c14kevincardenas commited on
Commit
f6012f4
·
verified ·
1 Parent(s): 7a18482

End of training

Browse files
Files changed (5) hide show
  1. README.md +6 -4
  2. all_results.json +13 -0
  3. eval_results.json +8 -0
  4. train_results.json +8 -0
  5. trainer_state.json +726 -0
README.md CHANGED
@@ -1,8 +1,10 @@
1
  ---
2
  library_name: transformers
3
  license: apache-2.0
4
- base_model: facebook/deit-small-distilled-patch16-224
5
  tags:
 
 
6
  - generated_from_trainer
7
  metrics:
8
  - accuracy
@@ -16,10 +18,10 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  # deit-small-distilled-patch16-224_alpha0.5_temp3.0
18
 
19
- This model is a fine-tuned version of [facebook/deit-small-distilled-patch16-224](https://huggingface.co/facebook/deit-small-distilled-patch16-224) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.7007
22
- - Accuracy: 0.7460
23
 
24
  ## Model description
25
 
 
1
  ---
2
  library_name: transformers
3
  license: apache-2.0
4
+ base_model: c14kevincardenas/ClimBEiTv2
5
  tags:
6
+ - knowledge_distillation
7
+ - vision
8
  - generated_from_trainer
9
  metrics:
10
  - accuracy
 
18
 
19
  # deit-small-distilled-patch16-224_alpha0.5_temp3.0
20
 
21
+ This model is a fine-tuned version of [c14kevincardenas/ClimBEiTv2](https://huggingface.co/c14kevincardenas/ClimBEiTv2) on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb_2 dataset.
22
  It achieves the following results on the evaluation set:
23
+ - Loss: 0.6994
24
+ - Accuracy: 0.7352
25
 
26
  ## Model description
27
 
all_results.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 20.0,
3
+ "eval_accuracy": 0.7351778656126482,
4
+ "eval_loss": 0.6994230151176453,
5
+ "eval_runtime": 31.0063,
6
+ "eval_samples_per_second": 32.639,
7
+ "eval_steps_per_second": 0.516,
8
+ "total_flos": 0.0,
9
+ "train_loss": 0.31672255052460563,
10
+ "train_runtime": 10514.5803,
11
+ "train_samples_per_second": 10.899,
12
+ "train_steps_per_second": 0.171
13
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 20.0,
3
+ "eval_accuracy": 0.7351778656126482,
4
+ "eval_loss": 0.6994230151176453,
5
+ "eval_runtime": 31.0063,
6
+ "eval_samples_per_second": 32.639,
7
+ "eval_steps_per_second": 0.516
8
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 20.0,
3
+ "total_flos": 0.0,
4
+ "train_loss": 0.31672255052460563,
5
+ "train_runtime": 10514.5803,
6
+ "train_samples_per_second": 10.899,
7
+ "train_steps_per_second": 0.171
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,726 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.6994230151176453,
3
+ "best_model_checkpoint": "./kd_results/facebook/deit-small-distilled-patch16-224_alpha0.5_temp3.0/checkpoint-1620",
4
+ "epoch": 20.0,
5
+ "eval_steps": 500,
6
+ "global_step": 1800,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.2777777777777778,
13
+ "grad_norm": 4.104116916656494,
14
+ "learning_rate": 6.944444444444445e-06,
15
+ "loss": 1.3245,
16
+ "step": 25
17
+ },
18
+ {
19
+ "epoch": 0.5555555555555556,
20
