Hugging Face's logo

SpireLab
/
RESPIN_LanguageModels

Fill-Mask
Transformers
Safetensors
muril
bert
mlm
low-resource
indic-languages
Model card Files
xet
 Community
SpireLab commited on
Commit
31f5e2f
·
verified ·
1 Parent(s): 358268d

Delete train_data/muril_ch_domain/checkpoint-30663/trainer_state.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. train_data/muril_ch_domain/checkpoint-30663/trainer_state.json +0 -2602
train_data/muril_ch_domain/checkpoint-30663/trainer_state.json DELETED
@@ -1,2602 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 2.999853250501394,
5
- "eval_steps": 500,
6
- "global_step": 30663,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.009783299907058651,
13
- "grad_norm": 33.780609130859375,
14
- "learning_rate": 1.6302575806977503e-06,
15
- "loss": 6.574,
16
- "step": 100
17
- },
18
- {
19
- "epoch": 0.019566599814117302,
20
- "grad_norm": 33.439022064208984,
21
- "learning_rate": 3.2605151613955006e-06,
22
- "loss": 6.1653,
23
- "step": 200
24
- },
25
- {
26
- "epoch": 0.02934989972117595,
27
- "grad_norm": 27.25751304626465,
28
- "learning_rate": 4.890772742093251e-06,
29
- "loss": 5.5515,
30
- "step": 300
31
- },
32
- {
33
- "epoch": 0.039133199628234604,
34
- "grad_norm": 38.36979675292969,
35
- "learning_rate": 6.521030322791001e-06,
36
- "loss": 5.0531,
37
- "step": 400
38
- },
39
- {
40
- "epoch": 0.048916499535293256,
41
- "grad_norm": 29.350488662719727,
42
- "learning_rate": 8.15128790348875e-06,
43
- "loss": 4.9225,
44
- "step": 500
45
- },
46
- {
47
- "epoch": 0.048916499535293256,
48
- "eval_runtime": 181.5812,
49
- "eval_samples_per_second": 112.578,
50
- "eval_steps_per_second": 14.076,
51
- "step": 500
52
- },
53
- {
54
- "epoch": 0.0586997994423519,
55
- "grad_norm": 33.02122497558594,
56
- "learning_rate": 9.781545484186502e-06,
57
- "loss": 4.8186,
58
- "step": 600
59
- },
60
- {
61
- "epoch": 0.06848309934941056,
62
- "grad_norm": 42.41593933105469,
63
- "learning_rate": 1.1411803064884251e-05,
64
- "loss": 4.5769,
65
- "step": 700
66
- },
67
- {
68
- "epoch": 0.07826639925646921,
69
- "grad_norm": 40.29044723510742,
70
- "learning_rate": 1.3042060645582003e-05,
71
- "loss": 4.3963,
72
- "step": 800
73
- },
74
- {
75
- "epoch": 0.08804969916352785,
76
- "grad_norm": 38.0811653137207,
77
- "learning_rate": 1.4672318226279752e-05,
78
- "loss": 4.3393,
79
- "step": 900
80
- },
81
- {
82
- "epoch": 0.09783299907058651,
83
- "grad_norm": 36.08370590209961,
84
- "learning_rate": 1.63025758069775e-05,
85
- "loss": 4.2421,
86
- "step": 1000
87
- },
88
- {
89
- "epoch": 0.09783299907058651,
90
- "eval_runtime": 181.886,
91
- "eval_samples_per_second": 112.389,
92
- "eval_steps_per_second": 14.053,
93
- "step": 1000
94
- },
95
- {
96
- "epoch": 0.10761629897764516,
97
- "grad_norm": 37.253684997558594,
98
- "learning_rate": 1.7932833387675256e-05,
99
- "loss": 4.1156,
100
- "step": 1100
101
- },
102
- {
103
- "epoch": 0.1173995988847038,
104
- "grad_norm": 33.003475189208984,
105
- "learning_rate": 1.9563090968373004e-05,
106
- "loss": 4.0112,
107
- "step": 1200
108
- },
109
- {
110
- "epoch": 0.12718289879176245,
111
- "grad_norm": 30.727867126464844,
112
- "learning_rate": 2.1193348549070755e-05,
113
- "loss": 3.9969,
114
- "step": 1300
115
- },
116
- {
117
- "epoch": 0.13696619869882112,
118
- "grad_norm": 37.471092224121094,
119
- "learning_rate": 2.2823606129768503e-05,
120
- "loss": 3.874,
121
- "step": 1400
122
- },
123
- {
124
- "epoch": 0.14674949860587977,
125
- "grad_norm": 42.32167434692383,
126
- "learning_rate": 2.4453863710466254e-05,
127
- "loss": 3.8518,
128
- "step": 1500
129
- },
130
- {
131
- "epoch": 0.14674949860587977,
132
- "eval_runtime": 181.9332,
133
- "eval_samples_per_second": 112.36,
134
- "eval_steps_per_second": 14.049,
135
- "step": 1500
136
- },
137
- {
138
- "epoch": 0.15653279851293841,
139
- "grad_norm": 38.00124740600586,
140
- "learning_rate": 2.6084121291164005e-05,
141
- "loss": 3.918,
142
- "step": 1600
143
- },
144
- {
145
- "epoch": 0.16631609841999706,
146
- "grad_norm": 44.637386322021484,
147
- "learning_rate": 2.7714378871861756e-05,
148
- "loss": 3.9134,
149
- "step": 1700
150
- },
151
- {
152
- "epoch": 0.1760993983270557,
153
- "grad_norm": 49.578609466552734,
154
- "learning_rate": 2.9344636452559504e-05,
155
- "loss": 3.7507,
156
- "step": 1800
157
- },
158
- {
159
- "epoch": 0.18588269823411438,
160
- "grad_norm": 36.65715789794922,
161
- "learning_rate": 3.0974894033257255e-05,
162
- "loss": 3.7551,
163
- "step": 1900
164
- },
165
- {
166
- "epoch": 0.19566599814117303,
167
- "grad_norm": 36.873443603515625,
168
- "learning_rate": 3.2605151613955e-05,
169
- "loss": 3.6951,
170
- "step": 2000
171
- },
172
- {
173
- "epoch": 0.19566599814117303,
174
- "eval_runtime": 181.8273,
175
- "eval_samples_per_second": 112.425,
176
- "eval_steps_per_second": 14.057,
177
- "step": 2000
178
- },
179
- {
180
- "epoch": 0.20544929804823167,
181
- "grad_norm": 33.025413513183594,
182
- "learning_rate": 3.423540919465276e-05,
183
- "loss": 3.6603,
184
- "step": 2100
185
- },
186
- {
187
- "epoch": 0.21523259795529032,
188
- "grad_norm": 30.105051040649414,
189
- "learning_rate": 3.586566677535051e-05,
190
- "loss": 3.525,
191
- "step": 2200
192
- },
193
- {
194
- "epoch": 0.22501589786234896,
195
- "grad_norm": 34.5129280090332,
196
- "learning_rate": 3.749592435604825e-05,
197
- "loss": 3.6454,
198
- "step": 2300
199
- },
200
- {
201
- "epoch": 0.2347991977694076,
202
- "grad_norm": 33.16934585571289,
203
- "learning_rate": 3.912618193674601e-05,
204
- "loss": 3.6356,
205
- "step": 2400
206
- },
207
- {
208
- "epoch": 0.24458249767646628,
209
- "grad_norm": 33.5789794921875,
210
- "learning_rate": 4.0756439517443756e-05,
211
- "loss": 3.5605,
212
- "step": 2500
213
- },
214
- {
215
- "epoch": 0.24458249767646628,
216
- "eval_runtime": 181.7254,
217
- "eval_samples_per_second": 112.488,
218
- "eval_steps_per_second": 14.065,
219
- "step": 2500
220
- },
221
- {
222
- "epoch": 0.2543657975835249,
223
- "grad_norm": 34.30876159667969,
224
- "learning_rate": 4.238669709814151e-05,
225
- "loss": 3.5447,
226
- "step": 2600
227
- },
228
- {
229
- "epoch": 0.2641490974905836,
230
- "grad_norm": 29.907989501953125,
231
- "learning_rate": 4.401695467883926e-05,
232
- "loss": 3.5116,
233
- "step": 2700
234
- },
235
- {
236
- "epoch": 0.27393239739764225,
237
- "grad_norm": 34.08231735229492,
238
- "learning_rate": 4.5647212259537006e-05,
239
- "loss": 3.4941,
240
- "step": 2800
241
- },
242
- {
243
- "epoch": 0.28371569730470086,
244
- "grad_norm": 25.034149169921875,
245
- "learning_rate": 4.727746984023476e-05,
246
- "loss": 3.4863,
247
- "step": 2900
248
- },
249
- {
250
- "epoch": 0.29349899721175954,
251
- "grad_norm": 32.21685028076172,
252
- "learning_rate": 4.890772742093251e-05,
253
- "loss": 3.5096,
254
- "step": 3000
255
- },
256
- {
257
- "epoch": 0.29349899721175954,
258
- "eval_runtime": 181.6612,
259
- "eval_samples_per_second": 112.528,
260
- "eval_steps_per_second": 14.07,
261
- "step": 3000
262
- },
263
- {
264
- "epoch": 0.30328229711881816,
265
- "grad_norm": 24.290380477905273,
266
- "learning_rate": 4.9940208725902305e-05,
267
- "loss": 3.3867,
268
- "step": 3100
269
- },
270
- {
271
- "epoch": 0.31306559702587683,
272
- "grad_norm": 22.924575805664062,
273
- "learning_rate": 4.975902304681838e-05,
274
- "loss": 3.398,
275
- "step": 3200
276
- },
277
- {
278
- "epoch": 0.3228488969329355,
279
- "grad_norm": 19.540430068969727,
280
- "learning_rate": 4.957783736773446e-05,
281
- "loss": 3.3727,
282
- "step": 3300
283
- },
284
- {
285
- "epoch": 0.3326321968399941,
286
- "grad_norm": 22.529376983642578,
287
- "learning_rate": 4.939665168865053e-05,
288
- "loss": 3.3364,
289
- "step": 3400
290
- },
291
- {
292
- "epoch": 0.3424154967470528,
293
- "grad_norm": 20.821264266967773,
294
- "learning_rate": 4.921546600956661e-05,
295
- "loss": 3.3126,
296
- "step": 3500
297
- },
298
- {
299
- "epoch": 0.3424154967470528,
300
- "eval_runtime": 181.7582,
301
- "eval_samples_per_second": 112.468,
302
- "eval_steps_per_second": 14.063,
303
- "step": 3500
304
- },
305
- {
306
- "epoch": 0.3521987966541114,
307
- "grad_norm": 24.346153259277344,
308
- "learning_rate": 4.903428033048268e-05,
309
- "loss": 3.2678,
310
- "step": 3600
311
- },
312
- {
313
- "epoch": 0.3619820965611701,
314
- "grad_norm": 19.89035415649414,
315
- "learning_rate": 4.8853094651398754e-05,
316
- "loss": 3.3233,
317
- "step": 3700
318
- },
319
- {
320
- "epoch": 0.37176539646822876,
321
- "grad_norm": 17.938880920410156,
322
- "learning_rate": 4.8671908972314825e-05,
