Text Generation
Transformers
PyTorch
English
llama
code
text-generation-inference
Inference Endpoints
Pasta009 commited on
Commit
243669d
1 Parent(s): 8f9b3a2

Delete trainer_state.json

Browse files
Files changed (1) hide show
  1. trainer_state.json +0 -1012
trainer_state.json DELETED
@@ -1,1012 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 1.9987871437234688,
5
- "eval_steps": 500,
6
- "global_step": 1648,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.01,
13
- "learning_rate": 1.9998183053318547e-05,
14
- "loss": 0.6602,
15
- "step": 10
16
- },
17
- {
18
- "epoch": 0.02,
19
- "learning_rate": 1.9992732873533223e-05,
20
- "loss": 0.5998,
21
- "step": 20
22
- },
23
- {
24
- "epoch": 0.04,
25
- "learning_rate": 1.9983651441181253e-05,
26
- "loss": 0.5727,
27
- "step": 30
28
- },
29
- {
30
- "epoch": 0.05,
31
- "learning_rate": 1.997094205635831e-05,
32
- "loss": 0.5677,
33
- "step": 40
34
- },
35
- {
36
- "epoch": 0.06,
37
- "learning_rate": 1.99546093375193e-05,
38
- "loss": 0.545,
39
- "step": 50
40
- },
41
- {
42
- "epoch": 0.07,
43
- "learning_rate": 1.99346592198001e-05,
44
- "loss": 0.5564,
45
- "step": 60
46
- },
47
- {
48
- "epoch": 0.08,
49
- "learning_rate": 1.9911098952860726e-05,
50
- "loss": 0.5393,
51
- "step": 70
52
- },
53
- {
54
- "epoch": 0.1,
55
- "learning_rate": 1.9883937098250962e-05,
56
- "loss": 0.52,
57
- "step": 80
58
- },
59
- {
60
- "epoch": 0.11,
61
- "learning_rate": 1.985318352629912e-05,
62
- "loss": 0.5319,
63
- "step": 90
64
- },
65
- {
66
- "epoch": 0.12,
67
- "learning_rate": 1.9818849412525294e-05,
68
- "loss": 0.5382,
69
- "step": 100
70
- },
71
- {
72
- "epoch": 0.13,
73
- "learning_rate": 1.978094723358031e-05,
74
- "loss": 0.5302,
75
- "step": 110
76
- },
77
- {
78
- "epoch": 0.15,
79
- "learning_rate": 1.9739490762711812e-05,
80
- "loss": 0.5377,
81
- "step": 120
82
- },
83
- {
84
- "epoch": 0.16,
85
- "learning_rate": 1.969449506475924e-05,
86
- "loss": 0.5253,
87
- "step": 130
88
- },
89
- {
90
- "epoch": 0.17,
91
- "learning_rate": 1.9645976490679402e-05,
92
- "loss": 0.5364,
93
- "step": 140
94
- },
95
- {
96
- "epoch": 0.18,
97
- "learning_rate": 1.9593952671604737e-05,
98
- "loss": 0.5277,
99
- "step": 150
100
- },
101
- {
102
- "epoch": 0.19,
103
- "learning_rate": 1.953844251243633e-05,
104
- "loss": 0.5193,
105
- "step": 160
106
- },
107
- {
108
- "epoch": 0.21,
109
- "learning_rate": 1.947946618497407e-05,
110
- "loss": 0.5161,
111
- "step": 170
112
- },
113
- {
114
- "epoch": 0.22,
115
- "learning_rate": 1.941704512058646e-05,
116
- "loss": 0.5225,
117
- "step": 180
118
- },
119
- {
120
- "epoch": 0.23,
121
- "learning_rate": 1.9351202002422654e-05,
122
- "loss": 0.5313,
123
- "step": 190
124
- },
125
- {
126
- "epoch": 0.24,
127
- "learning_rate": 1.928196075716966e-05,
128
- "loss": 0.5152,
129
- "step": 200
130
- },
131
- {
132
- "epoch": 0.25,
133
- "learning_rate": 1.920934654635764e-05,
134
- "loss": 0.5242,
135
- "step": 210
136
- },
137
- {
138
- "epoch": 0.27,
139
- "learning_rate": 1.9133385757216458e-05,
140
