dkqjrm commited on
Commit
35e2134
1 Parent(s): 082ab64

End of training

Browse files
Files changed (4) hide show
  1. all_results.json +7 -0
  2. best_results.json +5 -0
  3. train_results.json +7 -0
  4. trainer_state.json +1249 -0
all_results.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 80.0,
3
+ "train_loss": 0.5144821853637696,
4
+ "train_runtime": 1625.7247,
5
+ "train_samples_per_second": 19.684,
6
+ "train_steps_per_second": 1.23
7
+ }
best_results.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "best_epoch": 49,
3
+ "best_eval_accuracy": 0.74,
4
+ "epoch": 80.0
5
+ }
train_results.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 80.0,
3
+ "train_loss": 0.5144821853637696,
4
+ "train_runtime": 1625.7247,
5
+ "train_samples_per_second": 19.684,
6
+ "train_steps_per_second": 1.23
7
+ }
trainer_state.json ADDED
@@ -0,0 +1,1249 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 80.0,
5
+ "global_step": 2000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 1.0,
12
+ "eval_accuracy": 0.46,
13
+ "eval_loss": 0.49578723311424255,
14
+ "eval_runtime": 2.9005,
15
+ "eval_samples_per_second": 34.477,
16
+ "eval_steps_per_second": 4.482,
17
+ "step": 25
18
+ },
19
+ {
20
+ "best_epoch": 0,
21
+ "best_eval_accuracy": 0.46,
22
+ "epoch": 1.0,
23
+ "step": 25
24
+ },
25
+ {
26
+ "epoch": 2.0,
27
+ "eval_accuracy": 0.54,
28
+ "eval_loss": 0.5955537557601929,
29
+ "eval_runtime": 2.9308,
30
+ "eval_samples_per_second": 34.12,
31
+ "eval_steps_per_second": 4.436,
32
+ "step": 50
33
+ },
34
+ {
35
+ "best_epoch": 1,
36
+ "best_eval_accuracy": 0.54,
37
+ "epoch": 2.0,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 3.0,
42
+ "eval_accuracy": 0.45,
43
+ "eval_loss": 0.5376533269882202,
44
+ "eval_runtime": 2.9692,
45
+ "eval_samples_per_second": 33.679,
46
+ "eval_steps_per_second": 4.378,
47
+ "step": 75
48
+ },
49
+ {
50
+ "best_epoch": 1,
51
+ "best_eval_accuracy": 0.54,
52
+ "epoch": 3.0,
53
+ "step": 75
54
+ },
55
+ {
56
+ "epoch": 4.0,
57
+ "eval_accuracy": 0.61,
58
+ "eval_loss": 0.4201565980911255,
59
+ "eval_runtime": 2.9986,
60
+ "eval_samples_per_second": 33.349,
61
+ "eval_steps_per_second": 4.335,
62
+ "step": 100
63
+ },
64
+ {
65
+ "best_epoch": 3,
66
+ "best_eval_accuracy": 0.61,
67
+ "epoch": 4.0,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 5.0,
72
+ "eval_accuracy": 0.44,
73
+ "eval_loss": 0.43667927384376526,
74
+ "eval_runtime": 3.0016,
75
+ "eval_samples_per_second": 33.316,
76
+ "eval_steps_per_second": 4.331,
77
+ "step": 125
78
+ },
79
+ {
80
+ "best_epoch": 3,
81
+ "best_eval_accuracy": 0.61,
82
+ "epoch": 5.0,
83
+ "step": 125
84
+ },
85
+ {
86
+ "epoch": 6.0,
87
+ "eval_accuracy": 0.51,
88
+ "eval_loss": 0.43700677156448364,
89
+ "eval_runtime": 3.0099,
90
+ "eval_samples_per_second": 33.223,
91
+ "eval_steps_per_second": 4.319,
92
+ "step": 150
93
+ },
94
+ {
95
+ "best_epoch": 3,
96
+ "best_eval_accuracy": 0.61,
97
+ "epoch": 6.0,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 7.0,
102
+ "eval_accuracy": 0.66,
103
+ "eval_loss": 0.42071959376335144,
104
+ "eval_runtime": 3.0221,
105
+ "eval_samples_per_second": 33.09,
106
+ "eval_steps_per_second": 4.302,
107
+ "step": 175
108
+ },
109
+ {
110
+ "best_epoch": 6,
111
+ "best_eval_accuracy": 0.66,
112
+ "epoch": 7.0,
113
+ "step": 175
114
+ },
115
+ {
116
+ "epoch": 8.0,
117
+ "eval_accuracy": 0.58,
118
+ "eval_loss": 0.4423440098762512,
119
+ "eval_runtime": 3.0262,
120
+ "eval_samples_per_second": 33.044,
121
+ "eval_steps_per_second": 4.296,
122
+ "step": 200
123
+ },
124
+ {
125
+ "best_epoch": 6,
126
+ "best_eval_accuracy": 0.66,
127
+ "epoch": 8.0,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 9.0,
132
+ "eval_accuracy": 0.61,
133
