antonjaragon commited on
Commit
7de6d75
1 Parent(s): a3f69ef

End of training

Browse files
Files changed (5) hide show
  1. README.md +4 -3
  2. all_results.json +12 -0
  3. eval_results.json +8 -0
  4. train_results.json +7 -0
  5. trainer_state.json +1267 -0
README.md CHANGED
@@ -2,6 +2,7 @@
2
  license: apache-2.0
3
  base_model: facebook/wav2vec2-base
4
  tags:
 
5
  - generated_from_trainer
6
  datasets:
7
  - superb
@@ -22,7 +23,7 @@ model-index:
22
  metrics:
23
  - name: Accuracy
24
  type: accuracy
25
- value: 0.9830832597822889
26
  ---
27
 
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -32,8 +33,8 @@ should probably proofread and complete it, then remove this comment. -->
32
 
33
  This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the superb dataset.
34
  It achieves the following results on the evaluation set:
35
- - Loss: 0.0789
36
- - Accuracy: 0.9831
37
 
38
  ## Model description
39
 
 
2
  license: apache-2.0
3
  base_model: facebook/wav2vec2-base
4
  tags:
5
+ - audio-classification
6
  - generated_from_trainer
7
  datasets:
8
  - superb
 
23
  metrics:
24
  - name: Accuracy
25
  type: accuracy
26
+ value: 0.9832303618711385
27
  ---
28
 
29
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
33
 
34
  This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the superb dataset.
35
  It achieves the following results on the evaluation set:
36
+ - Loss: 0.0827
37
+ - Accuracy: 0.9832
38
 
39
  ## Model description
40
 
all_results.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "eval_accuracy": 0.9832303618711385,
4
+ "eval_loss": 0.0826701819896698,
5
+ "eval_runtime": 13.7161,
6
+ "eval_samples_per_second": 495.623,
7
+ "eval_steps_per_second": 15.529,
8
+ "train_loss": 0.39961676239071037,
9
+ "train_runtime": 1167.3183,
10
+ "train_samples_per_second": 218.852,
11
+ "train_steps_per_second": 1.709
12
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "eval_accuracy": 0.9832303618711385,
4
+ "eval_loss": 0.0826701819896698,
5
+ "eval_runtime": 13.7161,
6
+ "eval_samples_per_second": 495.623,
7
+ "eval_steps_per_second": 15.529
8
+ }
train_results.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 5.0,
3
+ "train_loss": 0.39961676239071037,
4
+ "train_runtime": 1167.3183,
5
+ "train_samples_per_second": 218.852,
6
+ "train_steps_per_second": 1.709
7
+ }
trainer_state.json ADDED
@@ -0,0 +1,1267 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.9832303618711385,
3
+ "best_model_checkpoint": "wav2vec2-base-ft-keyword-spotting/checkpoint-1597",
4
+ "epoch": 4.996869129618034,
5
+ "eval_steps": 500,
6
+ "global_step": 1995,
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
+ "learning_rate": 1.5e-06,
14
+ "loss": 2.4941,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 0.05,
19
+ "learning_rate": 3e-06,
20
+ "loss": 2.4759,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 0.08,
25
+ "learning_rate": 4.5e-06,
26
+ "loss": 2.4215,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 0.1,
31