323
- "loss": 3.2822,
324
- "step": 3800
325
- },
326
- {
327
- "epoch": 0.3815486963752874,
328
- "grad_norm": 16.92071533203125,
329
- "learning_rate": 4.84907232932309e-05,
330
- "loss": 3.2254,
331
- "step": 3900
332
- },
333
- {
334
- "epoch": 0.39133199628234605,
335
- "grad_norm": 18.241249084472656,
336
- "learning_rate": 4.830953761414698e-05,
337
- "loss": 3.2116,
338
- "step": 4000
339
- },
340
- {
341
- "epoch": 0.39133199628234605,
342
- "eval_runtime": 182.8906,
343
- "eval_samples_per_second": 111.772,
344
- "eval_steps_per_second": 13.976,
345
- "step": 4000
346
- },
347
- {
348
- "epoch": 0.40111529618940467,
349
- "grad_norm": 17.56020736694336,
350
- "learning_rate": 4.812835193506305e-05,
351
- "loss": 3.2232,
352
- "step": 4100
353
- },
354
- {
355
- "epoch": 0.41089859609646334,
356
- "grad_norm": 17.81117057800293,
357
- "learning_rate": 4.794716625597913e-05,
358
- "loss": 3.1936,
359
- "step": 4200
360
- },
361
- {
362
- "epoch": 0.420681896003522,
363
- "grad_norm": 19.89581871032715,
364
- "learning_rate": 4.77659805768952e-05,
365
- "loss": 3.1443,
366
- "step": 4300
367
- },
368
- {
369
- "epoch": 0.43046519591058063,
370
- "grad_norm": 22.968582153320312,
371
- "learning_rate": 4.758479489781128e-05,
372
- "loss": 3.2084,
373
- "step": 4400
374
- },
375
- {
376
- "epoch": 0.4402484958176393,
377
- "grad_norm": 17.119598388671875,
378
- "learning_rate": 4.740360921872735e-05,
379
- "loss": 3.1263,
380
- "step": 4500
381
- },
382
- {
383
- "epoch": 0.4402484958176393,
384
- "eval_runtime": 182.3246,
385
- "eval_samples_per_second": 112.119,
386
- "eval_steps_per_second": 14.019,
387
- "step": 4500
388
- },
389
- {
390
- "epoch": 0.4500317957246979,
391
- "grad_norm": 19.294527053833008,
392
- "learning_rate": 4.722242353964343e-05,
393
- "loss": 3.1327,
394
- "step": 4600
395
- },
396
- {
397
- "epoch": 0.4598150956317566,
398
- "grad_norm": 16.941057205200195,
399
- "learning_rate": 4.704123786055951e-05,
400
- "loss": 3.0944,
401
- "step": 4700
402
- },
403
- {
404
- "epoch": 0.4695983955388152,
405
- "grad_norm": 22.43411636352539,
406
- "learning_rate": 4.686005218147558e-05,
407
- "loss": 3.1093,
408
- "step": 4800
409
- },
410
- {
411
- "epoch": 0.4793816954458739,
412
- "grad_norm": 19.64097023010254,
413
- "learning_rate": 4.667886650239166e-05,
414
- "loss": 3.0597,
415
- "step": 4900
416
- },
417
- {
418
- "epoch": 0.48916499535293256,
419
- "grad_norm": 19.343788146972656,
420
- "learning_rate": 4.649768082330773e-05,
421
- "loss": 3.1659,
422
- "step": 5000
423
- },
424
- {
425
- "epoch": 0.48916499535293256,
426
- "eval_runtime": 181.8771,
427
- "eval_samples_per_second": 112.395,
428
- "eval_steps_per_second": 14.053,
429
- "step": 5000
430
- },
431
- {
432
- "epoch": 0.4989482952599912,
433
- "grad_norm": 19.657760620117188,
434
- "learning_rate": 4.63164951442238e-05,
435
- "loss": 3.0506,
436
- "step": 5100
437
- },
438
- {
439
- "epoch": 0.5087315951670498,
440
- "grad_norm": 16.2425537109375,
441
- "learning_rate": 4.613530946513987e-05,
442
- "loss": 3.0524,
443
- "step": 5200
444
- },
445
- {
446
- "epoch": 0.5185148950741085,
447
- "grad_norm": 19.64779281616211,
448
- "learning_rate": 4.595412378605595e-05,
449
- "loss": 2.9995,
450
- "step": 5300
451
- },
452
- {
453
- "epoch": 0.5282981949811671,
454
- "grad_norm": 17.29520606994629,
455
- "learning_rate": 4.577293810697203e-05,
456
- "loss": 3.0932,
457
- "step": 5400
458
- },
459
- {
460
- "epoch": 0.5380814948882258,
461
- "grad_norm": 17.694602966308594,
462
- "learning_rate": 4.55917524278881e-05,
463
- "loss": 3.0309,
464
- "step": 5500
465
- },
466
- {
467
- "epoch": 0.5380814948882258,
468
- "eval_runtime": 181.7231,
469
- "eval_samples_per_second": 112.49,
470
- "eval_steps_per_second": 14.065,
471
- "step": 5500
472
- },
473
- {
474
- "epoch": 0.5478647947952845,
475
- "grad_norm": 21.030174255371094,
476
- "learning_rate": 4.541056674880418e-05,
477
- "loss": 3.0313,
478
- "step": 5600
479
- },
480
- {
481
- "epoch": 0.5576480947023431,
482
- "grad_norm": 12.339129447937012,
483
- "learning_rate": 4.522938106972025e-05,
484
- "loss": 3.047,
485
- "step": 5700
486
- },
487
- {
488
- "epoch": 0.5674313946094017,
489
- "grad_norm": 16.496389389038086,
490
- "learning_rate": 4.504819539063633e-05,
491
- "loss": 2.9961,
492
- "step": 5800
493
- },
494
- {
495
- "epoch": 0.5772146945164603,
496
- "grad_norm": 15.456297874450684,
497
- "learning_rate": 4.48670097115524e-05,
498
- "loss": 2.9821,
499
- "step": 5900
500
- },
501
- {
502
- "epoch": 0.5869979944235191,
503
- "grad_norm": 17.8603572845459,
504
- "learning_rate": 4.468582403246848e-05,
505
- "loss": 2.9294,
506
- "step": 6000
507
- },
508
- {
509
- "epoch": 0.5869979944235191,
510
- "eval_runtime": 181.8258,
511
- "eval_samples_per_second": 112.426,
512
- "eval_steps_per_second": 14.057,
513
- "step": 6000
514
- },
515
- {
516
- "epoch": 0.5967812943305777,
517
- "grad_norm": 18.85349464416504,
518
- "learning_rate": 4.450463835338455e-05,
519
- "loss": 2.9929,
520
- "step": 6100
521
- },
522
- {
523
- "epoch": 0.6065645942376363,
524
- "grad_norm": 22.971813201904297,
525
- "learning_rate": 4.432345267430063e-05,
526
- "loss": 2.9684,
527
- "step": 6200
528
- },
529
- {
530
- "epoch": 0.616347894144695,
531
- "grad_norm": 15.877230644226074,
532
- "learning_rate": 4.4142266995216706e-05,
533
- "loss": 2.9399,
534
- "step": 6300
535
- },
536
- {
537
- "epoch": 0.6261311940517537,
538
- "grad_norm": 19.847482681274414,
539
- "learning_rate": 4.396108131613278e-05,
540
- "loss": 2.88,
541
- "step": 6400
542
- },
543
- {
544
- "epoch": 0.6359144939588123,
545
- "grad_norm": 15.004170417785645,
546
- "learning_rate": 4.377989563704885e-05,
547
- "loss": 2.9719,
548
- "step": 6500
549
- },
550
- {
551
- "epoch": 0.6359144939588123,
552
- "eval_runtime": 182.6045,
553
- "eval_samples_per_second": 111.947,
554
- "eval_steps_per_second": 13.997,
555
- "step": 6500
556
- },
557
- {
558
- "epoch": 0.645697793865871,
559
- "grad_norm": 19.473665237426758,
560
- "learning_rate": 4.359870995796492e-05,
561
- "loss": 2.9246,
562
- "step": 6600
563
- },
564
- {
565
- "epoch": 0.6554810937729296,
566
- "grad_norm": 18.071683883666992,
567
- "learning_rate": 4.3417524278881e-05,
568
- "loss": 2.9031,
569
- "step": 6700
570
- },
571
- {
572
- "epoch": 0.6652643936799882,
573
- "grad_norm": 17.544504165649414,
574
- "learning_rate": 4.323633859979707e-05,
575
- "loss": 2.8313,
576
- "step": 6800
577
- },
578
- {
579
- "epoch": 0.6750476935870469,
580
- "grad_norm": 18.936140060424805,
581
- "learning_rate": 4.305515292071315e-05,
582
- "loss": 2.8536,
583
- "step": 6900
584
- },
585
- {
586
- "epoch": 0.6848309934941056,
587
- "grad_norm": 14.77696418762207,
588
- "learning_rate": 4.2873967241629226e-05,
589
- "loss": 2.9104,
590
- "step": 7000
591
- },
592
- {
593
- "epoch": 0.6848309934941056,
594
- "eval_runtime": 181.938,
595
- "eval_samples_per_second": 112.357,
596
- "eval_steps_per_second": 14.049,
597
- "step": 7000
598
- },
599
- {
600
- "epoch": 0.6946142934011642,
601
- "grad_norm": 14.303226470947266,
602
- "learning_rate": 4.26927815625453e-05,
603
- "loss": 2.8386,
604
- "step": 7100
605
- },
606
- {
607
- "epoch": 0.7043975933082228,
608
- "grad_norm": 17.11782455444336,
609
- "learning_rate": 4.2511595883461376e-05,
610
- "loss": 2.9013,
611
- "step": 7200
612
- },
613
- {
614
- "epoch": 0.7141808932152816,
615
- "grad_norm": 18.661100387573242,
616
- "learning_rate": 4.233041020437745e-05,
617
- "loss": 2.9428,
618
- "step": 7300
619
- },
620
- {
621
- "epoch": 0.7239641931223402,
622
- "grad_norm": 15.535719871520996,
623
- "learning_rate": 4.2149224525293525e-05,
624
- "loss": 2.8582,
625
- "step": 7400
626
- },
627
- {
628
- "epoch": 0.7337474930293988,
629
- "grad_norm": 15.3306303024292,
630
- "learning_rate": 4.19680388462096e-05,
631
- "loss": 2.8896,
632
- "step": 7500
633
- },
634
- {
635
- "epoch": 0.7337474930293988,
636
- "eval_runtime": 181.8938,
637
- "eval_samples_per_second": 112.384,
638
- "eval_steps_per_second": 14.052,
639
- "step": 7500
640
- },
641
- {
642
- "epoch": 0.7435307929364575,
643
- "grad_norm": 16.730344772338867,
644
- "learning_rate": 4.1786853167125675e-05,
645
- "loss": 2.9097,
646
- "step": 7600
647
- },
648
- {
649
- "epoch": 0.7533140928435161,
650
- "grad_norm": 18.755483627319336,
651
- "learning_rate": 4.1605667488041746e-05,
652
- "loss": 2.8815,
653
- "step": 7700
654
- },
655
- {
656
- "epoch": 0.7630973927505748,
657
- "grad_norm": 18.737581253051758,
658