+ "grad_norm": 2.875807523727417,
21
+ "learning_rate": 1.388888888888889e-05,
22
+ "loss": 1.2378,
23
+ "step": 50
24
+ },
25
+ {
26
+ "epoch": 0.8333333333333334,
27
+ "grad_norm": 3.105369806289673,
28
+ "learning_rate": 2.0833333333333336e-05,
29
+ "loss": 1.2256,
30
+ "step": 75
31
+ },
32
+ {
33
+ "epoch": 1.0,
34
+ "eval_accuracy": 0.2707509881422925,
35
+ "eval_loss": 1.420397400856018,
36
+ "eval_runtime": 29.6146,
37
+ "eval_samples_per_second": 34.172,
38
+ "eval_steps_per_second": 0.54,
39
+ "step": 90
40
+ },
41
+ {
42
+ "epoch": 1.1111111111111112,
43
+ "grad_norm": 3.8822920322418213,
44
+ "learning_rate": 2.777777777777778e-05,
45
+ "loss": 1.1963,
46
+ "step": 100
47
+ },
48
+ {
49
+ "epoch": 1.3888888888888888,
50
+ "grad_norm": 3.240900754928589,
51
+ "learning_rate": 3.472222222222222e-05,
52
+ "loss": 1.1576,
53
+ "step": 125
54
+ },
55
+ {
56
+ "epoch": 1.6666666666666665,
57
+ "grad_norm": 3.8485963344573975,
58
+ "learning_rate": 4.166666666666667e-05,
59
+ "loss": 1.1301,
60
+ "step": 150
61
+ },
62
+ {
63
+ "epoch": 1.9444444444444444,
64
+ "grad_norm": 4.880220890045166,
65
+ "learning_rate": 4.8611111111111115e-05,
66
+ "loss": 1.0697,
67
+ "step": 175
68
+ },
69
+ {
70
+ "epoch": 2.0,
71
+ "eval_accuracy": 0.43774703557312256,
72
+ "eval_loss": 1.2418382167816162,
73
+ "eval_runtime": 30.2043,
74
+ "eval_samples_per_second": 33.505,
75
+ "eval_steps_per_second": 0.53,
76
+ "step": 180
77
+ },
78
+ {
79
+ "epoch": 2.2222222222222223,
80
+ "grad_norm": 8.320867538452148,
81
+ "learning_rate": 4.938271604938271e-05,
82
+ "loss": 0.9319,
83
+ "step": 200
84
+ },
85
+ {
86
+ "epoch": 2.5,
87
+ "grad_norm": 6.680625915527344,
88
+ "learning_rate": 4.8611111111111115e-05,
89
+ "loss": 0.8352,
90
+ "step": 225
91
+ },
92
+ {
93
+ "epoch": 2.7777777777777777,
94
+ "grad_norm": 4.338289260864258,
95
+ "learning_rate": 4.783950617283951e-05,
96
+ "loss": 0.745,
97
+ "step": 250
98
+ },
99
+ {
100
+ "epoch": 3.0,
101
+ "eval_accuracy": 0.6343873517786561,
102
+ "eval_loss": 0.8559542298316956,
103
+ "eval_runtime": 29.7805,
104
+ "eval_samples_per_second": 33.982,
105
+ "eval_steps_per_second": 0.537,
106
+ "step": 270
107
+ },
108
+ {
109
+ "epoch": 3.0555555555555554,
110
+ "grad_norm": 4.460063934326172,
111
+ "learning_rate": 4.70679012345679e-05,
112
+ "loss": 0.673,
113
+ "step": 275
114
+ },
115
+ {
116
+ "epoch": 3.3333333333333335,
117
+ "grad_norm": 3.742156982421875,
118
+ "learning_rate": 4.62962962962963e-05,
119
+ "loss": 0.5085,
120
+ "step": 300
121
+ },
122
+ {
123
+ "epoch": 3.611111111111111,
124
+ "grad_norm": 3.4500081539154053,
125
+ "learning_rate": 4.5524691358024696e-05,
126
+ "loss": 0.4901,
127
+ "step": 325
128
+ },
129
+ {
130
+ "epoch": 3.888888888888889,
131
+ "grad_norm": 3.695544958114624,
132
+ "learning_rate": 4.4753086419753084e-05,
133
+ "loss": 0.4822,
134
+ "step": 350
135
+ },
136
+ {
137
+ "epoch": 4.0,
138
+ "eval_accuracy": 0.7055335968379447,
139
+ "eval_loss": 0.7596805691719055,
140
+ "eval_runtime": 29.2784,