+ "learning_rate": 6e-06,
32
+ "loss": 2.3219,
33
+ "step": 40
34
+ },
35
+ {
36
+ "epoch": 0.13,
37
+ "learning_rate": 7.5e-06,
38
+ "loss": 2.1799,
39
+ "step": 50
40
+ },
41
+ {
42
+ "epoch": 0.15,
43
+ "learning_rate": 9e-06,
44
+ "loss": 2.0087,
45
+ "step": 60
46
+ },
47
+ {
48
+ "epoch": 0.18,
49
+ "learning_rate": 1.05e-05,
50
+ "loss": 1.8646,
51
+ "step": 70
52
+ },
53
+ {
54
+ "epoch": 0.2,
55
+ "learning_rate": 1.2e-05,
56
+ "loss": 1.7391,
57
+ "step": 80
58
+ },
59
+ {
60
+ "epoch": 0.23,
61
+ "learning_rate": 1.3500000000000001e-05,
62
+ "loss": 1.6466,
63
+ "step": 90
64
+ },
65
+ {
66
+ "epoch": 0.25,
67
+ "learning_rate": 1.5e-05,
68
+ "loss": 1.6398,
69
+ "step": 100
70
+ },
71
+ {
72
+ "epoch": 0.28,
73
+ "learning_rate": 1.65e-05,
74
+ "loss": 1.519,
75
+ "step": 110
76
+ },
77
+ {
78
+ "epoch": 0.3,
79
+ "learning_rate": 1.8e-05,
80
+ "loss": 1.5813,
81
+ "step": 120
82
+ },
83
+ {
84
+ "epoch": 0.33,
85
+ "learning_rate": 1.95e-05,
86
+ "loss": 1.4891,
87
+ "step": 130
88
+ },
89
+ {
90
+ "epoch": 0.35,
91
+ "learning_rate": 2.1e-05,
92
+ "loss": 1.5947,
93
+ "step": 140
94
+ },
95
+ {
96
+ "epoch": 0.38,
97
+ "learning_rate": 2.25e-05,
98
+ "loss": 1.542,
99
+ "step": 150
100
+ },
101
+ {
102
+ "epoch": 0.4,
103
+ "learning_rate": 2.4e-05,
104
+ "loss": 1.4833,
105
+ "step": 160
106
+ },
107
+ {
108
+ "epoch": 0.43,
109
+ "learning_rate": 2.55e-05,
110
+ "loss": 1.431,
111
+ "step": 170
112
+ },
113
+ {
114
+ "epoch": 0.45,
115
+ "learning_rate": 2.7000000000000002e-05,
116
+ "loss": 1.353,
117
+ "step": 180
118
+ },
119
+ {
120
+ "epoch": 0.48,
121
+ "learning_rate": 2.8499999999999998e-05,
122
+ "loss": 1.265,
123
+ "step": 190
124
+ },
125
+ {
126
+ "epoch": 0.5,
127
+ "learning_rate": 3e-05,
128
+ "loss": 1.2248,
129
+ "step": 200
130
+ },
131
+ {
132
+ "epoch": 0.53,
133
+ "learning_rate": 2.9832869080779945e-05,
134
+ "loss": 1.0998,
135
+ "step": 210
136
+ },
137
+ {
138
+ "epoch": 0.55,
139
+ "learning_rate": 2.9665738161559886e-05,
140
+ "loss": 1.1259,
141
+ "step": 220
142
+ },
143
+ {
144
+ "epoch": 0.58,
145
+ "learning_rate": 2.9498607242339834e-05,
146
+ "loss": 1.0713,
147
+ "step": 230
148
+ },
149
+ {
150
+ "epoch": 0.6,
151
+ "learning_rate": 2.933147632311978e-05,
152
+ "loss": 1.0161,
153
+ "step": 240
154
+ },
155
+ {
156
+ "epoch": 0.63,
157
+ "learning_rate": 2.916434540389972e-05,
158
+ "loss": 0.9614,
159
+ "step": 250
160
+ },
161
+ {
162
+ "epoch": 0.65,
163
+ "learning_rate": 2.8997214484679665e-05,
164
+ "loss": 0.8767,
165
+ "step": 260
166
+ },
167
+ {
168
+ "epoch": 0.68,
169
+ "learning_rate": 2.8830083565459613e-05,
170
+ "loss": 0.8643,
171
+ "step": 270
172
+ },
173
+ {
174
+ "epoch": 0.7,
175
+ "learning_rate": 2.8662952646239554e-05,
176
+ "loss": 0.8183,
177
+ "step": 280
178
+ },
179
+ {
180
+ "epoch": 0.73,
181
+ "learning_rate": 2.84958217270195e-05,
182
+ "loss": 0.7784,
183
+ "step": 290
184
+ },
185
+ {
186
+ "epoch": 0.75,
187
+ "learning_rate": 2.8328690807799443e-05,
188
+ "loss": 0.6894,