- "learning_rate": 4.1424481808957824e-05,
659
- "loss": 2.9202,
660
- "step": 7800
661
- },
662
- {
663
- "epoch": 0.7728806926576334,
664
- "grad_norm": 14.711681365966797,
665
- "learning_rate": 4.1243296129873896e-05,
666
- "loss": 2.806,
667
- "step": 7900
668
- },
669
- {
670
- "epoch": 0.7826639925646921,
671
- "grad_norm": 17.5069580078125,
672
- "learning_rate": 4.106211045078997e-05,
673
- "loss": 2.8576,
674
- "step": 8000
675
- },
676
- {
677
- "epoch": 0.7826639925646921,
678
- "eval_runtime": 181.9442,
679
- "eval_samples_per_second": 112.353,
680
- "eval_steps_per_second": 14.048,
681
- "step": 8000
682
- },
683
- {
684
- "epoch": 0.7924472924717507,
685
- "grad_norm": 17.678852081298828,
686
- "learning_rate": 4.0880924771706046e-05,
687
- "loss": 2.8035,
688
- "step": 8100
689
- },
690
- {
691
- "epoch": 0.8022305923788093,
692
- "grad_norm": 17.644638061523438,
693
- "learning_rate": 4.069973909262212e-05,
694
- "loss": 2.7958,
695
- "step": 8200
696
- },
697
- {
698
- "epoch": 0.8120138922858681,
699
- "grad_norm": 18.377134323120117,
700
- "learning_rate": 4.0518553413538195e-05,
701
- "loss": 2.8055,
702
- "step": 8300
703
- },
704
- {
705
- "epoch": 0.8217971921929267,
706
- "grad_norm": 18.026033401489258,
707
- "learning_rate": 4.0337367734454273e-05,
708
- "loss": 2.7334,
709
- "step": 8400
710
- },
711
- {
712
- "epoch": 0.8315804920999853,
713
- "grad_norm": 14.77315616607666,
714
- "learning_rate": 4.0156182055370345e-05,
715
- "loss": 2.8082,
716
- "step": 8500
717
- },
718
- {
719
- "epoch": 0.8315804920999853,
720
- "eval_runtime": 182.4176,
721
- "eval_samples_per_second": 112.062,
722
- "eval_steps_per_second": 14.012,
723
- "step": 8500
724
- },
725
- {
726
- "epoch": 0.841363792007044,
727
- "grad_norm": 13.729479789733887,
728
- "learning_rate": 3.997499637628642e-05,
729
- "loss": 2.7939,
730
- "step": 8600
731
- },
732
- {
733
- "epoch": 0.8511470919141026,
734
- "grad_norm": 16.34333610534668,
735
- "learning_rate": 3.9793810697202494e-05,
736
- "loss": 2.8517,
737
- "step": 8700
738
- },
739
- {
740
- "epoch": 0.8609303918211613,
741
- "grad_norm": 22.484411239624023,
742
- "learning_rate": 3.961262501811857e-05,
743
- "loss": 2.776,
744
- "step": 8800
745
- },
746
- {
747
- "epoch": 0.8707136917282199,
748
- "grad_norm": 15.922870635986328,
749
- "learning_rate": 3.9431439339034644e-05,
750
- "loss": 2.7909,
751
- "step": 8900
752
- },
753
- {
754
- "epoch": 0.8804969916352786,
755
- "grad_norm": 15.06955623626709,
756
- "learning_rate": 3.925025365995072e-05,
757
- "loss": 2.8416,
758
- "step": 9000
759
- },
760
- {
761
- "epoch": 0.8804969916352786,
762
- "eval_runtime": 181.9314,
763
- "eval_samples_per_second": 112.361,
764
- "eval_steps_per_second": 14.049,
765
- "step": 9000
766
- },
767
- {
768
- "epoch": 0.8902802915423372,
769
- "grad_norm": 16.060428619384766,
770
- "learning_rate": 3.9069067980866794e-05,
771
- "loss": 2.7803,
772
- "step": 9100
773
- },
774
- {
775
- "epoch": 0.9000635914493959,
776
- "grad_norm": 16.80124855041504,
777
- "learning_rate": 3.888788230178287e-05,
778
- "loss": 2.7548,
779
- "step": 9200
780
- },
781
- {
782
- "epoch": 0.9098468913564546,
783
- "grad_norm": 16.608434677124023,
784
- "learning_rate": 3.870669662269894e-05,
785
- "loss": 2.8606,
786
- "step": 9300
787
- },
788
- {
789
- "epoch": 0.9196301912635132,
790
- "grad_norm": 14.83870792388916,
791
- "learning_rate": 3.8525510943615015e-05,
792
- "loss": 2.7833,
793
- "step": 9400
794
- },
795
- {
796
- "epoch": 0.9294134911705718,
797
- "grad_norm": 25.778181076049805,
798
- "learning_rate": 3.834432526453109e-05,
799
- "loss": 2.7434,
800
- "step": 9500
801
- },
802
- {
803
- "epoch": 0.9294134911705718,
804
- "eval_runtime": 181.99,
805
- "eval_samples_per_second": 112.325,
806
- "eval_steps_per_second": 14.045,
807
- "step": 9500
808
- },
809
- {
810
- "epoch": 0.9391967910776304,
811
- "grad_norm": 17.374011993408203,
812
- "learning_rate": 3.8163139585447164e-05,
813
- "loss": 2.7258,
814
- "step": 9600
815
- },
816
- {
817
- "epoch": 0.9489800909846892,
818
- "grad_norm": 17.551128387451172,
819
- "learning_rate": 3.798195390636324e-05,
820
- "loss": 2.824,
821
- "step": 9700
822
- },
823
- {
824
- "epoch": 0.9587633908917478,
825
- "grad_norm": 14.35797119140625,
826
- "learning_rate": 3.7800768227279314e-05,
827
- "loss": 2.745,
828
- "step": 9800
829
- },
830
- {
831
- "epoch": 0.9685466907988064,
832
- "grad_norm": 20.098552703857422,
833
- "learning_rate": 3.761958254819539e-05,
834
- "loss": 2.7025,
835
- "step": 9900
836
- },
837
- {
838
- "epoch": 0.9783299907058651,
839
- "grad_norm": 16.218109130859375,
840
- "learning_rate": 3.743839686911147e-05,
841
- "loss": 2.8093,
842
- "step": 10000
843
- },
844
- {
845
- "epoch": 0.9783299907058651,
846
- "eval_runtime": 181.8987,
847
- "eval_samples_per_second": 112.381,
848
- "eval_steps_per_second": 14.052,
849
- "step": 10000
850
- },
851
- {
852
- "epoch": 0.9881132906129237,
853
- "grad_norm": 17.198423385620117,
854
- "learning_rate": 3.725721119002754e-05,
855
- "loss": 2.7124,
856
- "step": 10100
857
- },
858
- {
859
- "epoch": 0.9978965905199824,
860
- "grad_norm": 18.021198272705078,
861
- "learning_rate": 3.707602551094362e-05,
862
- "loss": 2.6922,
863
- "step": 10200
864
- },
865
- {
866
- "epoch": 1.007679890427041,
867
- "grad_norm": 15.27678108215332,
868
- "learning_rate": 3.689483983185969e-05,
869
- "loss": 2.6743,
870
- "step": 10300
871
- },
872
- {
873
- "epoch": 1.0174631903340996,
874
- "grad_norm": 16.770511627197266,
875
- "learning_rate": 3.671365415277577e-05,
876
- "loss": 2.857,
877
- "step": 10400
878
- },
879
- {
880
- "epoch": 1.0272464902411584,
881
- "grad_norm": 18.810932159423828,
882
- "learning_rate": 3.653246847369184e-05,
883
- "loss": 2.7269,
884
- "step": 10500
885
- },
886
- {
887
- "epoch": 1.0272464902411584,
888
- "eval_runtime": 181.8537,
889
- "eval_samples_per_second": 112.409,
890
- "eval_steps_per_second": 14.055,
891
- "step": 10500
892
- },
893
- {
894
- "epoch": 1.037029790148217,
895
- "grad_norm": 18.56201171875,
896
- "learning_rate": 3.635128279460791e-05,
897
- "loss": 2.7325,
898
- "step": 10600
899
- },
900
- {
901
- "epoch": 1.0468130900552757,
902
- "grad_norm": 15.063011169433594,
903
- "learning_rate": 3.617009711552399e-05,
904
- "loss": 2.7827,
905
- "step": 10700
906
- },
907
- {
908
- "epoch": 1.0565963899623343,
909
- "grad_norm": 15.339439392089844,
910
- "learning_rate": 3.598891143644006e-05,
911
- "loss": 2.7472,
912
- "step": 10800
913
- },
914
- {
915
- "epoch": 1.066379689869393,
916
- "grad_norm": 17.466033935546875,
917
- "learning_rate": 3.580772575735614e-05,
918
- "loss": 2.7859,
919
- "step": 10900
920
- },
921
- {
922
- "epoch": 1.0761629897764515,
923
- "grad_norm": 20.727872848510742,
924
- "learning_rate": 3.562654007827221e-05,
925
- "loss": 2.7278,
926
- "step": 11000
927
- },
928
- {
929
- "epoch": 1.0761629897764515,
930
- "eval_runtime": 181.8566,
931
- "eval_samples_per_second": 112.407,
932
- "eval_steps_per_second": 14.055,
933
- "step": 11000
934
- },
935
- {
936
- "epoch": 1.0859462896835101,
937
- "grad_norm": 16.02055549621582,
938
- "learning_rate": 3.544535439918829e-05,
939
- "loss": 2.6307,
940
- "step": 11100
941
- },
942
- {
943
- "epoch": 1.095729589590569,
944
- "grad_norm": 20.069686889648438,
945
- "learning_rate": 3.526416872010436e-05,
946
- "loss": 2.711,
947
- "step": 11200
948
- },
949
- {
950
- "epoch": 1.1055128894976276,
951
- "grad_norm": 14.833261489868164,
952
- "learning_rate": 3.508298304102044e-05,
953
- "loss": 2.6141,
954
- "step": 11300
955
- },
956
- {
957
- "epoch": 1.1152961894046862,
958
- "grad_norm": 14.86436653137207,
959
- "learning_rate": 3.490179736193652e-05,
960
- "loss": 2.6816,
961
- "step": 11400
962
- },
963
- {
964
- "epoch": 1.1250794893117448,
965
- "grad_norm": 17.955862045288086,
966
- "learning_rate": 3.472061168285259e-05,
967
- "loss": 2.6924,
968
- "step": 11500
969
- },
970
- {
971
- "epoch": 1.1250794893117448,
972
- "eval_runtime": 181.8085,
973
- "eval_samples_per_second": 112.437,
974
- "eval_steps_per_second": 14.059,
975
- "step": 11500
976
- },
977
- {
978
- "epoch": 1.1348627892188035,
979
- "grad_norm": 18.360109329223633,
980
- "learning_rate": 3.453942600376867e-05,
981
- "loss": 2.6181,
982
- "step": 11600
983
- },
984
- {
985
- "epoch": 1.144646089125862,
986
- "grad_norm": 17.547542572021484,
987
- "learning_rate": 3.435824032468474e-05,
988
- "loss": 2.6394,
989
- "step": 11700
990
- },
991
- {
992
- "epoch": 1.154429389032921,
993
- "grad_norm": 12.194833755493164,
