Theoreticallyhugo commited on
Commit
ed81781
1 Parent(s): ea5cbc7

Training in progress, epoch 13, checkpoint

Browse files
checkpoint-1053/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3589a1aeba89520f65657f7babd69a85b31ee7ff35758cf59fe138d4d1d17bfa
3
  size 592318676
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:005708d2a4323ee5bb6adf1329f8c24951816ddc42a04696728e247f976c5ef2
3
  size 592318676
checkpoint-1053/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:97e3299a4081d8405bb64d84e53cc94575aa5afe3f9cf5acad34fac9239e3053
3
  size 1014645498
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c56be6197e5aba69923a5d0966657df1f473084ca4128555fd91ee97b6dc7d23
3
  size 1014645498
checkpoint-1053/rng_state.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a2d0b6fd3e0e222c69d0c818ba4ff34a17c0844523a335fc5dcaffe1d7766cfb
3
  size 14244
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a8fed4ff2e0a1f5b060e9416aef90a5efb47a3301fd9fdfb8a444f0ad6d3aa3
3
  size 14244
checkpoint-1053/trainer_state.json CHANGED
@@ -11,521 +11,521 @@
11
  {
12
  "epoch": 1.0,
13
  "eval_B": {
14
- "f1-score": 0.8087666523420713,
15
- "precision": 0.8189730200174065,
16
- "recall": 0.7988115449915111,
17
- "support": 1178.0
18
  },
19
  "eval_I": {
20
- "f1-score": 0.9374597034171502,
21
- "precision": 0.9521418826739427,
22
- "recall": 0.9232234509762421,
23
- "support": 18899.0
24
  },
25
  "eval_O": {
26
- "f1-score": 0.8881362400419788,
27
- "precision": 0.863303347862376,
28
- "recall": 0.9144400785854617,
29
- "support": 10180.0
30
  },
31
- "eval_accuracy": 0.9154245298608586,
32
- "eval_loss": 0.2228638380765915,
33
  "eval_macro avg": {
34
- "f1-score": 0.8781208652670669,
35
- "precision": 0.8781394168512416,
36
- "recall": 0.8788250248510717,
37
- "support": 30257.0
38
- },
39
- "eval_runtime": 4.7167,
40
- "eval_samples_per_second": 17.173,
41
- "eval_steps_per_second": 2.332,
42
  "eval_weighted avg": {
43
- "f1-score": 0.9158543469268937,
44
- "precision": 0.9170673807540184,
45
- "recall": 0.9154245298608586,
46
- "support": 30257.0
47
  },
48
  "step": 81
49
  },
50
  {
51
  "epoch": 2.0,
52
  "eval_B": {
53
- "f1-score": 0.8401515151515152,
54
- "precision": 0.7585499316005472,
55
- "recall": 0.9414261460101867,
56
- "support": 1178.0
57
  },
58
  "eval_I": {
59
- "f1-score": 0.9568235603470944,
60
- "precision": 0.9510219016256338,
61
- "recall": 0.9626964389650247,
62
- "support": 18899.0
63
  },
64
  "eval_O": {
65
- "f1-score": 0.9185648054827655,
66
- "precision": 0.9430877483443708,
67
- "recall": 0.8952848722986247,
68
- "support": 10180.0
69
  },
70
- "eval_accuracy": 0.9391876260038999,
71
- "eval_loss": 0.16692574322223663,
72
  "eval_macro avg": {
73
- "f1-score": 0.905179960327125,
74
- "precision": 0.8842198605235172,
75
- "recall": 0.9331358190912787,
76
- "support": 30257.0
77
- },
78
- "eval_runtime": 4.7116,
79
- "eval_samples_per_second": 17.192,
80
- "eval_steps_per_second": 2.335,
81
  "eval_weighted avg": {
82
- "f1-score": 0.9394089523635119,
83
- "precision": 0.9408589092241132,
84
- "recall": 0.9391876260038999,
85
- "support": 30257.0
86
  },
87
  "step": 162
88
  },
89
  {
90
  "epoch": 3.0,
91
  "eval_B": {
92
- "f1-score": 0.8612244897959184,
93
- "precision": 0.8294025157232704,
94
- "recall": 0.8955857385398981,
95
- "support": 1178.0
96
  },
97
  "eval_I": {
98
- "f1-score": 0.951106371634231,
99
- "precision": 0.9584654236741712,
100
- "recall": 0.9438594634636753,
101
- "support": 18899.0
102
  },
103
  "eval_O": {
104
- "f1-score": 0.9106743212999903,
105
- "precision": 0.9021592442645074,
106
- "recall": 0.9193516699410609,
107
- "support": 10180.0
108
  },
109
- "eval_accuracy": 0.9337343424662061,
110
- "eval_loss": 0.1742253601551056,
111
  "eval_macro avg": {
112
- "f1-score": 0.9076683942433799,