189
+ "step": 300
190
+ },
191
+ {
192
+ "epoch": 0.78,
193
+ "learning_rate": 2.8161559888579388e-05,
194
+ "loss": 0.6612,
195
+ "step": 310
196
+ },
197
+ {
198
+ "epoch": 0.8,
199
+ "learning_rate": 2.7994428969359332e-05,
200
+ "loss": 0.6353,
201
+ "step": 320
202
+ },
203
+ {
204
+ "epoch": 0.83,
205
+ "learning_rate": 2.7827298050139277e-05,
206
+ "loss": 0.6632,
207
+ "step": 330
208
+ },
209
+ {
210
+ "epoch": 0.85,
211
+ "learning_rate": 2.7660167130919218e-05,
212
+ "loss": 0.5824,
213
+ "step": 340
214
+ },
215
+ {
216
+ "epoch": 0.88,
217
+ "learning_rate": 2.7493036211699166e-05,
218
+ "loss": 0.5914,
219
+ "step": 350
220
+ },
221
+ {
222
+ "epoch": 0.9,
223
+ "learning_rate": 2.732590529247911e-05,
224
+ "loss": 0.5481,
225
+ "step": 360
226
+ },
227
+ {
228
+ "epoch": 0.93,
229
+ "learning_rate": 2.7158774373259055e-05,
230
+ "loss": 0.5055,
231
+ "step": 370
232
+ },
233
+ {
234
+ "epoch": 0.95,
235
+ "learning_rate": 2.6991643454038996e-05,
236
+ "loss": 0.5035,
237
+ "step": 380
238
+ },
239
+ {
240
+ "epoch": 0.98,
241
+ "learning_rate": 2.6824512534818944e-05,
242
+ "loss": 0.4547,
243
+ "step": 390
244
+ },
245
+ {
246
+ "epoch": 1.0,
247
+ "eval_accuracy": 0.9710208884966166,
248
+ "eval_loss": 0.33722183108329773,
249
+ "eval_runtime": 13.726,
250
+ "eval_samples_per_second": 495.265,
251
+ "eval_steps_per_second": 15.518,
252
+ "step": 399
253
+ },
254
+ {
255
+ "epoch": 1.0,
256
+ "learning_rate": 2.665738161559889e-05,
257
+ "loss": 0.4363,
258
+ "step": 400
259
+ },
260
+ {
261
+ "epoch": 1.03,
262
+ "learning_rate": 2.649025069637883e-05,
263
+ "loss": 0.4134,
264
+ "step": 410
265
+ },
266
+ {
267
+ "epoch": 1.05,
268
+ "learning_rate": 2.6323119777158774e-05,
269
+ "loss": 0.4144,
270
+ "step": 420
271
+ },
272
+ {
273
+ "epoch": 1.08,
274
+ "learning_rate": 2.6155988857938722e-05,
275
+ "loss": 0.4239,
276
+ "step": 430
277
+ },
278
+ {
279
+ "epoch": 1.1,
280
+ "learning_rate": 2.5988857938718663e-05,
281
+ "loss": 0.3925,
282
+ "step": 440
283
+ },
284
+ {
285
+ "epoch": 1.13,
286
+ "learning_rate": 2.5821727019498608e-05,
287
+ "loss": 0.3468,
288
+ "step": 450
289
+ },
290
+ {
291
+ "epoch": 1.15,
292
+ "learning_rate": 2.5654596100278553e-05,
293
+ "loss": 0.3417,
294
+ "step": 460
295
+ },
296
+ {
297
+ "epoch": 1.18,
298
+ "learning_rate": 2.5487465181058497e-05,
299
+ "loss": 0.3667,
300
+ "step": 470
301
+ },
302
+ {
303
+ "epoch": 1.2,
304
+ "learning_rate": 2.532033426183844e-05,
305
+ "loss": 0.361,
306
+ "step": 480
307
+ },
308
+ {
309
+ "epoch": 1.23,
310
+ "learning_rate": 2.5153203342618386e-05,
311
+ "loss": 0.3563,
312
+ "step": 490
313
+ },
314
+ {
315
+ "epoch": 1.25,
316
+ "learning_rate": 2.4986072423398327e-05,
317
+ "loss": 0.3567,
318
+ "step": 500
319
+ },
320
+ {
321
+ "epoch": 1.28,
322
+ "learning_rate": 2.4818941504178275e-05,
323
+ "loss": 0.336,
324
+ "step": 510
325
+ },
326
+ {
327
+ "epoch": 1.3,
328
+ "learning_rate": 2.465181058495822e-05,
329
+ "loss": 0.3095,
330
