pavitemple commited on
Commit
25ac454
·
1 Parent(s): 3e357fb

End of training

Browse files
Files changed (3) hide show
  1. all_results.json +8 -0
  2. test_results.json +8 -0
  3. trainer_state.json +424 -0
all_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 29.03,
3
+ "eval_accuracy": 0.9032258064516129,
4
+ "eval_loss": 0.4718249440193176,
5
+ "eval_runtime": 80.7097,
6
+ "eval_samples_per_second": 0.384,
7
+ "eval_steps_per_second": 0.05
8
+ }
test_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 29.03,
3
+ "eval_accuracy": 0.9032258064516129,
4
+ "eval_loss": 0.4718249440193176,
5
+ "eval_runtime": 80.7097,
6
+ "eval_samples_per_second": 0.384,
7
+ "eval_steps_per_second": 0.05
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,424 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.92,
3
+ "best_model_checkpoint": "finetuned-Accident-Video-subsetv2/checkpoint-150",
4
+ "epoch": 29.033333333333335,
5
+ "eval_steps": 500,
6
+ "global_step": 180,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.03,
13
+ "eval_accuracy": 0.08,
14
+ "eval_loss": 1.6174906492233276,
15
+ "eval_runtime": 74.2855,
16
+ "eval_samples_per_second": 0.337,
17
+ "eval_steps_per_second": 0.054,
18
+ "step": 6
19
+ },
20
+ {
21
+ "epoch": 1.02,
22
+ "learning_rate": 2.777777777777778e-05,
23
+ "loss": 1.5831,
24
+ "step": 10
25
+ },
26
+ {
27
+ "epoch": 1.03,
28
+ "eval_accuracy": 0.52,
29
+ "eval_loss": 1.372893214225769,
30
+ "eval_runtime": 78.0357,
31
+ "eval_samples_per_second": 0.32,
32
+ "eval_steps_per_second": 0.051,
33
+ "step": 12
34
+ },
35
+ {
36
+ "epoch": 2.03,
37
+ "eval_accuracy": 0.52,
38
+ "eval_loss": 1.1953624486923218,
39
+ "eval_runtime": 78.5221,
40
+ "eval_samples_per_second": 0.318,
41
+ "eval_steps_per_second": 0.051,
42
+ "step": 18
43
+ },
44
+ {
45
+ "epoch": 3.01,
46
+ "learning_rate": 4.938271604938271e-05,
47
+ "loss": 1.3037,
48
+ "step": 20
49
+ },
50
+ {
51
+ "epoch": 3.03,
52
+ "eval_accuracy": 0.48,
53
+ "eval_loss": 1.176283597946167,
54
+ "eval_runtime": 74.6832,
55
+ "eval_samples_per_second": 0.335,
56
+ "eval_steps_per_second": 0.054,
57
+ "step": 24
58
+ },
59
+ {
60
+ "epoch": 4.03,
61
+ "learning_rate": 4.62962962962963e-05,
62
+ "loss": 1.1833,
63
+ "step": 30
64
+ },
65
+ {
66
+ "epoch": 4.03,
67
+ "eval_accuracy": 0.52,
68
+ "eval_loss": 1.1420832872390747,
69
+ "eval_runtime": 59.4182,
70
+ "eval_samples_per_second": 0.421,
71
+ "eval_steps_per_second": 0.067,
72
+ "step": 30
73
+ },
74
+ {
75
+ "epoch": 5.03,
76
+ "eval_accuracy": 0.52,
77
+ "eval_loss": 1.1220537424087524,
78
+ "eval_runtime": 59.6408,
79
+ "eval_samples_per_second": 0.419,
80
+ "eval_steps_per_second": 0.067,
81
+ "step": 36
82
+ },
83
+ {
84
+ "epoch": 6.02,
85
+ "learning_rate": 4.3209876543209875e-05,
86