113
- "precision": 0.8966757278873163,
114
- "recall": 0.9195989573148781,
115
- "support": 30257.0
116
- },
117
- "eval_runtime": 4.7094,
118
- "eval_samples_per_second": 17.2,
119
- "eval_steps_per_second": 2.336,
120
  "eval_weighted avg": {
121
- "f1-score": 0.9340035812317422,
122
- "precision": 0.9344963252190521,
123
- "recall": 0.9337343424662061,
124
- "support": 30257.0
125
  },
126
  "step": 243
127
  },
128
  {
129
  "epoch": 4.0,
130
  "eval_B": {
131
- "f1-score": 0.8717948717948719,
132
- "precision": 0.8254931714719271,
133
- "recall": 0.9235993208828522,
134
- "support": 1178.0
135
  },
136
  "eval_I": {
137
- "f1-score": 0.9561271586371415,
138
- "precision": 0.9374586871408959,
139
- "recall": 0.9755542621302714,
140
- "support": 18899.0
141
  },
142
  "eval_O": {
143
- "f1-score": 0.9117828500925355,
144
- "precision": 0.956427955133736,
145
- "recall": 0.8711198428290766,
146
- "support": 10180.0
147
  },
148
- "eval_accuracy": 0.9383944211256899,
149
- "eval_loss": 0.18792153894901276,
150
  "eval_macro avg": {
151
- "f1-score": 0.9132349601748496,
152
- "precision": 0.9064599379155197,
153
- "recall": 0.9234244752807333,
154
- "support": 30257.0
155
- },
156
- "eval_runtime": 4.7342,
157
- "eval_samples_per_second": 17.11,
158
- "eval_steps_per_second": 2.324,
159
  "eval_weighted avg": {
160
- "f1-score": 0.9379241479327001,
161
- "precision": 0.9394817486046585,
162
- "recall": 0.9383944211256899,
163
- "support": 30257.0
164
  },
165
  "step": 324
166
  },
167
  {
168
  "epoch": 5.0,
169
  "eval_B": {
170
- "f1-score": 0.8742949234488315,
171
- "precision": 0.8320552147239264,
172
- "recall": 0.9210526315789473,
173
- "support": 1178.0
174
  },
175
  "eval_I": {
176
- "f1-score": 0.9583957528450158,
177
- "precision": 0.9521148825065274,
178
- "recall": 0.9647600402137679,
179
- "support": 18899.0
180
  },
181
  "eval_O": {
182
- "f1-score": 0.9198818996146724,
183
- "precision": 0.9375701315923697,
184
- "recall": 0.9028487229862475,
185
- "support": 10180.0
186
  },
187
- "eval_accuracy": 0.9422282447037049,
188
- "eval_loss": 0.19990424811840057,
189
  "eval_macro avg": {
190
- "f1-score": 0.9175241919695066,
191
- "precision": 0.9072467429409411,
192
- "recall": 0.9295537982596542,
193
- "support": 30257.0
194
- },
195
- "eval_runtime": 4.7101,
196
- "eval_samples_per_second": 17.197,
197
- "eval_steps_per_second": 2.335,
198
  "eval_weighted avg": {
199
- "f1-score": 0.9421634164298524,
200
- "precision": 0.9425469857238316,
201
- "recall": 0.9422282447037049,
202
- "support": 30257.0
203
  },
204
  "step": 405
205
  },
206
  {
207
  "epoch": 6.0,
208
  "eval_B": {
209
- "f1-score": 0.8808247422680412,
210
- "precision": 0.8564554931836408,
211
- "recall": 0.9066213921901528,
212
- "support": 1178.0
213
  },
214
  "eval_I": {
215
- "f1-score": 0.957406827435014,
216
- "precision": 0.9446361435855178,
217
- "recall": 0.9705275411397428,
218
- "support": 18899.0
219
  },
220
  "eval_O": {
221
- "f1-score": 0.9163000050574014,
222
- "precision": 0.9443344105076619,
223
- "recall": 0.8898821218074656,
224
- "support": 10180.0
225
  },
226
- "eval_accuracy": 0.9409062365733549,
227
- "eval_loss": 0.20615626871585846,
228
  "eval_macro avg": {
229
- "f1-score": 0.9181771915868189,
230
- "precision": 0.9151420157589403,
231
- "recall": 0.922343685045787,
232
- "support": 30257.0
233
- },
234
- "eval_runtime": 4.6949,
235
- "eval_samples_per_second": 17.253,
236
- "eval_steps_per_second": 2.343,
237
  "eval_weighted avg": {
238
- "f1-score": 0.9405948120953971,
239
- "precision": 0.941101475610967,
240
- "recall": 0.9409062365733549,
241
- "support": 30257.0
242
  },
243
  "step": 486
244
  },
245
  {
246
  "epoch": 6.17,
247
- "grad_norm": 2.4104158878326416,
248
  "learning_rate": 1.7530864197530865e-05,
249
- "loss": 0.1662,