994
- "learning_rate": 3.417705464560082e-05,
995
- "loss": 2.6684,
996
- "step": 11800
997
- },
998
- {
999
- "epoch": 1.1642126889399795,
1000
- "grad_norm": 17.095104217529297,
1001
- "learning_rate": 3.399586896651689e-05,
1002
- "loss": 2.6129,
1003
- "step": 11900
1004
- },
1005
- {
1006
- "epoch": 1.1739959888470382,
1007
- "grad_norm": 20.788406372070312,
1008
- "learning_rate": 3.381468328743296e-05,
1009
- "loss": 2.5663,
1010
- "step": 12000
1011
- },
1012
- {
1013
- "epoch": 1.1739959888470382,
1014
- "eval_runtime": 181.8035,
1015
- "eval_samples_per_second": 112.44,
1016
- "eval_steps_per_second": 14.059,
1017
- "step": 12000
1018
- },
1019
- {
1020
- "epoch": 1.1837792887540968,
1021
- "grad_norm": 14.261167526245117,
1022
- "learning_rate": 3.363349760834904e-05,
1023
- "loss": 2.6544,
1024
- "step": 12100
1025
- },
1026
- {
1027
- "epoch": 1.1935625886611554,
1028
- "grad_norm": 24.68012046813965,
1029
- "learning_rate": 3.345231192926511e-05,
1030
- "loss": 2.6632,
1031
- "step": 12200
1032
- },
1033
- {
1034
- "epoch": 1.203345888568214,
1035
- "grad_norm": 16.10886573791504,
1036
- "learning_rate": 3.327112625018119e-05,
1037
- "loss": 2.6366,
1038
- "step": 12300
1039
- },
1040
- {
1041
- "epoch": 1.2131291884752726,
1042
- "grad_norm": 18.038848876953125,
1043
- "learning_rate": 3.308994057109726e-05,
1044
- "loss": 2.6563,
1045
- "step": 12400
1046
- },
1047
- {
1048
- "epoch": 1.2229124883823315,
1049
- "grad_norm": 17.40920639038086,
1050
- "learning_rate": 3.290875489201334e-05,
1051
- "loss": 2.718,
1052
- "step": 12500
1053
- },
1054
- {
1055
- "epoch": 1.2229124883823315,
1056
- "eval_runtime": 181.9491,
1057
- "eval_samples_per_second": 112.35,
1058
- "eval_steps_per_second": 14.048,
1059
- "step": 12500
1060
- },
1061
- {
1062
- "epoch": 1.23269578828939,
1063
- "grad_norm": 15.097307205200195,
1064
- "learning_rate": 3.272756921292941e-05,
1065
- "loss": 2.7282,
1066
- "step": 12600
1067
- },
1068
- {
1069
- "epoch": 1.2424790881964487,
1070
- "grad_norm": 17.63008689880371,
1071
- "learning_rate": 3.254638353384549e-05,
1072
- "loss": 2.7104,
1073
- "step": 12700
1074
- },
1075
- {
1076
- "epoch": 1.2522623881035073,
1077
- "grad_norm": 16.161130905151367,
1078
- "learning_rate": 3.236519785476156e-05,
1079
- "loss": 2.6427,
1080
- "step": 12800
1081
- },
1082
- {
1083
- "epoch": 1.262045688010566,
1084
- "grad_norm": 18.786882400512695,
1085
- "learning_rate": 3.218401217567764e-05,
1086
- "loss": 2.6105,
1087
- "step": 12900
1088
- },
1089
- {
1090
- "epoch": 1.2718289879176246,
1091
- "grad_norm": 24.145421981811523,
1092
- "learning_rate": 3.2002826496593715e-05,
1093
- "loss": 2.6322,
1094
- "step": 13000
1095
- },
1096
- {
1097
- "epoch": 1.2718289879176246,
1098
- "eval_runtime": 182.5613,
1099
- "eval_samples_per_second": 111.973,
1100
- "eval_steps_per_second": 14.001,
1101
- "step": 13000
1102
- },
1103
- {
1104
- "epoch": 1.2816122878246832,
1105
- "grad_norm": 15.286133766174316,
1106
- "learning_rate": 3.1821640817509786e-05,
1107
- "loss": 2.6465,
1108
- "step": 13100
1109
- },
1110
- {
1111
- "epoch": 1.291395587731742,
1112
- "grad_norm": 21.22935676574707,
1113
- "learning_rate": 3.1640455138425865e-05,
1114
- "loss": 2.6691,
1115
- "step": 13200
1116
- },
1117
- {
1118
- "epoch": 1.3011788876388006,
1119
- "grad_norm": 18.064428329467773,
1120
- "learning_rate": 3.1459269459341936e-05,
1121
- "loss": 2.5904,
1122
- "step": 13300
1123
- },
1124
- {
1125
- "epoch": 1.3109621875458592,
1126
- "grad_norm": 14.45976448059082,
1127
- "learning_rate": 3.127808378025801e-05,
1128
- "loss": 2.6602,
1129
- "step": 13400
1130
- },
1131
- {
1132
- "epoch": 1.3207454874529179,
1133
- "grad_norm": 19.72386360168457,
1134
- "learning_rate": 3.109689810117408e-05,
1135
- "loss": 2.6337,
1136
- "step": 13500
1137
- },
1138
- {
1139
- "epoch": 1.3207454874529179,
1140
- "eval_runtime": 182.4053,
1141
- "eval_samples_per_second": 112.069,
1142
- "eval_steps_per_second": 14.013,
1143
- "step": 13500
1144
- },
1145
- {
1146
- "epoch": 1.3305287873599765,
1147
- "grad_norm": 17.639583587646484,
1148
- "learning_rate": 3.091571242209016e-05,
1149
- "loss": 2.6135,
1150
- "step": 13600
1151
- },
1152
- {
1153
- "epoch": 1.340312087267035,
1154
- "grad_norm": 19.71700096130371,
1155
- "learning_rate": 3.0734526743006235e-05,
1156
- "loss": 2.6252,
1157
- "step": 13700
1158
- },
1159
- {
1160
- "epoch": 1.3500953871740937,
1161
- "grad_norm": 16.715856552124023,
1162
- "learning_rate": 3.055334106392231e-05,
1163
- "loss": 2.6475,
1164
- "step": 13800
1165
- },
1166
- {
1167
- "epoch": 1.3598786870811526,
1168
- "grad_norm": 12.645075798034668,
1169
- "learning_rate": 3.0372155384838385e-05,
1170
- "loss": 2.6199,
1171
- "step": 13900
1172
- },
1173
- {
1174
- "epoch": 1.3696619869882112,
1175
- "grad_norm": 20.150625228881836,
1176
- "learning_rate": 3.0190969705754456e-05,
1177
- "loss": 2.5567,
1178
- "step": 14000
1179
- },
1180
- {
1181
- "epoch": 1.3696619869882112,
1182
- "eval_runtime": 181.9086,
1183
- "eval_samples_per_second": 112.375,
1184
- "eval_steps_per_second": 14.051,
1185
- "step": 14000
1186
- },
1187
- {
1188
- "epoch": 1.3794452868952698,
1189
- "grad_norm": 19.111286163330078,
1190
- "learning_rate": 3.0009784026670535e-05,
1191
- "loss": 2.59,
1192
- "step": 14100
1193
- },
1194
- {
1195
- "epoch": 1.3892285868023284,
1196
- "grad_norm": 17.12226104736328,
1197
- "learning_rate": 2.9828598347586606e-05,
1198
- "loss": 2.5913,
1199
- "step": 14200
1200
- },
1201
- {
1202
- "epoch": 1.399011886709387,
1203
- "grad_norm": 19.741445541381836,
1204
- "learning_rate": 2.9647412668502684e-05,
1205
- "loss": 2.5617,
1206
- "step": 14300
1207
- },
1208
- {
1209
- "epoch": 1.4087951866164456,
1210
- "grad_norm": 17.605525970458984,
1211
- "learning_rate": 2.946622698941876e-05,
1212
- "loss": 2.6077,
1213
- "step": 14400
1214
- },
1215
- {
1216
- "epoch": 1.4185784865235043,
1217
- "grad_norm": 17.433218002319336,
1218
- "learning_rate": 2.928504131033483e-05,
1219
- "loss": 2.5713,
1220
- "step": 14500
1221
- },
1222
- {
1223
- "epoch": 1.4185784865235043,
1224
- "eval_runtime": 181.9305,
1225
- "eval_samples_per_second": 112.362,
1226
- "eval_steps_per_second": 14.049,
1227
- "step": 14500
1228
- },
1229
- {
1230
- "epoch": 1.428361786430563,
1231
- "grad_norm": 15.442538261413574,
1232
- "learning_rate": 2.910385563125091e-05,
1233
- "loss": 2.6499,
1234
- "step": 14600
1235
- },
1236
- {
1237
- "epoch": 1.4381450863376217,
1238
- "grad_norm": 15.078730583190918,
1239
- "learning_rate": 2.892266995216698e-05,
1240
- "loss": 2.6517,
1241
- "step": 14700
1242
- },
1243
- {
1244
- "epoch": 1.4479283862446803,
1245
- "grad_norm": 23.07891273498535,
1246
- "learning_rate": 2.874148427308306e-05,
1247
- "loss": 2.594,
1248
- "step": 14800
1249
- },
1250
- {
1251
- "epoch": 1.457711686151739,
1252
- "grad_norm": 16.707923889160156,
1253
- "learning_rate": 2.856029859399913e-05,
1254
- "loss": 2.6613,
1255
- "step": 14900
1256
- },
1257
- {
1258
- "epoch": 1.4674949860587976,
1259
- "grad_norm": 16.731164932250977,
1260
- "learning_rate": 2.8379112914915208e-05,
1261
- "loss": 2.5927,
1262
- "step": 15000
1263
- },
1264
- {
1265
- "epoch": 1.4674949860587976,
1266
- "eval_runtime": 181.9649,
1267
- "eval_samples_per_second": 112.34,
1268
- "eval_steps_per_second": 14.047,
1269
- "step": 15000
1270
- },
1271
- {
1272
- "epoch": 1.4772782859658564,
1273
- "grad_norm": 16.020864486694336,
1274
- "learning_rate": 2.819792723583128e-05,
1275
- "loss": 2.6464,
1276
- "step": 15100
1277
- },
1278
- {
1279
- "epoch": 1.4870615858729148,
1280
- "grad_norm": 16.674760818481445,
1281
- "learning_rate": 2.8016741556747354e-05,
1282
- "loss": 2.5853,
1283
- "step": 15200
1284
- },
1285
- {
1286
- "epoch": 1.4968448857799737,
1287
- "grad_norm": 16.890748977661133,
1288
- "learning_rate": 2.7835555877663432e-05,
1289
- "loss": 2.5748,
1290
- "step": 15300
1291
- },
1292
- {
1293
- "epoch": 1.5066281856870323,
1294
- "grad_norm": 20.217845916748047,
1295
- "learning_rate": 2.7654370198579504e-05,
1296
- "loss": 2.6204,
1297
- "step": 15400
1298
- },
1299
- {
1300
- "epoch": 1.516411485594091,
1301
- "grad_norm": 20.459087371826172,
1302
- "learning_rate": 2.7473184519495582e-05,
1303
- "loss": 2.6103,
1304
- "step": 15500
1305
- },
1306
- {
1307
- "epoch": 1.516411485594091,
1308
- "eval_runtime": 181.9454,
1309
- "eval_samples_per_second": 112.352,
1310
- "eval_steps_per_second": 14.048,
1311
- "step": 15500
1312
- },
1313
- {
1314
- "epoch": 1.5261947855011495,
1315
- "grad_norm": 18.207612991333008,