+ "loss": 1.125,
87
+ "step": 40
88
+ },
89
+ {
90
+ "epoch": 6.03,
91
+ "eval_accuracy": 0.68,
92
+ "eval_loss": 1.0468703508377075,
93
+ "eval_runtime": 59.1225,
94
+ "eval_samples_per_second": 0.423,
95
+ "eval_steps_per_second": 0.068,
96
+ "step": 42
97
+ },
98
+ {
99
+ "epoch": 7.03,
100
+ "eval_accuracy": 0.6,
101
+ "eval_loss": 1.0552805662155151,
102
+ "eval_runtime": 61.9333,
103
+ "eval_samples_per_second": 0.404,
104
+ "eval_steps_per_second": 0.065,
105
+ "step": 48
106
+ },
107
+ {
108
+ "epoch": 8.01,
109
+ "learning_rate": 4.012345679012346e-05,
110
+ "loss": 1.0967,
111
+ "step": 50
112
+ },
113
+ {
114
+ "epoch": 8.03,
115
+ "eval_accuracy": 0.72,
116
+ "eval_loss": 0.9645193219184875,
117
+ "eval_runtime": 59.8316,
118
+ "eval_samples_per_second": 0.418,
119
+ "eval_steps_per_second": 0.067,
120
+ "step": 54
121
+ },
122
+ {
123
+ "epoch": 9.03,
124
+ "learning_rate": 3.7037037037037037e-05,
125
+ "loss": 1.0025,
126
+ "step": 60
127
+ },
128
+ {
129
+ "epoch": 9.03,
130
+ "eval_accuracy": 0.6,
131
+ "eval_loss": 0.9435223340988159,
132
+ "eval_runtime": 60.4105,
133
+ "eval_samples_per_second": 0.414,
134
+ "eval_steps_per_second": 0.066,
135
+ "step": 60
136
+ },
137
+ {
138
+ "epoch": 10.03,
139
+ "eval_accuracy": 0.76,
140
+ "eval_loss": 0.8497534990310669,
141
+ "eval_runtime": 70.118,
142
+ "eval_samples_per_second": 0.357,
143
+ "eval_steps_per_second": 0.057,
144
+ "step": 66
145
+ },
146
+ {
147
+ "epoch": 11.02,
148
+ "learning_rate": 3.395061728395062e-05,
149
+ "loss": 0.9207,
150
+ "step": 70
151
+ },
152
+ {
153
+ "epoch": 11.03,
154
+ "eval_accuracy": 0.72,
155
+ "eval_loss": 0.8095592260360718,
156
+ "eval_runtime": 68.7785,
157
+ "eval_samples_per_second": 0.363,
158
+ "eval_steps_per_second": 0.058,
159
+ "step": 72
160
+ },
161
+ {
162
+ "epoch": 12.03,
163
+ "eval_accuracy": 0.68,
164
+ "eval_loss": 0.8176926374435425,
165
+ "eval_runtime": 68.3666,
166
+ "eval_samples_per_second": 0.366,
167
+ "eval_steps_per_second": 0.059,
168
+ "step": 78
169
+ },
170
+ {
171
+ "epoch": 13.01,
172
+ "learning_rate": 3.08641975308642e-05,
173
+ "loss": 0.7994,
174
+ "step": 80
175
+ },
176
+ {
177
+ "epoch": 13.03,
178
+ "eval_accuracy": 0.64,
179
+ "eval_loss": 0.7237111926078796,
180
+ "eval_runtime": 67.4729,
181
+ "eval_samples_per_second": 0.371,
182
+ "eval_steps_per_second": 0.059,
183
+ "step": 84
184
+ },
185
+ {
186
+ "epoch": 14.03,
187
+ "learning_rate": 2.777777777777778e-05,
188
+ "loss": 0.753,
189
+ "step": 90
190
+ },
191
+ {
192
+ "epoch": 14.03,
193
+ "eval_accuracy": 0.72,
194
+ "eval_loss": 0.6827337741851807,
195
+ "eval_runtime": 67.0259,
196
+ "eval_samples_per_second": 0.373,
197
+ "eval_steps_per_second": 0.06,
198
+ "step": 90
199
+ },
200
+ {
201
+ "epoch": 15.03,
202
