denocris commited on
Commit
51968db
1 Parent(s): e9807c2

Delete trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +0 -2894
trainer_state.json DELETED
@@ -1,2894 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 14.942528735632184,
5
- "global_step": 46800,
6
- "is_hyper_param_search": false,
7
- "is_local_process_zero": true,
8
- "is_world_process_zero": true,
9
- "log_history": [
10
- {
11
- "epoch": 0.031928480204342274,
12
- "learning_rate": 1.995742869306088e-05,
13
- "loss": 1.0424302673339845,
14
- "step": 100
15
- },
16
- {
17
- "epoch": 0.06385696040868455,
18
- "learning_rate": 1.9914857386121755e-05,
19
- "loss": 0.5240679550170898,
20
- "step": 200
21
- },
22
- {
23
- "epoch": 0.09578544061302682,
24
- "learning_rate": 1.9872286079182633e-05,
25
- "loss": 0.5014432144165039,
26
- "step": 300
27
- },
28
- {
29
- "epoch": 0.1277139208173691,
30
- "learning_rate": 1.9829714772243508e-05,
31
- "loss": 0.48912925720214845,
32
- "step": 400
33
- },
34
- {
35
- "epoch": 0.15964240102171137,
36
- "learning_rate": 1.9787143465304387e-05,
37
- "loss": 0.48597183227539065,
38
- "step": 500
39
- },
40
- {
41
- "epoch": 0.19157088122605365,
42
- "learning_rate": 1.974457215836526e-05,
43
- "loss": 0.4988706970214844,
44
- "step": 600
45
- },
46
- {
47
- "epoch": 0.22349936143039592,
48
- "learning_rate": 1.970200085142614e-05,
49
- "loss": 0.48521129608154295,
50
- "step": 700
51
- },
52
- {
53
- "epoch": 0.2554278416347382,
54
- "learning_rate": 1.9659429544487015e-05,
55
- "loss": 0.4737409210205078,
56
- "step": 800
57
- },
58
- {
59
- "epoch": 0.28735632183908044,
60
- "learning_rate": 1.9616858237547893e-05,
61
- "loss": 0.478267822265625,
62
- "step": 900
63
- },
64
- {
65
- "epoch": 0.31928480204342274,
66
- "learning_rate": 1.957428693060877e-05,
67
- "loss": 0.46842235565185547,
68
- "step": 1000
69
- },
70
- {
71
- "epoch": 0.351213282247765,
72
- "learning_rate": 1.953171562366965e-05,
73
- "loss": 0.47340740203857423,
74
- "step": 1100
75
- },
76
- {
77
- "epoch": 0.3831417624521073,
78
- "learning_rate": 1.9489144316730525e-05,
79
- "loss": 0.48022716522216796,
80
- "step": 1200
81
- },
82
- {
83
- "epoch": 0.41507024265644954,
84
- "learning_rate": 1.9446573009791403e-05,
85
- "loss": 0.4702701950073242,
86
- "step": 1300
87
- },
88
- {
89
- "epoch": 0.44699872286079184,
90
- "learning_rate": 1.9404001702852278e-05,
91
- "loss": 0.47483158111572266,
92
- "step": 1400
93
- },
94
- {
95
- "epoch": 0.4789272030651341,
96
- "learning_rate": 1.9361430395913157e-05,
97
- "loss": 0.45973930358886717,
98
- "step": 1500
99
- },
100
- {
101
- "epoch": 0.5108556832694764,
102
- "learning_rate": 1.9318859088974035e-05,
103
- "loss": 0.43192115783691404,
104
- "step": 1600
105
- },
106
- {
107
- "epoch": 0.5427841634738186,
108
- "learning_rate": 1.927628778203491e-05,
109
- "loss": 0.44062965393066406,
110
- "step": 1700
111
- },
112
- {
113
- "epoch": 0.5747126436781609,
114
- "learning_rate": 1.923371647509579e-05,
115
- "loss": 0.460767822265625,
116
- "step": 1800
117
- },
118
- {
119
- "epoch": 0.6066411238825032,
120
- "learning_rate": 1.9191145168156663e-05,
121
- "loss": 0.4533885192871094,
122
- "step": 1900
123
- },
124
- {
125
- "epoch": 0.6385696040868455,
126
- "learning_rate": 1.914857386121754e-05,
127
- "loss": 0.4435346221923828,
128
- "step": 2000
129
- },
130
- {
131
- "epoch": 0.6704980842911877,
132
- "learning_rate": 1.9106002554278417e-05,
133
- "loss": 0.4584750747680664,
134
- "step": 2100
135
- },
136
- {
137
- "epoch": 0.70242656449553,
138
- "learning_rate": 1.9063431247339295e-05,
139
- "loss": 0.43961280822753906,
140
- "step": 2200
141
- },
142
- {
143
- "epoch": 0.7343550446998723,
144
- "learning_rate": 1.902085994040017e-05,
145
- "loss": 0.43362281799316404,
146
- "step": 2300
147
- },
148
- {
149
- "epoch": 0.7662835249042146,
150
- "learning_rate": 1.8978288633461048e-05,
151
- "loss": 0.4443118286132812,
152
- "step": 2400
153
- },
154
- {
155
- "epoch": 0.7982120051085568,
156
- "learning_rate": 1.8935717326521927e-05,
157
- "loss": 0.4438581848144531,
158
- "step": 2500
159
- },
160
- {
161
- "epoch": 0.8301404853128991,
162
- "learning_rate": 1.88931460195828e-05,
163
- "loss": 0.4310215759277344,
164
- "step": 2600
165
- },
166
- {
167
- "epoch": 0.8620689655172413,
168
- "learning_rate": 1.885057471264368e-05,
169
- "loss": 0.4399332809448242,
170
- "step": 2700
171
- },
172
- {
173
- "epoch": 0.8939974457215837,
174
- "learning_rate": 1.8808003405704555e-05,
175
- "loss": 0.4511014175415039,
176
- "step": 2800
177
- },
178
- {
179
- "epoch": 0.9259259259259259,
180
- "learning_rate": 1.8765432098765433e-05,
181
- "loss": 0.44327774047851565,
182
- "step": 2900
183
- },
184
- {
185
- "epoch": 0.9578544061302682,
186
- "learning_rate": 1.872286079182631e-05,
187
- "loss": 0.43171138763427735,
188
- "step": 3000
189
- },
190
- {
191
- "epoch": 0.9897828863346104,
192
- "learning_rate": 1.868028948488719e-05,
193
- "loss": 0.4460626983642578,
194
- "step": 3100
195
- },
196
- {
197
- "epoch": 1.0,
198
- "eval_loss": 0.3910863995552063,
199
- "step": 3132
200
- },
201
- {
202
- "epoch": 1.0217113665389528,
203
- "learning_rate": 1.8637718177948065e-05,
204
- "loss": 0.43170944213867185,
205
- "step": 3200
206
- },
207
- {
208
- "epoch": 1.053639846743295,
209
- "learning_rate": 1.8595146871008943e-05,
210
- "loss": 0.42756095886230466,
211
- "step": 3300
212
- },
213
- {
214
- "epoch": 1.0855683269476373,
215
- "learning_rate": 1.8552575564069818e-05,
216
- "loss": 0.423099479675293,
217
- "step": 3400
218
- },
219
- {
220
- "epoch": 1.1174968071519795,
221
- "learning_rate": 1.8510004257130697e-05,
222
- "loss": 0.42962390899658204,
223
- "step": 3500
224
- },
225
- {
226
- "epoch": 1.1494252873563218,
227
- "learning_rate": 1.846743295019157e-05,
228
- "loss": 0.4373976898193359,
229
- "step": 3600
230
- },
231
- {
232
- "epoch": 1.181353767560664,
233
- "learning_rate": 1.842486164325245e-05,
234
- "loss": 0.4381977462768555,
235
- "step": 3700
236
- },
237
- {
238
- "epoch": 1.2132822477650063,
239
- "learning_rate": 1.8382290336313325e-05,
240
- "loss": 0.4158343887329102,
241
- "step": 3800
242
- },
243
- {
244
- "epoch": 1.2452107279693487,
245
- "learning_rate": 1.8339719029374203e-05,
246
- "loss": 0.4281564712524414,
247
- "step": 3900
248
- },
249
- {
250
- "epoch": 1.277139208173691,
251
- "learning_rate": 1.829714772243508e-05,
252
- "loss": 0.41806602478027344,
253
- "step": 4000
254
- },
255
- {
256
- "epoch": 1.3090676883780332,
257
- "learning_rate": 1.8254576415495957e-05,
258
- "loss": 0.4173674011230469,
259
- "step": 4100
260
- },
261
- {
262
- "epoch": 1.3409961685823755,
263
- "learning_rate": 1.8212005108556835e-05,
264
- "loss": 0.43225109100341796,
265
- "step": 4200
266
- },
267
- {
268
- "epoch": 1.3729246487867177,
269
- "learning_rate": 1.816943380161771e-05,
270
- "loss": 0.421945686340332,
271
- "step": 4300
272
- },
273
- {
274
- "epoch": 1.40485312899106,
275
- "learning_rate": 1.812686249467859e-05,
276
- "loss": 0.42005214691162107,
277
- "step": 4400
278
- },
279
- {
280
- "epoch": 1.4367816091954024,
281
- "learning_rate": 1.8084291187739463e-05,
282
- "loss": 0.40552581787109376,
283
- "step": 4500
284
- },
285
- {
286
- "epoch": 1.4687100893997447,
287
- "learning_rate": 1.804171988080034e-05,
288
- "loss": 0.41495433807373044,
289
- "step": 4600
290
- },
291
- {
292
- "epoch": 1.500638569604087,
293
- "learning_rate": 1.7999148573861217e-05,
294
- "loss": 0.4268427658081055,
295
- "step": 4700
296
- },
297
- {
298
- "epoch": 1.5325670498084292,
299
- "learning_rate": 1.7956577266922095e-05,
300
- "loss": 0.4178670120239258,
301
- "step": 4800
302
- },
303
- {
304
- "epoch": 1.5644955300127714,
305
- "learning_rate": 1.7914005959982973e-05,
306
- "loss": 0.4088083267211914,
307
- "step": 4900
308
- },
309
- {
310
- "epoch": 1.5964240102171137,
311
- "learning_rate": 1.787143465304385e-05,
312
- "loss": 0.4090264129638672,
313
- "step": 5000
314
- },
315
- {
316
- "epoch": 1.628352490421456,
317
- "learning_rate": 1.7828863346104727e-05,
318
- "loss": 0.39727077484130857,
319
- "step": 5100
320
- },
321
- {
322
- "epoch": 1.6602809706257982,
323
- "learning_rate": 1.7786292039165605e-05,
324
- "loss": 0.40877700805664063,
325
- "step": 5200
326
- },
327
- {
328
- "epoch": 1.6922094508301404,
329
- "learning_rate": 1.774372073222648e-05,
330
- "loss": 0.4055898666381836,
331
- "step": 5300
332
- },
333
- {
334
- "epoch": 1.7241379310344827,
335
- "learning_rate": 1.770114942528736e-05,
336
- "loss": 0.40825370788574217,
337
- "step": 5400
338
- },
339
- {
340
- "epoch": 1.756066411238825,
341
- "learning_rate": 1.7658578118348237e-05,
342
- "loss": 0.39790542602539064,
343
- "step": 5500
344
- },
345
- {
346
- "epoch": 1.7879948914431671,
347
- "learning_rate": 1.761600681140911e-05,
348
- "loss": 0.4129365158081055,
349
- "step": 5600
350
- },
351
- {
352
- "epoch": 1.8199233716475096,
353
- "learning_rate": 1.757343550446999e-05,
354
- "loss": 0.4137036895751953,
355
- "step": 5700
356
- },
357
- {
358
- "epoch": 1.8518518518518519,
359
- "learning_rate": 1.7530864197530865e-05,
360
- "loss": 0.41169502258300783,
361
- "step": 5800
362
- },
363
- {
364
- "epoch": 1.883780332056194,
365
- "learning_rate": 1.7488292890591743e-05,
366
- "loss": 0.3932318115234375,
367
- "step": 5900
368
- },