- "loss": 0.5219,
141
- "step": 220
142
- },
143
- {
144
- "epoch": 0.28,
145
- "learning_rate": 1.905410599308687e-05,
146
- "loss": 0.5212,
147
- "step": 230
148
- },
149
- {
150
- "epoch": 0.29,
151
- "learning_rate": 1.8971536063389745e-05,
152
- "loss": 0.5083,
153
- "step": 240
154
- },
155
- {
156
- "epoch": 0.3,
157
- "learning_rate": 1.888570597315703e-05,
158
- "loss": 0.509,
159
- "step": 250
160
- },
161
- {
162
- "epoch": 0.32,
163
- "learning_rate": 1.8796646912128247e-05,
164
- "loss": 0.513,
165
- "step": 260
166
- },
167
- {
168
- "epoch": 0.33,
169
- "learning_rate": 1.8704391243416478e-05,
170
- "loss": 0.5027,
171
- "step": 270
172
- },
173
- {
174
- "epoch": 0.34,
175
- "learning_rate": 1.8608972491747946e-05,
176
- "loss": 0.5167,
177
- "step": 280
178
- },
179
- {
180
- "epoch": 0.35,
181
- "learning_rate": 1.8510425331279488e-05,
182
- "loss": 0.5147,
183
- "step": 290
184
- },
185
- {
186
- "epoch": 0.36,
187
- "learning_rate": 1.8408785572998335e-05,
188
- "loss": 0.5045,
189
- "step": 300
190
- },
191
- {
192
- "epoch": 0.38,
193
- "learning_rate": 1.8304090151708797e-05,
194
- "loss": 0.5076,
195
- "step": 310
196
- },
197
- {
198
- "epoch": 0.39,
199
- "learning_rate": 1.8196377112610524e-05,
200
- "loss": 0.5066,
201
- "step": 320
202
- },
203
- {
204
- "epoch": 0.4,
205
- "learning_rate": 1.808568559747331e-05,
206
- "loss": 0.5125,
207
- "step": 330
208
- },
209
- {
210
- "epoch": 0.41,
211
- "learning_rate": 1.7972055830413372e-05,
212
- "loss": 0.5079,
213
- "step": 340
214
- },
215
- {
216
- "epoch": 0.42,
217
- "learning_rate": 1.7855529103276337e-05,
218
- "loss": 0.5085,
219
- "step": 350
220
- },
221
- {
222
- "epoch": 0.44,
223
- "learning_rate": 1.773614776063225e-05,
224
- "loss": 0.4952,
225
- "step": 360
226
- },
227
- {
228
- "epoch": 0.45,
229
- "learning_rate": 1.761395518438797e-05,
230
- "loss": 0.5107,
231
- "step": 370
232
- },
233
- {
234
- "epoch": 0.46,
235
- "learning_rate": 1.7488995778022687e-05,
236
- "loss": 0.4863,
237
- "step": 380
238
- },
239
- {
240
- "epoch": 0.47,
241
- "learning_rate": 1.7361314950452137e-05,
242
- "loss": 0.5035,
243
- "step": 390
244
- },
245
- {
246
- "epoch": 0.49,
247
- "learning_rate": 1.7230959099527512e-05,
248
- "loss": 0.514,
249
- "step": 400
250
- },
251
- {
252
- "epoch": 0.5,
253
- "learning_rate": 1.709797559517496e-05,
254
- "loss": 0.498,
255
- "step": 410
256
- },
257
- {
258
- "epoch": 0.51,
259
- "learning_rate": 1.6962412762181867e-05,
260
- "loss": 0.5048,
261
- "step": 420
262
- },
263
- {
264
- "epoch": 0.52,
265
- "learning_rate": 1.6824319862636137e-05,
266
- "loss": 0.5145,
267
- "step": 430
268
- },
269
- {
270
- "epoch": 0.53,
271
- "learning_rate": 1.6683747078024887e-05,
272
- "loss": 0.4913,
273
- "step": 440
274
- },
275
- {
276
- "epoch": 0.55,
277
- "learning_rate": 1.654074549099901e-05,
278
- "loss": 0.5129,
279
- "step": 450
280
- },
281
- {
282
- "epoch": 0.56,
283
- "learning_rate": 1.6395367066810312e-05,
284
- "loss": 0.5024,
285
- "step": 460
286
- },
287
- {