+ "eval_loss": 0.41069385409355164,
134
+ "eval_runtime": 3.0271,
135
+ "eval_samples_per_second": 33.035,
136
+ "eval_steps_per_second": 4.294,
137
+ "step": 225
138
+ },
139
+ {
140
+ "best_epoch": 6,
141
+ "best_eval_accuracy": 0.66,
142
+ "epoch": 9.0,
143
+ "step": 225
144
+ },
145
+ {
146
+ "epoch": 10.0,
147
+ "eval_accuracy": 0.64,
148
+ "eval_loss": 0.4332168996334076,
149
+ "eval_runtime": 3.0297,
150
+ "eval_samples_per_second": 33.007,
151
+ "eval_steps_per_second": 4.291,
152
+ "step": 250
153
+ },
154
+ {
155
+ "best_epoch": 6,
156
+ "best_eval_accuracy": 0.66,
157
+ "epoch": 10.0,
158
+ "step": 250
159
+ },
160
+ {
161
+ "epoch": 11.0,
162
+ "eval_accuracy": 0.6,
163
+ "eval_loss": 0.4055112898349762,
164
+ "eval_runtime": 3.031,
165
+ "eval_samples_per_second": 32.992,
166
+ "eval_steps_per_second": 4.289,
167
+ "step": 275
168
+ },
169
+ {
170
+ "best_epoch": 6,
171
+ "best_eval_accuracy": 0.66,
172
+ "epoch": 11.0,
173
+ "step": 275
174
+ },
175
+ {
176
+ "epoch": 12.0,
177
+ "eval_accuracy": 0.63,
178
+ "eval_loss": 0.4375588893890381,
179
+ "eval_runtime": 3.0325,
180
+ "eval_samples_per_second": 32.977,
181
+ "eval_steps_per_second": 4.287,
182
+ "step": 300
183
+ },
184
+ {
185
+ "best_epoch": 6,
186
+ "best_eval_accuracy": 0.66,
187
+ "epoch": 12.0,
188
+ "step": 300
189
+ },
190
+ {
191
+ "epoch": 13.0,
192
+ "eval_accuracy": 0.57,
193
+ "eval_loss": 0.40624508261680603,
194
+ "eval_runtime": 3.0305,
195
+ "eval_samples_per_second": 32.998,
196
+ "eval_steps_per_second": 4.29,
197
+ "step": 325
198
+ },
199
+ {
200
+ "best_epoch": 6,
201
+ "best_eval_accuracy": 0.66,
202
+ "epoch": 13.0,
203
+ "step": 325
204
+ },
205
+ {
206
+ "epoch": 14.0,
207
+ "eval_accuracy": 0.61,
208
+ "eval_loss": 0.4000048339366913,
209
+ "eval_runtime": 3.0315,
210
+ "eval_samples_per_second": 32.986,
211
+ "eval_steps_per_second": 4.288,
212
+ "step": 350
213
+ },
214
+ {
215
+ "best_epoch": 6,
216
+ "best_eval_accuracy": 0.66,
217
+ "epoch": 14.0,
218
+ "step": 350
219
+ },
220
+ {
221
+ "epoch": 15.0,
222
+ "eval_accuracy": 0.63,
223
+ "eval_loss": 0.40517398715019226,
224
+ "eval_runtime": 3.0336,
225
+ "eval_samples_per_second": 32.965,
226
+ "eval_steps_per_second": 4.285,
227
+ "step": 375
228
+ },
229
+ {
230
+ "best_epoch": 6,
231
+ "best_eval_accuracy": 0.66,
232
+ "epoch": 15.0,
233
+ "step": 375
234
+ },
235
+ {
236
+ "epoch": 16.0,
237
+ "eval_accuracy": 0.68,
238
+ "eval_loss": 0.3960806727409363,
239
+ "eval_runtime": 3.0345,
240
+ "eval_samples_per_second": 32.955,
241
+ "eval_steps_per_second": 4.284,
242
+ "step": 400
243
+ },
244
+ {
245
+ "best_epoch": 15,
246
+ "best_eval_accuracy": 0.68,
247
+ "epoch": 16.0,
248
+ "step": 400
249
+ },
250
+ {
251
+ "epoch": 17.0,
252
+ "eval_accuracy": 0.67,
253
+ "eval_loss": 0.3975551724433899,
254
+ "eval_runtime": 3.0333,
255
+ "eval_samples_per_second": 32.968,
256
+ "eval_steps_per_second": 4.286,
257
+ "step": 425
258
+ },
259
+ {
260
+ "best_epoch": 15,
261
+ "best_eval_accuracy": 0.68,
262
+ "epoch": 17.0,
263
+ "step": 425
264
+ },
265
+ {
266
+ "epoch": 18.0,
267
+ "eval_accuracy": 0.65,
268
+ "eval_loss": 0.41863301396369934,
269
+ "eval_runtime": 3.0326,
270
+ "eval_samples_per_second": 32.975,
271
+ "eval_steps_per_second": 4.287,
272
+ "step": 450
273
+ },
274
+ {
275
+ "best_epoch": 15,
276
+ "best_eval_accuracy": 0.68,
277
+ "epoch": 18.0,
278
+ "step": 450
279
+ },
280
+ {
281
+ "epoch": 19.0,
282
+ "eval_accuracy": 0.63,
283
+ "eval_loss": 0.4303971230983734,
284
+ "eval_runtime": 3.0325,
285
+ "eval_samples_per_second": 32.976,
286
+ "eval_steps_per_second": 4.287,
287
+ "step": 475
288
+ },
289
+ {
290
+ "best_epoch": 15,
291
+ "best_eval_accuracy": 0.68,
292
+ "epoch": 19.0,
293