141
+ "eval_samples_per_second": 34.565,
142
+ "eval_steps_per_second": 0.546,
143
+ "step": 360
144
+ },
145
+ {
146
+ "epoch": 4.166666666666667,
147
+ "grad_norm": 3.700939655303955,
148
+ "learning_rate": 4.3981481481481486e-05,
149
+ "loss": 0.3773,
150
+ "step": 375
151
+ },
152
+ {
153
+ "epoch": 4.444444444444445,
154
+ "grad_norm": 4.073795795440674,
155
+ "learning_rate": 4.3209876543209875e-05,
156
+ "loss": 0.2994,
157
+ "step": 400
158
+ },
159
+ {
160
+ "epoch": 4.722222222222222,
161
+ "grad_norm": 3.6610753536224365,
162
+ "learning_rate": 4.243827160493827e-05,
163
+ "loss": 0.2977,
164
+ "step": 425
165
+ },
166
+ {
167
+ "epoch": 5.0,
168
+ "grad_norm": 4.114968776702881,
169
+ "learning_rate": 4.166666666666667e-05,
170
+ "loss": 0.296,
171
+ "step": 450
172
+ },
173
+ {
174
+ "epoch": 5.0,
175
+ "eval_accuracy": 0.7035573122529645,
176
+ "eval_loss": 0.7598134875297546,
177
+ "eval_runtime": 29.5479,
178
+ "eval_samples_per_second": 34.25,
179
+ "eval_steps_per_second": 0.541,
180
+ "step": 450
181
+ },
182
+ {
183
+ "epoch": 5.277777777777778,
184
+ "grad_norm": 2.291775941848755,
185
+ "learning_rate": 4.089506172839506e-05,
186
+ "loss": 0.2227,
187
+ "step": 475
188
+ },
189
+ {
190
+ "epoch": 5.555555555555555,
191
+ "grad_norm": 3.2807505130767822,
192
+ "learning_rate": 4.012345679012346e-05,
193
+ "loss": 0.2138,
194
+ "step": 500
195
+ },
196
+ {
197
+ "epoch": 5.833333333333333,
198
+ "grad_norm": 2.3632428646087646,
199
+ "learning_rate": 3.935185185185186e-05,
200
+ "loss": 0.2083,
201
+ "step": 525
202
+ },
203
+ {
204
+ "epoch": 6.0,
205
+ "eval_accuracy": 0.6867588932806324,
206
+ "eval_loss": 0.7894434332847595,
207
+ "eval_runtime": 29.3598,
208
+ "eval_samples_per_second": 34.469,
209
+ "eval_steps_per_second": 0.545,
210
+ "step": 540
211
+ },
212
+ {
213
+ "epoch": 6.111111111111111,
214
+ "grad_norm": 2.4907870292663574,
215
+ "learning_rate": 3.8580246913580246e-05,
216
+ "loss": 0.2149,
217
+ "step": 550
218
+ },
219
+ {
220
+ "epoch": 6.388888888888889,
221
+ "grad_norm": 3.0467641353607178,
222
+ "learning_rate": 3.780864197530865e-05,
223
+ "loss": 0.1969,
224
+ "step": 575
225
+ },
226
+ {
227
+ "epoch": 6.666666666666667,
228
+ "grad_norm": 1.9196618795394897,
229
+ "learning_rate": 3.7037037037037037e-05,
230
+ "loss": 0.1929,
231
+ "step": 600
232
+ },
233
+ {
234
+ "epoch": 6.944444444444445,
235
+ "grad_norm": 1.7986650466918945,
236
+ "learning_rate": 3.626543209876543e-05,
237
+ "loss": 0.1911,
238
+ "step": 625
239
+ },
240
+ {
241
+ "epoch": 7.0,
242
+ "eval_accuracy": 0.7134387351778656,
243
+ "eval_loss": 0.7575205564498901,
244
+ "eval_runtime": 29.6909,
245
+ "eval_samples_per_second": 34.084,
246
+ "eval_steps_per_second": 0.539,
247
+ "step": 630
248
+ },
249
+ {
250
+ "epoch": 7.222222222222222,
251
+ "grad_norm": 1.0618268251419067,
252
+ "learning_rate": 3.5493827160493834e-05,
253
+ "loss": 0.1792,
254
+ "step": 650
255
+ },
256
+ {
257
+ "epoch": 7.5,
258
+ "grad_norm": 1.4524767398834229,