250
  "step": 500
251
  },
252
  {
253
  "epoch": 7.0,
254
  "eval_B": {
255
- "f1-score": 0.8836820083682009,
256
- "precision": 0.8712871287128713,
257
- "recall": 0.8964346349745331,
258
- "support": 1178.0
259
  },
260
  "eval_I": {
261
- "f1-score": 0.9614174269469787,
262
- "precision": 0.9609854091774159,
263
- "recall": 0.9618498333245146,
264
- "support": 18899.0
265
  },
266
  "eval_O": {
267
- "f1-score": 0.9271751440248166,
268
- "precision": 0.9295093296475466,
269
- "recall": 0.924852652259332,
270
- "support": 10180.0
271
  },
272
- "eval_accuracy": 0.9468552731599299,
273
- "eval_loss": 0.22925782203674316,
274
  "eval_macro avg": {
275
- "f1-score": 0.9240915264466655,
276
- "precision": 0.9205939558459445,
277
- "recall": 0.9277123735194599,
278
- "support": 30257.0
279
- },
280
- "eval_runtime": 4.7288,
281
- "eval_samples_per_second": 17.129,
282
- "eval_steps_per_second": 2.326,
283
  "eval_weighted avg": {
284
- "f1-score": 0.9468700903559945,
285
- "precision": 0.9469030129054358,
286
- "recall": 0.9468552731599299,
287
- "support": 30257.0
288
  },
289
  "step": 567
290
  },
291
  {
292
  "epoch": 8.0,
293
  "eval_B": {
294
- "f1-score": 0.8803630363036303,
295
- "precision": 0.8563402889245586,
296
- "recall": 0.9057724957555179,
297
- "support": 1178.0
298
  },
299
  "eval_I": {
300
- "f1-score": 0.9580651967507121,
301
- "precision": 0.9550928116947994,
302
- "recall": 0.9610561405365363,
303
- "support": 18899.0
304
  },
305
  "eval_O": {
306
- "f1-score": 0.9208882720333101,
307
- "precision": 0.9294576746047628,
308
- "recall": 0.912475442043222,
309
- "support": 10180.0
310
  },
311
- "eval_accuracy": 0.9425587467362925,
312
- "eval_loss": 0.22973769903182983,
313
  "eval_macro avg": {
314
- "f1-score": 0.9197721683625509,
315
- "precision": 0.9136302584080402,
316
- "recall": 0.9264346927784253,
317
- "support": 30257.0
318
- },
319
- "eval_runtime": 4.7191,
320
- "eval_samples_per_second": 17.164,
321
- "eval_steps_per_second": 2.331,
322
  "eval_weighted avg": {
323
- "f1-score": 0.9425317916335554,
324
- "precision": 0.942623096673485,
325
- "recall": 0.9425587467362925,
326
- "support": 30257.0
327
  },
328
  "step": 648
329
  },
330
  {
331
  "epoch": 9.0,
332
  "eval_B": {
333
- "f1-score": 0.8885191347753745,
334
- "precision": 0.8711256117455138,
335
- "recall": 0.9066213921901528,
336
- "support": 1178.0
337
  },
338
  "eval_I": {
339
- "f1-score": 0.9620326487625066,
340
- "precision": 0.9574445783763954,
341
- "recall": 0.9666649029049156,
342
- "support": 18899.0
343
  },
344
  "eval_O": {
345
- "f1-score": 0.9275707898658718,
346
- "precision": 0.9382914572864322,
347
- "recall": 0.9170923379174852,
348
- "support": 10180.0
349
  },
350
- "eval_accuracy": 0.9476484780381399,
351
- "eval_loss": 0.24401217699050903,
352
  "eval_macro avg": {
353
- "f1-score": 0.9260408578012509,
354
- "precision": 0.9222872158027805,
355
- "recall": 0.9301262110041845,
356
- "support": 30257.0
357
- },
358
- "eval_runtime": 4.701,
359
- "eval_samples_per_second": 17.23,
360
- "eval_steps_per_second": 2.34,
361
  "eval_weighted avg": {
362
- "f1-score": 0.94757580760031,
363
- "precision": 0.9476398219436029,
364
- "recall": 0.9476484780381399,
365
- "support": 30257.0
366
  },
367
  "step": 729
368
  },
369
  {
370
  "epoch": 10.0,
371
  "eval_B": {
372
- "f1-score": 0.875959595959596,
373
- "precision": 0.8357748650732459,
374
- "recall": 0.9202037351443124,
375
- "support": 1178.0
376
  },
377
  "eval_I": {
378
- "f1-score": 0.955652945166171,
379
- "precision": 0.9424749163879599,
380
- "recall": 0.9692047198264458,
381
- "support": 18899.0
382
  },
383
  "eval_O": {
384
- "f1-score": 0.9127632580563309,
385
- "precision": 0.9441469816272966,
386
- "recall": 0.8833988212180747,
387
- "support": 10180.0
388
  },