288
- "epoch": 0.57,
289
- "learning_rate": 1.6247664634427866e-05,
290
- "loss": 0.5127,
291
- "step": 470
292
- },
293
- {
294
- "epoch": 0.58,
295
- "learning_rate": 1.6097691867340547e-05,
296
- "loss": 0.4951,
297
- "step": 480
298
- },
299
- {
300
- "epoch": 0.59,
301
- "learning_rate": 1.5945503264052638e-05,
302
- "loss": 0.5059,
303
- "step": 490
304
- },
305
- {
306
- "epoch": 0.61,
307
- "learning_rate": 1.5791154128279694e-05,
308
- "loss": 0.4891,
309
- "step": 500
310
- },
311
- {
312
- "epoch": 0.62,
313
- "learning_rate": 1.5634700548851713e-05,
314
- "loss": 0.4874,
315
- "step": 510
316
- },
317
- {
318
- "epoch": 0.63,
319
- "learning_rate": 1.547619937933108e-05,
320
- "loss": 0.5008,
321
- "step": 520
322
- },
323
- {
324
- "epoch": 0.64,
325
- "learning_rate": 1.53157082173526e-05,
326
- "loss": 0.5038,
327
- "step": 530
328
- },
329
- {
330
- "epoch": 0.65,
331
- "learning_rate": 1.5153285383693091e-05,
332
- "loss": 0.5154,
333
- "step": 540
334
- },
335
- {
336
- "epoch": 0.67,
337
- "learning_rate": 1.4988989901078286e-05,
338
- "loss": 0.4846,
339
- "step": 550
340
- },
341
- {
342
- "epoch": 0.68,
343
- "learning_rate": 1.4822881472734563e-05,
344
- "loss": 0.5016,
345
- "step": 560
346
- },
347
- {
348
- "epoch": 0.69,
349
- "learning_rate": 1.4655020460693452e-05,
350
- "loss": 0.514,
351
- "step": 570
352
- },
353
- {
354
- "epoch": 0.7,
355
- "learning_rate": 1.4485467863856704e-05,
356
- "loss": 0.4998,
357
- "step": 580
358
- },
359
- {
360
- "epoch": 0.72,
361
- "learning_rate": 1.4314285295829957e-05,
362
- "loss": 0.4985,
363
- "step": 590
364
- },
365
- {
366
- "epoch": 0.73,
367
- "learning_rate": 1.4141534962532986e-05,
368
- "loss": 0.4833,
369
- "step": 600
370
- },
371
- {
372
- "epoch": 0.74,
373
- "learning_rate": 1.3967279639594753e-05,
374
- "loss": 0.5018,
375
- "step": 610
376
- },
377
- {
378
- "epoch": 0.75,
379
- "learning_rate": 1.3791582649541404e-05,
380
- "loss": 0.487,
381
- "step": 620
382
- },
383
- {
384
- "epoch": 0.76,
385
- "learning_rate": 1.3614507838785547e-05,
386
- "loss": 0.509,
387
- "step": 630
388
- },
389
- {
390
- "epoch": 0.78,
391
- "learning_rate": 1.3436119554425133e-05,
392
- "loss": 0.4856,
393
- "step": 640
394
- },
395
- {
396
- "epoch": 0.79,
397
- "learning_rate": 1.3256482620860415e-05,
398
- "loss": 0.5007,
399
- "step": 650
400
- },
401
- {
402
- "epoch": 0.8,
403
- "learning_rate": 1.3075662316237466e-05,
404
- "loss": 0.4985,
405
- "step": 660
406
- },
407
- {
408
- "epoch": 0.81,
409
- "learning_rate": 1.2893724348726757e-05,
410
- "loss": 0.4868,
411
- "step": 670
412
- },
413
- {
414
- "epoch": 0.82,
415
- "learning_rate": 1.2710734832645557e-05,
416
- "loss": 0.4888,
417
- "step": 680
418
- },
419
- {
420
- "epoch": 0.84,
421
- "learning_rate": 1.2526760264432658e-05,
422
- "loss": 0.4963,
423
- "step": 690
424
- },
425
- {
426
- "epoch": 0.85,
427
- "learning_rate": 1.2341867498484303e-05,
428
- "loss": 0.4898,
429
- "step": 700
430
- },
431
- {
432
- "epoch": 0.86,
433
