longluu commited on
Commit
005a529
·
verified ·
1 Parent(s): a30c55e

--per_device_train_batch_size 2 --learning_rate 3e-5 --num_train_epochs 2 --max_seq_length 512 --doc_stride 250 --max_answer_length 200

Browse files
Files changed (1) hide show
  1. trainer_state.json +612 -0
trainer_state.json ADDED
@@ -0,0 +1,612 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.0,
5
+ "eval_steps": 500,
6
+ "global_step": 48626,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.02,
13
+ "learning_rate": 2.9691523053510468e-05,
14
+ "loss": 0.5127,
15
+ "step": 500
16
+ },
17
+ {
18
+ "epoch": 0.04,
19
+ "learning_rate": 2.9383046107020938e-05,
20
+ "loss": 0.4648,
21
+ "step": 1000
22
+ },
23
+ {
24
+ "epoch": 0.06,
25
+ "learning_rate": 2.9074569160531405e-05,
26
+ "loss": 0.4986,
27
+ "step": 1500
28
+ },
29
+ {
30
+ "epoch": 0.08,
31
+ "learning_rate": 2.8766092214041872e-05,
32
+ "loss": 0.8909,
33
+ "step": 2000
34
+ },
35
+ {
36
+ "epoch": 0.1,
37
+ "learning_rate": 2.845761526755234e-05,
38
+ "loss": 0.6387,
39
+ "step": 2500
40
+ },
41
+ {
42
+ "epoch": 0.12,
43
+ "learning_rate": 2.8149138321062806e-05,
44
+ "loss": 0.5202,
45
+ "step": 3000
46
+ },
47
+ {
48
+ "epoch": 0.14,
49
+ "learning_rate": 2.7840661374573273e-05,
50
+ "loss": 0.5361,
51
+ "step": 3500
52
+ },
53
+ {
54
+ "epoch": 0.16,
55
+ "learning_rate": 2.753218442808374e-05,
56
+ "loss": 0.5472,
57
+ "step": 4000
58
+ },
59
+ {
60
+ "epoch": 0.19,
61
+ "learning_rate": 2.722370748159421e-05,
62
+ "loss": 0.6779,
63
+ "step": 4500
64
+ },
65
+ {
66
+ "epoch": 0.21,
67
+ "learning_rate": 2.6915230535104677e-05,
68
+ "loss": 0.5864,
69
+ "step": 5000
70
+ },
71
+ {
72
+ "epoch": 0.23,
73
+ "learning_rate": 2.6606753588615144e-05,
74
+ "loss": 0.4837,
75
+ "step": 5500
76
+ },
77
+ {
78
+ "epoch": 0.25,
79
+ "learning_rate": 2.6298276642125614e-05,
80
+ "loss": 0.6121,
81
+ "step": 6000
82
+ },
83
+ {
84
+ "epoch": 0.27,
85
+ "learning_rate": 2.598979969563608e-05,
86
+ "loss": 0.447,
87
+ "step": 6500
88
+ },
89
+ {
90
+ "epoch": 0.29,
91
+ "learning_rate": 2.5681322749146548e-05,
92
+ "loss": 0.473,
93
+ "step": 7000
94
+ },
95
+ {
96
+ "epoch": 0.31,
97
+ "learning_rate": 2.5372845802657015e-05,
98
+ "loss": 0.3499,
99
+ "step": 7500
100
+ },
101
+ {
102
+ "epoch": 0.33,
103
+ "learning_rate": 2.5064368856167485e-05,
104
+ "loss": 0.456,
105
+ "step": 8000
106
+ },
107
+ {
108
+ "epoch": 0.35,
109
+ "learning_rate": 2.4755891909677952e-05,
110
+ "loss": 0.4446,
111
+ "step": 8500
112
+ },
113
+ {
114
+ "epoch": 0.37,
115
+ "learning_rate": 2.4447414963188416e-05,
116
+ "loss": 0.3856,
117
+ "step": 9000
118
+ },