259
+ "learning_rate": 3.472222222222222e-05,
260
+ "loss": 0.1837,
261
+ "step": 675
262
+ },
263
+ {
264
+ "epoch": 7.777777777777778,
265
+ "grad_norm": 1.0937868356704712,
266
+ "learning_rate": 3.395061728395062e-05,
267
+ "loss": 0.1783,
268
+ "step": 700
269
+ },
270
+ {
271
+ "epoch": 8.0,
272
+ "eval_accuracy": 0.7302371541501976,
273
+ "eval_loss": 0.7299193143844604,
274
+ "eval_runtime": 29.5938,
275
+ "eval_samples_per_second": 34.196,
276
+ "eval_steps_per_second": 0.541,
277
+ "step": 720
278
+ },
279
+ {
280
+ "epoch": 8.055555555555555,
281
+ "grad_norm": 1.8283872604370117,
282
+ "learning_rate": 3.317901234567901e-05,
283
+ "loss": 0.1841,
284
+ "step": 725
285
+ },
286
+ {
287
+ "epoch": 8.333333333333334,
288
+ "grad_norm": 1.2189834117889404,
289
+ "learning_rate": 3.240740740740741e-05,
290
+ "loss": 0.1708,
291
+ "step": 750
292
+ },
293
+ {
294
+ "epoch": 8.61111111111111,
295
+ "grad_norm": 1.5531915426254272,
296
+ "learning_rate": 3.16358024691358e-05,
297
+ "loss": 0.1736,
298
+ "step": 775
299
+ },
300
+ {
301
+ "epoch": 8.88888888888889,
302
+ "grad_norm": 1.221135139465332,
303
+ "learning_rate": 3.08641975308642e-05,
304
+ "loss": 0.1743,
305
+ "step": 800
306
+ },
307
+ {
308
+ "epoch": 9.0,
309
+ "eval_accuracy": 0.7183794466403162,
310
+ "eval_loss": 0.7375351786613464,
311
+ "eval_runtime": 29.5919,
312
+ "eval_samples_per_second": 34.199,
313
+ "eval_steps_per_second": 0.541,
314
+ "step": 810
315
+ },
316
+ {
317
+ "epoch": 9.166666666666666,
318
+ "grad_norm": 1.11898672580719,
319
+ "learning_rate": 3.0092592592592593e-05,
320
+ "loss": 0.1633,
321
+ "step": 825
322
+ },
323
+ {
324
+ "epoch": 9.444444444444445,
325
+ "grad_norm": 1.2036685943603516,
326
+ "learning_rate": 2.9320987654320992e-05,
327
+ "loss": 0.1658,
328
+ "step": 850
329
+ },
330
+ {
331
+ "epoch": 9.722222222222221,
332
+ "grad_norm": 1.015676736831665,
333
+ "learning_rate": 2.8549382716049384e-05,
334
+ "loss": 0.1648,
335
+ "step": 875
336
+ },
337
+ {
338
+ "epoch": 10.0,
339
+ "grad_norm": 1.6817418336868286,
340
+ "learning_rate": 2.777777777777778e-05,
341
+ "loss": 0.1645,
342
+ "step": 900
343
+ },
344
+ {
345
+ "epoch": 10.0,
346
+ "eval_accuracy": 0.7272727272727273,
347
+ "eval_loss": 0.7215421795845032,
348
+ "eval_runtime": 29.8084,
349
+ "eval_samples_per_second": 33.95,
350
+ "eval_steps_per_second": 0.537,
351
+ "step": 900
352
+ },
353
+ {
354
+ "epoch": 10.277777777777779,
355
+ "grad_norm": 0.8497186303138733,
356
+ "learning_rate": 2.700617283950617e-05,
357
+ "loss": 0.161,
358
+ "step": 925
359
+ },
360
+ {
361
+ "epoch": 10.555555555555555,
362
+ "grad_norm": 0.9341030120849609,
363
+ "learning_rate": 2.623456790123457e-05,
364
+ "loss": 0.1614,
365
+ "step": 950
366
+ },
367
+ {
368
+ "epoch": 10.833333333333334,
369
+ "grad_norm": 1.0269076824188232,
370
+ "learning_rate": 2.5462962962962965e-05,
371
+ "loss": 0.1548,
372
+ "step": 975
373
+ },
374
+ {
375
+ "epoch": 11.0,
376
+ "eval_accuracy": 0.741106719367589,