369
- {
370
- "epoch": 1.9157088122605364,
371
- "learning_rate": 1.7445721583652618e-05,
372
- "loss": 0.40273929595947267,
373
- "step": 6000
374
- },
375
- {
376
- "epoch": 1.9476372924648788,
377
- "learning_rate": 1.7403150276713497e-05,
378
- "loss": 0.4069852066040039,
379
- "step": 6100
380
- },
381
- {
382
- "epoch": 1.979565772669221,
383
- "learning_rate": 1.736057896977437e-05,
384
- "loss": 0.40934764862060546,
385
- "step": 6200
386
- },
387
- {
388
- "epoch": 2.0,
389
- "eval_loss": 0.3667986989021301,
390
- "step": 6264
391
- },
392
- {
393
- "epoch": 2.0114942528735633,
394
- "learning_rate": 1.731800766283525e-05,
395
- "loss": 0.40322193145751956,
396
- "step": 6300
397
- },
398
- {
399
- "epoch": 2.0434227330779056,
400
- "learning_rate": 1.727543635589613e-05,
401
- "loss": 0.3913343048095703,
402
- "step": 6400
403
- },
404
- {
405
- "epoch": 2.075351213282248,
406
- "learning_rate": 1.7232865048957003e-05,
407
- "loss": 0.4131240081787109,
408
- "step": 6500
409
- },
410
- {
411
- "epoch": 2.10727969348659,
412
- "learning_rate": 1.719029374201788e-05,
413
- "loss": 0.394369010925293,
414
- "step": 6600
415
- },
416
- {
417
- "epoch": 2.1392081736909323,
418
- "learning_rate": 1.7147722435078757e-05,
419
- "loss": 0.3991780471801758,
420
- "step": 6700
421
- },
422
- {
423
- "epoch": 2.1711366538952745,
424
- "learning_rate": 1.7105151128139635e-05,
425
- "loss": 0.3874116134643555,
426
- "step": 6800
427
- },
428
- {
429
- "epoch": 2.203065134099617,
430
- "learning_rate": 1.706257982120051e-05,
431
- "loss": 0.387044792175293,
432
- "step": 6900
433
- },
434
- {
435
- "epoch": 2.234993614303959,
436
- "learning_rate": 1.702000851426139e-05,
437
- "loss": 0.39865818023681643,
438
- "step": 7000
439
- },
440
- {
441
- "epoch": 2.2669220945083013,
442
- "learning_rate": 1.6977437207322267e-05,
443
- "loss": 0.3999287414550781,
444
- "step": 7100
445
- },
446
- {
447
- "epoch": 2.2988505747126435,
448
- "learning_rate": 1.6934865900383145e-05,
449
- "loss": 0.4086351013183594,
450
- "step": 7200
451
- },
452
- {
453
- "epoch": 2.330779054916986,
454
- "learning_rate": 1.689229459344402e-05,
455
- "loss": 0.3991414642333984,
456
- "step": 7300
457
- },
458
- {
459
- "epoch": 2.362707535121328,
460
- "learning_rate": 1.68497232865049e-05,
461
- "loss": 0.3910430145263672,
462
- "step": 7400
463
- },
464
- {
465
- "epoch": 2.3946360153256707,
466
- "learning_rate": 1.6807151979565773e-05,
467
- "loss": 0.39477340698242186,
468
- "step": 7500
469
- },
470
- {
471
- "epoch": 2.4265644955300125,
472
- "learning_rate": 1.676458067262665e-05,
473
- "loss": 0.38976318359375,
474
- "step": 7600
475
- },
476
- {
477
- "epoch": 2.458492975734355,
478
- "learning_rate": 1.6722009365687527e-05,
479
- "loss": 0.38246253967285154,
480
- "step": 7700
481
- },
482
- {
483
- "epoch": 2.4904214559386975,
484
- "learning_rate": 1.6679438058748405e-05,
485
- "loss": 0.38488063812255857,
486
- "step": 7800
487
- },
488
- {
489
- "epoch": 2.5223499361430397,
490
- "learning_rate": 1.6636866751809283e-05,
491
- "loss": 0.3952900695800781,
492
- "step": 7900
493
- },
494
- {
495
- "epoch": 2.554278416347382,
496
- "learning_rate": 1.6594295444870158e-05,
497
- "loss": 0.3940334701538086,
498
- "step": 8000
499
- },
500
- {
501
- "epoch": 2.586206896551724,
502
- "learning_rate": 1.6551724137931037e-05,
503
- "loss": 0.3789644622802734,
504
- "step": 8100
505
- },
506
- {
507
- "epoch": 2.6181353767560664,
508
- "learning_rate": 1.650915283099191e-05,
509
- "loss": 0.38175716400146487,
510
- "step": 8200
511
- },
512
- {
513
- "epoch": 2.6500638569604087,
514
- "learning_rate": 1.646658152405279e-05,
515
- "loss": 0.3957417678833008,
516
- "step": 8300
517
- },
518
- {
519
- "epoch": 2.681992337164751,
520
- "learning_rate": 1.6424010217113665e-05,
521
- "loss": 0.3931695556640625,
522
- "step": 8400
523
- },
524
- {
525
- "epoch": 2.713920817369093,
526
- "learning_rate": 1.6381438910174543e-05,
527
- "loss": 0.3700098419189453,
528
- "step": 8500
529
- },
530
- {
531
- "epoch": 2.7458492975734354,
532
- "learning_rate": 1.6338867603235418e-05,
533
- "loss": 0.3986296463012695,
534
- "step": 8600
535
- },
536
- {
537
- "epoch": 2.7777777777777777,
538
- "learning_rate": 1.6296296296296297e-05,
539
- "loss": 0.377045783996582,
540
- "step": 8700
541
- },
542
- {
543
- "epoch": 2.80970625798212,
544
- "learning_rate": 1.6253724989357175e-05,
545
- "loss": 0.38455604553222655,
546
- "step": 8800
547
- },
548
- {
549
- "epoch": 2.841634738186462,
550
- "learning_rate": 1.6211153682418053e-05,
551
- "loss": 0.39248775482177733,
552
- "step": 8900
553
- },
554
- {
555
- "epoch": 2.873563218390805,
556
- "learning_rate": 1.6168582375478928e-05,
557
- "loss": 0.3952408599853516,
558
- "step": 9000
559
- },
560
- {
561
- "epoch": 2.9054916985951467,
562
- "learning_rate": 1.6126011068539807e-05,
563
- "loss": 0.38378463745117186,
564
- "step": 9100
565
- },
566
- {
567
- "epoch": 2.9374201787994894,
568
- "learning_rate": 1.608343976160068e-05,
569
- "loss": 0.3779494857788086,
570
- "step": 9200
571
- },
572
- {
573
- "epoch": 2.969348659003831,
574
- "learning_rate": 1.604086845466156e-05,
575
- "loss": 0.39154972076416017,
576
- "step": 9300
577
- },
578
- {
579
- "epoch": 3.0,
580
- "eval_loss": 0.3532629609107971,
581
- "step": 9396
582
- },
583
- {
584
- "epoch": 3.001277139208174,
585
- "learning_rate": 1.599829714772244e-05,
586
- "loss": 0.38505748748779295,
587
- "step": 9400
588
- },
589
- {
590
- "epoch": 3.033205619412516,
591
- "learning_rate": 1.5955725840783313e-05,
592
- "loss": 0.37334869384765623,
593
- "step": 9500
594
- },
595
- {
596
- "epoch": 3.0651340996168583,
597
- "learning_rate": 1.591315453384419e-05,
598
- "loss": 0.36475982666015627,
599
- "step": 9600
600
- },
601
- {
602
- "epoch": 3.0970625798212006,
603
- "learning_rate": 1.5870583226905067e-05,
604
- "loss": 0.3732810592651367,
605
- "step": 9700
606
- },
607
- {
608
- "epoch": 3.128991060025543,
609
- "learning_rate": 1.5828011919965945e-05,
610
- "loss": 0.36443687438964845,
611
- "step": 9800
612
- },
613
- {
614
- "epoch": 3.160919540229885,
615
- "learning_rate": 1.578544061302682e-05,
616
- "loss": 0.38134773254394533,
617
- "step": 9900
618
- },
619
- {
620
- "epoch": 3.1928480204342273,
621
- "learning_rate": 1.57428693060877e-05,
622
- "loss": 0.3660939407348633,
623
- "step": 10000
624
- },
625
- {
626
- "epoch": 3.2247765006385696,
627
- "learning_rate": 1.5700297999148573e-05,
628
- "loss": 0.38626991271972655,
629
- "step": 10100
630
- },
631
- {
632
- "epoch": 3.256704980842912,
633
- "learning_rate": 1.565772669220945e-05,
634
- "loss": 0.36326351165771487,
635
- "step": 10200
636
- },
637
- {
638
- "epoch": 3.288633461047254,
639
- "learning_rate": 1.561515538527033e-05,
640
- "loss": 0.3856014633178711,
641
- "step": 10300
642
- },
643
- {
644
- "epoch": 3.3205619412515963,
645
- "learning_rate": 1.5572584078331205e-05,
646
- "loss": 0.38314430236816405,
647
- "step": 10400
648
- },
649
- {
650
- "epoch": 3.3524904214559386,
651
- "learning_rate": 1.5530012771392083e-05,
652
- "loss": 0.3787594223022461,
653
- "step": 10500
654
- },
655
- {
656
- "epoch": 3.384418901660281,
657
- "learning_rate": 1.5487441464452958e-05,
658
- "loss": 0.37935165405273436,
659
- "step": 10600
660
- },
661
- {
662
- "epoch": 3.416347381864623,
663
- "learning_rate": 1.5444870157513837e-05,
664
- "loss": 0.3672695541381836,
665
- "step": 10700
666
- },
667
- {
668
- "epoch": 3.4482758620689653,
669
- "learning_rate": 1.540229885057471e-05,
670
- "loss": 0.3799928283691406,
671
- "step": 10800
672
- },
673
- {
674
- "epoch": 3.480204342273308,
675
- "learning_rate": 1.5359727543635593e-05,
676
- "loss": 0.381710090637207,
677
- "step": 10900
678
- },
679
- {
680
- "epoch": 3.51213282247765,
681
- "learning_rate": 1.531715623669647e-05,
682
- "loss": 0.35085960388183596,
683
- "step": 11000
684
- },
685
- {
686
- "epoch": 3.5440613026819925,
687
- "learning_rate": 1.5274584929757347e-05,
688
- "loss": 0.37504680633544923,
689
- "step": 11100
690
- },
691
- {
692
- "epoch": 3.5759897828863347,
693
- "learning_rate": 1.5232013622818223e-05,
694
- "loss": 0.36862171173095704,
695
- "step": 11200
696
- },
697
- {
698
- "epoch": 3.607918263090677,
699
- "learning_rate": 1.51894423158791e-05,
700
- "loss": 0.3727375793457031,
701
- "step": 11300
702
- },
703
- {
704
- "epoch": 3.6398467432950192,
705
- "learning_rate": 1.5146871008939977e-05,
706
- "loss": 0.37135467529296873,
707
- "step": 11400
708
- },
709
- {
710
- "epoch": 3.6717752234993615,
711
- "learning_rate": 1.5104299702000853e-05,
712
- "loss": 0.3693832778930664,
713
- "step": 11500
714
- },
715
- {
716
- "epoch": 3.7037037037037037,
717
- "learning_rate": 1.506172839506173e-05,
718
- "loss": 0.36289344787597655,
719
- "step": 11600
720
- },
721
- {
722
- "epoch": 3.735632183908046,
723
- "learning_rate": 1.5019157088122607e-05,
724
- "loss": 0.3692543411254883,
725
- "step": 11700
726
- },
727
- {
728
- "epoch": 3.767560664112388,
729
- "learning_rate": 1.4976585781183483e-05,
730
- "loss": 0.3623777770996094,
731
- "step": 11800
732
- },
733
- {
734
- "epoch": 3.7994891443167305,
735
- "learning_rate": 1.493401447424436e-05,
736
- "loss": 0.3629803848266602,
737
- "step": 11900
738
- },
739
- {
740
- "epoch": 3.8314176245210727,
741