119
+ {
120
+ "epoch": 0.39,
121
+ "learning_rate": 2.4138938016698886e-05,
122
+ "loss": 0.4949,
123
+ "step": 9500
124
+ },
125
+ {
126
+ "epoch": 0.41,
127
+ "learning_rate": 2.3830461070209353e-05,
128
+ "loss": 0.4161,
129
+ "step": 10000
130
+ },
131
+ {
132
+ "epoch": 0.43,
133
+ "learning_rate": 2.352198412371982e-05,
134
+ "loss": 0.4024,
135
+ "step": 10500
136
+ },
137
+ {
138
+ "epoch": 0.45,
139
+ "learning_rate": 2.321350717723029e-05,
140
+ "loss": 0.4108,
141
+ "step": 11000
142
+ },
143
+ {
144
+ "epoch": 0.47,
145
+ "learning_rate": 2.2905030230740757e-05,
146
+ "loss": 0.4503,
147
+ "step": 11500
148
+ },
149
+ {
150
+ "epoch": 0.49,
151
+ "learning_rate": 2.2596553284251224e-05,
152
+ "loss": 0.3935,
153
+ "step": 12000
154
+ },
155
+ {
156
+ "epoch": 0.51,
157
+ "learning_rate": 2.228807633776169e-05,
158
+ "loss": 0.4171,
159
+ "step": 12500
160
+ },
161
+ {
162
+ "epoch": 0.53,
163
+ "learning_rate": 2.197959939127216e-05,
164
+ "loss": 0.4416,
165
+ "step": 13000
166
+ },
167
+ {
168
+ "epoch": 0.56,
169
+ "learning_rate": 2.167112244478263e-05,
170
+ "loss": 0.526,
171
+ "step": 13500
172
+ },
173
+ {
174
+ "epoch": 0.58,
175
+ "learning_rate": 2.1362645498293095e-05,
176
+ "loss": 0.4721,
177
+ "step": 14000
178
+ },
179
+ {
180
+ "epoch": 0.6,
181
+ "learning_rate": 2.1054168551803562e-05,
182
+ "loss": 0.3916,
183
+ "step": 14500
184
+ },
185
+ {
186
+ "epoch": 0.62,
187
+ "learning_rate": 2.074569160531403e-05,
188
+ "loss": 0.3965,
189
+ "step": 15000
190
+ },
191
+ {
192
+ "epoch": 0.64,
193
+ "learning_rate": 2.0437214658824496e-05,
194
+ "loss": 0.3889,
195
+ "step": 15500
196
+ },
197
+ {
198
+ "epoch": 0.66,
199
+ "learning_rate": 2.0128737712334963e-05,
200
+ "loss": 0.3888,
201
+ "step": 16000
202
+ },
203
+ {
204
+ "epoch": 0.68,
205
+ "learning_rate": 1.9820260765845433e-05,
206
+ "loss": 0.4885,
207
+ "step": 16500
208
+ },
209
+ {
210
+ "epoch": 0.7,
211
+ "learning_rate": 1.95117838193559e-05,
212
+ "loss": 0.4247,
213
+ "step": 17000
214
+ },
215
+ {
216
+ "epoch": 0.72,
217
+ "learning_rate": 1.9203306872866367e-05,
218
+ "loss": 0.3837,
219
+ "step": 17500
220
+ },
221
+ {
222
+ "epoch": 0.74,
223
+ "learning_rate": 1.8894829926376838e-05,
224
+ "loss": 0.2933,
225
+ "step": 18000
226
+ },
227
+ {
228
+ "epoch": 0.76,
229
+ "learning_rate": 1.8586352979887305e-05,
230
+ "loss": 0.2972,
231
+ "step": 18500
232
+ },
233
+ {
234
+ "epoch": 0.78,
235
+ "learning_rate": 1.827787603339777e-05,
236
+ "loss": 0.4361,
237
+ "step": 19000
238
+ },
239
+ {
240
+ "epoch": 0.8,
241
+ "learning_rate": 1.7969399086908242e-05,
242
+ "loss": 0.3345,
243