377
+ "eval_loss": 0.7100472450256348,
378
+ "eval_runtime": 30.2305,
379
+ "eval_samples_per_second": 33.476,
380
+ "eval_steps_per_second": 0.529,
381
+ "step": 990
382
+ },
383
+ {
384
+ "epoch": 11.11111111111111,
385
+ "grad_norm": 0.7800092697143555,
386
+ "learning_rate": 2.4691358024691357e-05,
387
+ "loss": 0.1509,
388
+ "step": 1000
389
+ },
390
+ {
391
+ "epoch": 11.38888888888889,
392
+ "grad_norm": 0.8492220044136047,
393
+ "learning_rate": 2.3919753086419755e-05,
394
+ "loss": 0.15,
395
+ "step": 1025
396
+ },
397
+ {
398
+ "epoch": 11.666666666666666,
399
+ "grad_norm": 0.7080733180046082,
400
+ "learning_rate": 2.314814814814815e-05,
401
+ "loss": 0.1572,
402
+ "step": 1050
403
+ },
404
+ {
405
+ "epoch": 11.944444444444445,
406
+ "grad_norm": 0.579961895942688,
407
+ "learning_rate": 2.2376543209876542e-05,
408
+ "loss": 0.1544,
409
+ "step": 1075
410
+ },
411
+ {
412
+ "epoch": 12.0,
413
+ "eval_accuracy": 0.7351778656126482,
414
+ "eval_loss": 0.7043272852897644,
415
+ "eval_runtime": 29.3995,
416
+ "eval_samples_per_second": 34.422,
417
+ "eval_steps_per_second": 0.544,
418
+ "step": 1080
419
+ },
420
+ {
421
+ "epoch": 12.222222222222221,
422
+ "grad_norm": 0.7947866916656494,
423
+ "learning_rate": 2.1604938271604937e-05,
424
+ "loss": 0.1481,
425
+ "step": 1100
426
+ },
427
+ {
428
+ "epoch": 12.5,
429
+ "grad_norm": 0.7545654773712158,
430
+ "learning_rate": 2.0833333333333336e-05,
431
+ "loss": 0.1545,
432
+ "step": 1125
433
+ },
434
+ {
435
+ "epoch": 12.777777777777779,
436
+ "grad_norm": 0.6851534247398376,
437
+ "learning_rate": 2.006172839506173e-05,
438
+ "loss": 0.1461,
439
+ "step": 1150
440
+ },
441
+ {
442
+ "epoch": 13.0,
443
+ "eval_accuracy": 0.7351778656126482,
444
+ "eval_loss": 0.709749698638916,
445
+ "eval_runtime": 29.6843,
446
+ "eval_samples_per_second": 34.092,
447
+ "eval_steps_per_second": 0.539,
448
+ "step": 1170
449
+ },
450
+ {
451
+ "epoch": 13.055555555555555,
452
+ "grad_norm": 0.6691556572914124,
453
+ "learning_rate": 1.9290123456790123e-05,
454
+ "loss": 0.1434,
455
+ "step": 1175
456
+ },
457
+ {
458
+ "epoch": 13.333333333333334,
459
+ "grad_norm": 0.7591906189918518,
460
+ "learning_rate": 1.8518518518518518e-05,
461
+ "loss": 0.1493,
462
+ "step": 1200
463
+ },
464
+ {
465
+ "epoch": 13.61111111111111,
466
+ "grad_norm": 0.7788909673690796,
467
+ "learning_rate": 1.7746913580246917e-05,
468
+ "loss": 0.1472,
469
+ "step": 1225
470
+ },
471
+ {
472
+ "epoch": 13.88888888888889,
473
+ "grad_norm": 0.5040611624717712,
474
+ "learning_rate": 1.697530864197531e-05,
475
+ "loss": 0.1439,
476
+ "step": 1250
477
+ },
478
+ {
479
+ "epoch": 14.0,
480
+ "eval_accuracy": 0.7430830039525692,
481
+ "eval_loss": 0.7007103562355042,
482
+ "eval_runtime": 29.6092,
483
+ "eval_samples_per_second": 34.179,
484
+ "eval_steps_per_second": 0.54,
485
+ "step": 1260
486
+ },
487
+ {
488
+ "epoch": 14.166666666666666,
489
+ "grad_norm": 0.6750937700271606,
490
+ "learning_rate": 1.6203703703703704e-05,