- "learning_rate": 1.4891443167305237e-05,
742
- "loss": 0.3619497680664063,
743
- "step": 12000
744
- },
745
- {
746
- "epoch": 3.863346104725415,
747
- "learning_rate": 1.4848871860366115e-05,
748
- "loss": 0.35669734954833987,
749
- "step": 12100
750
- },
751
- {
752
- "epoch": 3.895274584929757,
753
- "learning_rate": 1.4806300553426992e-05,
754
- "loss": 0.359210205078125,
755
- "step": 12200
756
- },
757
- {
758
- "epoch": 3.9272030651340994,
759
- "learning_rate": 1.4763729246487868e-05,
760
- "loss": 0.3655078887939453,
761
- "step": 12300
762
- },
763
- {
764
- "epoch": 3.959131545338442,
765
- "learning_rate": 1.4721157939548745e-05,
766
- "loss": 0.3666878890991211,
767
- "step": 12400
768
- },
769
- {
770
- "epoch": 3.991060025542784,
771
- "learning_rate": 1.4678586632609622e-05,
772
- "loss": 0.3504248809814453,
773
- "step": 12500
774
- },
775
- {
776
- "epoch": 4.0,
777
- "eval_loss": 0.3464440107345581,
778
- "step": 12528
779
- },
780
- {
781
- "epoch": 4.022988505747127,
782
- "learning_rate": 1.4636015325670498e-05,
783
- "loss": 0.35299652099609374,
784
- "step": 12600
785
- },
786
- {
787
- "epoch": 4.054916985951468,
788
- "learning_rate": 1.4593444018731375e-05,
789
- "loss": 0.3569066619873047,
790
- "step": 12700
791
- },
792
- {
793
- "epoch": 4.086845466155811,
794
- "learning_rate": 1.4550872711792252e-05,
795
- "loss": 0.3493093490600586,
796
- "step": 12800
797
- },
798
- {
799
- "epoch": 4.118773946360153,
800
- "learning_rate": 1.4508301404853132e-05,
801
- "loss": 0.3634659957885742,
802
- "step": 12900
803
- },
804
- {
805
- "epoch": 4.150702426564496,
806
- "learning_rate": 1.4465730097914008e-05,
807
- "loss": 0.37093032836914064,
808
- "step": 13000
809
- },
810
- {
811
- "epoch": 4.182630906768837,
812
- "learning_rate": 1.4423158790974885e-05,
813
- "loss": 0.37300064086914064,
814
- "step": 13100
815
- },
816
- {
817
- "epoch": 4.21455938697318,
818
- "learning_rate": 1.4380587484035762e-05,
819
- "loss": 0.358408203125,
820
- "step": 13200
821
- },
822
- {
823
- "epoch": 4.246487867177523,
824
- "learning_rate": 1.4338016177096638e-05,
825
- "loss": 0.3569377517700195,
826
- "step": 13300
827
- },
828
- {
829
- "epoch": 4.278416347381865,
830
- "learning_rate": 1.4295444870157515e-05,
831
- "loss": 0.3640250015258789,
832
- "step": 13400
833
- },
834
- {
835
- "epoch": 4.310344827586207,
836
- "learning_rate": 1.4252873563218392e-05,
837
- "loss": 0.3725627517700195,
838
- "step": 13500
839
- },
840
- {
841
- "epoch": 4.342273307790549,
842
- "learning_rate": 1.421030225627927e-05,
843
- "loss": 0.3584425354003906,
844
- "step": 13600
845
- },
846
- {
847
- "epoch": 4.374201787994892,
848
- "learning_rate": 1.4167730949340147e-05,
849
- "loss": 0.37500988006591796,
850
- "step": 13700
851
- },
852
- {
853
- "epoch": 4.406130268199234,
854
- "learning_rate": 1.4125159642401023e-05,
855
- "loss": 0.35521396636962893,
856
- "step": 13800
857
- },
858
- {
859
- "epoch": 4.438058748403576,
860
- "learning_rate": 1.40825883354619e-05,
861
- "loss": 0.35183258056640626,
862
- "step": 13900
863
- },
864
- {
865
- "epoch": 4.469987228607918,
866
- "learning_rate": 1.4040017028522777e-05,
867
- "loss": 0.35960380554199217,
868
- "step": 14000
869
- },
870
- {
871
- "epoch": 4.501915708812261,
872
- "learning_rate": 1.3997445721583653e-05,
873
- "loss": 0.3571435546875,
874
- "step": 14100
875
- },
876
- {
877
- "epoch": 4.533844189016603,
878
- "learning_rate": 1.395487441464453e-05,
879
- "loss": 0.35867115020751955,
880
- "step": 14200
881
- },
882
- {
883
- "epoch": 4.565772669220945,
884
- "learning_rate": 1.3912303107705407e-05,
885
- "loss": 0.34017555236816405,
886
- "step": 14300
887
- },
888
- {
889
- "epoch": 4.597701149425287,
890
- "learning_rate": 1.3869731800766283e-05,
891
- "loss": 0.3735758209228516,
892
- "step": 14400
893
- },
894
- {
895
- "epoch": 4.62962962962963,
896
- "learning_rate": 1.3827160493827162e-05,
897
- "loss": 0.36705623626708983,
898
- "step": 14500
899
- },
900
- {
901
- "epoch": 4.661558109833972,
902
- "learning_rate": 1.3784589186888038e-05,
903
- "loss": 0.3525060272216797,
904
- "step": 14600
905
- },
906
- {
907
- "epoch": 4.693486590038314,
908
- "learning_rate": 1.3742017879948915e-05,
909
- "loss": 0.3783878326416016,
910
- "step": 14700
911
- },
912
- {
913
- "epoch": 4.725415070242656,
914
- "learning_rate": 1.3699446573009792e-05,
915
- "loss": 0.3659477615356445,
916
- "step": 14800
917
- },
918
- {
919
- "epoch": 4.757343550446999,
920
- "learning_rate": 1.365687526607067e-05,
921
- "loss": 0.3537181091308594,
922
- "step": 14900
923
- },
924
- {
925
- "epoch": 4.789272030651341,
926
- "learning_rate": 1.3614303959131547e-05,
927
- "loss": 0.36351833343505857,
928
- "step": 15000
929
- },
930
- {
931
- "epoch": 4.821200510855683,
932
- "learning_rate": 1.3571732652192425e-05,
933
- "loss": 0.3483928298950195,
934
- "step": 15100
935
- },
936
- {
937
- "epoch": 4.853128991060025,
938
- "learning_rate": 1.3529161345253302e-05,
939
- "loss": 0.35997581481933594,
940
- "step": 15200
941
- },
942
- {
943
- "epoch": 4.885057471264368,
944
- "learning_rate": 1.3486590038314178e-05,
945
- "loss": 0.36364505767822264,
946
- "step": 15300
947
- },
948
- {
949
- "epoch": 4.91698595146871,
950
- "learning_rate": 1.3444018731375055e-05,
951
- "loss": 0.3437914276123047,
952
- "step": 15400
953
- },
954
- {
955
- "epoch": 4.948914431673052,
956
- "learning_rate": 1.3401447424435932e-05,
957
- "loss": 0.3517444610595703,
958
- "step": 15500
959
- },
960
- {
961
- "epoch": 4.980842911877395,
962
- "learning_rate": 1.3358876117496808e-05,
963
- "loss": 0.3623521041870117,
964
- "step": 15600
965
- },
966
- {
967
- "epoch": 5.0,
968
- "eval_loss": 0.33480075001716614,
969
- "step": 15660
970
- },
971
- {
972
- "epoch": 5.012771392081737,
973
- "learning_rate": 1.3316304810557685e-05,
974
- "loss": 0.3527105712890625,
975
- "step": 15700
976
- },
977
- {
978
- "epoch": 5.044699872286079,
979
- "learning_rate": 1.3273733503618562e-05,
980
- "loss": 0.3470969009399414,
981
- "step": 15800
982
- },
983
- {
984
- "epoch": 5.076628352490421,
985
- "learning_rate": 1.3231162196679438e-05,
986
- "loss": 0.34361572265625,
987
- "step": 15900
988
- },
989
- {
990
- "epoch": 5.108556832694764,
991
- "learning_rate": 1.3188590889740317e-05,
992
- "loss": 0.35975650787353514,
993
- "step": 16000
994
- },
995
- {
996
- "epoch": 5.140485312899106,
997
- "learning_rate": 1.3146019582801193e-05,
998
- "loss": 0.34139663696289063,
999
- "step": 16100
1000
- },
1001
- {
1002
- "epoch": 5.172413793103448,
1003
- "learning_rate": 1.310344827586207e-05,
1004
- "loss": 0.34681896209716795,
1005
- "step": 16200
1006
- },
1007
- {
1008
- "epoch": 5.20434227330779,
1009
- "learning_rate": 1.3060876968922947e-05,
1010
- "loss": 0.34409351348876954,
1011
- "step": 16300
1012
- },
1013
- {
1014
- "epoch": 5.236270753512133,
1015
- "learning_rate": 1.3018305661983823e-05,
1016
- "loss": 0.3398979949951172,
1017
- "step": 16400
1018
- },
1019
- {
1020
- "epoch": 5.268199233716475,
1021
- "learning_rate": 1.29757343550447e-05,
1022
- "loss": 0.3437363815307617,
1023
- "step": 16500
1024
- },
1025
- {
1026
- "epoch": 5.300127713920817,
1027
- "learning_rate": 1.2933163048105577e-05,
1028
- "loss": 0.33456321716308596,
1029
- "step": 16600
1030
- },
1031
- {
1032
- "epoch": 5.33205619412516,
1033
- "learning_rate": 1.2890591741166453e-05,
1034
- "loss": 0.3437419128417969,
1035
- "step": 16700
1036
- },
1037
- {
1038
- "epoch": 5.363984674329502,
1039
- "learning_rate": 1.2848020434227333e-05,
1040
- "loss": 0.3656898880004883,
1041
- "step": 16800
1042
- },
1043
- {
1044
- "epoch": 5.395913154533845,
1045
- "learning_rate": 1.280544912728821e-05,
1046
- "loss": 0.33624221801757814,
1047
- "step": 16900
1048
- },
1049
- {
1050
- "epoch": 5.427841634738186,
1051
- "learning_rate": 1.2762877820349087e-05,
1052
- "loss": 0.35640827178955076,
1053
- "step": 17000
1054
- },
1055
- {
1056
- "epoch": 5.459770114942529,
1057
- "learning_rate": 1.2720306513409963e-05,
1058
- "loss": 0.34689849853515625,
1059
- "step": 17100
1060
- },
1061
- {
1062
- "epoch": 5.491698595146871,
1063
- "learning_rate": 1.267773520647084e-05,
1064
- "loss": 0.3492275238037109,
1065
- "step": 17200
1066
- },
1067
- {
1068
- "epoch": 5.5236270753512136,
1069
- "learning_rate": 1.2635163899531717e-05,
1070
- "loss": 0.3490084457397461,
1071
- "step": 17300
1072
- },
1073
- {
1074
- "epoch": 5.555555555555555,
1075
- "learning_rate": 1.2592592592592593e-05,
1076
- "loss": 0.33231266021728517,
1077
- "step": 17400
1078
- },
1079
- {
1080
- "epoch": 5.587484035759898,
1081
- "learning_rate": 1.2550021285653472e-05,
1082
- "loss": 0.3501285552978516,
1083
- "step": 17500
1084
- },
1085
- {
1086
- "epoch": 5.61941251596424,
1087
- "learning_rate": 1.2507449978714348e-05,
1088
- "loss": 0.3469379425048828,
1089
- "step": 17600
1090
- },
1091
- {
1092
- "epoch": 5.6513409961685825,
1093
- "learning_rate": 1.2464878671775225e-05,
1094
- "loss": 0.3600951766967773,
1095
- "step": 17700
1096
- },
1097
- {
1098
- "epoch": 5.683269476372924,
1099
- "learning_rate": 1.2422307364836102e-05,
1100
- "loss": 0.3439628982543945,
1101
- "step": 17800
1102
- },
1103
- {
1104
- "epoch": 5.715197956577267,
1105
- "learning_rate": 1.2379736057896978e-05,
1106
- "loss": 0.34368560791015623,
1107
- "step": 17900