- "learning_rate": 1.2156123722859989e-05,
434
- "loss": 0.4968,
435
- "step": 710
436
- },
437
- {
438
- "epoch": 0.87,
439
- "learning_rate": 1.1969596434867063e-05,
440
- "loss": 0.4901,
441
- "step": 720
442
- },
443
- {
444
- "epoch": 0.89,
445
- "learning_rate": 1.1782353416532908e-05,
446
- "loss": 0.5002,
447
- "step": 730
448
- },
449
- {
450
- "epoch": 0.9,
451
- "learning_rate": 1.1594462709973684e-05,
452
- "loss": 0.4862,
453
- "step": 740
454
- },
455
- {
456
- "epoch": 0.91,
457
- "learning_rate": 1.140599259266854e-05,
458
- "loss": 0.4839,
459
- "step": 750
460
- },
461
- {
462
- "epoch": 0.92,
463
- "learning_rate": 1.1217011552648316e-05,
464
- "loss": 0.4923,
465
- "step": 760
466
- },
467
- {
468
- "epoch": 0.93,
469
- "learning_rate": 1.102758826360772e-05,
470
- "loss": 0.4957,
471
- "step": 770
472
- },
473
- {
474
- "epoch": 0.95,
475
- "learning_rate": 1.0837791559950029e-05,
476
- "loss": 0.4864,
477
- "step": 780
478
- },
479
- {
480
- "epoch": 0.96,
481
- "learning_rate": 1.0647690411773415e-05,
482
- "loss": 0.4843,
483
- "step": 790
484
- },
485
- {
486
- "epoch": 0.97,
487
- "learning_rate": 1.0457353899807947e-05,
488
- "loss": 0.5063,
489
- "step": 800
490
- },
491
- {
492
- "epoch": 0.98,
493
- "learning_rate": 1.0266851190312375e-05,
494
- "loss": 0.4786,
495
- "step": 810
496
- },
497
- {
498
- "epoch": 0.99,
499
- "learning_rate": 1.0076251509939867e-05,
500
- "loss": 0.4892,
501
- "step": 820
502
- },
503
- {
504
- "epoch": 1.01,
505
- "learning_rate": 9.885624120581772e-06,
506
- "loss": 0.4588,
507
- "step": 830
508
- },
509
- {
510
- "epoch": 1.02,
511
- "learning_rate": 9.695038294198588e-06,
512
- "loss": 0.4106,
513
- "step": 840
514
- },
515
- {
516
- "epoch": 1.03,
517
- "learning_rate": 9.504563287647265e-06,
518
- "loss": 0.4155,
519
- "step": 850
520
- },
521
- {
522
- "epoch": 1.04,
523
- "learning_rate": 9.314268317514023e-06,
524
- "loss": 0.4123,
525
- "step": 860
526
- },
527
- {
528
- "epoch": 1.06,
529
- "learning_rate": 9.12422253496175e-06,
530
- "loss": 0.4064,
531
- "step": 870
532
- },
533
- {
534
- "epoch": 1.07,
535
- "learning_rate": 8.934495000601241e-06,
536
- "loss": 0.4115,
537
- "step": 880
538
- },
539
- {
540
- "epoch": 1.08,
541
- "learning_rate": 8.745154659395273e-06,
542
- "loss": 0.4155,
543
- "step": 890
544
- },
545
- {
546
- "epoch": 1.09,
547
- "learning_rate": 8.556270315604779e-06,
548
- "loss": 0.404,
549
- "step": 900
550
- },
551
- {
552
- "epoch": 1.1,
553
- "learning_rate": 8.36791060778608e-06,
554
- "loss": 0.4002,
555
- "step": 910
556
- },
557
- {
558
- "epoch": 1.12,
559
- "learning_rate": 8.180143983848388e-06,
560
- "loss": 0.4042,
561
- "step": 920
562
- },
563
- {
564
- "epoch": 1.13,
565
- "learning_rate": 7.993038676180546e-06,
566
- "loss": 0.402,
567
- "step": 930
568
- },
569
- {
570
- "epoch": 1.14,
571
- "learning_rate": 7.806662676856134e-06,
572
- "loss": 0.4041,
573
- "step": 940
574
- },
575
- {
576
- "epoch": 1.15,
577
- "learning_rate": 7.62108371292584e-06,
578
- "loss": 0.4162,
579
- "step": 950