+ "step": 475
294
+ },
295
+ {
296
+ "epoch": 20.0,
297
+ "learning_rate": 0.015,
298
+ "loss": 0.731,
299
+ "step": 500
300
+ },
301
+ {
302
+ "epoch": 20.0,
303
+ "eval_accuracy": 0.69,
304
+ "eval_loss": 0.4357996881008148,
305
+ "eval_runtime": 3.0345,
306
+ "eval_samples_per_second": 32.955,
307
+ "eval_steps_per_second": 4.284,
308
+ "step": 500
309
+ },
310
+ {
311
+ "best_epoch": 19,
312
+ "best_eval_accuracy": 0.69,
313
+ "epoch": 20.0,
314
+ "step": 500
315
+ },
316
+ {
317
+ "epoch": 21.0,
318
+ "eval_accuracy": 0.68,
319
+ "eval_loss": 0.41348427534103394,
320
+ "eval_runtime": 3.0336,
321
+ "eval_samples_per_second": 32.964,
322
+ "eval_steps_per_second": 4.285,
323
+ "step": 525
324
+ },
325
+ {
326
+ "best_epoch": 19,
327
+ "best_eval_accuracy": 0.69,
328
+ "epoch": 21.0,
329
+ "step": 525
330
+ },
331
+ {
332
+ "epoch": 22.0,
333
+ "eval_accuracy": 0.68,
334
+ "eval_loss": 0.4180167019367218,
335
+ "eval_runtime": 3.0311,
336
+ "eval_samples_per_second": 32.991,
337
+ "eval_steps_per_second": 4.289,
338
+ "step": 550
339
+ },
340
+ {
341
+ "best_epoch": 19,
342
+ "best_eval_accuracy": 0.69,
343
+ "epoch": 22.0,
344
+ "step": 550
345
+ },
346
+ {
347
+ "epoch": 23.0,
348
+ "eval_accuracy": 0.66,
349
+ "eval_loss": 0.46266496181488037,
350
+ "eval_runtime": 3.0317,
351
+ "eval_samples_per_second": 32.985,
352
+ "eval_steps_per_second": 4.288,
353
+ "step": 575
354
+ },
355
+ {
356
+ "best_epoch": 19,
357
+ "best_eval_accuracy": 0.69,
358
+ "epoch": 23.0,
359
+ "step": 575
360
+ },
361
+ {
362
+ "epoch": 24.0,
363
+ "eval_accuracy": 0.65,
364
+ "eval_loss": 0.4150283932685852,
365
+ "eval_runtime": 3.0391,
366
+ "eval_samples_per_second": 32.904,
367
+ "eval_steps_per_second": 4.278,
368
+ "step": 600
369
+ },
370
+ {
371
+ "best_epoch": 19,
372
+ "best_eval_accuracy": 0.69,
373
+ "epoch": 24.0,
374
+ "step": 600
375
+ },
376
+ {
377
+ "epoch": 25.0,
378
+ "eval_accuracy": 0.67,
379
+ "eval_loss": 0.4004598557949066,
380
+ "eval_runtime": 3.032,
381
+ "eval_samples_per_second": 32.982,
382
+ "eval_steps_per_second": 4.288,
383
+ "step": 625
384
+ },
385
+ {
386
+ "best_epoch": 19,
387
+ "best_eval_accuracy": 0.69,
388
+ "epoch": 25.0,
389
+ "step": 625
390
+ },
391
+ {
392
+ "epoch": 26.0,
393
+ "eval_accuracy": 0.7,
394
+ "eval_loss": 0.4123203158378601,
395
+ "eval_runtime": 3.0302,
396
+ "eval_samples_per_second": 33.001,
397
+ "eval_steps_per_second": 4.29,
398
+ "step": 650
399
+ },
400
+ {
401
+ "best_epoch": 25,
402
+ "best_eval_accuracy": 0.7,
403
+ "epoch": 26.0,
404
+ "step": 650
405
+ },
406
+ {
407
+ "epoch": 27.0,
408
+ "eval_accuracy": 0.69,
409
+ "eval_loss": 0.4341800808906555,
410
+ "eval_runtime": 3.0344,
411
+ "eval_samples_per_second": 32.955,
412
+ "eval_steps_per_second": 4.284,
413
+ "step": 675
414
+ },
415
+ {
416
+ "best_epoch": 25,
417
+ "best_eval_accuracy": 0.7,
418
+ "epoch": 27.0,
419
+ "step": 675
420
+ },
421
+ {
422
+ "epoch": 28.0,
423
+ "eval_accuracy": 0.67,
424
+ "eval_loss": 0.45509403944015503,
425
+ "eval_runtime": 3.0328,
426
+ "eval_samples_per_second": 32.973,
427
+ "eval_steps_per_second": 4.287,
428
+ "step": 700
429
+ },
430
+ {
431
+ "best_epoch": 25,
432
+ "best_eval_accuracy": 0.7,
433
+ "epoch": 28.0,
434
+ "step": 700
435
+ },
436
+ {
437
+ "epoch": 29.0,
438
+ "eval_accuracy": 0.69,
439
+ "eval_loss": 0.42222651839256287,
440
+ "eval_runtime": 3.0333,
441
+ "eval_samples_per_second": 32.967,
442
+ "eval_steps_per_second": 4.286,
443
+ "step": 725
444
+ },
445
+ {
446
+ "best_epoch": 25,
447
+ "best_eval_accuracy": 0.7,
448
+ "epoch": 29.0,
449
+ "step": 725
450
+ },
451
+ {
452
+ "epoch": 30.0,
453
+ "eval_accuracy": 0.71,
454
+ "eval_loss": 0.4225788116455078,
455
+ "eval_runtime": 3.0312,
456