1108
- },
1109
- {
1110
- "epoch": 5.747126436781609,
1111
- "learning_rate": 1.2337164750957855e-05,
1112
- "loss": 0.3624436950683594,
1113
- "step": 18000
1114
- },
1115
- {
1116
- "epoch": 5.7790549169859515,
1117
- "learning_rate": 1.2294593444018732e-05,
1118
- "loss": 0.3493986129760742,
1119
- "step": 18100
1120
- },
1121
- {
1122
- "epoch": 5.810983397190293,
1123
- "learning_rate": 1.2252022137079608e-05,
1124
- "loss": 0.33745758056640623,
1125
- "step": 18200
1126
- },
1127
- {
1128
- "epoch": 5.842911877394636,
1129
- "learning_rate": 1.2209450830140485e-05,
1130
- "loss": 0.3514177703857422,
1131
- "step": 18300
1132
- },
1133
- {
1134
- "epoch": 5.874840357598979,
1135
- "learning_rate": 1.2166879523201363e-05,
1136
- "loss": 0.3409608459472656,
1137
- "step": 18400
1138
- },
1139
- {
1140
- "epoch": 5.9067688378033205,
1141
- "learning_rate": 1.212430821626224e-05,
1142
- "loss": 0.3416262054443359,
1143
- "step": 18500
1144
- },
1145
- {
1146
- "epoch": 5.938697318007663,
1147
- "learning_rate": 1.2081736909323117e-05,
1148
- "loss": 0.33906620025634765,
1149
- "step": 18600
1150
- },
1151
- {
1152
- "epoch": 5.970625798212005,
1153
- "learning_rate": 1.2039165602383993e-05,
1154
- "loss": 0.3565713119506836,
1155
- "step": 18700
1156
- },
1157
- {
1158
- "epoch": 6.0,
1159
- "eval_loss": 0.32608646154403687,
1160
- "step": 18792
1161
- },
1162
- {
1163
- "epoch": 6.002554278416348,
1164
- "learning_rate": 1.1996594295444872e-05,
1165
- "loss": 0.3423441314697266,
1166
- "step": 18800
1167
- },
1168
- {
1169
- "epoch": 6.0344827586206895,
1170
- "learning_rate": 1.1954022988505748e-05,
1171
- "loss": 0.3439017105102539,
1172
- "step": 18900
1173
- },
1174
- {
1175
- "epoch": 6.066411238825032,
1176
- "learning_rate": 1.1911451681566627e-05,
1177
- "loss": 0.33277828216552735,
1178
- "step": 19000
1179
- },
1180
- {
1181
- "epoch": 6.098339719029374,
1182
- "learning_rate": 1.1868880374627503e-05,
1183
- "loss": 0.3349918365478516,
1184
- "step": 19100
1185
- },
1186
- {
1187
- "epoch": 6.130268199233717,
1188
- "learning_rate": 1.182630906768838e-05,
1189
- "loss": 0.3411783981323242,
1190
- "step": 19200
1191
- },
1192
- {
1193
- "epoch": 6.1621966794380585,
1194
- "learning_rate": 1.1783737760749257e-05,
1195
- "loss": 0.34938507080078124,
1196
- "step": 19300
1197
- },
1198
- {
1199
- "epoch": 6.194125159642401,
1200
- "learning_rate": 1.1741166453810133e-05,
1201
- "loss": 0.3302972412109375,
1202
- "step": 19400
1203
- },
1204
- {
1205
- "epoch": 6.226053639846743,
1206
- "learning_rate": 1.169859514687101e-05,
1207
- "loss": 0.3324479293823242,
1208
- "step": 19500
1209
- },
1210
- {
1211
- "epoch": 6.257982120051086,
1212
- "learning_rate": 1.1656023839931887e-05,
1213
- "loss": 0.3325730133056641,
1214
- "step": 19600
1215
- },
1216
- {
1217
- "epoch": 6.2899106002554275,
1218
- "learning_rate": 1.1613452532992763e-05,
1219
- "loss": 0.32390750885009767,
1220
- "step": 19700
1221
- },
1222
- {
1223
- "epoch": 6.32183908045977,
1224
- "learning_rate": 1.157088122605364e-05,
1225
- "loss": 0.3410587692260742,
1226
- "step": 19800
1227
- },
1228
- {
1229
- "epoch": 6.353767560664112,
1230
- "learning_rate": 1.1528309919114518e-05,
1231
- "loss": 0.33700084686279297,
1232
- "step": 19900
1233
- },
1234
- {
1235
- "epoch": 6.385696040868455,
1236
- "learning_rate": 1.1485738612175395e-05,
1237
- "loss": 0.343173942565918,
1238
- "step": 20000
1239
- },
1240
- {
1241
- "epoch": 6.417624521072797,
1242
- "learning_rate": 1.1443167305236272e-05,
1243
- "loss": 0.34580535888671876,
1244
- "step": 20100
1245
- },
1246
- {
1247
- "epoch": 6.449553001277139,
1248
- "learning_rate": 1.1400595998297148e-05,
1249
- "loss": 0.3282489395141602,
1250
- "step": 20200
1251
- },
1252
- {
1253
- "epoch": 6.481481481481482,
1254
- "learning_rate": 1.1358024691358025e-05,
1255
- "loss": 0.32179367065429687,
1256
- "step": 20300
1257
- },
1258
- {
1259
- "epoch": 6.513409961685824,
1260
- "learning_rate": 1.1315453384418902e-05,
1261
- "loss": 0.34018295288085937,
1262
- "step": 20400
1263
- },
1264
- {
1265
- "epoch": 6.545338441890166,
1266
- "learning_rate": 1.1272882077479778e-05,
1267
- "loss": 0.3345378494262695,
1268
- "step": 20500
1269
- },
1270
- {
1271
- "epoch": 6.577266922094508,
1272
- "learning_rate": 1.1230310770540655e-05,
1273
- "loss": 0.34897972106933595,
1274
- "step": 20600
1275
- },
1276
- {
1277
- "epoch": 6.609195402298851,
1278
- "learning_rate": 1.1187739463601532e-05,
1279
- "loss": 0.3494709014892578,
1280
- "step": 20700
1281
- },
1282
- {
1283
- "epoch": 6.641123882503193,
1284
- "learning_rate": 1.1145168156662412e-05,
1285
- "loss": 0.3467377090454102,
1286
- "step": 20800
1287
- },
1288
- {
1289
- "epoch": 6.673052362707535,
1290
- "learning_rate": 1.1102596849723288e-05,
1291
- "loss": 0.3386357116699219,
1292
- "step": 20900
1293
- },
1294
- {
1295
- "epoch": 6.704980842911877,
1296
- "learning_rate": 1.1060025542784165e-05,
1297
- "loss": 0.3346070098876953,
1298
- "step": 21000
1299
- },
1300
- {
1301
- "epoch": 6.73690932311622,
1302
- "learning_rate": 1.1017454235845042e-05,
1303
- "loss": 0.3389591598510742,
1304
- "step": 21100
1305
- },
1306
- {
1307
- "epoch": 6.768837803320562,
1308
- "learning_rate": 1.0974882928905918e-05,
1309
- "loss": 0.33276222229003904,
1310
- "step": 21200
1311
- },
1312
- {
1313
- "epoch": 6.800766283524904,
1314
- "learning_rate": 1.0932311621966795e-05,
1315
- "loss": 0.3417454528808594,
1316
- "step": 21300
1317
- },
1318
- {
1319
- "epoch": 6.832694763729246,
1320
- "learning_rate": 1.0889740315027673e-05,
1321
- "loss": 0.3352804183959961,
1322
- "step": 21400
1323
- },
1324
- {
1325
- "epoch": 6.864623243933589,
1326
- "learning_rate": 1.084716900808855e-05,
1327
- "loss": 0.341392822265625,
1328
- "step": 21500
1329
- },
1330
- {
1331
- "epoch": 6.896551724137931,
1332
- "learning_rate": 1.0804597701149427e-05,
1333
- "loss": 0.3257337188720703,
1334
- "step": 21600
1335
- },
1336
- {
1337
- "epoch": 6.928480204342273,
1338
- "learning_rate": 1.0762026394210303e-05,
1339
- "loss": 0.340169792175293,
1340
- "step": 21700
1341
- },
1342
- {
1343
- "epoch": 6.960408684546616,
1344
- "learning_rate": 1.071945508727118e-05,
1345
- "loss": 0.3393547821044922,
1346
- "step": 21800
1347
- },
1348
- {
1349
- "epoch": 6.992337164750958,
1350
- "learning_rate": 1.0676883780332057e-05,
1351
- "loss": 0.3345566940307617,
1352
- "step": 21900
1353
- },
1354
- {
1355
- "epoch": 7.0,
1356
- "eval_loss": 0.3236748278141022,
1357
- "step": 21924
1358
- },
1359
- {
1360
- "epoch": 7.0242656449553005,
1361
- "learning_rate": 1.0634312473392933e-05,
1362
- "loss": 0.341474609375,
1363
- "step": 22000
1364
- },
1365
- {
1366
- "epoch": 7.056194125159642,
1367
- "learning_rate": 1.059174116645381e-05,
1368
- "loss": 0.33665103912353517,
1369
- "step": 22100
1370
- },
1371
- {
1372
- "epoch": 7.088122605363985,
1373
- "learning_rate": 1.0549169859514687e-05,
1374
- "loss": 0.33666282653808594,
1375
- "step": 22200
1376
- },
1377
- {
1378
- "epoch": 7.120051085568327,
1379
- "learning_rate": 1.0506598552575565e-05,
1380
- "loss": 0.32267608642578127,
1381
- "step": 22300
1382
- },
1383
- {
1384
- "epoch": 7.1519795657726695,
1385
- "learning_rate": 1.0464027245636442e-05,
1386
- "loss": 0.32230213165283206,
1387
- "step": 22400
1388
- },
1389
- {
1390
- "epoch": 7.183908045977011,
1391
- "learning_rate": 1.0421455938697318e-05,
1392
- "loss": 0.3335090637207031,
1393
- "step": 22500
1394
- },
1395
- {
1396
- "epoch": 7.215836526181354,
1397
- "learning_rate": 1.0378884631758195e-05,
1398
- "loss": 0.3402565002441406,
1399
- "step": 22600
1400
- },
1401
- {
1402
- "epoch": 7.247765006385696,
1403
- "learning_rate": 1.0336313324819072e-05,
1404
- "loss": 0.33222633361816406,
1405
- "step": 22700
1406
- },
1407
- {
1408
- "epoch": 7.2796934865900385,
1409
- "learning_rate": 1.029374201787995e-05,
1410
- "loss": 0.3273036956787109,
1411
- "step": 22800
1412
- },
1413
- {
1414
- "epoch": 7.31162196679438,
1415
- "learning_rate": 1.0251170710940828e-05,
1416
- "loss": 0.34040824890136717,
1417
- "step": 22900
1418
- },
1419
- {
1420
- "epoch": 7.343550446998723,
1421
- "learning_rate": 1.0208599404001705e-05,
1422
- "loss": 0.33351837158203124,
1423
- "step": 23000
1424
- },
1425
- {
1426
- "epoch": 7.375478927203065,
1427
- "learning_rate": 1.0166028097062582e-05,
1428
- "loss": 0.3304040145874023,
1429
- "step": 23100
1430
- },
1431
- {
1432
- "epoch": 7.407407407407407,
1433
- "learning_rate": 1.0123456790123458e-05,
1434
- "loss": 0.3389059829711914,
1435
- "step": 23200
1436
- },
1437
- {
1438
- "epoch": 7.439335887611749,
1439
- "learning_rate": 1.0080885483184335e-05,
1440
- "loss": 0.35038864135742187,
1441
- "step": 23300
1442
- },
1443
- {
1444
- "epoch": 7.471264367816092,
1445
- "learning_rate": 1.0038314176245212e-05,
1446
- "loss": 0.32688159942626954,
1447
- "step": 23400
1448
- },
1449
- {
1450
- "epoch": 7.503192848020435,
1451
- "learning_rate": 9.995742869306088e-06,
1452
- "loss": 0.3443561935424805,
1453
- "step": 23500
1454
- },
1455
- {
1456
- "epoch": 7.535121328224776,
1457
- "learning_rate": 9.953171562366965e-06,
1458
- "loss": 0.33195884704589845,
1459
- "step": 23600
1460
- },
1461
- {
1462
- "epoch": 7.567049808429119,
1463
- "learning_rate": 9.910600255427842e-06,
1464
- "loss": 0.3439883041381836,