580
- },
581
- {
582
- "epoch": 1.16,
583
- "learning_rate": 7.436369221806201e-06,
584
- "loss": 0.4206,
585
- "step": 960
586
- },
587
- {
588
- "epoch": 1.18,
589
- "learning_rate": 7.2525863267735405e-06,
590
- "loss": 0.4022,
591
- "step": 970
592
- },
593
- {
594
- "epoch": 1.19,
595
- "learning_rate": 7.069801812572117e-06,
596
- "loss": 0.3994,
597
- "step": 980
598
- },
599
- {
600
- "epoch": 1.2,
601
- "learning_rate": 6.888082101145222e-06,
602
- "loss": 0.4124,
603
- "step": 990
604
- },
605
- {
606
- "epoch": 1.21,
607
- "learning_rate": 6.707493227498187e-06,
608
- "loss": 0.4182,
609
- "step": 1000
610
- },
611
- {
612
- "epoch": 1.22,
613
- "learning_rate": 6.5281008157019425e-06,
614
- "loss": 0.4079,
615
- "step": 1010
616
- },
617
- {
618
- "epoch": 1.24,
619
- "learning_rate": 6.3499700550459554e-06,
620
- "loss": 0.4118,
621
- "step": 1020
622
- },
623
- {
624
- "epoch": 1.25,
625
- "learning_rate": 6.173165676349103e-06,
626
- "loss": 0.4144,
627
- "step": 1030
628
- },
629
- {
630
- "epoch": 1.26,
631
- "learning_rate": 5.99775192843722e-06,
632
- "loss": 0.4101,
633
- "step": 1040
634
- },
635
- {
636
- "epoch": 1.27,
637
- "learning_rate": 5.823792554795738e-06,
638
- "loss": 0.4041,
639
- "step": 1050
640
- },
641
- {
642
- "epoch": 1.29,
643
- "learning_rate": 5.6513507704059835e-06,
644
- "loss": 0.413,
645
- "step": 1060
646
- },
647
- {
648
- "epoch": 1.3,
649
- "learning_rate": 5.480489238773536e-06,
650
- "loss": 0.3996,
651
- "step": 1070
652
- },
653
- {
654
- "epoch": 1.31,
655
- "learning_rate": 5.311270049156967e-06,
656
- "loss": 0.4124,
657
- "step": 1080
658
- },
659
- {
660
- "epoch": 1.32,
661
- "learning_rate": 5.14375469400529e-06,
662
- "loss": 0.4026,
663
- "step": 1090
664
- },
665
- {
666
- "epoch": 1.33,
667
- "learning_rate": 4.978004046612224e-06,
668
- "loss": 0.4067,
669
- "step": 1100
670
- },
671
- {
672
- "epoch": 1.35,
673
- "learning_rate": 4.814078338995516e-06,
674
- "loss": 0.3976,
675
- "step": 1110
676
- },
677
- {
678
- "epoch": 1.36,
679
- "learning_rate": 4.652037140009259e-06,
680
- "loss": 0.3902,
681
- "step": 1120
682
- },
683
- {
684
- "epoch": 1.37,
685
- "learning_rate": 4.491939333697205e-06,
686
- "loss": 0.403,
687
- "step": 1130
688
- },
689
- {
690
- "epoch": 1.38,
691
- "learning_rate": 4.333843097894932e-06,
692
- "loss": 0.4138,
693
- "step": 1140
694
- },
695
- {
696
- "epoch": 1.39,
697
- "learning_rate": 4.177805883088641e-06,
698
- "loss": 0.4199,
699
- "step": 1150
700
- },
701
- {
702
- "epoch": 1.41,
703
- "learning_rate": 4.023884391538244e-06,
704
- "loss": 0.4104,
705
- "step": 1160
706
- },
707
- {
708
- "epoch": 1.42,
709
- "learning_rate": 3.8721345566724156e-06,
710
- "loss": 0.4091,
711
- "step": 1170
712
- },
713
- {
714
- "epoch": 1.43,
715
- "learning_rate": 3.722611522762917e-06,
716
- "loss": 0.4069,
717
- "step": 1180
718
- },
719
- {
720
- "epoch": 1.44,
721
- "learning_rate": 3.575369624885803e-06,
722
- "loss": 0.4105,
723
- "step": 1190
724
- },
725
- {
726
- "epoch": 1.46,
727