+ "step": 520
331
+ },
332
+ {
333
+ "epoch": 1.33,
334
+ "learning_rate": 2.448467966573816e-05,
335
+ "loss": 0.3213,
336
+ "step": 530
337
+ },
338
+ {
339
+ "epoch": 1.35,
340
+ "learning_rate": 2.4317548746518106e-05,
341
+ "loss": 0.281,
342
+ "step": 540
343
+ },
344
+ {
345
+ "epoch": 1.38,
346
+ "learning_rate": 2.415041782729805e-05,
347
+ "loss": 0.2936,
348
+ "step": 550
349
+ },
350
+ {
351
+ "epoch": 1.4,
352
+ "learning_rate": 2.3983286908077995e-05,
353
+ "loss": 0.2731,
354
+ "step": 560
355
+ },
356
+ {
357
+ "epoch": 1.43,
358
+ "learning_rate": 2.381615598885794e-05,
359
+ "loss": 0.255,
360
+ "step": 570
361
+ },
362
+ {
363
+ "epoch": 1.45,
364
+ "learning_rate": 2.3649025069637884e-05,
365
+ "loss": 0.2892,
366
+ "step": 580
367
+ },
368
+ {
369
+ "epoch": 1.48,
370
+ "learning_rate": 2.3481894150417825e-05,
371
+ "loss": 0.258,
372
+ "step": 590
373
+ },
374
+ {
375
+ "epoch": 1.5,
376
+ "learning_rate": 2.3314763231197773e-05,
377
+ "loss": 0.2621,
378
+ "step": 600
379
+ },
380
+ {
381
+ "epoch": 1.53,
382
+ "learning_rate": 2.3147632311977718e-05,
383
+ "loss": 0.2533,
384
+ "step": 610
385
+ },
386
+ {
387
+ "epoch": 1.55,
388
+ "learning_rate": 2.298050139275766e-05,
389
+ "loss": 0.2595,
390
+ "step": 620
391
+ },
392
+ {
393
+ "epoch": 1.58,
394
+ "learning_rate": 2.2813370473537603e-05,
395
+ "loss": 0.2685,
396
+ "step": 630
397
+ },
398
+ {
399
+ "epoch": 1.6,
400
+ "learning_rate": 2.264623955431755e-05,
401
+ "loss": 0.2603,
402
+ "step": 640
403
+ },
404
+ {
405
+ "epoch": 1.63,
406
+ "learning_rate": 2.2479108635097492e-05,
407
+ "loss": 0.286,
408
+ "step": 650
409
+ },
410
+ {
411
+ "epoch": 1.65,
412
+ "learning_rate": 2.2311977715877437e-05,
413
+ "loss": 0.2588,
414
+ "step": 660
415
+ },
416
+ {
417
+ "epoch": 1.68,
418
+ "learning_rate": 2.214484679665738e-05,
419
+ "loss": 0.2327,
420
+ "step": 670
421
+ },
422
+ {
423
+ "epoch": 1.7,
424
+ "learning_rate": 2.1977715877437326e-05,
425
+ "loss": 0.2604,
426
+ "step": 680
427
+ },
428
+ {
429
+ "epoch": 1.73,
430
+ "learning_rate": 2.181058495821727e-05,
431
+ "loss": 0.2328,
432
+ "step": 690
433
+ },
434
+ {
435
+ "epoch": 1.75,
436
+ "learning_rate": 2.1643454038997215e-05,
437
+ "loss": 0.2207,
438
+ "step": 700
439
+ },
440
+ {
441
+ "epoch": 1.78,
442
+ "learning_rate": 2.147632311977716e-05,
443
+ "loss": 0.2125,
444
+ "step": 710
445
+ },
446
+ {
447
+ "epoch": 1.8,
448
+ "learning_rate": 2.1309192200557104e-05,
449
+ "loss": 0.2212,
450
+ "step": 720
451
+ },
452
+ {
453
+ "epoch": 1.83,
454
+ "learning_rate": 2.114206128133705e-05,
455
+ "loss": 0.1926,
456
+ "step": 730
457
+ },
458
+ {
459
+ "epoch": 1.85,
460
+ "learning_rate": 2.0974930362116993e-05,
461
+ "loss": 0.2338,
462
+ "step": 740
463
+ },
464
+ {
465
+ "epoch": 1.88,
466
+ "learning_rate": 2.0807799442896935e-05,
467
+ "loss": 0.2278,
468
+ "step": 750
469
+ },
470
+ {
471
+ "epoch": 1.9,
472
+ "learning_rate": 2.0640668523676883e-05,
473
+ "loss": 0.2128,
474
+ "step": 760
475
+ },
476