+ "eval_accuracy": 0.88,
203
+ "eval_loss": 0.5696862936019897,
204
+ "eval_runtime": 69.113,
205
+ "eval_samples_per_second": 0.362,
206
+ "eval_steps_per_second": 0.058,
207
+ "step": 96
208
+ },
209
+ {
210
+ "epoch": 16.02,
211
+ "learning_rate": 2.4691358024691357e-05,
212
+ "loss": 0.6998,
213
+ "step": 100
214
+ },
215
+ {
216
+ "epoch": 16.03,
217
+ "eval_accuracy": 0.76,
218
+ "eval_loss": 0.7594789266586304,
219
+ "eval_runtime": 68.4666,
220
+ "eval_samples_per_second": 0.365,
221
+ "eval_steps_per_second": 0.058,
222
+ "step": 102
223
+ },
224
+ {
225
+ "epoch": 17.03,
226
+ "eval_accuracy": 0.88,
227
+ "eval_loss": 0.5631096363067627,
228
+ "eval_runtime": 68.5985,
229
+ "eval_samples_per_second": 0.364,
230
+ "eval_steps_per_second": 0.058,
231
+ "step": 108
232
+ },
233
+ {
234
+ "epoch": 18.01,
235
+ "learning_rate": 2.1604938271604937e-05,
236
+ "loss": 0.6564,
237
+ "step": 110
238
+ },
239
+ {
240
+ "epoch": 18.03,
241
+ "eval_accuracy": 0.8,
242
+ "eval_loss": 0.5146105885505676,
243
+ "eval_runtime": 64.7403,
244
+ "eval_samples_per_second": 0.386,
245
+ "eval_steps_per_second": 0.062,
246
+ "step": 114
247
+ },
248
+ {
249
+ "epoch": 19.03,
250
+ "learning_rate": 1.8518518518518518e-05,
251
+ "loss": 0.5929,
252
+ "step": 120
253
+ },
254
+ {
255
+ "epoch": 19.03,
256
+ "eval_accuracy": 0.6,
257
+ "eval_loss": 0.7956848740577698,
258
+ "eval_runtime": 79.3384,
259
+ "eval_samples_per_second": 0.315,
260
+ "eval_steps_per_second": 0.05,
261
+ "step": 120
262
+ },
263
+ {
264
+ "epoch": 20.03,
265
+ "eval_accuracy": 0.88,
266
+ "eval_loss": 0.5254407525062561,
267
+ "eval_runtime": 63.7552,
268
+ "eval_samples_per_second": 0.392,
269
+ "eval_steps_per_second": 0.063,
270
+ "step": 126
271
+ },
272
+ {
273
+ "epoch": 21.02,
274
+ "learning_rate": 1.54320987654321e-05,
275
+ "loss": 0.5148,
276
+ "step": 130
277
+ },
278
+ {
279
+ "epoch": 21.03,
280
+ "eval_accuracy": 0.88,
281
+ "eval_loss": 0.4102124273777008,
282
+ "eval_runtime": 65.1408,
283
+ "eval_samples_per_second": 0.384,
284
+ "eval_steps_per_second": 0.061,
285
+ "step": 132
286
+ },
287
+ {
288
+ "epoch": 22.03,
289
+ "eval_accuracy": 0.88,
290
+ "eval_loss": 0.4555157721042633,
291
+ "eval_runtime": 65.856,
292
+ "eval_samples_per_second": 0.38,
293
+ "eval_steps_per_second": 0.061,
294
+ "step": 138
295
+ },
296
+ {
297
+ "epoch": 23.01,
298
+ "learning_rate": 1.2345679012345678e-05,
299
+ "loss": 0.3351,
300
+ "step": 140
301
+ },
302
+ {
303
+ "epoch": 23.03,
304
+ "eval_accuracy": 0.72,
305
+ "eval_loss": 0.773196280002594,
306
+ "eval_runtime": 70.5357,
307
+ "eval_samples_per_second": 0.354,
308
+ "eval_steps_per_second": 0.057,
309
+ "step": 144
310
+ },
311
+ {
312
+ "epoch": 24.03,
313
+ "learning_rate": 9.259259259259259e-06,
314
+ "loss": 0.4489,