491
+ "loss": 0.1432,
492
+ "step": 1275
493
+ },
494
+ {
495
+ "epoch": 14.444444444444445,
496
+ "grad_norm": 0.4908043146133423,
497
+ "learning_rate": 1.54320987654321e-05,
498
+ "loss": 0.1412,
499
+ "step": 1300
500
+ },
501
+ {
502
+ "epoch": 14.722222222222221,
503
+ "grad_norm": 0.5028871297836304,
504
+ "learning_rate": 1.4660493827160496e-05,
505
+ "loss": 0.1403,
506
+ "step": 1325
507
+ },
508
+ {
509
+ "epoch": 15.0,
510
+ "grad_norm": 0.579842746257782,
511
+ "learning_rate": 1.388888888888889e-05,
512
+ "loss": 0.1428,
513
+ "step": 1350
514
+ },
515
+ {
516
+ "epoch": 15.0,
517
+ "eval_accuracy": 0.7401185770750988,
518
+ "eval_loss": 0.7012726068496704,
519
+ "eval_runtime": 29.6572,
520
+ "eval_samples_per_second": 34.123,
521
+ "eval_steps_per_second": 0.539,
522
+ "step": 1350
523
+ },
524
+ {
525
+ "epoch": 15.277777777777779,
526
+ "grad_norm": 0.372830331325531,
527
+ "learning_rate": 1.3117283950617285e-05,
528
+ "loss": 0.142,
529
+ "step": 1375
530
+ },
531
+ {
532
+ "epoch": 15.555555555555555,
533
+ "grad_norm": 0.30693933367729187,
534
+ "learning_rate": 1.2345679012345678e-05,
535
+ "loss": 0.1399,
536
+ "step": 1400
537
+ },
538
+ {
539
+ "epoch": 15.833333333333334,
540
+ "grad_norm": 0.4288826286792755,
541
+ "learning_rate": 1.1574074074074075e-05,
542
+ "loss": 0.1409,
543
+ "step": 1425
544
+ },
545
+ {
546
+ "epoch": 16.0,
547
+ "eval_accuracy": 0.7401185770750988,
548
+ "eval_loss": 0.7025883197784424,
549
+ "eval_runtime": 29.677,
550
+ "eval_samples_per_second": 34.1,
551
+ "eval_steps_per_second": 0.539,
552
+ "step": 1440
553
+ },
554
+ {
555
+ "epoch": 16.11111111111111,
556
+ "grad_norm": 0.3597729504108429,
557
+ "learning_rate": 1.0802469135802469e-05,
558
+ "loss": 0.1359,
559
+ "step": 1450
560
+ },
561
+ {
562
+ "epoch": 16.38888888888889,
563
+ "grad_norm": 0.36627763509750366,
564
+ "learning_rate": 1.0030864197530866e-05,
565
+ "loss": 0.1388,
566
+ "step": 1475
567
+ },
568
+ {
569
+ "epoch": 16.666666666666668,
570
+ "grad_norm": 0.40269795060157776,
571
+ "learning_rate": 9.259259259259259e-06,
572
+ "loss": 0.1417,
573
+ "step": 1500
574
+ },
575
+ {
576
+ "epoch": 16.944444444444443,
577
+ "grad_norm": 0.42748990654945374,
578
+ "learning_rate": 8.487654320987654e-06,
579
+ "loss": 0.1402,
580
+ "step": 1525
581
+ },
582
+ {
583
+ "epoch": 17.0,
584
+ "eval_accuracy": 0.7430830039525692,
585
+ "eval_loss": 0.704167902469635,
586
+ "eval_runtime": 29.7414,
587
+ "eval_samples_per_second": 34.027,
588
+ "eval_steps_per_second": 0.538,
589
+ "step": 1530
590
+ },
591
+ {
592
+ "epoch": 17.22222222222222,
593
+ "grad_norm": 0.40921568870544434,
594
+ "learning_rate": 7.71604938271605e-06,
595
+ "loss": 0.1402,
596
+ "step": 1550
597
+ },
598
+ {
599
+ "epoch": 17.5,
600
+ "grad_norm": 0.37299108505249023,
601
+ "learning_rate": 6.944444444444445e-06,
602
+ "loss": 0.1379,
603
+ "step": 1575
604
+ },
605
+ {
606
+ "epoch": 17.77777777777778,
607
+ "grad_norm": 0.39989030361175537,
608