- "learning_rate": 3.4304623691766193e-06,
728
- "loss": 0.399,
729
- "step": 1200
730
- },
731
- {
732
- "epoch": 1.47,
733
- "learning_rate": 3.287942413386841e-06,
734
- "loss": 0.4018,
735
- "step": 1210
736
- },
737
- {
738
- "epoch": 1.48,
739
- "learning_rate": 3.147861547748612e-06,
740
- "loss": 0.4119,
741
- "step": 1220
742
- },
743
- {
744
- "epoch": 1.49,
745
- "learning_rate": 3.0102706761547264e-06,
746
- "loss": 0.3934,
747
- "step": 1230
748
- },
749
- {
750
- "epoch": 1.5,
751
- "learning_rate": 2.875219797660681e-06,
752
- "loss": 0.3954,
753
- "step": 1240
754
- },
755
- {
756
- "epoch": 1.52,
757
- "learning_rate": 2.7427579883155895e-06,
758
- "loss": 0.4046,
759
- "step": 1250
760
- },
761
- {
762
- "epoch": 1.53,
763
- "learning_rate": 2.612933383328432e-06,
764
- "loss": 0.4042,
765
- "step": 1260
766
- },
767
- {
768
- "epoch": 1.54,
769
- "learning_rate": 2.4857931595762406e-06,
770
- "loss": 0.4089,
771
- "step": 1270
772
- },
773
- {
774
- "epoch": 1.55,
775
- "learning_rate": 2.3613835184605527e-06,
776
- "loss": 0.4,
777
- "step": 1280
778
- },
779
- {
780
- "epoch": 1.56,
781
- "learning_rate": 2.239749669118272e-06,
782
- "loss": 0.4033,
783
- "step": 1290
784
- },
785
- {
786
- "epoch": 1.58,
787
- "learning_rate": 2.1209358119931845e-06,
788
- "loss": 0.4108,
789
- "step": 1300
790
- },
791
- {
792
- "epoch": 1.59,
793
- "learning_rate": 2.0049851227739746e-06,
794
- "loss": 0.4039,
795
- "step": 1310
796
- },
797
- {
798
- "epoch": 1.6,
799
- "learning_rate": 1.891939736704641e-06,
800
- "loss": 0.4134,
801
- "step": 1320
802
- },
803
- {
804
- "epoch": 1.61,
805
- "learning_rate": 1.7818407332729914e-06,
806
- "loss": 0.4096,
807
- "step": 1330
808
- },
809
- {
810
- "epoch": 1.63,
811
- "learning_rate": 1.6747281212828193e-06,
812
- "loss": 0.4128,
813
- "step": 1340
814
- },
815
- {
816
- "epoch": 1.64,
817
- "learning_rate": 1.570640824315095e-06,
818
- "loss": 0.4065,
819
- "step": 1350
820
- },
821
- {
822
- "epoch": 1.65,
823
- "learning_rate": 1.4696166665835853e-06,
824
- "loss": 0.4054,
825
- "step": 1360
826
- },
827
- {
828
- "epoch": 1.66,
829
- "learning_rate": 1.3716923591899167e-06,
830
- "loss": 0.408,
831
- "step": 1370
832
- },
833
- {
834
- "epoch": 1.67,
835
- "learning_rate": 1.2769034867831588e-06,
836
- "loss": 0.4029,
837
- "step": 1380
838
- },
839
- {
840
- "epoch": 1.69,
841
- "learning_rate": 1.1852844946287434e-06,
842
- "loss": 0.4044,
843
- "step": 1390
844
- },
845
- {
846
- "epoch": 1.7,
847
- "learning_rate": 1.0968686760914248e-06,
848
- "loss": 0.4138,
849
- "step": 1400
850
- },
851
- {
852
- "epoch": 1.71,
853
- "learning_rate": 1.0116881605368112e-06,
854
- "loss": 0.4042,
855
- "step": 1410
856
- },
857
- {
858
- "epoch": 1.72,
859
- "learning_rate": 9.297739016559226e-07,
860
- "loss": 0.3991,
861
- "step": 1420
862
- },
863
- {
864
- "epoch": 1.73,
865
- "learning_rate": 8.511556662169218e-07,
866
- "loss": 0.4039,
867
- "step": 1430
868
- },
869
- {
870
- "epoch": 1.75,
871
- "learning_rate": 7.758620232482083e-07,