+ "step": 19500
244
+ },
245
+ {
246
+ "epoch": 0.82,
247
+ "learning_rate": 1.7660922140418705e-05,
248
+ "loss": 0.4121,
249
+ "step": 20000
250
+ },
251
+ {
252
+ "epoch": 0.84,
253
+ "learning_rate": 1.7352445193929172e-05,
254
+ "loss": 0.3455,
255
+ "step": 20500
256
+ },
257
+ {
258
+ "epoch": 0.86,
259
+ "learning_rate": 1.704396824743964e-05,
260
+ "loss": 0.3316,
261
+ "step": 21000
262
+ },
263
+ {
264
+ "epoch": 0.88,
265
+ "learning_rate": 1.673549130095011e-05,
266
+ "loss": 0.3792,
267
+ "step": 21500
268
+ },
269
+ {
270
+ "epoch": 0.9,
271
+ "learning_rate": 1.6427014354460576e-05,
272
+ "loss": 0.3612,
273
+ "step": 22000
274
+ },
275
+ {
276
+ "epoch": 0.93,
277
+ "learning_rate": 1.6118537407971043e-05,
278
+ "loss": 0.2907,
279
+ "step": 22500
280
+ },
281
+ {
282
+ "epoch": 0.95,
283
+ "learning_rate": 1.5810060461481514e-05,
284
+ "loss": 0.3382,
285
+ "step": 23000
286
+ },
287
+ {
288
+ "epoch": 0.97,
289
+ "learning_rate": 1.550158351499198e-05,
290
+ "loss": 0.347,
291
+ "step": 23500
292
+ },
293
+ {
294
+ "epoch": 0.99,
295
+ "learning_rate": 1.5193106568502448e-05,
296
+ "loss": 0.3149,
297
+ "step": 24000
298
+ },
299
+ {
300
+ "epoch": 1.01,
301
+ "learning_rate": 1.4884629622012915e-05,
302
+ "loss": 0.2851,
303
+ "step": 24500
304
+ },
305
+ {
306
+ "epoch": 1.03,
307
+ "learning_rate": 1.4576152675523383e-05,
308
+ "loss": 0.1878,
309
+ "step": 25000
310
+ },
311
+ {
312
+ "epoch": 1.05,
313
+ "learning_rate": 1.426767572903385e-05,
314
+ "loss": 0.2873,
315
+ "step": 25500
316
+ },
317
+ {
318
+ "epoch": 1.07,
319
+ "learning_rate": 1.3959198782544319e-05,
320
+ "loss": 0.2119,
321
+ "step": 26000
322
+ },
323
+ {
324
+ "epoch": 1.09,
325
+ "learning_rate": 1.3650721836054786e-05,
326
+ "loss": 0.3332,
327
+ "step": 26500
328
+ },
329
+ {
330
+ "epoch": 1.11,
331
+ "learning_rate": 1.3342244889565253e-05,
332
+ "loss": 0.2739,
333
+ "step": 27000
334
+ },
335
+ {
336
+ "epoch": 1.13,
337
+ "learning_rate": 1.3033767943075721e-05,
338
+ "loss": 0.2526,
339
+ "step": 27500
340
+ },
341
+ {
342
+ "epoch": 1.15,
343
+ "learning_rate": 1.2725290996586188e-05,
344
+ "loss": 0.2085,
345
+ "step": 28000
346
+ },
347
+ {
348
+ "epoch": 1.17,
349
+ "learning_rate": 1.2416814050096657e-05,
350
+ "loss": 0.2634,
351
+ "step": 28500
352
+ },
353
+ {
354
+ "epoch": 1.19,
355
+ "learning_rate": 1.2108337103607124e-05,
356
+ "loss": 0.2051,
357
+ "step": 29000
358
+ },
359
+ {
360
+ "epoch": 1.21,
361
+ "learning_rate": 1.1799860157117592e-05,
362
+ "loss": 0.2745,
363
+ "step": 29500
364
+ },
365
+ {
366
+ "epoch": 1.23,
367
+ "learning_rate": 1.149138321062806e-05,