315
+ "step": 150
316
+ },
317
+ {
318
+ "epoch": 24.03,
319
+ "eval_accuracy": 0.92,
320
+ "eval_loss": 0.4464223086833954,
321
+ "eval_runtime": 67.1272,
322
+ "eval_samples_per_second": 0.372,
323
+ "eval_steps_per_second": 0.06,
324
+ "step": 150
325
+ },
326
+ {
327
+ "epoch": 25.03,
328
+ "eval_accuracy": 0.92,
329
+ "eval_loss": 0.3872789144515991,
330
+ "eval_runtime": 67.9637,
331
+ "eval_samples_per_second": 0.368,
332
+ "eval_steps_per_second": 0.059,
333
+ "step": 156
334
+ },
335
+ {
336
+ "epoch": 26.02,
337
+ "learning_rate": 6.172839506172839e-06,
338
+ "loss": 0.3652,
339
+ "step": 160
340
+ },
341
+ {
342
+ "epoch": 26.03,
343
+ "eval_accuracy": 0.92,
344
+ "eval_loss": 0.3817351460456848,
345
+ "eval_runtime": 67.6243,
346
+ "eval_samples_per_second": 0.37,
347
+ "eval_steps_per_second": 0.059,
348
+ "step": 162
349
+ },
350
+ {
351
+ "epoch": 27.03,
352
+ "eval_accuracy": 0.92,
353
+ "eval_loss": 0.39160746335983276,
354
+ "eval_runtime": 67.5822,
355
+ "eval_samples_per_second": 0.37,
356
+ "eval_steps_per_second": 0.059,
357
+ "step": 168
358
+ },
359
+ {
360
+ "epoch": 28.01,
361
+ "learning_rate": 3.0864197530864196e-06,
362
+ "loss": 0.2532,
363
+ "step": 170
364
+ },
365
+ {
366
+ "epoch": 28.03,
367
+ "eval_accuracy": 0.92,
368
+ "eval_loss": 0.39454135298728943,
369
+ "eval_runtime": 63.3977,
370
+ "eval_samples_per_second": 0.394,
371
+ "eval_steps_per_second": 0.063,
372
+ "step": 174
373
+ },
374
+ {
375
+ "epoch": 29.03,
376
+ "learning_rate": 0.0,
377
+ "loss": 0.1712,
378
+ "step": 180
379
+ },
380
+ {
381
+ "epoch": 29.03,
382
+ "eval_accuracy": 0.92,
383
+ "eval_loss": 0.39993801712989807,
384
+ "eval_runtime": 64.7389,
385
+ "eval_samples_per_second": 0.386,
386
+ "eval_steps_per_second": 0.062,
387
+ "step": 180
388
+ },
389
+ {
390
+ "epoch": 29.03,
391
+ "step": 180,
392
+ "total_flos": 1.682232201732096e+18,
393
+ "train_loss": 0.7669431573814816,
394
+ "train_runtime": 7558.2473,
395
+ "train_samples_per_second": 0.191,
396
+ "train_steps_per_second": 0.024
397
+ },
398
+ {
399
+ "epoch": 29.03,
400
+ "eval_accuracy": 0.9032258064516129,
401
+ "eval_loss": 0.47182491421699524,
402
+ "eval_runtime": 77.6157,
403
+ "eval_samples_per_second": 0.399,
404
+ "eval_steps_per_second": 0.052,
405
+ "step": 180
406
+ },
407
+ {
408
+ "epoch": 29.03,
409
+ "eval_accuracy": 0.9032258064516129,
410
+ "eval_loss": 0.4718249440193176,
411
+ "eval_runtime": 80.7097,
412
+ "eval_samples_per_second": 0.384,
413
+ "eval_steps_per_second": 0.05,
414
+ "step": 180
415
+ }
416
+ ],
417
+ "logging_steps": 10,
418
+ "max_steps": 180,
419
+ "num_train_epochs": 9223372036854775807,
420
+ "save_steps": 500,
421
+ "total_flos": 1.682232201732096e+18,
422
+ "trial_name": null,
423
+ "trial_params": null
424
+ }