389
- "eval_accuracy": 0.9384274713289487,
390
- "eval_loss": 0.29471203684806824,
391
  "eval_macro avg": {
392
- "f1-score": 0.9147919330606994,
393
- "precision": 0.9074655876961675,
394
- "recall": 0.9242690920629443,
395
- "support": 30257.0
396
- },
397
- "eval_runtime": 4.6985,
398
- "eval_samples_per_second": 17.24,
399
- "eval_steps_per_second": 2.341,
400
  "eval_weighted avg": {
401
- "f1-score": 0.9381199518045186,
402
- "precision": 0.9388833165494999,
403
- "recall": 0.9384274713289487,
404
- "support": 30257.0
405
  },
406
  "step": 810
407
  },
408
  {
409
  "epoch": 11.0,
410
  "eval_B": {
411
- "f1-score": 0.875515251442704,
412
- "precision": 0.8509615384615384,
413
- "recall": 0.9015280135823429,
414
- "support": 1178.0
415
  },
416
  "eval_I": {
417
- "f1-score": 0.9575287538250502,
418
- "precision": 0.9547582723972855,
419
- "recall": 0.96031536060109,
420
- "support": 18899.0
421
  },
422
  "eval_O": {
423
- "f1-score": 0.9195242814667988,
424
- "precision": 0.9278,
425
- "recall": 0.9113948919449901,
426
- "support": 10180.0
427
  },
428
- "eval_accuracy": 0.9415672406385299,
429
- "eval_loss": 0.30941879749298096,
430
  "eval_macro avg": {
431
- "f1-score": 0.9175227622448511,
432
- "precision": 0.9111732702862746,
433
- "recall": 0.9244127553761411,
434
- "support": 30257.0
435
- },
436
- "eval_runtime": 4.7017,
437
- "eval_samples_per_second": 17.228,
438
- "eval_steps_per_second": 2.34,
439
  "eval_weighted avg": {
440
- "f1-score": 0.9415490653426031,
441
- "precision": 0.9416470001105195,
442
- "recall": 0.9415672406385299,
443
- "support": 30257.0
444
  },
445
  "step": 891
446
  },
447
  {
448
  "epoch": 12.0,
449
  "eval_B": {
450
- "f1-score": 0.8818181818181817,
451
- "precision": 0.8590982286634461,
452
- "recall": 0.9057724957555179,
453
- "support": 1178.0
454
  },
455
  "eval_I": {
456
- "f1-score": 0.9542487087386278,
457
- "precision": 0.9457436856875585,
458
- "recall": 0.9629080903751521,
459
- "support": 18899.0
460
  },
461
  "eval_O": {
462
- "f1-score": 0.910740239562973,
463
- "precision": 0.9297042873222142,
464
- "recall": 0.8925343811394892,
465
- "support": 10180.0
466
  },
467
- "eval_accuracy": 0.9370063125888224,
468
- "eval_loss": 0.28610122203826904,
469
  "eval_macro avg": {
470
- "f1-score": 0.9156023767065942,
471
- "precision": 0.9115154005577396,
472
- "recall": 0.9204049890900531,
473
- "support": 30257.0
474
- },
475
- "eval_runtime": 4.6911,
476
- "eval_samples_per_second": 17.267,
477
- "eval_steps_per_second": 2.345,
478
  "eval_weighted avg": {
479
- "f1-score": 0.9367902899621314,
480
- "precision": 0.9369738332985704,
481
- "recall": 0.9370063125888224,
482
- "support": 30257.0
483
  },
484
  "step": 972
485
  },
486
  {
487
  "epoch": 12.35,
488
- "grad_norm": 0.9797911643981934,
489
  "learning_rate": 1.506172839506173e-05,
490
- "loss": 0.0233,
491
  "step": 1000
492
  },
493
  {
494
  "epoch": 13.0,
495
  "eval_B": {
496
- "f1-score": 0.8875154511742892,
497
- "precision": 0.8622898318654924,
498
- "recall": 0.9142614601018676,
499
- "support": 1178.0
500
  },
501
  "eval_I": {
502
- "f1-score": 0.9602009576889865,
503
- "precision": 0.9497877627083549,
504
- "recall": 0.9708450182549341,
505
- "support": 18899.0
506
  },
507
  "eval_O": {
508
- "f1-score": 0.9229994967287367,
509
- "precision": 0.9463364293085655,
510
- "recall": 0.9007858546168959,
511
- "support": 10180.0
512
  },
513
- "eval_accuracy": 0.9450705621839575,
514
- "eval_loss": 0.3311145305633545,
515
  "eval_macro avg": {
516
- "f1-score": 0.9235719685306708,
517
- "precision": 0.9194713412941375,
518
- "recall": 0.9286307776578991,
519
- "support": 30257.0
520
- },
521
- "eval_runtime": 4.7187,
522
- "eval_samples_per_second": 17.166,
523
- "eval_steps_per_second": 2.331,
524
  "eval_weighted avg": {
525