368
+ "loss": 0.2785,
369
+ "step": 30000
370
+ },
371
+ {
372
+ "epoch": 1.25,
373
+ "learning_rate": 1.1182906264138526e-05,
374
+ "loss": 0.2079,
375
+ "step": 30500
376
+ },
377
+ {
378
+ "epoch": 1.28,
379
+ "learning_rate": 1.0874429317648995e-05,
380
+ "loss": 0.1581,
381
+ "step": 31000
382
+ },
383
+ {
384
+ "epoch": 1.3,
385
+ "learning_rate": 1.0565952371159462e-05,
386
+ "loss": 0.2053,
387
+ "step": 31500
388
+ },
389
+ {
390
+ "epoch": 1.32,
391
+ "learning_rate": 1.025747542466993e-05,
392
+ "loss": 0.271,
393
+ "step": 32000
394
+ },
395
+ {
396
+ "epoch": 1.34,
397
+ "learning_rate": 9.948998478180399e-06,
398
+ "loss": 0.226,
399
+ "step": 32500
400
+ },
401
+ {
402
+ "epoch": 1.36,
403
+ "learning_rate": 9.640521531690864e-06,
404
+ "loss": 0.2281,
405
+ "step": 33000
406
+ },
407
+ {
408
+ "epoch": 1.38,
409
+ "learning_rate": 9.332044585201333e-06,
410
+ "loss": 0.2202,
411
+ "step": 33500
412
+ },
413
+ {
414
+ "epoch": 1.4,
415
+ "learning_rate": 9.0235676387118e-06,
416
+ "loss": 0.1796,
417
+ "step": 34000
418
+ },
419
+ {
420
+ "epoch": 1.42,
421
+ "learning_rate": 8.715090692222269e-06,
422
+ "loss": 0.2394,
423
+ "step": 34500
424
+ },
425
+ {
426
+ "epoch": 1.44,
427
+ "learning_rate": 8.406613745732735e-06,
428
+ "loss": 0.2211,
429
+ "step": 35000
430
+ },
431
+ {
432
+ "epoch": 1.46,
433
+ "learning_rate": 8.098136799243202e-06,
434
+ "loss": 0.3344,
435
+ "step": 35500
436
+ },
437
+ {
438
+ "epoch": 1.48,
439
+ "learning_rate": 7.789659852753671e-06,
440
+ "loss": 0.2179,
441
+ "step": 36000
442
+ },
443
+ {
444
+ "epoch": 1.5,
445
+ "learning_rate": 7.481182906264139e-06,
446
+ "loss": 0.1424,
447
+ "step": 36500
448
+ },
449
+ {
450
+ "epoch": 1.52,
451
+ "learning_rate": 7.172705959774607e-06,
452
+ "loss": 0.2399,
453
+ "step": 37000
454
+ },
455
+ {
456
+ "epoch": 1.54,
457
+ "learning_rate": 6.864229013285074e-06,
458
+ "loss": 0.2546,
459
+ "step": 37500
460
+ },
461
+ {
462
+ "epoch": 1.56,
463
+ "learning_rate": 6.555752066795541e-06,
464
+ "loss": 0.2332,
465
+ "step": 38000
466
+ },
467
+ {
468
+ "epoch": 1.58,
469
+ "learning_rate": 6.247275120306009e-06,
470
+ "loss": 0.2694,
471
+ "step": 38500
472
+ },
473
+ {
474
+ "epoch": 1.6,
475
+ "learning_rate": 5.938798173816477e-06,
476
+ "loss": 0.1963,
477
+ "step": 39000
478
+ },
479
+ {
480
+ "epoch": 1.62,
481
+ "learning_rate": 5.630321227326945e-06,
482
+ "loss": 0.1356,
483
+ "step": 39500
484
+ },
485
+ {
486
+ "epoch": 1.65,
487
+ "learning_rate": 5.3218442808374125e-06,
488
+ "loss": 0.243,
489
+ "step": 40000
490
+ },
491
+ {
492
+ "epoch": 1.67,
493
+ "learning_rate": 5.013367334347879e-06,