+ "eval_samples_per_second": 32.99,
457
+ "eval_steps_per_second": 4.289,
458
+ "step": 750
459
+ },
460
+ {
461
+ "best_epoch": 29,
462
+ "best_eval_accuracy": 0.71,
463
+ "epoch": 30.0,
464
+ "step": 750
465
+ },
466
+ {
467
+ "epoch": 31.0,
468
+ "eval_accuracy": 0.69,
469
+ "eval_loss": 0.47024014592170715,
470
+ "eval_runtime": 3.0363,
471
+ "eval_samples_per_second": 32.934,
472
+ "eval_steps_per_second": 4.281,
473
+ "step": 775
474
+ },
475
+ {
476
+ "best_epoch": 29,
477
+ "best_eval_accuracy": 0.71,
478
+ "epoch": 31.0,
479
+ "step": 775
480
+ },
481
+ {
482
+ "epoch": 32.0,
483
+ "eval_accuracy": 0.7,
484
+ "eval_loss": 0.41000667214393616,
485
+ "eval_runtime": 3.0339,
486
+ "eval_samples_per_second": 32.961,
487
+ "eval_steps_per_second": 4.285,
488
+ "step": 800
489
+ },
490
+ {
491
+ "best_epoch": 29,
492
+ "best_eval_accuracy": 0.71,
493
+ "epoch": 32.0,
494
+ "step": 800
495
+ },
496
+ {
497
+ "epoch": 33.0,
498
+ "eval_accuracy": 0.69,
499
+ "eval_loss": 0.4317716658115387,
500
+ "eval_runtime": 3.0312,
501
+ "eval_samples_per_second": 32.99,
502
+ "eval_steps_per_second": 4.289,
503
+ "step": 825
504
+ },
505
+ {
506
+ "best_epoch": 29,
507
+ "best_eval_accuracy": 0.71,
508
+ "epoch": 33.0,
509
+ "step": 825
510
+ },
511
+ {
512
+ "epoch": 34.0,
513
+ "eval_accuracy": 0.71,
514
+ "eval_loss": 0.4446772634983063,
515
+ "eval_runtime": 3.0338,
516
+ "eval_samples_per_second": 32.962,
517
+ "eval_steps_per_second": 4.285,
518
+ "step": 850
519
+ },
520
+ {
521
+ "best_epoch": 29,
522
+ "best_eval_accuracy": 0.71,
523
+ "epoch": 34.0,
524
+ "step": 850
525
+ },
526
+ {
527
+ "epoch": 35.0,
528
+ "eval_accuracy": 0.72,
529
+ "eval_loss": 0.3881460130214691,
530
+ "eval_runtime": 3.0331,
531
+ "eval_samples_per_second": 32.969,
532
+ "eval_steps_per_second": 4.286,
533
+ "step": 875
534
+ },
535
+ {
536
+ "best_epoch": 34,
537
+ "best_eval_accuracy": 0.72,
538
+ "epoch": 35.0,
539
+ "step": 875
540
+ },
541
+ {
542
+ "epoch": 36.0,
543
+ "eval_accuracy": 0.69,
544
+ "eval_loss": 0.42335742712020874,
545
+ "eval_runtime": 3.0313,
546
+ "eval_samples_per_second": 32.989,
547
+ "eval_steps_per_second": 4.289,
548
+ "step": 900
549
+ },
550
+ {
551
+ "best_epoch": 34,
552
+ "best_eval_accuracy": 0.72,
553
+ "epoch": 36.0,
554
+ "step": 900
555
+ },
556
+ {
557
+ "epoch": 37.0,
558
+ "eval_accuracy": 0.69,
559
+ "eval_loss": 0.4869128465652466,
560
+ "eval_runtime": 3.0313,
561
+ "eval_samples_per_second": 32.989,
562
+ "eval_steps_per_second": 4.289,
563
+ "step": 925
564
+ },
565
+ {
566
+ "best_epoch": 34,
567
+ "best_eval_accuracy": 0.72,
568
+ "epoch": 37.0,
569
+ "step": 925
570
+ },
571
+ {
572
+ "epoch": 38.0,
573
+ "eval_accuracy": 0.71,
574
+ "eval_loss": 0.4352262020111084,
575
+ "eval_runtime": 3.0335,
576
+ "eval_samples_per_second": 32.965,
577
+ "eval_steps_per_second": 4.285,
578
+ "step": 950
579
+ },
580
+ {
581
+ "best_epoch": 34,
582
+ "best_eval_accuracy": 0.72,
583
+ "epoch": 38.0,
584
+ "step": 950
585
+ },
586
+ {
587
+ "epoch": 39.0,
588
+ "eval_accuracy": 0.71,
589
+ "eval_loss": 0.4464748501777649,
590
+ "eval_runtime": 3.0309,
591
+ "eval_samples_per_second": 32.993,
592
+ "eval_steps_per_second": 4.289,
593
+ "step": 975
594
+ },
595
+ {
596
+ "best_epoch": 34,
597
+ "best_eval_accuracy": 0.72,
598
+ "epoch": 39.0,
599
+ "step": 975
600
+ },
601
+ {
602
+ "epoch": 40.0,
603
+ "learning_rate": 0.01,
604
+ "loss": 0.5086,
605
+ "step": 1000
606
+ },
607
+ {
608
+ "epoch": 40.0,
609
+ "eval_accuracy": 0.7,
610
+ "eval_loss": 0.413508802652359,
611
+ "eval_runtime": 3.0326,
612
+ "eval_samples_per_second": 32.974,
613
+ "eval_steps_per_second": 4.287,
614
+ "step": 1000
615
+ },
616
+ {
617
+ "best_epoch": 34,
618