1316
- "learning_rate": 2.7291998840411654e-05,
1317
- "loss": 2.5786,
1318
- "step": 15600
1319
- },
1320
- {
1321
- "epoch": 1.5359780854082081,
1322
- "grad_norm": 18.084758758544922,
1323
- "learning_rate": 2.7110813161327732e-05,
1324
- "loss": 2.6535,
1325
- "step": 15700
1326
- },
1327
- {
1328
- "epoch": 1.545761385315267,
1329
- "grad_norm": 15.03881549835205,
1330
- "learning_rate": 2.6929627482243803e-05,
1331
- "loss": 2.6061,
1332
- "step": 15800
1333
- },
1334
- {
1335
- "epoch": 1.5555446852223254,
1336
- "grad_norm": 16.99995231628418,
1337
- "learning_rate": 2.6748441803159878e-05,
1338
- "loss": 2.6151,
1339
- "step": 15900
1340
- },
1341
- {
1342
- "epoch": 1.5653279851293842,
1343
- "grad_norm": 15.581089973449707,
1344
- "learning_rate": 2.6567256124075956e-05,
1345
- "loss": 2.6163,
1346
- "step": 16000
1347
- },
1348
- {
1349
- "epoch": 1.5653279851293842,
1350
- "eval_runtime": 181.8152,
1351
- "eval_samples_per_second": 112.433,
1352
- "eval_steps_per_second": 14.058,
1353
- "step": 16000
1354
- },
1355
- {
1356
- "epoch": 1.5751112850364428,
1357
- "grad_norm": 21.4382266998291,
1358
- "learning_rate": 2.6386070444992028e-05,
1359
- "loss": 2.5975,
1360
- "step": 16100
1361
- },
1362
- {
1363
- "epoch": 1.5848945849435014,
1364
- "grad_norm": 15.874536514282227,
1365
- "learning_rate": 2.6204884765908106e-05,
1366
- "loss": 2.5851,
1367
- "step": 16200
1368
- },
1369
- {
1370
- "epoch": 1.59467788485056,
1371
- "grad_norm": 17.902137756347656,
1372
- "learning_rate": 2.6023699086824177e-05,
1373
- "loss": 2.6027,
1374
- "step": 16300
1375
- },
1376
- {
1377
- "epoch": 1.6044611847576187,
1378
- "grad_norm": 17.04872703552246,
1379
- "learning_rate": 2.5842513407740255e-05,
1380
- "loss": 2.5854,
1381
- "step": 16400
1382
- },
1383
- {
1384
- "epoch": 1.6142444846646775,
1385
- "grad_norm": 15.406013488769531,
1386
- "learning_rate": 2.5661327728656327e-05,
1387
- "loss": 2.5158,
1388
- "step": 16500
1389
- },
1390
- {
1391
- "epoch": 1.6142444846646775,
1392
- "eval_runtime": 181.8647,
1393
- "eval_samples_per_second": 112.402,
1394
- "eval_steps_per_second": 14.054,
1395
- "step": 16500
1396
- },
1397
- {
1398
- "epoch": 1.624027784571736,
1399
- "grad_norm": 19.62627601623535,
1400
- "learning_rate": 2.5480142049572402e-05,
1401
- "loss": 2.5378,
1402
- "step": 16600
1403
- },
1404
- {
1405
- "epoch": 1.6338110844787948,
1406
- "grad_norm": 17.825178146362305,
1407
- "learning_rate": 2.529895637048848e-05,
1408
- "loss": 2.6162,
1409
- "step": 16700
1410
- },
1411
- {
1412
- "epoch": 1.6435943843858534,
1413
- "grad_norm": 15.442023277282715,
1414
- "learning_rate": 2.511777069140455e-05,
1415
- "loss": 2.5802,
1416
- "step": 16800
1417
- },
1418
- {
1419
- "epoch": 1.653377684292912,
1420
- "grad_norm": 18.695241928100586,
1421
- "learning_rate": 2.4936585012320626e-05,
1422
- "loss": 2.585,
1423
- "step": 16900
1424
- },
1425
- {
1426
- "epoch": 1.6631609841999706,
1427
- "grad_norm": 18.992969512939453,
1428
- "learning_rate": 2.4755399333236704e-05,
1429
- "loss": 2.5448,
1430
- "step": 17000
1431
- },
1432
- {
1433
- "epoch": 1.6631609841999706,
1434
- "eval_runtime": 181.91,
1435
- "eval_samples_per_second": 112.374,
1436
- "eval_steps_per_second": 14.051,
1437
- "step": 17000
1438
- },
1439
- {
1440
- "epoch": 1.6729442841070292,
1441
- "grad_norm": 19.065349578857422,
1442
- "learning_rate": 2.457421365415278e-05,
1443
- "loss": 2.6565,
1444
- "step": 17100
1445
- },
1446
- {
1447
- "epoch": 1.682727584014088,
1448
- "grad_norm": 20.110734939575195,
1449
- "learning_rate": 2.439302797506885e-05,
1450
- "loss": 2.5519,
1451
- "step": 17200
1452
- },
1453
- {
1454
- "epoch": 1.6925108839211465,
1455
- "grad_norm": 15.886931419372559,
1456
- "learning_rate": 2.4211842295984925e-05,
1457
- "loss": 2.5589,
1458
- "step": 17300
1459
- },
1460
- {
1461
- "epoch": 1.7022941838282053,
1462
- "grad_norm": 19.213207244873047,
1463
- "learning_rate": 2.4030656616901e-05,
1464
- "loss": 2.5714,
1465
- "step": 17400
1466
- },
1467
- {
1468
- "epoch": 1.712077483735264,
1469
- "grad_norm": 17.117481231689453,
1470
- "learning_rate": 2.3849470937817075e-05,
1471
- "loss": 2.6682,
1472
- "step": 17500
1473
- },
1474
- {
1475
- "epoch": 1.712077483735264,
1476
- "eval_runtime": 181.766,
1477
- "eval_samples_per_second": 112.463,
1478
- "eval_steps_per_second": 14.062,
1479
- "step": 17500
1480
- },
1481
- {
1482
- "epoch": 1.7218607836423225,
1483
- "grad_norm": 17.19162940979004,
1484
- "learning_rate": 2.366828525873315e-05,
1485
- "loss": 2.5591,
1486
- "step": 17600
1487
- },
1488
- {
1489
- "epoch": 1.7316440835493812,
1490
- "grad_norm": 15.454411506652832,
1491
- "learning_rate": 2.3487099579649225e-05,
1492
- "loss": 2.469,
1493
- "step": 17700
1494
- },
1495
- {
1496
- "epoch": 1.7414273834564398,
1497
- "grad_norm": 15.227791786193848,
1498
- "learning_rate": 2.3305913900565303e-05,
1499
- "loss": 2.664,
1500
- "step": 17800
1501
- },
1502
- {
1503
- "epoch": 1.7512106833634986,
1504
- "grad_norm": 18.5739688873291,
1505
- "learning_rate": 2.3124728221481374e-05,
1506
- "loss": 2.5991,
1507
- "step": 17900
1508
- },
1509
- {
1510
- "epoch": 1.760993983270557,
1511
- "grad_norm": 12.589066505432129,
1512
- "learning_rate": 2.294354254239745e-05,
1513
- "loss": 2.6593,
1514
- "step": 18000
1515
- },
1516
- {
1517
- "epoch": 1.760993983270557,
1518
- "eval_runtime": 181.9699,
1519
- "eval_samples_per_second": 112.337,
1520
- "eval_steps_per_second": 14.046,
1521
- "step": 18000
1522
- },
1523
- {
1524
- "epoch": 1.7707772831776158,
1525
- "grad_norm": 20.695772171020508,
1526
- "learning_rate": 2.2762356863313524e-05,
1527
- "loss": 2.5555,
1528
- "step": 18100
1529
- },
1530
- {
1531
- "epoch": 1.7805605830846745,
1532
- "grad_norm": 12.731703758239746,
1533
- "learning_rate": 2.25811711842296e-05,
1534
- "loss": 2.4617,
1535
- "step": 18200
1536
- },
1537
- {
1538
- "epoch": 1.790343882991733,
1539
- "grad_norm": 18.506074905395508,
1540
- "learning_rate": 2.2399985505145674e-05,
1541
- "loss": 2.6061,
1542
- "step": 18300
1543
- },
1544
- {
1545
- "epoch": 1.800127182898792,
1546
- "grad_norm": 14.8694486618042,
1547
- "learning_rate": 2.221879982606175e-05,
1548
- "loss": 2.5779,
1549
- "step": 18400
1550
- },
1551
- {
1552
- "epoch": 1.8099104828058503,
1553
- "grad_norm": 22.47985076904297,
1554
- "learning_rate": 2.2037614146977827e-05,
1555
- "loss": 2.5012,
1556
- "step": 18500
1557
- },
1558
- {
1559
- "epoch": 1.8099104828058503,
1560
- "eval_runtime": 182.3919,
1561
- "eval_samples_per_second": 112.077,
1562
- "eval_steps_per_second": 14.014,
1563
- "step": 18500
1564
- },
1565
- {
1566
- "epoch": 1.8196937827129092,
1567
- "grad_norm": 25.74334144592285,
1568
- "learning_rate": 2.1856428467893898e-05,
1569
- "loss": 2.5265,
1570
- "step": 18600
1571
- },
1572
- {
1573
- "epoch": 1.8294770826199676,
1574
- "grad_norm": 18.477630615234375,
1575
- "learning_rate": 2.1675242788809973e-05,
1576
- "loss": 2.5555,
1577
- "step": 18700
1578
- },
1579
- {
1580
- "epoch": 1.8392603825270264,
1581
- "grad_norm": 14.832316398620605,
1582
- "learning_rate": 2.1494057109726048e-05,
1583
- "loss": 2.4609,
1584
- "step": 18800
1585
- },
1586
- {
1587
- "epoch": 1.849043682434085,
1588
- "grad_norm": 17.025096893310547,
1589
- "learning_rate": 2.1312871430642123e-05,
1590
- "loss": 2.5119,
1591
- "step": 18900
1592
- },
1593
- {
1594
- "epoch": 1.8588269823411436,
1595
- "grad_norm": 16.852436065673828,
1596
- "learning_rate": 2.1131685751558197e-05,
1597
- "loss": 2.5369,
1598
- "step": 19000
1599
- },
1600
- {
1601
- "epoch": 1.8588269823411436,
1602
- "eval_runtime": 181.7443,
1603
- "eval_samples_per_second": 112.477,
1604
- "eval_steps_per_second": 14.064,
1605
- "step": 19000
1606
- },
1607
- {
1608
- "epoch": 1.8686102822482025,
1609
- "grad_norm": 15.160259246826172,
1610
- "learning_rate": 2.0950500072474272e-05,
1611
- "loss": 2.6297,
1612
- "step": 19100
1613
- },
1614
- {
1615
- "epoch": 1.8783935821552609,
1616
- "grad_norm": 15.909671783447266,
1617
- "learning_rate": 2.0769314393390347e-05,
1618
- "loss": 2.4696,
1619
- "step": 19200
1620
- },
1621
- {
1622
- "epoch": 1.8881768820623197,
1623
- "grad_norm": 14.201844215393066,
1624
- "learning_rate": 2.0588128714306422e-05,
1625
- "loss": 2.5653,
1626
- "step": 19300
1627
- },
1628
- {
1629
- "epoch": 1.8979601819693783,
1630
- "grad_norm": 16.351415634155273,
1631
- "learning_rate": 2.0406943035222497e-05,
1632
- "loss": 2.4962,
1633
- "step": 19400
1634
- },
1635
- {
1636
- "epoch": 1.907743481876437,
1637