+ "learning_rate": 6.172839506172839e-06,
609
+ "loss": 0.1325,
610
+ "step": 1600
611
+ },
612
+ {
613
+ "epoch": 18.0,
614
+ "eval_accuracy": 0.7351778656126482,
615
+ "eval_loss": 0.6994230151176453,
616
+ "eval_runtime": 29.536,
617
+ "eval_samples_per_second": 34.263,
618
+ "eval_steps_per_second": 0.542,
619
+ "step": 1620
620
+ },
621
+ {
622
+ "epoch": 18.055555555555557,
623
+ "grad_norm": 0.4737277925014496,
624
+ "learning_rate": 5.401234567901234e-06,
625
+ "loss": 0.1447,
626
+ "step": 1625
627
+ },
628
+ {
629
+ "epoch": 18.333333333333332,
630
+ "grad_norm": 0.2812809944152832,
631
+ "learning_rate": 4.6296296296296296e-06,
632
+ "loss": 0.1418,
633
+ "step": 1650
634
+ },
635
+ {
636
+ "epoch": 18.61111111111111,
637
+ "grad_norm": 0.3522038161754608,
638
+ "learning_rate": 3.858024691358025e-06,
639
+ "loss": 0.1337,
640
+ "step": 1675
641
+ },
642
+ {
643
+ "epoch": 18.88888888888889,
644
+ "grad_norm": 0.34057098627090454,
645
+ "learning_rate": 3.0864197530864196e-06,
646
+ "loss": 0.1364,
647
+ "step": 1700
648
+ },
649
+ {
650
+ "epoch": 19.0,
651
+ "eval_accuracy": 0.7460474308300395,
652
+ "eval_loss": 0.7015388607978821,
653
+ "eval_runtime": 29.1817,
654
+ "eval_samples_per_second": 34.679,
655
+ "eval_steps_per_second": 0.548,
656
+ "step": 1710
657
+ },
658
+ {
659
+ "epoch": 19.166666666666668,
660
+ "grad_norm": 0.4172286093235016,
661
+ "learning_rate": 2.3148148148148148e-06,
662
+ "loss": 0.1344,
663
+ "step": 1725
664
+ },
665
+ {
666
+ "epoch": 19.444444444444443,
667
+ "grad_norm": 0.5193507671356201,
668
+ "learning_rate": 1.5432098765432098e-06,
669
+ "loss": 0.1415,
670
+ "step": 1750
671
+ },
672
+ {
673
+ "epoch": 19.72222222222222,
674
+ "grad_norm": 0.3823259174823761,
675
+ "learning_rate": 7.716049382716049e-07,
676
+ "loss": 0.1379,
677
+ "step": 1775
678
+ },
679
+ {
680
+ "epoch": 20.0,
681
+ "grad_norm": 0.4077320694923401,
682
+ "learning_rate": 0.0,
683
+ "loss": 0.1329,
684
+ "step": 1800
685
+ },
686
+ {
687
+ "epoch": 20.0,
688
+ "eval_accuracy": 0.7460474308300395,
689
+ "eval_loss": 0.7007161378860474,
690
+ "eval_runtime": 29.7968,
691
+ "eval_samples_per_second": 33.963,
692
+ "eval_steps_per_second": 0.537,
693
+ "step": 1800
694
+ },
695
+ {
696
+ "epoch": 20.0,
697
+ "step": 1800,
698
+ "total_flos": 0.0,
699
+ "train_loss": 0.31672255052460563,
700
+ "train_runtime": 10514.5803,
701
+ "train_samples_per_second": 10.899,
702
+ "train_steps_per_second": 0.171
703
+ }
704
+ ],
705
+ "logging_steps": 25,
706
+ "max_steps": 1800,
707
+ "num_input_tokens_seen": 0,
708
+ "num_train_epochs": 20,
709
+ "save_steps": 500,
710
+ "stateful_callbacks": {
711
+ "TrainerControl": {
712
+ "args": {
713
+ "should_epoch_stop": false,
714
+ "should_evaluate": false,
715
+ "should_log": false,
716
+ "should_save": true,
717
+ "should_training_stop": true
718
+ },
719
+ "attributes": {}
720
+ }
721
+ },
722
+ "total_flos": 0.0,
723
+ "train_batch_size": 64,
724
+ "trial_name": null,
725
+ "trial_params": null
726
+ }