- "f1-score": 0.9448546114137558,
526
- "precision": 0.9452199887538072,
527
- "recall": 0.9450705621839575,
528
- "support": 30257.0
529
  },
530
  "step": 1053
531
  }
@@ -535,7 +535,7 @@
535
  "num_input_tokens_seen": 0,
536
  "num_train_epochs": 50,
537
  "save_steps": 500,
538
- "total_flos": 1863421434365400.0,
539
  "train_batch_size": 4,
540
  "trial_name": null,
541
  "trial_params": null
 
11
  {
12
  "epoch": 1.0,
13
  "eval_B": {
14
+ "f1-score": 0.8266764382557715,
15
+ "precision": 0.7866108786610879,
16
+ "recall": 0.871042471042471,
17
+ "support": 1295.0
18
  },
19
  "eval_I": {
20
+ "f1-score": 0.9410658617447671,
21
+ "precision": 0.9344077040239768,
22
+ "recall": 0.9478195863443808,
23
+ "support": 20065.0
24
  },
25
  "eval_O": {
26
+ "f1-score": 0.8562443302086484,
27
+ "precision": 0.8789421405512788,
28
+ "recall": 0.8346893055064262,
29
+ "support": 8481.0
30
  },
31
+ "eval_accuracy": 0.912335377500754,
32
+ "eval_loss": 0.2450982630252838,
33
  "eval_macro avg": {
34
+ "f1-score": 0.8746622100697291,
35
+ "precision": 0.8666535744121145,
36
+ "recall": 0.8845171209644261,
37
+ "support": 29841.0
38
+ },
39
+ "eval_runtime": 4.6269,
40
+ "eval_samples_per_second": 17.29,
41
+ "eval_steps_per_second": 2.161,
42
  "eval_weighted avg": {
43
+ "f1-score": 0.9119949287205363,
44
+ "precision": 0.9122301519092053,
45
+ "recall": 0.912335377500754,
46
+ "support": 29841.0
47
  },
48
  "step": 81
49
  },
50
  {
51
  "epoch": 2.0,
52
  "eval_B": {
53
+ "f1-score": 0.8833395452851286,
54
+ "precision": 0.8537463976945245,
55
+ "recall": 0.915057915057915,
56
+ "support": 1295.0
57
  },
58
  "eval_I": {
59
+ "f1-score": 0.9549518227873249,
60
+ "precision": 0.9516903430183636,
61
+ "recall": 0.958235733864939,
62
+ "support": 20065.0
63
  },
64
  "eval_O": {
65
+ "f1-score": 0.88924750463212,
66
+ "precision": 0.9016969696969697,
67
+ "recall": 0.877137130055418,
68
+ "support": 8481.0
69
  },
70
+ "eval_accuracy": 0.9333132267685399,
71
+ "eval_loss": 0.20076116919517517,
72
  "eval_macro avg": {
73
+ "f1-score": 0.9091796242348579,
74
+ "precision": 0.9023779034699526,
75
+ "recall": 0.916810259659424,
76
+ "support": 29841.0
77
+ },
78
+ "eval_runtime": 4.6524,
79
+ "eval_samples_per_second": 17.195,
80
+ "eval_steps_per_second": 2.149,
81
  "eval_weighted avg": {
82
+ "f1-score": 0.9331705077630417,
83
+ "precision": 0.9332314707173981,
84
+ "recall": 0.9333132267685399,
85
+ "support": 29841.0
86
  },
87
  "step": 162
88
  },
89
  {
90
  "epoch": 3.0,
91
  "eval_B": {
92
+ "f1-score": 0.8927335640138409,
93
+ "precision": 0.8889739663093414,
94
+ "recall": 0.8965250965250965,
95
+ "support": 1295.0
96
  },
97
  "eval_I": {
98
+ "f1-score": 0.9590235396687009,
99
+ "precision": 0.9586653386454184,
100
+ "recall": 0.9593820084724645,
101
+ "support": 20065.0
102
  },
103
  "eval_O": {
104
+ "f1-score": 0.9006849315068494,
105
+ "precision": 0.9020697811945594,
106
+ "recall": 0.8993043273198915,
107
+ "support": 8481.0
108
  },
109
+ "eval_accuracy": 0.9395797727958178,
110
+ "eval_loss": 0.19863653182983398,
111
  "eval_macro avg": {
112
+ "f1-score": 0.9174806783964637,
113
+ "precision": 0.9165696953831063,
114
+ "recall": 0.9184038107724842,
115
+ "support": 29841.0
116
+ },
117
+ "eval_runtime": 4.655,
118
+ "eval_samples_per_second": 17.186,
119
+ "eval_steps_per_second": 2.148,
120
  "eval_weighted avg": {
121
+ "f1-score": 0.9395665759512081,
122
+ "precision": 0.9395561516236713,
123
+ "recall": 0.9395797727958178,
124
+ "support": 29841.0
125
  },
126
  "step": 243
127
  },
128
  {
129
  "epoch": 4.0,
130
  "eval_B": {
131
+ "f1-score": 0.8999616711383672,
132
+ "precision": 0.893455098934551,
133
+ "recall": 0.9065637065637066,
134
+ "support": 1295.0
135
  },
136
  "eval_I": {
137
+ "f1-score": 0.9612657113512985,