1465
- "step": 23700
1466
- },
1467
- {
1468
- "epoch": 7.598978288633461,
1469
- "learning_rate": 9.86802894848872e-06,
1470
- "loss": 0.32301868438720704,
1471
- "step": 23800
1472
- },
1473
- {
1474
- "epoch": 7.630906768837804,
1475
- "learning_rate": 9.825457641549597e-06,
1476
- "loss": 0.3437799072265625,
1477
- "step": 23900
1478
- },
1479
- {
1480
- "epoch": 7.662835249042145,
1481
- "learning_rate": 9.782886334610473e-06,
1482
- "loss": 0.3296714401245117,
1483
- "step": 24000
1484
- },
1485
- {
1486
- "epoch": 7.694763729246488,
1487
- "learning_rate": 9.74031502767135e-06,
1488
- "loss": 0.3370498275756836,
1489
- "step": 24100
1490
- },
1491
- {
1492
- "epoch": 7.72669220945083,
1493
- "learning_rate": 9.697743720732228e-06,
1494
- "loss": 0.3219729232788086,
1495
- "step": 24200
1496
- },
1497
- {
1498
- "epoch": 7.758620689655173,
1499
- "learning_rate": 9.655172413793105e-06,
1500
- "loss": 0.3258438491821289,
1501
- "step": 24300
1502
- },
1503
- {
1504
- "epoch": 7.790549169859514,
1505
- "learning_rate": 9.612601106853982e-06,
1506
- "loss": 0.33826839447021484,
1507
- "step": 24400
1508
- },
1509
- {
1510
- "epoch": 7.822477650063857,
1511
- "learning_rate": 9.570029799914858e-06,
1512
- "loss": 0.3263700866699219,
1513
- "step": 24500
1514
- },
1515
- {
1516
- "epoch": 7.854406130268199,
1517
- "learning_rate": 9.527458492975735e-06,
1518
- "loss": 0.3350722122192383,
1519
- "step": 24600
1520
- },
1521
- {
1522
- "epoch": 7.886334610472542,
1523
- "learning_rate": 9.484887186036612e-06,
1524
- "loss": 0.334184455871582,
1525
- "step": 24700
1526
- },
1527
- {
1528
- "epoch": 7.918263090676884,
1529
- "learning_rate": 9.442315879097488e-06,
1530
- "loss": 0.3267938232421875,
1531
- "step": 24800
1532
- },
1533
- {
1534
- "epoch": 7.950191570881226,
1535
- "learning_rate": 9.399744572158365e-06,
1536
- "loss": 0.3196305465698242,
1537
- "step": 24900
1538
- },
1539
- {
1540
- "epoch": 7.982120051085568,
1541
- "learning_rate": 9.357173265219243e-06,
1542
- "loss": 0.3230043029785156,
1543
- "step": 25000
1544
- },
1545
- {
1546
- "epoch": 8.0,
1547
- "eval_loss": 0.32138851284980774,
1548
- "step": 25056
1549
- },
1550
- {
1551
- "epoch": 8.01404853128991,
1552
- "learning_rate": 9.31460195828012e-06,
1553
- "loss": 0.3258121109008789,
1554
- "step": 25100
1555
- },
1556
- {
1557
- "epoch": 8.045977011494253,
1558
- "learning_rate": 9.272030651340997e-06,
1559
- "loss": 0.32868377685546873,
1560
- "step": 25200
1561
- },
1562
- {
1563
- "epoch": 8.077905491698596,
1564
- "learning_rate": 9.229459344401875e-06,
1565
- "loss": 0.31930324554443357,
1566
- "step": 25300
1567
- },
1568
- {
1569
- "epoch": 8.109833971902937,
1570
- "learning_rate": 9.186888037462752e-06,
1571
- "loss": 0.3216040420532227,
1572
- "step": 25400
1573
- },
1574
- {
1575
- "epoch": 8.14176245210728,
1576
- "learning_rate": 9.144316730523628e-06,
1577
- "loss": 0.3392459869384766,
1578
- "step": 25500
1579
- },
1580
- {
1581
- "epoch": 8.173690932311622,
1582
- "learning_rate": 9.101745423584505e-06,
1583
- "loss": 0.3244002914428711,
1584
- "step": 25600
1585
- },
1586
- {
1587
- "epoch": 8.205619412515965,
1588
- "learning_rate": 9.059174116645382e-06,
1589
- "loss": 0.329796142578125,
1590
- "step": 25700
1591
- },
1592
- {
1593
- "epoch": 8.237547892720306,
1594
- "learning_rate": 9.016602809706258e-06,
1595
- "loss": 0.310886344909668,
1596
- "step": 25800
1597
- },
1598
- {
1599
- "epoch": 8.269476372924649,
1600
- "learning_rate": 8.974031502767135e-06,
1601
- "loss": 0.3182815361022949,
1602
- "step": 25900
1603
- },
1604
- {
1605
- "epoch": 8.301404853128991,
1606
- "learning_rate": 8.931460195828012e-06,
1607
- "loss": 0.3267630386352539,
1608
- "step": 26000
1609
- },
1610
- {
1611
- "epoch": 8.333333333333334,
1612
- "learning_rate": 8.888888888888888e-06,
1613
- "loss": 0.31576622009277344,
1614
- "step": 26100
1615
- },
1616
- {
1617
- "epoch": 8.365261813537675,
1618
- "learning_rate": 8.846317581949767e-06,
1619
- "loss": 0.33119094848632813,
1620
- "step": 26200
1621
- },
1622
- {
1623
- "epoch": 8.397190293742018,
1624
- "learning_rate": 8.803746275010643e-06,
1625
- "loss": 0.3301513671875,
1626
- "step": 26300
1627
- },
1628
- {
1629
- "epoch": 8.42911877394636,
1630
- "learning_rate": 8.76117496807152e-06,
1631
- "loss": 0.3321194839477539,
1632
- "step": 26400
1633
- },
1634
- {
1635
- "epoch": 8.461047254150703,
1636
- "learning_rate": 8.718603661132398e-06,
1637
- "loss": 0.332429313659668,
1638
- "step": 26500
1639
- },
1640
- {
1641
- "epoch": 8.492975734355046,
1642
- "learning_rate": 8.676032354193275e-06,
1643
- "loss": 0.3258559036254883,
1644
- "step": 26600
1645
- },
1646
- {
1647
- "epoch": 8.524904214559387,
1648
- "learning_rate": 8.633461047254152e-06,
1649
- "loss": 0.32335933685302737,
1650
- "step": 26700
1651
- },
1652
- {
1653
- "epoch": 8.55683269476373,
1654
- "learning_rate": 8.590889740315028e-06,
1655
- "loss": 0.3441028594970703,
1656
- "step": 26800
1657
- },
1658
- {
1659
- "epoch": 8.588761174968072,
1660
- "learning_rate": 8.548318433375905e-06,
1661
- "loss": 0.3302944564819336,
1662
- "step": 26900
1663
- },
1664
- {
1665
- "epoch": 8.620689655172415,
1666
- "learning_rate": 8.505747126436782e-06,
1667
- "loss": 0.3203293228149414,
1668
- "step": 27000
1669
- },
1670
- {
1671
- "epoch": 8.652618135376756,
1672
- "learning_rate": 8.463175819497658e-06,
1673
- "loss": 0.3180072593688965,
1674
- "step": 27100
1675
- },
1676
- {
1677
- "epoch": 8.684546615581098,
1678
- "learning_rate": 8.420604512558537e-06,
1679
- "loss": 0.3143235015869141,
1680
- "step": 27200
1681
- },
1682
- {
1683
- "epoch": 8.71647509578544,
1684
- "learning_rate": 8.378033205619413e-06,
1685
- "loss": 0.31695037841796875,
1686
- "step": 27300
1687
- },
1688
- {
1689
- "epoch": 8.748403575989784,
1690
- "learning_rate": 8.33546189868029e-06,
1691
- "loss": 0.31933542251586916,
1692
- "step": 27400
1693
- },
1694
- {
1695
- "epoch": 8.780332056194124,
1696
- "learning_rate": 8.292890591741167e-06,
1697
- "loss": 0.3094404983520508,
1698
- "step": 27500
1699
- },
1700
- {
1701
- "epoch": 8.812260536398467,
1702
- "learning_rate": 8.250319284802043e-06,
1703
- "loss": 0.3251906204223633,
1704
- "step": 27600
1705
- },
1706
- {
1707
- "epoch": 8.84418901660281,
1708
- "learning_rate": 8.207747977862922e-06,
1709
- "loss": 0.31949323654174805,
1710
- "step": 27700
1711
- },
1712
- {
1713
- "epoch": 8.876117496807153,
1714
- "learning_rate": 8.165176670923798e-06,
1715
- "loss": 0.32407459259033206,
1716
- "step": 27800
1717
- },
1718
- {
1719
- "epoch": 8.908045977011493,
1720
- "learning_rate": 8.122605363984675e-06,
1721
- "loss": 0.34080780029296875,
1722
- "step": 27900
1723
- },
1724
- {
1725
- "epoch": 8.939974457215836,
1726
- "learning_rate": 8.080034057045552e-06,
1727
- "loss": 0.3133597183227539,
1728
- "step": 28000
1729
- },
1730
- {
1731
- "epoch": 8.971902937420179,
1732
- "learning_rate": 8.037462750106428e-06,
1733
- "loss": 0.31053606033325193,
1734
- "step": 28100
1735
- },
1736
- {
1737
- "epoch": 9.0,
1738
- "eval_loss": 0.3119257688522339,
1739
- "step": 28188
1740
- },
1741
- {
1742
- "epoch": 9.003831417624522,
1743
- "learning_rate": 7.994891443167307e-06,
1744
- "loss": 0.3158924293518066,
1745
- "step": 28200
1746
- },
1747
- {
1748
- "epoch": 9.035759897828862,
1749
- "learning_rate": 7.952320136228183e-06,
1750
- "loss": 0.3117160987854004,
1751
- "step": 28300
1752
- },
1753
- {
1754
- "epoch": 9.067688378033205,
1755
- "learning_rate": 7.90974882928906e-06,
1756
- "loss": 0.32216869354248046,
1757
- "step": 28400
1758
- },
1759
- {
1760
- "epoch": 9.099616858237548,
1761
- "learning_rate": 7.867177522349937e-06,
1762
- "loss": 0.314890079498291,
1763
- "step": 28500
1764
- },
1765
- {
1766
- "epoch": 9.13154533844189,
1767
- "learning_rate": 7.824606215410813e-06,
1768
- "loss": 0.3113277626037598,
1769
- "step": 28600
1770
- },
1771
- {
1772
- "epoch": 9.163473818646233,
1773
- "learning_rate": 7.78203490847169e-06,
1774
- "loss": 0.324998779296875,
1775
- "step": 28700
1776
- },
1777
- {
1778
- "epoch": 9.195402298850574,
1779
- "learning_rate": 7.739463601532567e-06,
1780
- "loss": 0.32509784698486327,
1781
- "step": 28800
1782
- },
1783
- {
1784
- "epoch": 9.227330779054917,
1785
- "learning_rate": 7.696892294593445e-06,
1786
- "loss": 0.3238474273681641,
1787
- "step": 28900
1788
- },
1789
- {
1790
- "epoch": 9.25925925925926,
1791
- "learning_rate": 7.654320987654322e-06,
1792
- "loss": 0.33168052673339843,
1793
- "step": 29000
1794
- },
1795
- {
1796
- "epoch": 9.291187739463602,
1797
- "learning_rate": 7.611749680715198e-06,
1798
- "loss": 0.3050485992431641,
1799
- "step": 29100
1800
- },
1801
- {
1802
- "epoch": 9.323116219667943,
1803
- "learning_rate": 7.569178373776076e-06,
1804
- "loss": 0.31826154708862303,
1805
- "step": 29200
1806
- },
1807
- {
1808
- "epoch": 9.355044699872286,
1809
- "learning_rate": 7.5266070668369525e-06,
1810
- "loss": 0.31972583770751956,
1811
- "step": 29300
1812
- },
1813
- {
1814
- "epoch": 9.386973180076629,
1815
- "learning_rate": 7.48403575989783e-06,
1816
- "loss": 0.3287076568603516,
1817
- "step": 29400
1818
- },
1819
- {
1820
- "epoch": 9.418901660280971,
1821
- "learning_rate": 7.441464452958707e-06,
1822
- "loss": 0.30761892318725587,
1823
- "step": 29500
1824
- },
1825
- {
1826