- "grad_norm": 16.943771362304688,
1638
- "learning_rate": 2.022575735613857e-05,
1639
- "loss": 2.5091,
1640
- "step": 19500
1641
- },
1642
- {
1643
- "epoch": 1.907743481876437,
1644
- "eval_runtime": 181.6486,
1645
- "eval_samples_per_second": 112.536,
1646
- "eval_steps_per_second": 14.071,
1647
- "step": 19500
1648
- },
1649
- {
1650
- "epoch": 1.9175267817834956,
1651
- "grad_norm": 15.006349563598633,
1652
- "learning_rate": 2.0044571677054646e-05,
1653
- "loss": 2.5214,
1654
- "step": 19600
1655
- },
1656
- {
1657
- "epoch": 1.9273100816905542,
1658
- "grad_norm": 17.305580139160156,
1659
- "learning_rate": 1.986338599797072e-05,
1660
- "loss": 2.4989,
1661
- "step": 19700
1662
- },
1663
- {
1664
- "epoch": 1.937093381597613,
1665
- "grad_norm": 17.28044891357422,
1666
- "learning_rate": 1.9682200318886796e-05,
1667
- "loss": 2.4008,
1668
- "step": 19800
1669
- },
1670
- {
1671
- "epoch": 1.9468766815046714,
1672
- "grad_norm": 18.25079917907715,
1673
- "learning_rate": 1.950101463980287e-05,
1674
- "loss": 2.6015,
1675
- "step": 19900
1676
- },
1677
- {
1678
- "epoch": 1.9566599814117303,
1679
- "grad_norm": 20.741668701171875,
1680
- "learning_rate": 1.9319828960718946e-05,
1681
- "loss": 2.4081,
1682
- "step": 20000
1683
- },
1684
- {
1685
- "epoch": 1.9566599814117303,
1686
- "eval_runtime": 181.7745,
1687
- "eval_samples_per_second": 112.458,
1688
- "eval_steps_per_second": 14.061,
1689
- "step": 20000
1690
- },
1691
- {
1692
- "epoch": 1.9664432813187889,
1693
- "grad_norm": 16.1226863861084,
1694
- "learning_rate": 1.913864328163502e-05,
1695
- "loss": 2.5418,
1696
- "step": 20100
1697
- },
1698
- {
1699
- "epoch": 1.9762265812258475,
1700
- "grad_norm": 13.914982795715332,
1701
- "learning_rate": 1.8957457602551095e-05,
1702
- "loss": 2.5248,
1703
- "step": 20200
1704
- },
1705
- {
1706
- "epoch": 1.986009881132906,
1707
- "grad_norm": 15.072690963745117,
1708
- "learning_rate": 1.877627192346717e-05,
1709
- "loss": 2.5488,
1710
- "step": 20300
1711
- },
1712
- {
1713
- "epoch": 1.9957931810399647,
1714
- "grad_norm": 15.510763168334961,
1715
- "learning_rate": 1.8595086244383245e-05,
1716
- "loss": 2.4605,
1717
- "step": 20400
1718
- },
1719
- {
1720
- "epoch": 2.0055764809470236,
1721
- "grad_norm": 18.463842391967773,
1722
- "learning_rate": 1.841390056529932e-05,
1723
- "loss": 2.522,
1724
- "step": 20500
1725
- },
1726
- {
1727
- "epoch": 2.0055764809470236,
1728
- "eval_runtime": 182.07,
1729
- "eval_samples_per_second": 112.276,
1730
- "eval_steps_per_second": 14.039,
1731
- "step": 20500
1732
- },
1733
- {
1734
- "epoch": 2.015359780854082,
1735
- "grad_norm": 16.670269012451172,
1736
- "learning_rate": 1.8232714886215394e-05,
1737
- "loss": 2.5585,
1738
- "step": 20600
1739
- },
1740
- {
1741
- "epoch": 2.025143080761141,
1742
- "grad_norm": 20.60368537902832,
1743
- "learning_rate": 1.805152920713147e-05,
1744
- "loss": 2.5381,
1745
- "step": 20700
1746
- },
1747
- {
1748
- "epoch": 2.034926380668199,
1749
- "grad_norm": 15.686981201171875,
1750
- "learning_rate": 1.7870343528047544e-05,
1751
- "loss": 2.5721,
1752
- "step": 20800
1753
- },
1754
- {
1755
- "epoch": 2.044709680575258,
1756
- "grad_norm": 14.691718101501465,
1757
- "learning_rate": 1.768915784896362e-05,
1758
- "loss": 2.5187,
1759
- "step": 20900
1760
- },
1761
- {
1762
- "epoch": 2.054492980482317,
1763
- "grad_norm": 16.31734848022461,
1764
- "learning_rate": 1.7507972169879694e-05,
1765
- "loss": 2.5202,
1766
- "step": 21000
1767
- },
1768
- {
1769
- "epoch": 2.054492980482317,
1770
- "eval_runtime": 181.9896,
1771
- "eval_samples_per_second": 112.325,
1772
- "eval_steps_per_second": 14.045,
1773
- "step": 21000
1774
- },
1775
- {
1776
- "epoch": 2.0642762803893753,
1777
- "grad_norm": 12.698554992675781,
1778
- "learning_rate": 1.732678649079577e-05,
1779
- "loss": 2.4228,
1780
- "step": 21100
1781
- },
1782
- {
1783
- "epoch": 2.074059580296434,
1784
- "grad_norm": 16.34201431274414,
1785
- "learning_rate": 1.7145600811711843e-05,
1786
- "loss": 2.3963,
1787
- "step": 21200
1788
- },
1789
- {
1790
- "epoch": 2.0838428802034925,
1791
- "grad_norm": 16.52840232849121,
1792
- "learning_rate": 1.6964415132627918e-05,
1793
- "loss": 2.4759,
1794
- "step": 21300
1795
- },
1796
- {
1797
- "epoch": 2.0936261801105513,
1798
- "grad_norm": 14.856452941894531,
1799
- "learning_rate": 1.6783229453543993e-05,
1800
- "loss": 2.4675,
1801
- "step": 21400
1802
- },
1803
- {
1804
- "epoch": 2.1034094800176097,
1805
- "grad_norm": 19.68895721435547,
1806
- "learning_rate": 1.6602043774460068e-05,
1807
- "loss": 2.5324,
1808
- "step": 21500
1809
- },
1810
- {
1811
- "epoch": 2.1034094800176097,
1812
- "eval_runtime": 182.1877,
1813
- "eval_samples_per_second": 112.203,
1814
- "eval_steps_per_second": 14.029,
1815
- "step": 21500
1816
- },
1817
- {
1818
- "epoch": 2.1131927799246686,
1819
- "grad_norm": 23.248056411743164,
1820
- "learning_rate": 1.6420858095376143e-05,
1821
- "loss": 2.5231,
1822
- "step": 21600
1823
- },
1824
- {
1825
- "epoch": 2.1229760798317274,
1826
- "grad_norm": 25.471004486083984,
1827
- "learning_rate": 1.6239672416292217e-05,
1828
- "loss": 2.5871,
1829
- "step": 21700
1830
- },
1831
- {
1832
- "epoch": 2.132759379738786,
1833
- "grad_norm": 17.794851303100586,
1834
- "learning_rate": 1.6058486737208292e-05,
1835
- "loss": 2.5008,
1836
- "step": 21800
1837
- },
1838
- {
1839
- "epoch": 2.1425426796458447,
1840
- "grad_norm": 15.450346946716309,
1841
- "learning_rate": 1.5877301058124367e-05,
1842
- "loss": 2.4194,
1843
- "step": 21900
1844
- },
1845
- {
1846
- "epoch": 2.152325979552903,
1847
- "grad_norm": 13.243645668029785,
1848
- "learning_rate": 1.5696115379040442e-05,
1849
- "loss": 2.5018,
1850
- "step": 22000
1851
- },
1852
- {
1853
- "epoch": 2.152325979552903,
1854
- "eval_runtime": 181.9841,
1855
- "eval_samples_per_second": 112.328,
1856
- "eval_steps_per_second": 14.045,
1857
- "step": 22000
1858
- },
1859
- {
1860
- "epoch": 2.162109279459962,
1861
- "grad_norm": 16.996198654174805,
1862
- "learning_rate": 1.5514929699956517e-05,
1863
- "loss": 2.4492,
1864
- "step": 22100
1865
- },
1866
- {
1867
- "epoch": 2.1718925793670203,
1868
- "grad_norm": 20.05558967590332,
1869
- "learning_rate": 1.5333744020872588e-05,
1870
- "loss": 2.489,
1871
- "step": 22200
1872
- },
1873
- {
1874
- "epoch": 2.181675879274079,
1875
- "grad_norm": 15.66326904296875,
1876
- "learning_rate": 1.5152558341788666e-05,
1877
- "loss": 2.5089,
1878
- "step": 22300
1879
- },
1880
- {
1881
- "epoch": 2.191459179181138,
1882
- "grad_norm": 17.83564567565918,
1883
- "learning_rate": 1.4971372662704741e-05,
1884
- "loss": 2.4945,
1885
- "step": 22400
1886
- },
1887
- {
1888
- "epoch": 2.2012424790881964,
1889
- "grad_norm": 21.466899871826172,
1890
- "learning_rate": 1.4790186983620816e-05,
1891
- "loss": 2.5467,
1892
- "step": 22500
1893
- },
1894
- {
1895
- "epoch": 2.2012424790881964,
1896
- "eval_runtime": 182.8328,
1897
- "eval_samples_per_second": 111.807,
1898
- "eval_steps_per_second": 13.98,
1899
- "step": 22500
1900
- },
1901
- {
1902
- "epoch": 2.211025778995255,
1903
- "grad_norm": 17.91064453125,
1904
- "learning_rate": 1.4609001304536891e-05,
1905
- "loss": 2.5144,
1906
- "step": 22600
1907
- },
1908
- {
1909
- "epoch": 2.2208090789023136,
1910
- "grad_norm": 17.678396224975586,
1911
- "learning_rate": 1.4427815625452964e-05,
1912
- "loss": 2.5018,
1913
- "step": 22700
1914
- },
1915
- {
1916
- "epoch": 2.2305923788093724,
1917
- "grad_norm": 17.510461807250977,
1918
- "learning_rate": 1.4246629946369039e-05,
1919
- "loss": 2.4228,
1920
- "step": 22800
1921
- },
1922
- {
1923
- "epoch": 2.240375678716431,
1924
- "grad_norm": 24.923967361450195,
1925
- "learning_rate": 1.4065444267285114e-05,
1926
- "loss": 2.5249,
1927
- "step": 22900
1928
- },
1929
- {
1930
- "epoch": 2.2501589786234897,
1931
- "grad_norm": 17.82384490966797,
1932
- "learning_rate": 1.388425858820119e-05,
1933
- "loss": 2.4282,
1934
- "step": 23000
1935
- },
1936
- {
1937
- "epoch": 2.2501589786234897,
1938
- "eval_runtime": 182.0459,
1939
- "eval_samples_per_second": 112.29,
1940
- "eval_steps_per_second": 14.04,
1941
- "step": 23000
1942
- },
1943
- {
1944
- "epoch": 2.2599422785305485,
1945
- "grad_norm": 16.13028335571289,
1946
- "learning_rate": 1.3703072909117265e-05,
1947
- "loss": 2.4472,
1948
- "step": 23100
1949
- },
1950
- {
1951
- "epoch": 2.269725578437607,
1952
- "grad_norm": 15.137242317199707,
1953
- "learning_rate": 1.352188723003334e-05,
1954
- "loss": 2.5985,
1955
- "step": 23200
1956
- },
1957
- {
1958
- "epoch": 2.2795088783446658,