138
+ "precision": 0.9436362763454799,
139
+ "recall": 0.9795664091701969,
140
+ "support": 20065.0
141
  },
142
  "eval_O": {
143
+ "f1-score": 0.8996847765622102,
144
+ "precision": 0.9454403741231489,
145
+ "recall": 0.8581535196321188,
146
+ "support": 8481.0
147
  },
148
+ "eval_accuracy": 0.9418920277470594,
149
+ "eval_loss": 0.19743724167346954,
150
  "eval_macro avg": {
151
+ "f1-score": 0.920304053017292,
152
+ "precision": 0.9275105831343932,
153
+ "recall": 0.9147612117886741,
154
+ "support": 29841.0
155
+ },
156
+ "eval_runtime": 4.6615,
157
+ "eval_samples_per_second": 17.162,
158
+ "eval_steps_per_second": 2.145,
159
  "eval_weighted avg": {
160
+ "f1-score": 0.9411036309913238,
161
+ "precision": 0.9419713163409645,
162
+ "recall": 0.9418920277470594,
163
+ "support": 29841.0
164
  },
165
  "step": 324
166
  },
167
  {
168
  "epoch": 5.0,
169
  "eval_B": {
170
+ "f1-score": 0.8953846153846153,
171
+ "precision": 0.8919540229885058,
172
+ "recall": 0.8988416988416988,
173
+ "support": 1295.0
174
  },
175
  "eval_I": {
176
+ "f1-score": 0.9602426713742261,
177
+ "precision": 0.9581224570804803,
178
+ "recall": 0.9623722900573137,
179
+ "support": 20065.0
180
  },
181
  "eval_O": {
182
+ "f1-score": 0.9022119433078337,
183
+ "precision": 0.9075399665950847,
184
+ "recall": 0.8969461148449476,
185
+ "support": 8481.0
186
  },
187
+ "eval_accuracy": 0.9410207432726785,
188
+ "eval_loss": 0.2340461015701294,
189
  "eval_macro avg": {
190
+ "f1-score": 0.9192797433555584,
191
+ "precision": 0.9192054822213569,
192
+ "recall": 0.9193867012479867,
193
+ "support": 29841.0
194
+ },
195
+ "eval_runtime": 4.6719,
196
+ "eval_samples_per_second": 17.124,
197
+ "eval_steps_per_second": 2.14,
198
  "eval_weighted avg": {
199
+ "f1-score": 0.9409353496612266,
200
+ "precision": 0.9408751053176122,
201
+ "recall": 0.9410207432726785,
202
+ "support": 29841.0
203
  },
204
  "step": 405
205
  },
206
  {
207
  "epoch": 6.0,
208
  "eval_B": {
209
+ "f1-score": 0.8929831438651511,
210
+ "precision": 0.9068471337579618,
211
+ "recall": 0.8795366795366796,
212
+ "support": 1295.0
213
  },
214
  "eval_I": {
215
+ "f1-score": 0.9522293798155866,
216
+ "precision": 0.9655225409836066,
217
+ "recall": 0.9392972838275604,
218
+ "support": 20065.0
219
  },
220
  "eval_O": {
221
+ "f1-score": 0.8885216003647556,
222
+ "precision": 0.8599007170435742,
223
+ "recall": 0.919113312109421,
224
+ "support": 8481.0
225
  },
226
+ "eval_accuracy": 0.930967460875976,
227
+ "eval_loss": 0.21774999797344208,
228
  "eval_macro avg": {
229
+ "f1-score": 0.9112447080151643,
230
+ "precision": 0.9107567972617142,
231
+ "recall": 0.9126490918245537,
232
+ "support": 29841.0
233
+ },
234
+ "eval_runtime": 4.647,
235
+ "eval_samples_per_second": 17.215,
236
+ "eval_steps_per_second": 2.152,
237
  "eval_weighted avg": {
238
+ "f1-score": 0.9315521386682287,
239
+ "precision": 0.9329578366776978,
240
+ "recall": 0.930967460875976,
241
+ "support": 29841.0
242
  },
243
  "step": 486
244
  },
245
  {
246
  "epoch": 6.17,
247
+ "grad_norm": 9.310327529907227,
248
  "learning_rate": 1.7530864197530865e-05,
249
+ "loss": 0.1627,
250
  "step": 500
251
  },
252
  {
253
  "epoch": 7.0,
254
  "eval_B": {
255
+ "f1-score": 0.8936490041337843,
256
+ "precision": 0.8704245973645681,
257
+ "recall": 0.9181467181467181,
258
+ "support": 1295.0
259
  },
260
  "eval_I": {
261
+ "f1-score": 0.9599271850630982,
262
+ "precision": 0.9477800446905664,
263
+ "recall": 0.9723897333665587,
264
+ "support": 20065.0
265
  },
266
  "eval_O": {
267
+ "f1-score": 0.8979841172877214,
268
+ "precision": 0.9316770186335404,
269
+ "recall": 0.8666430845419172,
270
+ "support": 8481.0
271
  },
272
+ "eval_accuracy": 0.9399819040916859,
273
+ "eval_loss": 0.27376389503479004,
274
  "eval_macro avg": {
275