+ "best_eval_accuracy": 0.72,
619
+ "epoch": 40.0,
620
+ "step": 1000
621
+ },
622
+ {
623
+ "epoch": 41.0,
624
+ "eval_accuracy": 0.7,
625
+ "eval_loss": 0.4060712456703186,
626
+ "eval_runtime": 3.0312,
627
+ "eval_samples_per_second": 32.99,
628
+ "eval_steps_per_second": 4.289,
629
+ "step": 1025
630
+ },
631
+ {
632
+ "best_epoch": 34,
633
+ "best_eval_accuracy": 0.72,
634
+ "epoch": 41.0,
635
+ "step": 1025
636
+ },
637
+ {
638
+ "epoch": 42.0,
639
+ "eval_accuracy": 0.72,
640
+ "eval_loss": 0.44368666410446167,
641
+ "eval_runtime": 3.0306,
642
+ "eval_samples_per_second": 32.997,
643
+ "eval_steps_per_second": 4.29,
644
+ "step": 1050
645
+ },
646
+ {
647
+ "best_epoch": 34,
648
+ "best_eval_accuracy": 0.72,
649
+ "epoch": 42.0,
650
+ "step": 1050
651
+ },
652
+ {
653
+ "epoch": 43.0,
654
+ "eval_accuracy": 0.72,
655
+ "eval_loss": 0.4461361765861511,
656
+ "eval_runtime": 3.0306,
657
+ "eval_samples_per_second": 32.997,
658
+ "eval_steps_per_second": 4.29,
659
+ "step": 1075
660
+ },
661
+ {
662
+ "best_epoch": 34,
663
+ "best_eval_accuracy": 0.72,
664
+ "epoch": 43.0,
665
+ "step": 1075
666
+ },
667
+ {
668
+ "epoch": 44.0,
669
+ "eval_accuracy": 0.69,
670
+ "eval_loss": 0.41435202956199646,
671
+ "eval_runtime": 3.0303,
672
+ "eval_samples_per_second": 33.0,
673
+ "eval_steps_per_second": 4.29,
674
+ "step": 1100
675
+ },
676
+ {
677
+ "best_epoch": 34,
678
+ "best_eval_accuracy": 0.72,
679
+ "epoch": 44.0,
680
+ "step": 1100
681
+ },
682
+ {
683
+ "epoch": 45.0,
684
+ "eval_accuracy": 0.71,
685
+ "eval_loss": 0.3973015248775482,
686
+ "eval_runtime": 3.0301,
687
+ "eval_samples_per_second": 33.002,
688
+ "eval_steps_per_second": 4.29,
689
+ "step": 1125
690
+ },
691
+ {
692
+ "best_epoch": 34,
693
+ "best_eval_accuracy": 0.72,
694
+ "epoch": 45.0,
695
+ "step": 1125
696
+ },
697
+ {
698
+ "epoch": 46.0,
699
+ "eval_accuracy": 0.73,
700
+ "eval_loss": 0.45110881328582764,
701
+ "eval_runtime": 3.0313,
702
+ "eval_samples_per_second": 32.989,
703
+ "eval_steps_per_second": 4.289,
704
+ "step": 1150
705
+ },
706
+ {
707
+ "best_epoch": 45,
708
+ "best_eval_accuracy": 0.73,
709
+ "epoch": 46.0,
710
+ "step": 1150
711
+ },
712
+ {
713
+ "epoch": 47.0,
714
+ "eval_accuracy": 0.71,
715
+ "eval_loss": 0.4273070991039276,
716
+ "eval_runtime": 3.0301,
717
+ "eval_samples_per_second": 33.002,
718
+ "eval_steps_per_second": 4.29,
719
+ "step": 1175
720
+ },
721
+ {
722
+ "best_epoch": 45,
723
+ "best_eval_accuracy": 0.73,
724
+ "epoch": 47.0,
725
+ "step": 1175
726
+ },
727
+ {
728
+ "epoch": 48.0,
729
+ "eval_accuracy": 0.71,
730
+ "eval_loss": 0.4100435972213745,
731
+ "eval_runtime": 3.0299,
732
+ "eval_samples_per_second": 33.004,
733
+ "eval_steps_per_second": 4.291,
734
+ "step": 1200
735
+ },
736
+ {
737
+ "best_epoch": 45,
738
+ "best_eval_accuracy": 0.73,
739
+ "epoch": 48.0,
740
+ "step": 1200
741
+ },
742
+ {
743
+ "epoch": 49.0,
744
+ "eval_accuracy": 0.72,
745
+ "eval_loss": 0.4208601415157318,
746
+ "eval_runtime": 3.0294,
747
+ "eval_samples_per_second": 33.01,
748
+ "eval_steps_per_second": 4.291,
749
+ "step": 1225
750
+ },
751
+ {
752
+ "best_epoch": 45,
753
+ "best_eval_accuracy": 0.73,
754
+ "epoch": 49.0,
755
+ "step": 1225
756
+ },
757
+ {
758
+ "epoch": 50.0,
759
+ "eval_accuracy": 0.74,
760
+ "eval_loss": 0.41906824707984924,
761
+ "eval_runtime": 3.0315,
762
+ "eval_samples_per_second": 32.987,
763
+ "eval_steps_per_second": 4.288,
764
+ "step": 1250
765
+ },
766
+ {
767
+ "best_epoch": 49,
768
+ "best_eval_accuracy": 0.74,
769
+ "epoch": 50.0,
770
+ "step": 1250
771
+ },
772
+ {
773
+ "epoch": 51.0,
774
+ "eval_accuracy": 0.74,
775
+ "eval_loss": 0.40232476592063904,
776
+ "eval_runtime": 3.0299,
777
+ "eval_samples_per_second": 33.004,
778