1959
- "grad_norm": 16.187530517578125,
1960
- "learning_rate": 1.3340701550949415e-05,
1961
- "loss": 2.4862,
1962
- "step": 23300
1963
- },
1964
- {
1965
- "epoch": 2.289292178251724,
1966
- "grad_norm": 18.84433937072754,
1967
- "learning_rate": 1.3159515871865488e-05,
1968
- "loss": 2.516,
1969
- "step": 23400
1970
- },
1971
- {
1972
- "epoch": 2.299075478158783,
1973
- "grad_norm": 20.209121704101562,
1974
- "learning_rate": 1.2978330192781563e-05,
1975
- "loss": 2.5031,
1976
- "step": 23500
1977
- },
1978
- {
1979
- "epoch": 2.299075478158783,
1980
- "eval_runtime": 181.9806,
1981
- "eval_samples_per_second": 112.331,
1982
- "eval_steps_per_second": 14.045,
1983
- "step": 23500
1984
- },
1985
- {
1986
- "epoch": 2.308858778065842,
1987
- "grad_norm": 67.4502182006836,
1988
- "learning_rate": 1.2797144513697637e-05,
1989
- "loss": 2.4491,
1990
- "step": 23600
1991
- },
1992
- {
1993
- "epoch": 2.3186420779729002,
1994
- "grad_norm": 14.940401077270508,
1995
- "learning_rate": 1.2615958834613712e-05,
1996
- "loss": 2.5669,
1997
- "step": 23700
1998
- },
1999
- {
2000
- "epoch": 2.328425377879959,
2001
- "grad_norm": 16.591793060302734,
2002
- "learning_rate": 1.2434773155529787e-05,
2003
- "loss": 2.4565,
2004
- "step": 23800
2005
- },
2006
- {
2007
- "epoch": 2.3382086777870175,
2008
- "grad_norm": 16.798791885375977,
2009
- "learning_rate": 1.2253587476445862e-05,
2010
- "loss": 2.4046,
2011
- "step": 23900
2012
- },
2013
- {
2014
- "epoch": 2.3479919776940763,
2015
- "grad_norm": 17.712255477905273,
2016
- "learning_rate": 1.2072401797361937e-05,
2017
- "loss": 2.4453,
2018
- "step": 24000
2019
- },
2020
- {
2021
- "epoch": 2.3479919776940763,
2022
- "eval_runtime": 182.0401,
2023
- "eval_samples_per_second": 112.294,
2024
- "eval_steps_per_second": 14.041,
2025
- "step": 24000
2026
- },
2027
- {
2028
- "epoch": 2.3577752776011347,
2029
- "grad_norm": 18.64284324645996,
2030
- "learning_rate": 1.1891216118278011e-05,
2031
- "loss": 2.3973,
2032
- "step": 24100
2033
- },
2034
- {
2035
- "epoch": 2.3675585775081935,
2036
- "grad_norm": 18.185895919799805,
2037
- "learning_rate": 1.1710030439194086e-05,
2038
- "loss": 2.5045,
2039
- "step": 24200
2040
- },
2041
- {
2042
- "epoch": 2.377341877415252,
2043
- "grad_norm": 23.201522827148438,
2044
- "learning_rate": 1.1528844760110163e-05,
2045
- "loss": 2.5402,
2046
- "step": 24300
2047
- },
2048
- {
2049
- "epoch": 2.3871251773223108,
2050
- "grad_norm": 21.606412887573242,
2051
- "learning_rate": 1.1347659081026236e-05,
2052
- "loss": 2.4285,
2053
- "step": 24400
2054
- },
2055
- {
2056
- "epoch": 2.3969084772293696,
2057
- "grad_norm": 16.318761825561523,
2058
- "learning_rate": 1.116647340194231e-05,
2059
- "loss": 2.5509,
2060
- "step": 24500
2061
- },
2062
- {
2063
- "epoch": 2.3969084772293696,
2064
- "eval_runtime": 182.0431,
2065
- "eval_samples_per_second": 112.292,
2066
- "eval_steps_per_second": 14.041,
2067
- "step": 24500
2068
- },
2069
- {
2070
- "epoch": 2.406691777136428,
2071
- "grad_norm": 17.779014587402344,
2072
- "learning_rate": 1.0985287722858386e-05,
2073
- "loss": 2.4245,
2074
- "step": 24600
2075
- },
2076
- {
2077
- "epoch": 2.416475077043487,
2078
- "grad_norm": 18.44321060180664,
2079
- "learning_rate": 1.080410204377446e-05,
2080
- "loss": 2.5223,
2081
- "step": 24700
2082
- },
2083
- {
2084
- "epoch": 2.4262583769505452,
2085
- "grad_norm": 24.017047882080078,
2086
- "learning_rate": 1.0622916364690535e-05,
2087
- "loss": 2.4846,
2088
- "step": 24800
2089
- },
2090
- {
2091
- "epoch": 2.436041676857604,
2092
- "grad_norm": 14.89560604095459,
2093
- "learning_rate": 1.044173068560661e-05,
2094
- "loss": 2.5922,
2095
- "step": 24900
2096
- },
2097
- {
2098
- "epoch": 2.445824976764663,
2099
- "grad_norm": 15.532561302185059,
2100
- "learning_rate": 1.0260545006522685e-05,
2101
- "loss": 2.3976,
2102
- "step": 25000
2103
- },
2104
- {
2105
- "epoch": 2.445824976764663,
2106
- "eval_runtime": 182.1033,
2107
- "eval_samples_per_second": 112.255,
2108
- "eval_steps_per_second": 14.036,
2109
- "step": 25000
2110
- },
2111
- {
2112
- "epoch": 2.4556082766717213,
2113
- "grad_norm": 18.041282653808594,
2114
- "learning_rate": 1.007935932743876e-05,
2115
- "loss": 2.4731,
2116
- "step": 25100
2117
- },
2118
- {
2119
- "epoch": 2.46539157657878,
2120
- "grad_norm": 13.40858268737793,
2121
- "learning_rate": 9.898173648354834e-06,
2122
- "loss": 2.4838,
2123
- "step": 25200
2124
- },
2125
- {
2126
- "epoch": 2.4751748764858386,
2127
- "grad_norm": 17.450841903686523,
2128
- "learning_rate": 9.71698796927091e-06,
2129
- "loss": 2.3999,
2130
- "step": 25300
2131
- },
2132
- {
2133
- "epoch": 2.4849581763928974,
2134
- "grad_norm": 17.556467056274414,
2135
- "learning_rate": 9.535802290186984e-06,
2136
- "loss": 2.3867,
2137
- "step": 25400
2138
- },
2139
- {
2140
- "epoch": 2.494741476299956,
2141
- "grad_norm": 18.578310012817383,
2142
- "learning_rate": 9.354616611103059e-06,
2143
- "loss": 2.4546,
2144
- "step": 25500
2145
- },
2146
- {
2147
- "epoch": 2.494741476299956,
2148
- "eval_runtime": 182.0338,
2149
- "eval_samples_per_second": 112.298,
2150
- "eval_steps_per_second": 14.041,
2151
- "step": 25500
2152
- },
2153
- {
2154
- "epoch": 2.5045247762070146,
2155
- "grad_norm": 14.936469078063965,
2156
- "learning_rate": 9.173430932019134e-06,
2157
- "loss": 2.5562,
2158
- "step": 25600
2159
- },
2160
- {
2161
- "epoch": 2.514308076114073,
2162
- "grad_norm": 17.527040481567383,
2163
- "learning_rate": 8.992245252935209e-06,
2164
- "loss": 2.4008,
2165
- "step": 25700
2166
- },
2167
- {
2168
- "epoch": 2.524091376021132,
2169
- "grad_norm": 12.91336727142334,
2170
- "learning_rate": 8.811059573851283e-06,
2171
- "loss": 2.4655,
2172
- "step": 25800
2173
- },
2174
- {
2175
- "epoch": 2.5338746759281907,
2176
- "grad_norm": 15.168461799621582,
2177
- "learning_rate": 8.629873894767358e-06,
2178
- "loss": 2.4468,
2179
- "step": 25900
2180
- },
2181
- {
2182
- "epoch": 2.543657975835249,
2183
- "grad_norm": 17.5390682220459,
2184
- "learning_rate": 8.448688215683433e-06,
2185
- "loss": 2.4836,
2186
- "step": 26000
2187
- },
2188
- {
2189
- "epoch": 2.543657975835249,
2190
- "eval_runtime": 182.1148,
2191
- "eval_samples_per_second": 112.248,
2192
- "eval_steps_per_second": 14.035,
2193
- "step": 26000
2194
- },
2195
- {
2196
- "epoch": 2.553441275742308,
2197
- "grad_norm": 15.126510620117188,
2198
- "learning_rate": 8.267502536599508e-06,
2199
- "loss": 2.387,
2200
- "step": 26100
2201
- },
2202
- {
2203
- "epoch": 2.5632245756493663,
2204
- "grad_norm": 15.374293327331543,
2205
- "learning_rate": 8.086316857515583e-06,
2206
- "loss": 2.3652,
2207
- "step": 26200
2208
- },
2209
- {
2210
- "epoch": 2.573007875556425,
2211
- "grad_norm": 15.498108863830566,
2212
- "learning_rate": 7.905131178431657e-06,
2213
- "loss": 2.4749,
2214
- "step": 26300
2215
- },
2216
- {
2217
- "epoch": 2.582791175463484,
2218
- "grad_norm": 16.221315383911133,
2219
- "learning_rate": 7.723945499347732e-06,
2220
- "loss": 2.4567,
2221
- "step": 26400
2222
- },
2223
- {
2224
- "epoch": 2.5925744753705424,
2225
- "grad_norm": 18.839122772216797,
2226
- "learning_rate": 7.542759820263806e-06,
2227
- "loss": 2.3554,
2228
- "step": 26500
2229
- },
2230
- {
2231
- "epoch": 2.5925744753705424,
2232
- "eval_runtime": 181.9597,
2233
- "eval_samples_per_second": 112.344,
2234
- "eval_steps_per_second": 14.047,
2235
- "step": 26500
2236
- },
2237
- {
2238
- "epoch": 2.6023577752776013,
2239
- "grad_norm": 22.626708984375,
2240
- "learning_rate": 7.361574141179882e-06,
2241
- "loss": 2.502,
2242
- "step": 26600
2243
- },
2244
- {
2245
- "epoch": 2.6121410751846597,
2246
- "grad_norm": 16.519880294799805,
2247
- "learning_rate": 7.180388462095957e-06,
2248
- "loss": 2.5034,
2249
- "step": 26700
2250
- },
2251
- {
2252
- "epoch": 2.6219243750917185,
2253
- "grad_norm": 27.421489715576172,
2254
- "learning_rate": 6.999202783012031e-06,
2255
- "loss": 2.5276,
2256
- "step": 26800
2257
- },
2258
- {
2259
- "epoch": 2.6317076749987773,
2260
- "grad_norm": 15.274630546569824,
2261
- "learning_rate": 6.8180171039281055e-06,
2262
- "loss": 2.4121,
2263
- "step": 26900
2264
- },
2265
- {
2266
- "epoch": 2.6414909749058357,
2267
- "grad_norm": 15.751582145690918,
2268
- "learning_rate": 6.636831424844181e-06,
2269
- "loss": 2.5799,
2270
- "step": 27000
2271
- },
2272
- {
2273
- "epoch": 2.6414909749058357,
2274
- "eval_runtime": 182.0873,
2275
- "eval_samples_per_second": 112.265,
2276
- "eval_steps_per_second": 14.037,
2277
- "step": 27000
2278
- },
2279
- {
2280