872
- "loss": 0.3994,
873
- "step": 1440
874
- },
875
- {
876
- "epoch": 1.76,
877
- "learning_rate": 7.039203336567247e-07,
878
- "loss": 0.4016,
879
- "step": 1450
880
- },
881
- {
882
- "epoch": 1.77,
883
- "learning_rate": 6.353567402853056e-07,
884
- "loss": 0.3923,
885
- "step": 1460
886
- },
887
- {
888
- "epoch": 1.78,
889
- "learning_rate": 5.701961584126392e-07,
890
- "loss": 0.4143,
891
- "step": 1470
892
- },
893
- {
894
- "epoch": 1.8,
895
- "learning_rate": 5.084622666993244e-07,
896
- "loss": 0.3955,
897
- "step": 1480
898
- },
899
- {
900
- "epoch": 1.81,
901
- "learning_rate": 4.501774985832974e-07,
902
- "loss": 0.4004,
903
- "step": 1490
904
- },
905
- {
906
- "epoch": 1.82,
907
- "learning_rate": 3.953630341277603e-07,
908
- "loss": 0.4099,
909
- "step": 1500
910
- },
911
- {
912
- "epoch": 1.83,
913
- "learning_rate": 3.440387923245714e-07,
914
- "loss": 0.4063,
915
- "step": 1510
916
- },
917
- {
918
- "epoch": 1.84,
919
- "learning_rate": 2.9622342385589256e-07,
920
- "loss": 0.3992,
921
- "step": 1520
922
- },
923
- {
924
- "epoch": 1.86,
925
- "learning_rate": 2.519343043167399e-07,
926
- "loss": 0.4101,
927
- "step": 1530
928
- },
929
- {
930
- "epoch": 1.87,
931
- "learning_rate": 2.111875279008657e-07,
932
- "loss": 0.3898,
933
- "step": 1540
934
- },
935
- {
936
- "epoch": 1.88,
937
- "learning_rate": 1.7399790155230633e-07,
938
- "loss": 0.412,
939
- "step": 1550
940
- },
941
- {
942
- "epoch": 1.89,
943
- "learning_rate": 1.4037893958469994e-07,
944
- "loss": 0.4033,
945
- "step": 1560
946
- },
947
- {
948
- "epoch": 1.9,
949
- "learning_rate": 1.1034285877032147e-07,
950
- "loss": 0.3981,
951
- "step": 1570
952
- },
953
- {
954
- "epoch": 1.92,
955
- "learning_rate": 8.390057390064266e-08,
956
- "loss": 0.4058,
957
- "step": 1580
958
- },
959
- {
960
- "epoch": 1.93,
961
- "learning_rate": 6.10616938200137e-08,
962
- "loss": 0.3935,
963
- "step": 1590
964
- },
965
- {
966
- "epoch": 1.94,
967
- "learning_rate": 4.183451793390747e-08,
968
- "loss": 0.4147,
969
- "step": 1600
970
- },
971
- {
972
- "epoch": 1.95,
973
- "learning_rate": 2.6226033193007538e-08,
974
- "loss": 0.4007,
975
- "step": 1610
976
- },
977
- {
978
- "epoch": 1.96,
979
- "learning_rate": 1.424191155422583e-08,
980
- "loss": 0.4087,
981
- "step": 1620
982
- },
983
- {
984
- "epoch": 1.98,
985
- "learning_rate": 5.886507919570239e-09,
986
- "loss": 0.4052,
987
- "step": 1630
988
- },
989
- {
990
- "epoch": 1.99,
991
- "learning_rate": 1.1628585536216374e-09,
992
- "loss": 0.4062,
993
- "step": 1640
994
- },
995
- {
996
- "epoch": 2.0,
997
- "step": 1648,
998
- "total_flos": 1323567688974336.0,
999
- "train_loss": 0.4584712987964593,
1000
- "train_runtime": 26272.8035,
1001
- "train_samples_per_second": 16.064,
1002
- "train_steps_per_second": 0.063
1003
- }
1004
- ],
1005
- "logging_steps": 10,
1006
- "max_steps": 1648,
1007
- "num_train_epochs": 2,
1008
- "save_steps": 500,
1009
- "total_flos": 1323567688974336.0,
1010
- "trial_name": null,
1011
- "trial_params": null
1012
- }