494
+ "loss": 0.2149,
495
+ "step": 40500
496
+ },
497
+ {
498
+ "epoch": 1.69,
499
+ "learning_rate": 4.704890387858347e-06,
500
+ "loss": 0.2202,
501
+ "step": 41000
502
+ },
503
+ {
504
+ "epoch": 1.71,
505
+ "learning_rate": 4.396413441368815e-06,
506
+ "loss": 0.1844,
507
+ "step": 41500
508
+ },
509
+ {
510
+ "epoch": 1.73,
511
+ "learning_rate": 4.087936494879283e-06,
512
+ "loss": 0.166,
513
+ "step": 42000
514
+ },
515
+ {
516
+ "epoch": 1.75,
517
+ "learning_rate": 3.7794595483897506e-06,
518
+ "loss": 0.1799,
519
+ "step": 42500
520
+ },
521
+ {
522
+ "epoch": 1.77,
523
+ "learning_rate": 3.470982601900218e-06,
524
+ "loss": 0.0921,
525
+ "step": 43000
526
+ },
527
+ {
528
+ "epoch": 1.79,
529
+ "learning_rate": 3.1625056554106857e-06,
530
+ "loss": 0.1504,
531
+ "step": 43500
532
+ },
533
+ {
534
+ "epoch": 1.81,
535
+ "learning_rate": 2.8540287089211535e-06,
536
+ "loss": 0.2312,
537
+ "step": 44000
538
+ },
539
+ {
540
+ "epoch": 1.83,
541
+ "learning_rate": 2.545551762431621e-06,
542
+ "loss": 0.1744,
543
+ "step": 44500
544
+ },
545
+ {
546
+ "epoch": 1.85,
547
+ "learning_rate": 2.2370748159420886e-06,
548
+ "loss": 0.0672,
549
+ "step": 45000
550
+ },
551
+ {
552
+ "epoch": 1.87,
553
+ "learning_rate": 1.9285978694525564e-06,
554
+ "loss": 0.1917,
555
+ "step": 45500
556
+ },
557
+ {
558
+ "epoch": 1.89,
559
+ "learning_rate": 1.6201209229630238e-06,
560
+ "loss": 0.2083,
561
+ "step": 46000
562
+ },
563
+ {
564
+ "epoch": 1.91,
565
+ "learning_rate": 1.3116439764734916e-06,
566
+ "loss": 0.1929,
567
+ "step": 46500
568
+ },
569
+ {
570
+ "epoch": 1.93,
571
+ "learning_rate": 1.0031670299839593e-06,
572
+ "loss": 0.1463,
573
+ "step": 47000
574
+ },
575
+ {
576
+ "epoch": 1.95,
577
+ "learning_rate": 6.946900834944268e-07,
578
+ "loss": 0.2275,
579
+ "step": 47500
580
+ },
581
+ {
582
+ "epoch": 1.97,
583
+ "learning_rate": 3.862131370048945e-07,
584
+ "loss": 0.1238,
585
+ "step": 48000
586
+ },
587
+ {
588
+ "epoch": 1.99,
589
+ "learning_rate": 7.773619051536215e-08,
590
+ "loss": 0.1673,
591
+ "step": 48500
592
+ },
593
+ {
594
+ "epoch": 2.0,
595
+ "step": 48626,
596
+ "total_flos": 9.031922425511117e+16,
597
+ "train_loss": 0.3265183397838395,
598
+ "train_runtime": 46558.184,
599
+ "train_samples_per_second": 2.089,
600
+ "train_steps_per_second": 1.044
601
+ }
602
+ ],
603
+ "logging_steps": 500,
604
+ "max_steps": 48626,
605
+ "num_input_tokens_seen": 0,
606
+ "num_train_epochs": 2,
607
+ "save_steps": 500,
608
+ "total_flos": 9.031922425511117e+16,
609
+ "train_batch_size": 2,
610
+ "trial_name": null,
611
+ "trial_params": null
612
+ }