+ {
477
+ "epoch": 1.93,
478
+ "learning_rate": 2.0473537604456827e-05,
479
+ "loss": 0.2214,
480
+ "step": 770
481
+ },
482
+ {
483
+ "epoch": 1.95,
484
+ "learning_rate": 2.0306406685236768e-05,
485
+ "loss": 0.199,
486
+ "step": 780
487
+ },
488
+ {
489
+ "epoch": 1.98,
490
+ "learning_rate": 2.0139275766016713e-05,
491
+ "loss": 0.2625,
492
+ "step": 790
493
+ },
494
+ {
495
+ "epoch": 2.0,
496
+ "eval_accuracy": 0.9779346866725508,
497
+ "eval_loss": 0.11822522431612015,
498
+ "eval_runtime": 13.9175,
499
+ "eval_samples_per_second": 488.448,
500
+ "eval_steps_per_second": 15.304,
501
+ "step": 798
502
+ },
503
+ {
504
+ "epoch": 2.0,
505
+ "learning_rate": 1.997214484679666e-05,
506
+ "loss": 0.1903,
507
+ "step": 800
508
+ },
509
+ {
510
+ "epoch": 2.03,
511
+ "learning_rate": 1.9805013927576602e-05,
512
+ "loss": 0.1891,
513
+ "step": 810
514
+ },
515
+ {
516
+ "epoch": 2.05,
517
+ "learning_rate": 1.9637883008356546e-05,
518
+ "loss": 0.2066,
519
+ "step": 820
520
+ },
521
+ {
522
+ "epoch": 2.08,
523
+ "learning_rate": 1.947075208913649e-05,
524
+ "loss": 0.1839,
525
+ "step": 830
526
+ },
527
+ {
528
+ "epoch": 2.1,
529
+ "learning_rate": 1.9303621169916436e-05,
530
+ "loss": 0.1799,
531
+ "step": 840
532
+ },
533
+ {
534
+ "epoch": 2.13,
535
+ "learning_rate": 1.9153203342618386e-05,
536
+ "loss": 0.2192,
537
+ "step": 850
538
+ },
539
+ {
540
+ "epoch": 2.15,
541
+ "learning_rate": 1.898607242339833e-05,
542
+ "loss": 0.2021,
543
+ "step": 860
544
+ },
545
+ {
546
+ "epoch": 2.18,
547
+ "learning_rate": 1.8818941504178272e-05,
548
+ "loss": 0.2315,
549
+ "step": 870
550
+ },
551
+ {
552
+ "epoch": 2.2,
553
+ "learning_rate": 1.8651810584958216e-05,
554
+ "loss": 0.2357,
555
+ "step": 880
556
+ },
557
+ {
558
+ "epoch": 2.23,
559
+ "learning_rate": 1.8484679665738164e-05,
560
+ "loss": 0.1723,
561
+ "step": 890
562
+ },
563
+ {
564
+ "epoch": 2.25,
565
+ "learning_rate": 1.8317548746518105e-05,
566
+ "loss": 0.2169,
567
+ "step": 900
568
+ },
569
+ {
570
+ "epoch": 2.28,
571
+ "learning_rate": 1.815041782729805e-05,
572
+ "loss": 0.1875,
573
+ "step": 910
574
+ },
575
+ {
576
+ "epoch": 2.3,
577
+ "learning_rate": 1.7983286908077995e-05,
578
+ "loss": 0.1807,
579
+ "step": 920
580
+ },
581
+ {
582
+ "epoch": 2.33,
583
+ "learning_rate": 1.781615598885794e-05,
584
+ "loss": 0.1777,
585
+ "step": 930
586
+ },
587
+ {
588
+ "epoch": 2.35,
589
+ "learning_rate": 1.7649025069637884e-05,
590
+ "loss": 0.187,
591
+ "step": 940
592
+ },
593
+ {
594
+ "epoch": 2.38,
595
+ "learning_rate": 1.7481894150417828e-05,
596
+ "loss": 0.1855,
597
+ "step": 950
598
+ },
599
+ {
600
+ "epoch": 2.4,
601
+ "learning_rate": 1.731476323119777e-05,
602
+ "loss": 0.1486,
603
+ "step": 960
604
+ },
605
+ {
606
+ "epoch": 2.43,
607
+ "learning_rate": 1.7147632311977717e-05,
608
+ "loss": 0.1879,
609
+ "step": 970
610
+ },
611
+ {
612
+ "epoch": 2.45,
613
+ "learning_rate": 1.6980501392757662e-05,
614
+ "loss": 0.1662,
615
+ "step": 980
616
+ },
617
+ {
618
