  {
137
+ "epoch": 2.02,
138
+ "learning_rate": 1.628662498467574e-05,
139
+ "loss": 0.6733,
140
  "step": 11000
141
  },
142
  {
143
+ "epoch": 2.11,
144
+ "learning_rate": 1.4754198847615546e-05,
145
+ "loss": 0.5524,
146
  "step": 11500
147
  },
148
  {
149
+ "epoch": 2.21,
150
+ "learning_rate": 1.3221772710555353e-05,
151
+ "loss": 0.5488,
152
  "step": 12000
153
  },
154
  {
155
+ "epoch": 2.3,
156
+ "learning_rate": 1.1689346573495157e-05,
157
+ "loss": 0.5507,
158
  "step": 12500
159
  },
160
  {
161
+ "epoch": 2.39,
162
+ "learning_rate": 1.0156920436434965e-05,
163
+ "loss": 0.553,
164
  "step": 13000
165
  },
166
  {
167
+ "epoch": 2.48,
168
+ "learning_rate": 8.62449429937477e-06,
169
+ "loss": 0.5558,
170
  "step": 13500
171
  },
172
  {
173
+ "epoch": 2.57,
174
+ "learning_rate": 7.092068162314578e-06,
175
+ "loss": 0.5554,
176
  "step": 14000
177
  },
178
  {
179
+ "epoch": 2.67,
180
+ "learning_rate": 5.559642025254383e-06,
181
+ "loss": 0.5457,
182
  "step": 14500
183
  },
184
  {
185
+ "epoch": 2.76,
186
+ "learning_rate": 4.027215888194189e-06,
187
+ "loss": 0.5548,
188
  "step": 15000
189
  },
190
  {
191
+ "epoch": 2.85,
192
+ "learning_rate": 2.4947897511339955e-06,
193
+ "loss": 0.5538,
194
  "step": 15500
195
  },
196
  {
197
+ "epoch": 2.94,
198
+ "learning_rate": 9.623636140738016e-07,
199
+ "loss": 0.5554,
200
  "step": 16000
201
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
202
  {
203
  "epoch": 3.0,
204
+ "step": 16314,
205
+ "total_flos": 246245231169216000,
206
+ "train_runtime": 35196.4998,
207
+ "train_samples_per_second": 0.464
208
  }
209
  ],
210
+ "max_steps": 16314,
211
  "num_train_epochs": 3,
212
+ "total_flos": 246245231169216000,
213
  "trial_name": null,
214
  "trial_params": null
215
  }