- "epoch": 2.651274274812894,
2281
- "grad_norm": 16.674850463867188,
2282
- "learning_rate": 6.455645745760255e-06,
2283
- "loss": 2.3872,
2284
- "step": 27100
2285
- },
2286
- {
2287
- "epoch": 2.661057574719953,
2288
- "grad_norm": 12.62803840637207,
2289
- "learning_rate": 6.27446006667633e-06,
2290
- "loss": 2.4,
2291
- "step": 27200
2292
- },
2293
- {
2294
- "epoch": 2.670840874627012,
2295
- "grad_norm": 18.055158615112305,
2296
- "learning_rate": 6.093274387592405e-06,
2297
- "loss": 2.4681,
2298
- "step": 27300
2299
- },
2300
- {
2301
- "epoch": 2.68062417453407,
2302
- "grad_norm": 17.21278190612793,
2303
- "learning_rate": 5.91208870850848e-06,
2304
- "loss": 2.5441,
2305
- "step": 27400
2306
- },
2307
- {
2308
- "epoch": 2.690407474441129,
2309
- "grad_norm": 20.945236206054688,
2310
- "learning_rate": 5.7309030294245544e-06,
2311
- "loss": 2.4388,
2312
- "step": 27500
2313
- },
2314
- {
2315
- "epoch": 2.690407474441129,
2316
- "eval_runtime": 182.1279,
2317
- "eval_samples_per_second": 112.24,
2318
- "eval_steps_per_second": 14.034,
2319
- "step": 27500
2320
- },
2321
- {
2322
- "epoch": 2.7001907743481874,
2323
- "grad_norm": 23.483661651611328,
2324
- "learning_rate": 5.549717350340629e-06,
2325
- "loss": 2.4589,
2326
- "step": 27600
2327
- },
2328
- {
2329
- "epoch": 2.7099740742552463,
2330
- "grad_norm": 17.954036712646484,
2331
- "learning_rate": 5.368531671256704e-06,
2332
- "loss": 2.4477,
2333
- "step": 27700
2334
- },
2335
- {
2336
- "epoch": 2.719757374162305,
2337
- "grad_norm": 16.187314987182617,
2338
- "learning_rate": 5.187345992172779e-06,
2339
- "loss": 2.4967,
2340
- "step": 27800
2341
- },
2342
- {
2343
- "epoch": 2.7295406740693635,
2344
- "grad_norm": 14.324910163879395,
2345
- "learning_rate": 5.006160313088854e-06,
2346
- "loss": 2.3921,
2347
- "step": 27900
2348
- },
2349
- {
2350
- "epoch": 2.7393239739764224,
2351
- "grad_norm": 20.81557846069336,
2352
- "learning_rate": 4.8249746340049285e-06,
2353
- "loss": 2.5201,
2354
- "step": 28000
2355
- },
2356
- {
2357
- "epoch": 2.7393239739764224,
2358
- "eval_runtime": 182.146,
2359
- "eval_samples_per_second": 112.229,
2360
- "eval_steps_per_second": 14.033,
2361
- "step": 28000
2362
- },
2363
- {
2364
- "epoch": 2.7491072738834808,
2365
- "grad_norm": 18.682844161987305,
2366
- "learning_rate": 4.643788954921003e-06,
2367
- "loss": 2.4325,
2368
- "step": 28100
2369
- },
2370
- {
2371
- "epoch": 2.7588905737905396,
2372
- "grad_norm": 16.227272033691406,
2373
- "learning_rate": 4.462603275837078e-06,
2374
- "loss": 2.3864,
2375
- "step": 28200
2376
- },
2377
- {
2378
- "epoch": 2.7686738736975984,
2379
- "grad_norm": 16.20302963256836,
2380
- "learning_rate": 4.281417596753152e-06,
2381
- "loss": 2.5296,
2382
- "step": 28300
2383
- },
2384
- {
2385
- "epoch": 2.778457173604657,
2386
- "grad_norm": 18.634096145629883,
2387
- "learning_rate": 4.100231917669228e-06,
2388
- "loss": 2.4514,
2389
- "step": 28400
2390
- },
2391
- {
2392
- "epoch": 2.7882404735117157,
2393
- "grad_norm": 13.040008544921875,
2394
- "learning_rate": 3.919046238585303e-06,
2395
- "loss": 2.3661,
2396
- "step": 28500
2397
- },
2398
- {
2399
- "epoch": 2.7882404735117157,
2400
- "eval_runtime": 181.9164,
2401
- "eval_samples_per_second": 112.37,
2402
- "eval_steps_per_second": 14.05,
2403
- "step": 28500
2404
- },
2405
- {
2406
- "epoch": 2.798023773418774,
2407
- "grad_norm": 14.142943382263184,
2408
- "learning_rate": 3.737860559501377e-06,
2409
- "loss": 2.5074,
2410
- "step": 28600
2411
- },
2412
- {
2413
- "epoch": 2.807807073325833,
2414
- "grad_norm": 17.934324264526367,
2415
- "learning_rate": 3.5566748804174523e-06,
2416
- "loss": 2.4224,
2417
- "step": 28700
2418
- },
2419
- {
2420
- "epoch": 2.8175903732328913,
2421
- "grad_norm": 14.450194358825684,
2422
- "learning_rate": 3.3754892013335267e-06,
2423
- "loss": 2.4949,
2424
- "step": 28800
2425
- },
2426
- {
2427
- "epoch": 2.82737367313995,
2428
- "grad_norm": 17.746837615966797,
2429
- "learning_rate": 3.194303522249602e-06,
2430
- "loss": 2.4153,
2431
- "step": 28900
2432
- },
2433
- {
2434
- "epoch": 2.8371569730470085,
2435
- "grad_norm": 13.962541580200195,
2436
- "learning_rate": 3.0131178431656763e-06,
2437
- "loss": 2.4804,
2438
- "step": 29000
2439
- },
2440
- {
2441
- "epoch": 2.8371569730470085,
2442
- "eval_runtime": 182.0262,
2443
- "eval_samples_per_second": 112.303,
2444
- "eval_steps_per_second": 14.042,
2445
- "step": 29000
2446
- },
2447
- {
2448
- "epoch": 2.8469402729540674,
2449
- "grad_norm": 16.669286727905273,
2450
- "learning_rate": 2.831932164081751e-06,
2451
- "loss": 2.5397,
2452
- "step": 29100
2453
- },
2454
- {
2455
- "epoch": 2.856723572861126,
2456
- "grad_norm": 15.421733856201172,
2457
- "learning_rate": 2.650746484997826e-06,
2458
- "loss": 2.4175,
2459
- "step": 29200
2460
- },
2461
- {
2462
- "epoch": 2.8665068727681846,
2463
- "grad_norm": 14.135702133178711,
2464
- "learning_rate": 2.4695608059139007e-06,
2465
- "loss": 2.5069,
2466
- "step": 29300
2467
- },
2468
- {
2469
- "epoch": 2.8762901726752435,
2470
- "grad_norm": 17.41412925720215,
2471
- "learning_rate": 2.2883751268299756e-06,
2472
- "loss": 2.3997,
2473
- "step": 29400
2474
- },
2475
- {
2476
- "epoch": 2.886073472582302,
2477
- "grad_norm": 14.824533462524414,
2478
- "learning_rate": 2.1071894477460504e-06,
2479
- "loss": 2.3945,
2480
- "step": 29500
2481
- },
2482
- {
2483
- "epoch": 2.886073472582302,
2484
- "eval_runtime": 181.9299,
2485
- "eval_samples_per_second": 112.362,
2486
- "eval_steps_per_second": 14.049,
2487
- "step": 29500
2488
- },
2489
- {
2490
- "epoch": 2.8958567724893607,
2491
- "grad_norm": 27.31865119934082,
2492
- "learning_rate": 1.926003768662125e-06,
2493
- "loss": 2.45,
2494
- "step": 29600
2495
- },
2496
- {
2497
- "epoch": 2.9056400723964195,
2498
- "grad_norm": 18.966655731201172,
2499
- "learning_rate": 1.7448180895781998e-06,
2500
- "loss": 2.3916,
2501
- "step": 29700
2502
- },
2503
- {
2504
- "epoch": 2.915423372303478,
2505
- "grad_norm": 18.538440704345703,
2506
- "learning_rate": 1.5636324104942746e-06,
2507
- "loss": 2.4625,
2508
- "step": 29800
2509
- },
2510
- {
2511
- "epoch": 2.9252066722105368,
2512
- "grad_norm": 21.757272720336914,
2513
- "learning_rate": 1.3824467314103494e-06,
2514
- "loss": 2.3722,
2515
- "step": 29900
2516
- },
2517
- {
2518
- "epoch": 2.934989972117595,
2519
- "grad_norm": 16.907358169555664,
2520
- "learning_rate": 1.201261052326424e-06,
2521
- "loss": 2.464,
2522
- "step": 30000
2523
- },
2524
- {
2525
- "epoch": 2.934989972117595,
2526
- "eval_runtime": 181.9148,
2527
- "eval_samples_per_second": 112.371,
2528
- "eval_steps_per_second": 14.051,
2529
- "step": 30000
2530
- },
2531
- {
2532
- "epoch": 2.944773272024654,
2533
- "grad_norm": 13.88399600982666,
2534
- "learning_rate": 1.0200753732424989e-06,
2535
- "loss": 2.5005,
2536
- "step": 30100
2537
- },
2538
- {
2539
- "epoch": 2.954556571931713,
2540
- "grad_norm": 19.77507781982422,
2541
- "learning_rate": 8.388896941585737e-07,
2542
- "loss": 2.3829,
2543
- "step": 30200
2544
- },
2545
- {
2546
- "epoch": 2.9643398718387712,
2547
- "grad_norm": 16.535932540893555,
2548
- "learning_rate": 6.577040150746485e-07,
2549
- "loss": 2.4788,
2550
- "step": 30300
2551
- },
2552
- {
2553
- "epoch": 2.9741231717458296,
2554
- "grad_norm": 15.027000427246094,
2555
- "learning_rate": 4.765183359907233e-07,
2556
- "loss": 2.5007,
2557
- "step": 30400
2558
- },
2559
- {
2560
- "epoch": 2.9839064716528885,
2561
- "grad_norm": 14.9392671585083,
2562
- "learning_rate": 2.953326569067981e-07,
2563
- "loss": 2.4847,
2564
- "step": 30500
2565
- },
2566
- {
2567
- "epoch": 2.9839064716528885,
2568
- "eval_runtime": 181.9853,
2569
- "eval_samples_per_second": 112.328,
2570
- "eval_steps_per_second": 14.045,
2571
- "step": 30500
2572
- },
2573
- {
2574
- "epoch": 2.9936897715599473,
2575
- "grad_norm": 15.128337860107422,
2576
- "learning_rate": 1.1414697782287289e-07,
2577
- "loss": 2.4209,
2578
- "step": 30600
2579
- }
2580
- ],
2581
- "logging_steps": 100,
2582
- "max_steps": 30663,
2583
- "num_input_tokens_seen": 0,
2584
- "num_train_epochs": 3,
2585
- "save_steps": 500,
2586
- "stateful_callbacks": {
2587
- "TrainerControl": {
2588
- "args": {
2589
- "should_epoch_stop": false,
2590
- "should_evaluate": false,
2591
- "should_log": false,
2592
- "should_save": true,
2593
- "should_training_stop": true
2594
- },
2595
- "attributes": {}
2596
- }
2597
- },
2598
- "total_flos": 1.0701267610290972e+16,
2599
- "train_batch_size": 8,
2600
- "trial_name": null,
2601
- "trial_params": null
2602
- }