+ "epoch": 2.48,
619
+ "learning_rate": 1.6813370473537606e-05,
620
+ "loss": 0.1908,
621
+ "step": 990
622
+ },
623
+ {
624
+ "epoch": 2.5,
625
+ "learning_rate": 1.6646239554317548e-05,
626
+ "loss": 0.1785,
627
+ "step": 1000
628
+ },
629
+ {
630
+ "epoch": 2.53,
631
+ "learning_rate": 1.6479108635097496e-05,
632
+ "loss": 0.1643,
633
+ "step": 1010
634
+ },
635
+ {
636
+ "epoch": 2.55,
637
+ "learning_rate": 1.631197771587744e-05,
638
+ "loss": 0.1606,
639
+ "step": 1020
640
+ },
641
+ {
642
+ "epoch": 2.58,
643
+ "learning_rate": 1.614484679665738e-05,
644
+ "loss": 0.1504,
645
+ "step": 1030
646
+ },
647
+ {
648
+ "epoch": 2.6,
649
+ "learning_rate": 1.5977715877437326e-05,
650
+ "loss": 0.1806,
651
+ "step": 1040
652
+ },
653
+ {
654
+ "epoch": 2.63,
655
+ "learning_rate": 1.581058495821727e-05,
656
+ "loss": 0.1574,
657
+ "step": 1050
658
+ },
659
+ {
660
+ "epoch": 2.65,
661
+ "learning_rate": 1.5643454038997215e-05,
662
+ "loss": 0.1769,
663
+ "step": 1060
664
+ },
665
+ {
666
+ "epoch": 2.68,
667
+ "learning_rate": 1.547632311977716e-05,
668
+ "loss": 0.1853,
669
+ "step": 1070
670
+ },
671
+ {
672
+ "epoch": 2.71,
673
+ "learning_rate": 1.5309192200557104e-05,
674
+ "loss": 0.1916,
675
+ "step": 1080
676
+ },
677
+ {
678
+ "epoch": 2.73,
679
+ "learning_rate": 1.5142061281337047e-05,
680
+ "loss": 0.1796,
681
+ "step": 1090
682
+ },
683
+ {
684
+ "epoch": 2.76,
685
+ "learning_rate": 1.4974930362116992e-05,
686
+ "loss": 0.172,
687
+ "step": 1100
688
+ },
689
+ {
690
+ "epoch": 2.78,
691
+ "learning_rate": 1.4807799442896936e-05,
692
+ "loss": 0.1934,
693
+ "step": 1110
694
+ },
695
+ {
696
+ "epoch": 2.81,
697
+ "learning_rate": 1.464066852367688e-05,
698
+ "loss": 0.1946,
699
+ "step": 1120
700
+ },
701
+ {
702
+ "epoch": 2.83,
703
+ "learning_rate": 1.4473537604456825e-05,
704
+ "loss": 0.159,
705
+ "step": 1130
706
+ },
707
+ {
708
+ "epoch": 2.86,
709
+ "learning_rate": 1.4306406685236768e-05,
710
+ "loss": 0.1769,
711
+ "step": 1140
712
+ },
713
+ {
714
+ "epoch": 2.88,
715
+ "learning_rate": 1.4139275766016714e-05,
716
+ "loss": 0.1501,
717
+ "step": 1150
718
+ },
719
+ {
720
+ "epoch": 2.91,
721
+ "learning_rate": 1.3972144846796657e-05,
722
+ "loss": 0.1731,
723
+ "step": 1160
724
+ },
725
+ {
726
+ "epoch": 2.93,
727
+ "learning_rate": 1.3805013927576602e-05,
728
+ "loss": 0.1645,
729
+ "step": 1170
730
+ },
731
+ {
732
+ "epoch": 2.96,
733
+ "learning_rate": 1.3637883008356546e-05,
734
+ "loss": 0.1756,
735
+ "step": 1180
736
+ },
737
+ {
738
+ "epoch": 2.98,
739
+ "learning_rate": 1.3470752089136491e-05,
740
+ "loss": 0.1575,
741
+ "step": 1190
742
+ },
743
+ {
744
+ "epoch": 3.0,
745
+ "eval_accuracy": 0.978670197116799,
746
+ "eval_loss": 0.09626787155866623,
747
+ "eval_runtime": 13.6857,
748
+ "eval_samples_per_second": 496.721,
749
+ "eval_steps_per_second": 15.564,
750
+ "step": 1197
751
+ },
752
+ {
753
+ "epoch": 3.01,
754
+ "learning_rate": 1.3303621169916434e-05,
755
+ "loss": 0.1462,
756
+ "step": 1200
757
+ },
758
+ {
759