+ "eval_steps_per_second": 4.291,
779
+ "step": 1275
780
+ },
781
+ {
782
+ "best_epoch": 49,
783
+ "best_eval_accuracy": 0.74,
784
+ "epoch": 51.0,
785
+ "step": 1275
786
+ },
787
+ {
788
+ "epoch": 52.0,
789
+ "eval_accuracy": 0.72,
790
+ "eval_loss": 0.40375789999961853,
791
+ "eval_runtime": 3.0318,
792
+ "eval_samples_per_second": 32.984,
793
+ "eval_steps_per_second": 4.288,
794
+ "step": 1300
795
+ },
796
+ {
797
+ "best_epoch": 49,
798
+ "best_eval_accuracy": 0.74,
799
+ "epoch": 52.0,
800
+ "step": 1300
801
+ },
802
+ {
803
+ "epoch": 53.0,
804
+ "eval_accuracy": 0.73,
805
+ "eval_loss": 0.41481122374534607,
806
+ "eval_runtime": 3.0329,
807
+ "eval_samples_per_second": 32.972,
808
+ "eval_steps_per_second": 4.286,
809
+ "step": 1325
810
+ },
811
+ {
812
+ "best_epoch": 49,
813
+ "best_eval_accuracy": 0.74,
814
+ "epoch": 53.0,
815
+ "step": 1325
816
+ },
817
+ {
818
+ "epoch": 54.0,
819
+ "eval_accuracy": 0.72,
820
+ "eval_loss": 0.42626023292541504,
821
+ "eval_runtime": 3.0313,
822
+ "eval_samples_per_second": 32.989,
823
+ "eval_steps_per_second": 4.289,
824
+ "step": 1350
825
+ },
826
+ {
827
+ "best_epoch": 49,
828
+ "best_eval_accuracy": 0.74,
829
+ "epoch": 54.0,
830
+ "step": 1350
831
+ },
832
+ {
833
+ "epoch": 55.0,
834
+ "eval_accuracy": 0.73,
835
+ "eval_loss": 0.43311816453933716,
836
+ "eval_runtime": 3.035,
837
+ "eval_samples_per_second": 32.949,
838
+ "eval_steps_per_second": 4.283,
839
+ "step": 1375
840
+ },
841
+ {
842
+ "best_epoch": 49,
843
+ "best_eval_accuracy": 0.74,
844
+ "epoch": 55.0,
845
+ "step": 1375
846
+ },
847
+ {
848
+ "epoch": 56.0,
849
+ "eval_accuracy": 0.71,
850
+ "eval_loss": 0.4372536838054657,
851
+ "eval_runtime": 3.0311,
852
+ "eval_samples_per_second": 32.991,
853
+ "eval_steps_per_second": 4.289,
854
+ "step": 1400
855
+ },
856
+ {
857
+ "best_epoch": 49,
858
+ "best_eval_accuracy": 0.74,
859
+ "epoch": 56.0,
860
+ "step": 1400
861
+ },
862
+ {
863
+ "epoch": 57.0,
864
+ "eval_accuracy": 0.72,
865
+ "eval_loss": 0.40811866521835327,
866
+ "eval_runtime": 3.0315,
867
+ "eval_samples_per_second": 32.987,
868
+ "eval_steps_per_second": 4.288,
869
+ "step": 1425
870
+ },
871
+ {
872
+ "best_epoch": 49,
873
+ "best_eval_accuracy": 0.74,
874
+ "epoch": 57.0,
875
+ "step": 1425
876
+ },
877
+ {
878
+ "epoch": 58.0,
879
+ "eval_accuracy": 0.71,
880
+ "eval_loss": 0.40780752897262573,
881
+ "eval_runtime": 3.0299,
882
+ "eval_samples_per_second": 33.004,
883
+ "eval_steps_per_second": 4.291,
884
+ "step": 1450
885
+ },
886
+ {
887
+ "best_epoch": 49,
888
+ "best_eval_accuracy": 0.74,
889
+ "epoch": 58.0,
890
+ "step": 1450
891
+ },
892
+ {
893
+ "epoch": 59.0,
894
+ "eval_accuracy": 0.72,
895
+ "eval_loss": 0.42498713731765747,
896
+ "eval_runtime": 3.0363,
897
+ "eval_samples_per_second": 32.935,
898
+ "eval_steps_per_second": 4.282,
899
+ "step": 1475
900
+ },
901
+ {
902
+ "best_epoch": 49,
903
+ "best_eval_accuracy": 0.74,
904
+ "epoch": 59.0,
905
+ "step": 1475
906
+ },
907
+ {
908
+ "epoch": 60.0,
909
+ "learning_rate": 0.005,
910
+ "loss": 0.4268,
911
+ "step": 1500
912
+ },
913
+ {
914
+ "epoch": 60.0,
915
+ "eval_accuracy": 0.7,
916
+ "eval_loss": 0.4223678708076477,
917
+ "eval_runtime": 3.0288,
918
+ "eval_samples_per_second": 33.016,
919
+ "eval_steps_per_second": 4.292,
920
+ "step": 1500
921
+ },
922
+ {
923
+ "best_epoch": 49,
924
+ "best_eval_accuracy": 0.74,
925
+ "epoch": 60.0,
926
+ "step": 1500
927
+ },
928
+ {
929
+ "epoch": 61.0,
930
+ "eval_accuracy": 0.7,
931
+ "eval_loss": 0.4255259335041046,
932
+ "eval_runtime": 3.0305,
933
+ "eval_samples_per_second": 32.997,
934
+ "eval_steps_per_second": 4.29,
935
+ "step": 1525
936
+ },
937
+ {
938
+ "best_epoch": 49,
939
+ "best_eval_accuracy": 0.74,