- "epoch": 9.450830140485312,
1827
- "learning_rate": 7.398893146019583e-06,
1828
- "loss": 0.3097171401977539,
1829
- "step": 29600
1830
- },
1831
- {
1832
- "epoch": 9.482758620689655,
1833
- "learning_rate": 7.35632183908046e-06,
1834
- "loss": 0.30806644439697267,
1835
- "step": 29700
1836
- },
1837
- {
1838
- "epoch": 9.514687100893997,
1839
- "learning_rate": 7.3137505321413375e-06,
1840
- "loss": 0.3174296760559082,
1841
- "step": 29800
1842
- },
1843
- {
1844
- "epoch": 9.54661558109834,
1845
- "learning_rate": 7.271179225202214e-06,
1846
- "loss": 0.31353973388671874,
1847
- "step": 29900
1848
- },
1849
- {
1850
- "epoch": 9.578544061302683,
1851
- "learning_rate": 7.228607918263091e-06,
1852
- "loss": 0.30578601837158204,
1853
- "step": 30000
1854
- },
1855
- {
1856
- "epoch": 9.610472541507024,
1857
- "learning_rate": 7.1860366113239675e-06,
1858
- "loss": 0.3045210838317871,
1859
- "step": 30100
1860
- },
1861
- {
1862
- "epoch": 9.642401021711366,
1863
- "learning_rate": 7.143465304384846e-06,
1864
- "loss": 0.3149559211730957,
1865
- "step": 30200
1866
- },
1867
- {
1868
- "epoch": 9.67432950191571,
1869
- "learning_rate": 7.1008939974457225e-06,
1870
- "loss": 0.3207520294189453,
1871
- "step": 30300
1872
- },
1873
- {
1874
- "epoch": 9.706257982120052,
1875
- "learning_rate": 7.058322690506599e-06,
1876
- "loss": 0.32308254241943357,
1877
- "step": 30400
1878
- },
1879
- {
1880
- "epoch": 9.738186462324393,
1881
- "learning_rate": 7.015751383567476e-06,
1882
- "loss": 0.3183433723449707,
1883
- "step": 30500
1884
- },
1885
- {
1886
- "epoch": 9.770114942528735,
1887
- "learning_rate": 6.973180076628353e-06,
1888
- "loss": 0.3200767135620117,
1889
- "step": 30600
1890
- },
1891
- {
1892
- "epoch": 9.802043422733078,
1893
- "learning_rate": 6.93060876968923e-06,
1894
- "loss": 0.3113987922668457,
1895
- "step": 30700
1896
- },
1897
- {
1898
- "epoch": 9.83397190293742,
1899
- "learning_rate": 6.888037462750107e-06,
1900
- "loss": 0.33167327880859376,
1901
- "step": 30800
1902
- },
1903
- {
1904
- "epoch": 9.865900383141762,
1905
- "learning_rate": 6.845466155810983e-06,
1906
- "loss": 0.3172581100463867,
1907
- "step": 30900
1908
- },
1909
- {
1910
- "epoch": 9.897828863346104,
1911
- "learning_rate": 6.802894848871861e-06,
1912
- "loss": 0.3124402046203613,
1913
- "step": 31000
1914
- },
1915
- {
1916
- "epoch": 9.929757343550447,
1917
- "learning_rate": 6.760323541932738e-06,
1918
- "loss": 0.30687171936035157,
1919
- "step": 31100
1920
- },
1921
- {
1922
- "epoch": 9.96168582375479,
1923
- "learning_rate": 6.717752234993615e-06,
1924
- "loss": 0.304738712310791,
1925
- "step": 31200
1926
- },
1927
- {
1928
- "epoch": 9.99361430395913,
1929
- "learning_rate": 6.6751809280544925e-06,
1930
- "loss": 0.3069480514526367,
1931
- "step": 31300
1932
- },
1933
- {
1934
- "epoch": 10.0,
1935
- "eval_loss": 0.31222018599510193,
1936
- "step": 31320
1937
- },
1938
- {
1939
- "epoch": 10.025542784163473,
1940
- "learning_rate": 6.632609621115369e-06,
1941
- "loss": 0.30982017517089844,
1942
- "step": 31400
1943
- },
1944
- {
1945
- "epoch": 10.057471264367816,
1946
- "learning_rate": 6.590038314176246e-06,
1947
- "loss": 0.29576118469238283,
1948
- "step": 31500
1949
- },
1950
- {
1951
- "epoch": 10.089399744572159,
1952
- "learning_rate": 6.5474670072371225e-06,
1953
- "loss": 0.318297004699707,
1954
- "step": 31600
1955
- },
1956
- {
1957
- "epoch": 10.121328224776502,
1958
- "learning_rate": 6.504895700297999e-06,
1959
- "loss": 0.2947650337219238,
1960
- "step": 31700
1961
- },
1962
- {
1963
- "epoch": 10.153256704980842,
1964
- "learning_rate": 6.462324393358877e-06,
1965
- "loss": 0.31912431716918943,
1966
- "step": 31800
1967
- },
1968
- {
1969
- "epoch": 10.185185185185185,
1970
- "learning_rate": 6.419753086419753e-06,
1971
- "loss": 0.31549993515014646,
1972
- "step": 31900
1973
- },
1974
- {
1975
- "epoch": 10.217113665389528,
1976
- "learning_rate": 6.37718177948063e-06,
1977
- "loss": 0.3185459327697754,
1978
- "step": 32000
1979
- },
1980
- {
1981
- "epoch": 10.24904214559387,
1982
- "learning_rate": 6.334610472541508e-06,
1983
- "loss": 0.3061997413635254,
1984
- "step": 32100
1985
- },
1986
- {
1987
- "epoch": 10.280970625798211,
1988
- "learning_rate": 6.292039165602385e-06,
1989
- "loss": 0.30356922149658205,
1990
- "step": 32200
1991
- },
1992
- {
1993
- "epoch": 10.312899106002554,
1994
- "learning_rate": 6.249467858663262e-06,
1995
- "loss": 0.30677152633666993,
1996
- "step": 32300
1997
- },
1998
- {
1999
- "epoch": 10.344827586206897,
2000
- "learning_rate": 6.206896551724138e-06,
2001
- "loss": 0.3148806190490723,
2002
- "step": 32400
2003
- },
2004
- {
2005
- "epoch": 10.37675606641124,
2006
- "learning_rate": 6.164325244785016e-06,
2007
- "loss": 0.3114926528930664,
2008
- "step": 32500
2009
- },
2010
- {
2011
- "epoch": 10.40868454661558,
2012
- "learning_rate": 6.1217539378458925e-06,
2013
- "loss": 0.31096630096435546,
2014
- "step": 32600
2015
- },
2016
- {
2017
- "epoch": 10.440613026819923,
2018
- "learning_rate": 6.079182630906769e-06,
2019
- "loss": 0.3131294822692871,
2020
- "step": 32700
2021
- },
2022
- {
2023
- "epoch": 10.472541507024266,
2024
- "learning_rate": 6.036611323967646e-06,
2025
- "loss": 0.31513975143432615,
2026
- "step": 32800
2027
- },
2028
- {
2029
- "epoch": 10.504469987228608,
2030
- "learning_rate": 5.9940400170285225e-06,
2031
- "loss": 0.3035664939880371,
2032
- "step": 32900
2033
- },
2034
- {
2035
- "epoch": 10.53639846743295,
2036
- "learning_rate": 5.9514687100894e-06,
2037
- "loss": 0.30514934539794925,
2038
- "step": 33000
2039
- },
2040
- {
2041
- "epoch": 10.568326947637292,
2042
- "learning_rate": 5.9088974031502775e-06,
2043
- "loss": 0.3236639404296875,
2044
- "step": 33100
2045
- },
2046
- {
2047
- "epoch": 10.600255427841635,
2048
- "learning_rate": 5.866326096211154e-06,
2049
- "loss": 0.32729095458984375,
2050
- "step": 33200
2051
- },
2052
- {
2053
- "epoch": 10.632183908045977,
2054
- "learning_rate": 5.823754789272032e-06,
2055
- "loss": 0.30390745162963867,
2056
- "step": 33300
2057
- },
2058
- {
2059
- "epoch": 10.66411238825032,
2060
- "learning_rate": 5.781183482332908e-06,
2061
- "loss": 0.30320255279541014,
2062
- "step": 33400
2063
- },
2064
- {
2065
- "epoch": 10.696040868454661,
2066
- "learning_rate": 5.738612175393785e-06,
2067
- "loss": 0.31617319107055664,
2068
- "step": 33500
2069
- },
2070
- {
2071
- "epoch": 10.727969348659004,
2072
- "learning_rate": 5.696040868454662e-06,
2073
- "loss": 0.31349088668823244,
2074
- "step": 33600
2075
- },
2076
- {
2077
- "epoch": 10.759897828863346,
2078
- "learning_rate": 5.653469561515539e-06,
2079
- "loss": 0.3055162620544434,
2080
- "step": 33700
2081
- },
2082
- {
2083
- "epoch": 10.79182630906769,
2084
- "learning_rate": 5.610898254576416e-06,
2085
- "loss": 0.29683116912841795,
2086
- "step": 33800
2087
- },
2088
- {
2089
- "epoch": 10.82375478927203,
2090
- "learning_rate": 5.5683269476372925e-06,
2091
- "loss": 0.2972592926025391,
2092
- "step": 33900
2093
- },
2094
- {
2095
- "epoch": 10.855683269476373,
2096
- "learning_rate": 5.525755640698169e-06,
2097
- "loss": 0.30813514709472656,
2098
- "step": 34000
2099
- },
2100
- {
2101
- "epoch": 10.887611749680715,
2102
- "learning_rate": 5.4831843337590475e-06,
2103
- "loss": 0.311496696472168,
2104
- "step": 34100
2105
- },
2106
- {
2107
- "epoch": 10.919540229885058,
2108
- "learning_rate": 5.440613026819924e-06,
2109
- "loss": 0.3147770881652832,
2110
- "step": 34200
2111
- },
2112
- {
2113
- "epoch": 10.951468710089399,
2114
- "learning_rate": 5.398041719880801e-06,
2115
- "loss": 0.2977629852294922,
2116
- "step": 34300
2117
- },
2118
- {
2119
- "epoch": 10.983397190293742,
2120
- "learning_rate": 5.3554704129416775e-06,
2121
- "loss": 0.30722129821777344,
2122
- "step": 34400
2123
- },
2124
- {
2125
- "epoch": 11.0,
2126
- "eval_loss": 0.30927398800849915,
2127
- "step": 34452
2128
- },
2129
- {
2130
- "epoch": 11.015325670498084,
2131
- "learning_rate": 5.312899106002555e-06,
2132
- "loss": 0.30775304794311525,
2133
- "step": 34500
2134
- },
2135
- {
2136
- "epoch": 11.047254150702427,
2137
- "learning_rate": 5.270327799063432e-06,
2138
- "loss": 0.31844793319702147,
2139
- "step": 34600
2140
- },
2141
- {
2142
- "epoch": 11.079182630906768,
2143
- "learning_rate": 5.227756492124308e-06,
2144
- "loss": 0.3163930511474609,
2145
- "step": 34700
2146
- },
2147
- {
2148
- "epoch": 11.11111111111111,
2149
- "learning_rate": 5.185185185185185e-06,
2150
- "loss": 0.3178179359436035,
2151
- "step": 34800
2152
- },
2153
- {
2154
- "epoch": 11.143039591315453,
2155
- "learning_rate": 5.1426138782460625e-06,
2156
- "loss": 0.3107015609741211,
2157
- "step": 34900
2158
- },
2159
- {
2160
- "epoch": 11.174968071519796,
2161
- "learning_rate": 5.100042571306939e-06,
2162
- "loss": 0.31198001861572267,
2163
- "step": 35000
2164
- },
2165
- {
2166
- "epoch": 11.206896551724139,
2167
- "learning_rate": 5.057471264367817e-06,
2168
- "loss": 0.30287263870239256,
2169
- "step": 35100
2170
- },
2171
- {
2172
- "epoch": 11.23882503192848,
2173
- "learning_rate": 5.014899957428694e-06,
2174
- "loss": 0.3046586036682129,
2175
- "step": 35200
2176
- },
2177
- {
2178
- "epoch": 11.270753512132822,
2179
- "learning_rate": 4.972328650489571e-06,
2180
- "loss": 0.29926385879516604,
2181
- "step": 35300
2182
- },
2183