+ "epoch": 3.03,
760
+ "learning_rate": 1.313649025069638e-05,
761
+ "loss": 0.1298,
762
+ "step": 1210
763
+ },
764
+ {
765
+ "epoch": 3.06,
766
+ "learning_rate": 1.2969359331476323e-05,
767
+ "loss": 0.1755,
768
+ "step": 1220
769
+ },
770
+ {
771
+ "epoch": 3.08,
772
+ "learning_rate": 1.2802228412256267e-05,
773
+ "loss": 0.1677,
774
+ "step": 1230
775
+ },
776
+ {
777
+ "epoch": 3.11,
778
+ "learning_rate": 1.2635097493036212e-05,
779
+ "loss": 0.1487,
780
+ "step": 1240
781
+ },
782
+ {
783
+ "epoch": 3.13,
784
+ "learning_rate": 1.2467966573816157e-05,
785
+ "loss": 0.1359,
786
+ "step": 1250
787
+ },
788
+ {
789
+ "epoch": 3.16,
790
+ "learning_rate": 1.2300835654596101e-05,
791
+ "loss": 0.1499,
792
+ "step": 1260
793
+ },
794
+ {
795
+ "epoch": 3.18,
796
+ "learning_rate": 1.2133704735376046e-05,
797
+ "loss": 0.1593,
798
+ "step": 1270
799
+ },
800
+ {
801
+ "epoch": 3.21,
802
+ "learning_rate": 1.1966573816155989e-05,
803
+ "loss": 0.1632,
804
+ "step": 1280
805
+ },
806
+ {
807
+ "epoch": 3.23,
808
+ "learning_rate": 1.1799442896935935e-05,
809
+ "loss": 0.1769,
810
+ "step": 1290
811
+ },
812
+ {
813
+ "epoch": 3.26,
814
+ "learning_rate": 1.1632311977715878e-05,
815
+ "loss": 0.159,
816
+ "step": 1300
817
+ },
818
+ {
819
+ "epoch": 3.28,
820
+ "learning_rate": 1.1465181058495822e-05,
821
+ "loss": 0.1489,
822
+ "step": 1310
823
+ },
824
+ {
825
+ "epoch": 3.31,
826
+ "learning_rate": 1.1298050139275767e-05,
827
+ "loss": 0.1684,
828
+ "step": 1320
829
+ },
830
+ {
831
+ "epoch": 3.33,
832
+ "learning_rate": 1.1130919220055711e-05,
833
+ "loss": 0.1515,
834
+ "step": 1330
835
+ },
836
+ {
837
+ "epoch": 3.36,
838
+ "learning_rate": 1.0963788300835654e-05,
839
+ "loss": 0.1421,
840
+ "step": 1340
841
+ },
842
+ {
843
+ "epoch": 3.38,
844
+ "learning_rate": 1.0796657381615599e-05,
845
+ "loss": 0.143,
846
+ "step": 1350
847
+ },
848
+ {
849
+ "epoch": 3.41,
850
+ "learning_rate": 1.0629526462395543e-05,
851
+ "loss": 0.1674,
852
+ "step": 1360
853
+ },
854
+ {
855
+ "epoch": 3.43,
856
+ "learning_rate": 1.0462395543175486e-05,
857
+ "loss": 0.1255,
858
+ "step": 1370
859
+ },
860
+ {
861
+ "epoch": 3.46,
862
+ "learning_rate": 1.0295264623955432e-05,
863
+ "loss": 0.1431,
864
+ "step": 1380
865
+ },
866
+ {
867
+ "epoch": 3.48,
868
+ "learning_rate": 1.0128133704735375e-05,
869
+ "loss": 0.1066,
870
+ "step": 1390
871
+ },
872
+ {
873
+ "epoch": 3.51,
874
+ "learning_rate": 9.961002785515322e-06,
875
+ "loss": 0.1471,
876
+ "step": 1400
877
+ },
878
+ {
879
+ "epoch": 3.53,
880
+ "learning_rate": 9.793871866295264e-06,
881
+ "loss": 0.1384,
882
+ "step": 1410
883
+ },
884
+ {
885
+ "epoch": 3.56,
886
+ "learning_rate": 9.626740947075209e-06,
887
+ "loss": 0.1366,
888
+ "step": 1420
889
+ },
890
+ {
891
+ "epoch": 3.58,
892
+ "learning_rate": 9.459610027855154e-06,
893
+ "loss": 0.1433,
894
+ "step": 1430
895
+ },
896
+ {
897
+ "epoch": 3.61,
898
+ "learning_rate": 9.292479108635098e-06,
899
+ "loss": 0.1387,
900
+ "step": 1440
901
+ },