+ "f1-score": 0.9171867688282013,
276
+ "precision": 0.9166272202295582,
277
+ "recall": 0.9190598453517312,
278
+ "support": 29841.0
279
+ },
280
+ "eval_runtime": 4.6671,
281
+ "eval_samples_per_second": 17.141,
282
+ "eval_steps_per_second": 2.143,
283
  "eval_weighted avg": {
284
+ "f1-score": 0.9394463230910989,
285
+ "precision": 0.9398464945991887,
286
+ "recall": 0.9399819040916859,
287
+ "support": 29841.0
288
  },
289
  "step": 567
290
  },
291
  {
292
  "epoch": 8.0,
293
  "eval_B": {
294
+ "f1-score": 0.8948545861297539,
295
+ "precision": 0.8651766402307137,
296
+ "recall": 0.9266409266409267,
297
+ "support": 1295.0
298
  },
299
  "eval_I": {
300
+ "f1-score": 0.9590139880661254,
301
+ "precision": 0.9414702069429106,
302
+ "recall": 0.9772240219287316,
303
+ "support": 20065.0
304
  },
305
  "eval_O": {
306
+ "f1-score": 0.8927241122423641,
307
+ "precision": 0.9427035531663825,
308
+ "recall": 0.8477773847423653,
309
+ "support": 8481.0
310
  },
311
+ "eval_accuracy": 0.9382393351429241,
312
+ "eval_loss": 0.3103167414665222,
313
  "eval_macro avg": {
314
+ "f1-score": 0.9155308954794145,
315
+ "precision": 0.916450133446669,
316
+ "recall": 0.9172141111040079,
317
+ "support": 29841.0
318
+ },
319
+ "eval_runtime": 4.6339,
320
+ "eval_samples_per_second": 17.264,
321
+ "eval_steps_per_second": 2.158,
322
  "eval_weighted avg": {
323
+ "f1-score": 0.9373896838414373,
324
+ "precision": 0.9385098450391195,
325
+ "recall": 0.9382393351429241,
326
+ "support": 29841.0
327
  },
328
  "step": 648
329
  },
330
  {
331
  "epoch": 9.0,
332
  "eval_B": {
333
+ "f1-score": 0.9011235955056179,
334
+ "precision": 0.8749090909090909,
335
+ "recall": 0.9289575289575289,
336
+ "support": 1295.0
337
  },
338
  "eval_I": {
339
+ "f1-score": 0.9615119937082186,
340
+ "precision": 0.9485040973670175,
341
+ "recall": 0.9748816346872664,
342
+ "support": 20065.0
343
  },
344
  "eval_O": {
345
+ "f1-score": 0.9013722126929674,
346
+ "precision": 0.9380339155935229,
347
+ "recall": 0.8674684589081476,
348
+ "support": 8481.0
349
  },
350
+ "eval_accuracy": 0.9423611809255722,
351
+ "eval_loss": 0.32883673906326294,
352
  "eval_macro avg": {
353
+ "f1-score": 0.9213359339689347,
354
+ "precision": 0.9204823679565438,
355
+ "recall": 0.9237692075176476,
356
+ "support": 29841.0
357
+ },
358
+ "eval_runtime": 4.6714,
359
+ "eval_samples_per_second": 17.126,
360
+ "eval_steps_per_second": 2.141,
361
  "eval_weighted avg": {
362
+ "f1-score": 0.9417992341337168,
363
+ "precision": 0.9423346276781992,
364
+ "recall": 0.9423611809255722,
365
+ "support": 29841.0
366
  },
367
  "step": 729
368
  },
369
  {
370
  "epoch": 10.0,
371
  "eval_B": {
372
+ "f1-score": 0.8935361216730039,
373
+ "precision": 0.8801498127340824,
374
+ "recall": 0.9073359073359073,
375
+ "support": 1295.0
376
  },
377
  "eval_I": {
378
+ "f1-score": 0.9632332861646931,
379
+ "precision": 0.9596359319351009,
380
+ "recall": 0.9668577124345876,
381
+ "support": 20065.0
382
  },
383
  "eval_O": {
384
+ "f1-score": 0.9095462405342556,
385
+ "precision": 0.9200241254523522,
386
+ "recall": 0.8993043273198915,
387
+ "support": 8481.0
388
  },
389
+ "eval_accuracy": 0.9450755671726819,
390
+ "eval_loss": 0.3041447699069977,
391
  "eval_macro avg": {
392
+ "f1-score": 0.9221052161239841,
393
+ "precision": 0.9199366233738452,
394
+ "recall": 0.9244993156967954,
395
+ "support": 29841.0
396
+ },
397
+ "eval_runtime": 4.6532,
398
+ "eval_samples_per_second": 17.193,
399
+ "eval_steps_per_second": 2.149,
400
  "eval_weighted avg": {
401
+ "f1-score": 0.9449504651463465,
402
+ "precision": 0.9449285744355027,
403
+ "recall": 0.9450755671726819,
404
+ "support": 29841.0
405
  },
406
  "step": 810
407
  },
408
  {
409
  "epoch": 11.0,
410
  "eval_B": {
411
+ "f1-score": 0.8924122310305775,
412