- {
2184
- "epoch": 11.302681992337165,
2185
- "learning_rate": 4.9297573435504475e-06,
2186
- "loss": 0.31370662689208983,
2187
- "step": 35400
2188
- },
2189
- {
2190
- "epoch": 11.334610472541508,
2191
- "learning_rate": 4.887186036611324e-06,
2192
- "loss": 0.3177505874633789,
2193
- "step": 35500
2194
- },
2195
- {
2196
- "epoch": 11.366538952745849,
2197
- "learning_rate": 4.844614729672202e-06,
2198
- "loss": 0.3028862190246582,
2199
- "step": 35600
2200
- },
2201
- {
2202
- "epoch": 11.398467432950191,
2203
- "learning_rate": 4.802043422733078e-06,
2204
- "loss": 0.2965104293823242,
2205
- "step": 35700
2206
- },
2207
- {
2208
- "epoch": 11.430395913154534,
2209
- "learning_rate": 4.759472115793956e-06,
2210
- "loss": 0.31520273208618166,
2211
- "step": 35800
2212
- },
2213
- {
2214
- "epoch": 11.462324393358877,
2215
- "learning_rate": 4.7169008088548325e-06,
2216
- "loss": 0.2961687469482422,
2217
- "step": 35900
2218
- },
2219
- {
2220
- "epoch": 11.494252873563218,
2221
- "learning_rate": 4.674329501915709e-06,
2222
- "loss": 0.30442037582397463,
2223
- "step": 36000
2224
- },
2225
- {
2226
- "epoch": 11.52618135376756,
2227
- "learning_rate": 4.631758194976586e-06,
2228
- "loss": 0.3178094863891602,
2229
- "step": 36100
2230
- },
2231
- {
2232
- "epoch": 11.558109833971903,
2233
- "learning_rate": 4.589186888037463e-06,
2234
- "loss": 0.3036604118347168,
2235
- "step": 36200
2236
- },
2237
- {
2238
- "epoch": 11.590038314176246,
2239
- "learning_rate": 4.54661558109834e-06,
2240
- "loss": 0.3115557861328125,
2241
- "step": 36300
2242
- },
2243
- {
2244
- "epoch": 11.621966794380587,
2245
- "learning_rate": 4.5040442741592175e-06,
2246
- "loss": 0.30345108032226564,
2247
- "step": 36400
2248
- },
2249
- {
2250
- "epoch": 11.65389527458493,
2251
- "learning_rate": 4.461472967220094e-06,
2252
- "loss": 0.30634918212890627,
2253
- "step": 36500
2254
- },
2255
- {
2256
- "epoch": 11.685823754789272,
2257
- "learning_rate": 4.418901660280971e-06,
2258
- "loss": 0.3051659774780273,
2259
- "step": 36600
2260
- },
2261
- {
2262
- "epoch": 11.717752234993615,
2263
- "learning_rate": 4.3763303533418475e-06,
2264
- "loss": 0.3069002342224121,
2265
- "step": 36700
2266
- },
2267
- {
2268
- "epoch": 11.749680715197957,
2269
- "learning_rate": 4.333759046402725e-06,
2270
- "loss": 0.30947404861450195,
2271
- "step": 36800
2272
- },
2273
- {
2274
- "epoch": 11.781609195402298,
2275
- "learning_rate": 4.291187739463602e-06,
2276
- "loss": 0.3073232650756836,
2277
- "step": 36900
2278
- },
2279
- {
2280
- "epoch": 11.813537675606641,
2281
- "learning_rate": 4.248616432524479e-06,
2282
- "loss": 0.3001542472839355,
2283
- "step": 37000
2284
- },
2285
- {
2286
- "epoch": 11.845466155810984,
2287
- "learning_rate": 4.206045125585356e-06,
2288
- "loss": 0.31570695877075194,
2289
- "step": 37100
2290
- },
2291
- {
2292
- "epoch": 11.877394636015326,
2293
- "learning_rate": 4.1634738186462325e-06,
2294
- "loss": 0.311122932434082,
2295
- "step": 37200
2296
- },
2297
- {
2298
- "epoch": 11.909323116219667,
2299
- "learning_rate": 4.12090251170711e-06,
2300
- "loss": 0.301647891998291,
2301
- "step": 37300
2302
- },
2303
- {
2304
- "epoch": 11.94125159642401,
2305
- "learning_rate": 4.078331204767987e-06,
2306
- "loss": 0.3205462646484375,
2307
- "step": 37400
2308
- },
2309
- {
2310
- "epoch": 11.973180076628353,
2311
- "learning_rate": 4.035759897828863e-06,
2312
- "loss": 0.29651784896850586,
2313
- "step": 37500
2314
- },
2315
- {
2316
- "epoch": 12.0,
2317
- "eval_loss": 0.3090454041957855,
2318
- "step": 37584
2319
- },
2320
- {
2321
- "epoch": 12.005108556832695,
2322
- "learning_rate": 3.993188590889741e-06,
2323
- "loss": 0.31057783126831057,
2324
- "step": 37600
2325
- },
2326
- {
2327
- "epoch": 12.037037037037036,
2328
- "learning_rate": 3.9506172839506175e-06,
2329
- "loss": 0.305908145904541,
2330
- "step": 37700
2331
- },
2332
- {
2333
- "epoch": 12.068965517241379,
2334
- "learning_rate": 3.908045977011495e-06,
2335
- "loss": 0.3145115280151367,
2336
- "step": 37800
2337
- },
2338
- {
2339
- "epoch": 12.100893997445722,
2340
- "learning_rate": 3.865474670072372e-06,
2341
- "loss": 0.3086430168151855,
2342
- "step": 37900
2343
- },
2344
- {
2345
- "epoch": 12.132822477650064,
2346
- "learning_rate": 3.822903363133248e-06,
2347
- "loss": 0.3069817733764648,
2348
- "step": 38000
2349
- },
2350
- {
2351
- "epoch": 12.164750957854405,
2352
- "learning_rate": 3.7803320561941254e-06,
2353
- "loss": 0.3044874954223633,
2354
- "step": 38100
2355
- },
2356
- {
2357
- "epoch": 12.196679438058748,
2358
- "learning_rate": 3.737760749255002e-06,
2359
- "loss": 0.3003558731079102,
2360
- "step": 38200
2361
- },
2362
- {
2363
- "epoch": 12.22860791826309,
2364
- "learning_rate": 3.6951894423158796e-06,
2365
- "loss": 0.2989999961853027,
2366
- "step": 38300
2367
- },
2368
- {
2369
- "epoch": 12.260536398467433,
2370
- "learning_rate": 3.6526181353767567e-06,
2371
- "loss": 0.30233255386352537,
2372
- "step": 38400
2373
- },
2374
- {
2375
- "epoch": 12.292464878671776,
2376
- "learning_rate": 3.6100468284376333e-06,
2377
- "loss": 0.30123531341552734,
2378
- "step": 38500
2379
- },
2380
- {
2381
- "epoch": 12.324393358876117,
2382
- "learning_rate": 3.56747552149851e-06,
2383
- "loss": 0.3095419502258301,
2384
- "step": 38600
2385
- },
2386
- {
2387
- "epoch": 12.35632183908046,
2388
- "learning_rate": 3.524904214559387e-06,
2389
- "loss": 0.32312957763671873,
2390
- "step": 38700
2391
- },
2392
- {
2393
- "epoch": 12.388250319284802,
2394
- "learning_rate": 3.4823329076202646e-06,
2395
- "loss": 0.31157236099243163,
2396
- "step": 38800
2397
- },
2398
- {
2399
- "epoch": 12.420178799489145,
2400
- "learning_rate": 3.4397616006811412e-06,
2401
- "loss": 0.3003998374938965,
2402
- "step": 38900
2403
- },
2404
- {
2405
- "epoch": 12.452107279693486,
2406
- "learning_rate": 3.3971902937420183e-06,
2407
- "loss": 0.3086379051208496,
2408
- "step": 39000
2409
- },
2410
- {
2411
- "epoch": 12.484035759897829,
2412
- "learning_rate": 3.354618986802895e-06,
2413
- "loss": 0.28147794723510744,
2414
- "step": 39100
2415
- },
2416
- {
2417
- "epoch": 12.515964240102171,
2418
- "learning_rate": 3.3120476798637717e-06,
2419
- "loss": 0.29010528564453125,
2420
- "step": 39200
2421
- },
2422
- {
2423
- "epoch": 12.547892720306514,
2424
- "learning_rate": 3.269476372924649e-06,
2425
- "loss": 0.30009984970092773,
2426
- "step": 39300
2427
- },
2428
- {
2429
- "epoch": 12.579821200510855,
2430
- "learning_rate": 3.2269050659855262e-06,
2431
- "loss": 0.3059814834594727,
2432
- "step": 39400
2433
- },
2434
- {
2435
- "epoch": 12.611749680715198,
2436
- "learning_rate": 3.184333759046403e-06,
2437
- "loss": 0.2934641456604004,
2438
- "step": 39500
2439
- },
2440
- {
2441
- "epoch": 12.64367816091954,
2442
- "learning_rate": 3.14176245210728e-06,
2443
- "loss": 0.30141252517700196,
2444
- "step": 39600
2445
- },
2446
- {
2447
- "epoch": 12.675606641123883,
2448
- "learning_rate": 3.0991911451681567e-06,
2449
- "loss": 0.3053057289123535,
2450
- "step": 39700
2451
- },
2452
- {
2453
- "epoch": 12.707535121328224,
2454
- "learning_rate": 3.056619838229034e-06,
2455
- "loss": 0.2935024261474609,
2456
- "step": 39800
2457
- },
2458
- {
2459
- "epoch": 12.739463601532567,
2460
- "learning_rate": 3.014048531289911e-06,
2461
- "loss": 0.2974138069152832,
2462
- "step": 39900
2463
- },
2464
- {
2465
- "epoch": 12.77139208173691,
2466
- "learning_rate": 2.971477224350788e-06,
2467
- "loss": 0.29385158538818357,
2468
- "step": 40000
2469
- },
2470
- {
2471
- "epoch": 12.803320561941252,
2472
- "learning_rate": 2.9289059174116646e-06,
2473
- "loss": 0.309177303314209,
2474
- "step": 40100
2475
- },
2476
- {
2477
- "epoch": 12.835249042145595,
2478
- "learning_rate": 2.8863346104725417e-06,
2479
- "loss": 0.30561391830444334,
2480
- "step": 40200
2481
- },
2482
- {
2483
- "epoch": 12.867177522349936,
2484
- "learning_rate": 2.8437633035334187e-06,
2485
- "loss": 0.30923063278198243,
2486
- "step": 40300
2487
- },
2488
- {
2489
- "epoch": 12.899106002554278,
2490
- "learning_rate": 2.801191996594296e-06,
2491
- "loss": 0.30836896896362304,
2492
- "step": 40400
2493
- },
2494
- {
2495
- "epoch": 12.931034482758621,
2496
- "learning_rate": 2.7586206896551725e-06,
2497
- "loss": 0.31578615188598635,
2498
- "step": 40500
2499
- },
2500
- {
2501
- "epoch": 12.962962962962964,
2502
- "learning_rate": 2.7160493827160496e-06,
2503
- "loss": 0.31291526794433594,
2504
- "step": 40600
2505
- },
2506
- {
2507
- "epoch": 12.994891443167305,
2508
- "learning_rate": 2.6734780757769262e-06,
2509
- "loss": 0.3064906311035156,
2510
- "step": 40700
2511
- },
2512
- {
2513
- "epoch": 13.0,
2514
- "eval_loss": 0.30278804898262024,
2515
- "step": 40716
2516
- },
2517
- {
2518
- "epoch": 13.026819923371647,
2519
- "learning_rate": 2.6309067688378037e-06,
2520
- "loss": 0.2991274642944336,
2521
- "step": 40800
2522
- },
2523
- {
2524
- "epoch": 13.05874840357599,
2525
- "learning_rate": 2.5883354618986804e-06,
2526
- "loss": 0.3081726837158203,
2527
- "step": 40900
2528
- },
2529
- {
2530
- "epoch": 13.090676883780333,
2531
- "learning_rate": 2.5457641549595575e-06,
2532
- "loss": 0.28322860717773435,
2533
- "step": 41000
2534
- },
2535
- {
2536
- "epoch": 13.122605363984674,
2537
- "learning_rate": 2.503192848020434e-06,
2538
- "loss": 0.3077671813964844,