940
+ "epoch": 61.0,
941
+ "step": 1525
942
+ },
943
+ {
944
+ "epoch": 62.0,
945
+ "eval_accuracy": 0.72,
946
+ "eval_loss": 0.41144871711730957,
947
+ "eval_runtime": 3.0322,
948
+ "eval_samples_per_second": 32.98,
949
+ "eval_steps_per_second": 4.287,
950
+ "step": 1550
951
+ },
952
+ {
953
+ "best_epoch": 49,
954
+ "best_eval_accuracy": 0.74,
955
+ "epoch": 62.0,
956
+ "step": 1550
957
+ },
958
+ {
959
+ "epoch": 63.0,
960
+ "eval_accuracy": 0.72,
961
+ "eval_loss": 0.42656296491622925,
962
+ "eval_runtime": 3.0304,
963
+ "eval_samples_per_second": 32.999,
964
+ "eval_steps_per_second": 4.29,
965
+ "step": 1575
966
+ },
967
+ {
968
+ "best_epoch": 49,
969
+ "best_eval_accuracy": 0.74,
970
+ "epoch": 63.0,
971
+ "step": 1575
972
+ },
973
+ {
974
+ "epoch": 64.0,
975
+ "eval_accuracy": 0.72,
976
+ "eval_loss": 0.4097367227077484,
977
+ "eval_runtime": 3.03,
978
+ "eval_samples_per_second": 33.004,
979
+ "eval_steps_per_second": 4.29,
980
+ "step": 1600
981
+ },
982
+ {
983
+ "best_epoch": 49,
984
+ "best_eval_accuracy": 0.74,
985
+ "epoch": 64.0,
986
+ "step": 1600
987
+ },
988
+ {
989
+ "epoch": 65.0,
990
+ "eval_accuracy": 0.72,
991
+ "eval_loss": 0.40528982877731323,
992
+ "eval_runtime": 3.0318,
993
+ "eval_samples_per_second": 32.983,
994
+ "eval_steps_per_second": 4.288,
995
+ "step": 1625
996
+ },
997
+ {
998
+ "best_epoch": 49,
999
+ "best_eval_accuracy": 0.74,
1000
+ "epoch": 65.0,
1001
+ "step": 1625
1002
+ },
1003
+ {
1004
+ "epoch": 66.0,
1005
+ "eval_accuracy": 0.71,
1006
+ "eval_loss": 0.4050586223602295,
1007
+ "eval_runtime": 3.0289,
1008
+ "eval_samples_per_second": 33.015,
1009
+ "eval_steps_per_second": 4.292,
1010
+ "step": 1650
1011
+ },
1012
+ {
1013
+ "best_epoch": 49,
1014
+ "best_eval_accuracy": 0.74,
1015
+ "epoch": 66.0,
1016
+ "step": 1650
1017
+ },
1018
+ {
1019
+ "epoch": 67.0,
1020
+ "eval_accuracy": 0.73,
1021
+ "eval_loss": 0.41345199942588806,
1022
+ "eval_runtime": 3.031,
1023
+ "eval_samples_per_second": 32.993,
1024
+ "eval_steps_per_second": 4.289,
1025
+ "step": 1675
1026
+ },
1027
+ {
1028
+ "best_epoch": 49,
1029
+ "best_eval_accuracy": 0.74,
1030
+ "epoch": 67.0,
1031
+ "step": 1675
1032
+ },
1033
+ {
1034
+ "epoch": 68.0,
1035
+ "eval_accuracy": 0.74,
1036
+ "eval_loss": 0.3958780765533447,
1037
+ "eval_runtime": 3.0359,
1038
+ "eval_samples_per_second": 32.939,
1039
+ "eval_steps_per_second": 4.282,
1040
+ "step": 1700
1041
+ },
1042
+ {
1043
+ "best_epoch": 49,
1044
+ "best_eval_accuracy": 0.74,
1045
+ "epoch": 68.0,
1046
+ "step": 1700
1047
+ },
1048
+ {
1049
+ "epoch": 69.0,
1050
+ "eval_accuracy": 0.72,
1051
+ "eval_loss": 0.4162384867668152,
1052
+ "eval_runtime": 3.0305,
1053
+ "eval_samples_per_second": 32.998,
1054
+ "eval_steps_per_second": 4.29,
1055
+ "step": 1725
1056
+ },
1057
+ {
1058
+ "best_epoch": 49,
1059
+ "best_eval_accuracy": 0.74,
1060
+ "epoch": 69.0,
1061
+ "step": 1725
1062
+ },
1063
+ {
1064
+ "epoch": 70.0,
1065
+ "eval_accuracy": 0.73,
1066
+ "eval_loss": 0.4060596227645874,
1067
+ "eval_runtime": 3.0315,
1068
+ "eval_samples_per_second": 32.986,
1069
+ "eval_steps_per_second": 4.288,
1070
+ "step": 1750
1071
+ },
1072
+ {
1073
+ "best_epoch": 49,
1074
+ "best_eval_accuracy": 0.74,
1075
+ "epoch": 70.0,
1076
+ "step": 1750
1077
+ },
1078
+ {
1079
+ "epoch": 71.0,
1080
+ "eval_accuracy": 0.71,
1081
+ "eval_loss": 0.4015595614910126,
1082
+ "eval_runtime": 3.03,
1083
+ "eval_samples_per_second": 33.003,
1084
+ "eval_steps_per_second": 4.29,
1085
+ "step": 1775
1086
+ },
1087
+ {
1088
+ "best_epoch": 49,
1089
+ "best_eval_accuracy": 0.74,
1090
+ "epoch": 71.0,
1091
+ "step": 1775
1092
+ },
1093
+ {
1094
+ "epoch": 72.0,
1095
+ "eval_accuracy": 0.71,
1096
+ "eval_loss": 0.4193567931652069,