+ "precision": 0.8729689807976366,
413
+ "recall": 0.9127413127413128,
414
+ "support": 1295.0
415
  },
416
  "eval_I": {
417
+ "f1-score": 0.9588723786993913,
418
+ "precision": 0.9483330083837005,
419
+ "recall": 0.9696486419137802,
420
+ "support": 20065.0
421
  },
422
  "eval_O": {
423
+ "f1-score": 0.8956965718453683,
424
+ "precision": 0.9243507715468574,
425
+ "recall": 0.8687654757693668,
426
+ "support": 8481.0
427
  },
428
+ "eval_accuracy": 0.9385074226735028,
429
+ "eval_loss": 0.3531416654586792,
430
  "eval_macro avg": {
431
+ "f1-score": 0.9156603938584458,
432
+ "precision": 0.9152175869093981,
433
+ "recall": 0.9170518101414866,
434
+ "support": 29841.0
435
+ },
436
+ "eval_runtime": 4.6508,
437
+ "eval_samples_per_second": 17.201,
438
+ "eval_steps_per_second": 2.15,
439
  "eval_weighted avg": {
440
+ "f1-score": 0.938033267772811,
441
+ "precision": 0.9382465579853486,
442
+ "recall": 0.9385074226735028,
443
+ "support": 29841.0
444
  },
445
  "step": 891
446
  },
447
  {
448
  "epoch": 12.0,
449
  "eval_B": {
450
+ "f1-score": 0.8970199924556771,
451
+ "precision": 0.8768436578171092,
452
+ "recall": 0.9181467181467181,
453
+ "support": 1295.0
454
  },
455
  "eval_I": {
456
+ "f1-score": 0.9595004897159647,
457
+ "precision": 0.9431046931407943,
458
+ "recall": 0.9764764515325193,
459
+ "support": 20065.0
460
  },
461
  "eval_O": {
462
+ "f1-score": 0.8953122104873078,
463
+ "precision": 0.9400778210116731,
464
+ "recall": 0.8546162009197029,
465
+ "support": 8481.0
466
  },
467
+ "eval_accuracy": 0.9393116852652391,
468
+ "eval_loss": 0.4237186312675476,
469
  "eval_macro avg": {
470
+ "f1-score": 0.9172775642196499,
471
+ "precision": 0.9200087239898588,
472
+ "recall": 0.9164131235329801,
473
+ "support": 29841.0
474
+ },
475
+ "eval_runtime": 4.6388,
476
+ "eval_samples_per_second": 17.246,
477
+ "eval_steps_per_second": 2.156,
478
  "eval_weighted avg": {
479
+ "f1-score": 0.9385463313402296,
480
+ "precision": 0.9393689288141548,
481
+ "recall": 0.9393116852652391,
482
+ "support": 29841.0
483
  },
484
  "step": 972
485
  },
486
  {
487
  "epoch": 12.35,
488
+ "grad_norm": 0.5223535895347595,
489
  "learning_rate": 1.506172839506173e-05,
490
+ "loss": 0.0196,
491
  "step": 1000
492
  },
493
  {
494
  "epoch": 13.0,
495
  "eval_B": {
496
+ "f1-score": 0.8891411275066211,
497
+ "precision": 0.8716617210682492,
498
+ "recall": 0.9073359073359073,
499
+ "support": 1295.0
500
  },
501
  "eval_I": {
502
+ "f1-score": 0.9610692322951795,
503
+ "precision": 0.9563736860287223,
504
+ "recall": 0.9658111138798904,
505
+ "support": 20065.0
506
  },
507
  "eval_O": {
508
+ "f1-score": 0.9043145233678416,
509
+ "precision": 0.918104495747266,
510
+ "recall": 0.8909326730338404,
511
+ "support": 8481.0
512
  },
513
+ "eval_accuracy": 0.9419925605710264,
514
+ "eval_loss": 0.4308999180793762,
515
  "eval_macro avg": {
516
+ "f1-score": 0.9181749610565474,
517
+ "precision": 0.9153799676147458,
518
+ "recall": 0.9213598980832126,
519
+ "support": 29841.0
520
+ },
521
+ "eval_runtime": 4.6555,
522
+ "eval_samples_per_second": 17.184,
523
+ "eval_steps_per_second": 2.148,
524
  "eval_weighted avg": {
525
+ "f1-score": 0.9418177466843106,
526
+ "precision": 0.9418211242043584,
527
+ "recall": 0.9419925605710264,
528
+ "support": 29841.0
529
  },
530
  "step": 1053
531
  }
 
535
  "num_input_tokens_seen": 0,
536
  "num_train_epochs": 50,
537
  "save_steps": 500,
538
+ "total_flos": 1869226485562800.0,
539
  "train_batch_size": 4,
540
  "trial_name": null,
541
  "trial_params": null
checkpoint-1053/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a9f81281de988dbf2616f1c96e725c00f6e4415b289b670a7cb21136eb66e95f
3
  size 4856
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d229d7bc9127d222c880edfd3fa9418a9ccede656fe27f7f0aa04990c9605be4
3
  size 4856