2539
- "step": 41100
2540
- },
2541
- {
2542
- "epoch": 13.154533844189016,
2543
- "learning_rate": 2.4606215410813112e-06,
2544
- "loss": 0.3038086128234863,
2545
- "step": 41200
2546
- },
2547
- {
2548
- "epoch": 13.186462324393359,
2549
- "learning_rate": 2.4180502341421883e-06,
2550
- "loss": 0.3015581703186035,
2551
- "step": 41300
2552
- },
2553
- {
2554
- "epoch": 13.218390804597702,
2555
- "learning_rate": 2.3754789272030654e-06,
2556
- "loss": 0.30243860244750975,
2557
- "step": 41400
2558
- },
2559
- {
2560
- "epoch": 13.250319284802043,
2561
- "learning_rate": 2.332907620263942e-06,
2562
- "loss": 0.3104331398010254,
2563
- "step": 41500
2564
- },
2565
- {
2566
- "epoch": 13.282247765006385,
2567
- "learning_rate": 2.290336313324819e-06,
2568
- "loss": 0.30330204010009765,
2569
- "step": 41600
2570
- },
2571
- {
2572
- "epoch": 13.314176245210728,
2573
- "learning_rate": 2.2477650063856962e-06,
2574
- "loss": 0.2972829818725586,
2575
- "step": 41700
2576
- },
2577
- {
2578
- "epoch": 13.34610472541507,
2579
- "learning_rate": 2.205193699446573e-06,
2580
- "loss": 0.300672607421875,
2581
- "step": 41800
2582
- },
2583
- {
2584
- "epoch": 13.378033205619413,
2585
- "learning_rate": 2.1626223925074504e-06,
2586
- "loss": 0.30183706283569334,
2587
- "step": 41900
2588
- },
2589
- {
2590
- "epoch": 13.409961685823754,
2591
- "learning_rate": 2.120051085568327e-06,
2592
- "loss": 0.3096357536315918,
2593
- "step": 42000
2594
- },
2595
- {
2596
- "epoch": 13.441890166028097,
2597
- "learning_rate": 2.077479778629204e-06,
2598
- "loss": 0.3032659912109375,
2599
- "step": 42100
2600
- },
2601
- {
2602
- "epoch": 13.47381864623244,
2603
- "learning_rate": 2.0349084716900813e-06,
2604
- "loss": 0.2949547386169434,
2605
- "step": 42200
2606
- },
2607
- {
2608
- "epoch": 13.505747126436782,
2609
- "learning_rate": 1.992337164750958e-06,
2610
- "loss": 0.2889937973022461,
2611
- "step": 42300
2612
- },
2613
- {
2614
- "epoch": 13.537675606641123,
2615
- "learning_rate": 1.949765857811835e-06,
2616
- "loss": 0.30804216384887695,
2617
- "step": 42400
2618
- },
2619
- {
2620
- "epoch": 13.569604086845466,
2621
- "learning_rate": 1.9071945508727119e-06,
2622
- "loss": 0.30334890365600586,
2623
- "step": 42500
2624
- },
2625
- {
2626
- "epoch": 13.601532567049809,
2627
- "learning_rate": 1.864623243933589e-06,
2628
- "loss": 0.2941057586669922,
2629
- "step": 42600
2630
- },
2631
- {
2632
- "epoch": 13.633461047254151,
2633
- "learning_rate": 1.8220519369944658e-06,
2634
- "loss": 0.2950305938720703,
2635
- "step": 42700
2636
- },
2637
- {
2638
- "epoch": 13.665389527458492,
2639
- "learning_rate": 1.7794806300553427e-06,
2640
- "loss": 0.3061481285095215,
2641
- "step": 42800
2642
- },
2643
- {
2644
- "epoch": 13.697318007662835,
2645
- "learning_rate": 1.7369093231162198e-06,
2646
- "loss": 0.2870841217041016,
2647
- "step": 42900
2648
- },
2649
- {
2650
- "epoch": 13.729246487867178,
2651
- "learning_rate": 1.6943380161770967e-06,
2652
- "loss": 0.28978254318237306,
2653
- "step": 43000
2654
- },
2655
- {
2656
- "epoch": 13.76117496807152,
2657
- "learning_rate": 1.6517667092379737e-06,
2658
- "loss": 0.30341114044189454,
2659
- "step": 43100
2660
- },
2661
- {
2662
- "epoch": 13.793103448275861,
2663
- "learning_rate": 1.6091954022988506e-06,
2664
- "loss": 0.307810001373291,
2665
- "step": 43200
2666
- },
2667
- {
2668
- "epoch": 13.825031928480204,
2669
- "learning_rate": 1.5666240953597275e-06,
2670
- "loss": 0.30963399887084964,
2671
- "step": 43300
2672
- },
2673
- {
2674
- "epoch": 13.856960408684547,
2675
- "learning_rate": 1.5240527884206046e-06,
2676
- "loss": 0.3031142807006836,
2677
- "step": 43400
2678
- },
2679
- {
2680
- "epoch": 13.88888888888889,
2681
- "learning_rate": 1.4814814814814815e-06,
2682
- "loss": 0.3144196891784668,
2683
- "step": 43500
2684
- },
2685
- {
2686
- "epoch": 13.920817369093232,
2687
- "learning_rate": 1.4389101745423588e-06,
2688
- "loss": 0.28602792739868166,
2689
- "step": 43600
2690
- },
2691
- {
2692
- "epoch": 13.952745849297573,
2693
- "learning_rate": 1.3963388676032354e-06,
2694
- "loss": 0.2984015083312988,
2695
- "step": 43700
2696
- },
2697
- {
2698
- "epoch": 13.984674329501916,
2699
- "learning_rate": 1.3537675606641127e-06,
2700
- "loss": 0.29513154983520506,
2701
- "step": 43800
2702
- },
2703
- {
2704
- "epoch": 14.0,
2705
- "eval_loss": 0.3065377175807953,
2706
- "step": 43848
2707
- },
2708
- {
2709
- "epoch": 14.016602809706258,
2710
- "learning_rate": 1.3111962537249896e-06,
2711
- "loss": 0.28543767929077146,
2712
- "step": 43900
2713
- },
2714
- {
2715
- "epoch": 14.048531289910601,
2716
- "learning_rate": 1.2686249467858662e-06,
2717
- "loss": 0.2963153839111328,
2718
- "step": 44000
2719
- },
2720
- {
2721
- "epoch": 14.080459770114942,
2722
- "learning_rate": 1.2260536398467433e-06,
2723
- "loss": 0.3030729293823242,
2724
- "step": 44100
2725
- },
2726
- {
2727
- "epoch": 14.112388250319285,
2728
- "learning_rate": 1.1834823329076204e-06,
2729
- "loss": 0.2889740180969238,
2730
- "step": 44200
2731
- },
2732
- {
2733
- "epoch": 14.144316730523627,
2734
- "learning_rate": 1.1409110259684973e-06,
2735
- "loss": 0.2915242385864258,
2736
- "step": 44300
2737
- },
2738
- {
2739
- "epoch": 14.17624521072797,
2740
- "learning_rate": 1.0983397190293744e-06,
2741
- "loss": 0.28075174331665037,
2742
- "step": 44400
2743
- },
2744
- {
2745
- "epoch": 14.20817369093231,
2746
- "learning_rate": 1.0557684120902512e-06,
2747
- "loss": 0.3046562194824219,
2748
- "step": 44500
2749
- },
2750
- {
2751
- "epoch": 14.240102171136654,
2752
- "learning_rate": 1.0131971051511283e-06,
2753
- "loss": 0.28045236587524414,
2754
- "step": 44600
2755
- },
2756
- {
2757
- "epoch": 14.272030651340996,
2758
- "learning_rate": 9.706257982120052e-07,
2759
- "loss": 0.2948748970031738,
2760
- "step": 44700
2761
- },
2762
- {
2763
- "epoch": 14.303959131545339,
2764
- "learning_rate": 9.280544912728822e-07,
2765
- "loss": 0.2956666946411133,
2766
- "step": 44800
2767
- },
2768
- {
2769
- "epoch": 14.33588761174968,
2770
- "learning_rate": 8.854831843337592e-07,
2771
- "loss": 0.2919887924194336,
2772
- "step": 44900
2773
- },
2774
- {
2775
- "epoch": 14.367816091954023,
2776
- "learning_rate": 8.429118773946361e-07,
2777
- "loss": 0.29414264678955077,
2778
- "step": 45000
2779
- },
2780
- {
2781
- "epoch": 14.399744572158365,
2782
- "learning_rate": 8.003405704555131e-07,
2783
- "loss": 0.2997250938415527,
2784
- "step": 45100
2785
- },
2786
- {
2787
- "epoch": 14.431673052362708,
2788
- "learning_rate": 7.5776926351639e-07,
2789
- "loss": 0.2927609634399414,
2790
- "step": 45200
2791
- },
2792
- {
2793
- "epoch": 14.46360153256705,
2794
- "learning_rate": 7.15197956577267e-07,
2795
- "loss": 0.30317195892333987,
2796
- "step": 45300
2797
- },
2798
- {
2799
- "epoch": 14.495530012771392,
2800
- "learning_rate": 6.72626649638144e-07,
2801
- "loss": 0.29728498458862307,
2802
- "step": 45400
2803
- },
2804
- {
2805
- "epoch": 14.527458492975734,
2806
- "learning_rate": 6.300553426990209e-07,
2807
- "loss": 0.2906969451904297,
2808
- "step": 45500
2809
- },
2810
- {
2811
- "epoch": 14.559386973180077,
2812
- "learning_rate": 5.874840357598978e-07,
2813
- "loss": 0.3137422752380371,
2814
- "step": 45600
2815
- },
2816
- {
2817
- "epoch": 14.59131545338442,
2818
- "learning_rate": 5.449127288207749e-07,
2819
- "loss": 0.3111775016784668,
2820
- "step": 45700
2821
- },
2822
- {
2823
- "epoch": 14.62324393358876,
2824
- "learning_rate": 5.023414218816518e-07,
2825
- "loss": 0.30036439895629885,
2826
- "step": 45800
2827
- },
2828
- {
2829
- "epoch": 14.655172413793103,
2830
- "learning_rate": 4.5977011494252875e-07,
2831
- "loss": 0.2988995361328125,
2832
- "step": 45900
2833
- },
2834
- {
2835
- "epoch": 14.687100893997446,
2836
- "learning_rate": 4.171988080034058e-07,
2837
- "loss": 0.306041259765625,
2838
- "step": 46000
2839
- },
2840
- {
2841
- "epoch": 14.719029374201789,
2842
- "learning_rate": 3.7462750106428265e-07,
2843
- "loss": 0.30521770477294924,
2844
- "step": 46100
2845
- },
2846
- {
2847
- "epoch": 14.75095785440613,
2848
- "learning_rate": 3.320561941251597e-07,
2849
- "loss": 0.28645307540893555,
2850
- "step": 46200
2851
- },
2852
- {
2853
- "epoch": 14.782886334610472,
2854
- "learning_rate": 2.894848871860366e-07,
2855
- "loss": 0.29830142974853513,
2856
- "step": 46300
2857
- },
2858
- {
2859
- "epoch": 14.814814814814815,
2860
- "learning_rate": 2.469135802469136e-07,
2861
- "loss": 0.29040414810180665,
2862
- "step": 46400
2863
- },
2864
- {
2865
- "epoch": 14.846743295019158,
2866
- "learning_rate": 2.0434227330779057e-07,
2867
- "loss": 0.2852139472961426,
2868
- "step": 46500
2869
- },
2870
- {
2871
- "epoch": 14.878671775223498,
2872
- "learning_rate": 1.6177096636866753e-07,
2873
- "loss": 0.291912841796875,
2874
- "step": 46600
2875
- },
2876
- {
2877
- "epoch": 14.910600255427841,
2878
- "learning_rate": 1.1919965942954449e-07,
2879
- "loss": 0.2825672149658203,
2880
- "step": 46700
2881
- },
2882
- {
2883
- "epoch": 14.942528735632184,
2884
- "learning_rate": 7.662835249042146e-08,
2885
- "loss": 0.3074253273010254,
2886
- "step": 46800
2887
- }
2888
- ],
2889
- "max_steps": 46980,
2890
- "num_train_epochs": 15,
2891
- "total_flos": 76612640706201600,
2892
- "trial_name": null,
2893
- "trial_params": null
2894
- }