1097
+ "eval_runtime": 3.0293,
1098
+ "eval_samples_per_second": 33.011,
1099
+ "eval_steps_per_second": 4.291,
1100
+ "step": 1800
1101
+ },
1102
+ {
1103
+ "best_epoch": 49,
1104
+ "best_eval_accuracy": 0.74,
1105
+ "epoch": 72.0,
1106
+ "step": 1800
1107
+ },
1108
+ {
1109
+ "epoch": 73.0,
1110
+ "eval_accuracy": 0.72,
1111
+ "eval_loss": 0.4097726345062256,
1112
+ "eval_runtime": 3.0293,
1113
+ "eval_samples_per_second": 33.011,
1114
+ "eval_steps_per_second": 4.291,
1115
+ "step": 1825
1116
+ },
1117
+ {
1118
+ "best_epoch": 49,
1119
+ "best_eval_accuracy": 0.74,
1120
+ "epoch": 73.0,
1121
+ "step": 1825
1122
+ },
1123
+ {
1124
+ "epoch": 74.0,
1125
+ "eval_accuracy": 0.71,
1126
+ "eval_loss": 0.41785815358161926,
1127
+ "eval_runtime": 3.0313,
1128
+ "eval_samples_per_second": 32.989,
1129
+ "eval_steps_per_second": 4.289,
1130
+ "step": 1850
1131
+ },
1132
+ {
1133
+ "best_epoch": 49,
1134
+ "best_eval_accuracy": 0.74,
1135
+ "epoch": 74.0,
1136
+ "step": 1850
1137
+ },
1138
+ {
1139
+ "epoch": 75.0,
1140
+ "eval_accuracy": 0.71,
1141
+ "eval_loss": 0.41046372056007385,
1142
+ "eval_runtime": 3.0316,
1143
+ "eval_samples_per_second": 32.986,
1144
+ "eval_steps_per_second": 4.288,
1145
+ "step": 1875
1146
+ },
1147
+ {
1148
+ "best_epoch": 49,
1149
+ "best_eval_accuracy": 0.74,
1150
+ "epoch": 75.0,
1151
+ "step": 1875
1152
+ },
1153
+ {
1154
+ "epoch": 76.0,
1155
+ "eval_accuracy": 0.72,
1156
+ "eval_loss": 0.413968950510025,
1157
+ "eval_runtime": 3.0295,
1158
+ "eval_samples_per_second": 33.008,
1159
+ "eval_steps_per_second": 4.291,
1160
+ "step": 1900
1161
+ },
1162
+ {
1163
+ "best_epoch": 49,
1164
+ "best_eval_accuracy": 0.74,
1165
+ "epoch": 76.0,
1166
+ "step": 1900
1167
+ },
1168
+ {
1169
+ "epoch": 77.0,
1170
+ "eval_accuracy": 0.73,
1171
+ "eval_loss": 0.4080766439437866,
1172
+ "eval_runtime": 3.0307,
1173
+ "eval_samples_per_second": 32.996,
1174
+ "eval_steps_per_second": 4.289,
1175
+ "step": 1925
1176
+ },
1177
+ {
1178
+ "best_epoch": 49,
1179
+ "best_eval_accuracy": 0.74,
1180
+ "epoch": 77.0,
1181
+ "step": 1925
1182
+ },
1183
+ {
1184
+ "epoch": 78.0,
1185
+ "eval_accuracy": 0.73,
1186
+ "eval_loss": 0.40437057614326477,
1187
+ "eval_runtime": 3.0312,
1188
+ "eval_samples_per_second": 32.99,
1189
+ "eval_steps_per_second": 4.289,
1190
+ "step": 1950
1191
+ },
1192
+ {
1193
+ "best_epoch": 49,
1194
+ "best_eval_accuracy": 0.74,
1195
+ "epoch": 78.0,
1196
+ "step": 1950
1197
+ },
1198
+ {
1199
+ "epoch": 79.0,
1200
+ "eval_accuracy": 0.72,
1201
+ "eval_loss": 0.3996027410030365,
1202
+ "eval_runtime": 3.033,
1203
+ "eval_samples_per_second": 32.971,
1204
+ "eval_steps_per_second": 4.286,
1205
+ "step": 1975
1206
+ },
1207
+ {
1208
+ "best_epoch": 49,
1209
+ "best_eval_accuracy": 0.74,
1210
+ "epoch": 79.0,
1211
+ "step": 1975
1212
+ },
1213
+ {
1214
+ "epoch": 80.0,
1215
+ "learning_rate": 0.0,
1216
+ "loss": 0.3915,
1217
+ "step": 2000
1218
+ },
1219
+ {
1220
+ "epoch": 80.0,
1221
+ "eval_accuracy": 0.72,
1222
+ "eval_loss": 0.40144088864326477,
1223
+ "eval_runtime": 3.0353,
1224
+ "eval_samples_per_second": 32.946,
1225
+ "eval_steps_per_second": 4.283,
1226
+ "step": 2000
1227
+ },
1228
+ {
1229
+ "best_epoch": 49,
1230
+ "best_eval_accuracy": 0.74,
1231
+ "epoch": 80.0,
1232
+ "step": 2000
1233
+ },
1234
+ {
1235
+ "epoch": 80.0,
1236
+ "step": 2000,
1237
+ "total_flos": 2.9821702864896e+16,
1238
+ "train_loss": 0.5144821853637696,
1239
+ "train_runtime": 1625.7247,
1240
+ "train_samples_per_second": 19.684,
1241
+ "train_steps_per_second": 1.23
1242
+ }
1243
+ ],
1244
+ "max_steps": 2000,
1245
+ "num_train_epochs": 80,
1246
+ "total_flos": 2.9821702864896e+16,
1247
+ "trial_name": null,
1248
+ "trial_params": null
1249
+ }