902
+ {
903
+ "epoch": 3.63,
904
+ "learning_rate": 9.125348189415041e-06,
905
+ "loss": 0.1336,
906
+ "step": 1450
907
+ },
908
+ {
909
+ "epoch": 3.66,
910
+ "learning_rate": 8.958217270194987e-06,
911
+ "loss": 0.1277,
912
+ "step": 1460
913
+ },
914
+ {
915
+ "epoch": 3.68,
916
+ "learning_rate": 8.79108635097493e-06,
917
+ "loss": 0.1762,
918
+ "step": 1470
919
+ },
920
+ {
921
+ "epoch": 3.71,
922
+ "learning_rate": 8.623955431754875e-06,
923
+ "loss": 0.1493,
924
+ "step": 1480
925
+ },
926
+ {
927
+ "epoch": 3.73,
928
+ "learning_rate": 8.45682451253482e-06,
929
+ "loss": 0.1595,
930
+ "step": 1490
931
+ },
932
+ {
933
+ "epoch": 3.76,
934
+ "learning_rate": 8.289693593314764e-06,
935
+ "loss": 0.154,
936
+ "step": 1500
937
+ },
938
+ {
939
+ "epoch": 3.78,
940
+ "learning_rate": 8.122562674094707e-06,
941
+ "loss": 0.1023,
942
+ "step": 1510
943
+ },
944
+ {
945
+ "epoch": 3.81,
946
+ "learning_rate": 7.955431754874653e-06,
947
+ "loss": 0.1103,
948
+ "step": 1520
949
+ },
950
+ {
951
+ "epoch": 3.83,
952
+ "learning_rate": 7.788300835654596e-06,
953
+ "loss": 0.1822,
954
+ "step": 1530
955
+ },
956
+ {
957
+ "epoch": 3.86,
958
+ "learning_rate": 7.621169916434541e-06,
959
+ "loss": 0.1238,
960
+ "step": 1540
961
+ },
962
+ {
963
+ "epoch": 3.88,
964
+ "learning_rate": 7.454038997214485e-06,
965
+ "loss": 0.1626,
966
+ "step": 1550
967
+ },
968
+ {
969
+ "epoch": 3.91,
970
+ "learning_rate": 7.2869080779944286e-06,
971
+ "loss": 0.1534,
972
+ "step": 1560
973
+ },
974
+ {
975
+ "epoch": 3.93,
976
+ "learning_rate": 7.119777158774373e-06,
977
+ "loss": 0.1365,
978
+ "step": 1570
979
+ },
980
+ {
981
+ "epoch": 3.96,
982
+ "learning_rate": 6.952646239554318e-06,
983
+ "loss": 0.1453,
984
+ "step": 1580
985
+ },
986
+ {
987
+ "epoch": 3.98,
988
+ "learning_rate": 6.785515320334261e-06,
989
+ "loss": 0.1314,
990
+ "step": 1590
991
+ },
992
+ {
993
+ "epoch": 4.0,
994
+ "eval_accuracy": 0.9832303618711385,
995
+ "eval_loss": 0.0826701819896698,
996
+ "eval_runtime": 14.003,
997
+ "eval_samples_per_second": 485.468,
998
+ "eval_steps_per_second": 15.211,
999
+ "step": 1597
1000
+ },
1001
+ {
1002
+ "epoch": 4.01,
1003
+ "learning_rate": 6.618384401114206e-06,
1004
+ "loss": 0.154,
1005
+ "step": 1600
1006
+ },
1007
+ {
1008
+ "epoch": 4.03,
1009
+ "learning_rate": 6.4512534818941505e-06,
1010
+ "loss": 0.1425,
1011
+ "step": 1610
1012
+ },
1013
+ {
1014
+ "epoch": 4.06,
1015
+ "learning_rate": 6.284122562674095e-06,
1016
+ "loss": 0.1525,
1017
+ "step": 1620
1018
+ },
1019
+ {
1020
+ "epoch": 4.08,
1021
+ "learning_rate": 6.116991643454039e-06,
1022
+ "loss": 0.1148,
1023
+ "step": 1630
1024
+ },
1025
+ {
1026
+ "epoch": 4.11,
1027
+ "learning_rate": 5.949860724233983e-06,
1028
+ "loss": 0.1506,
1029
+ "step": 1640
1030
+ },
1031
+ {
1032
+ "epoch": 4.13,
1033
+ "learning_rate": 5.782729805013928e-06,
1034
+ "loss": 0.1339,
1035
+ "step": 1650
1036
+ },
1037
+ {
1038
+ "epoch": 4.16,
1039
+ "learning_rate": 5.615598885793872e-06,
1040
+ "loss": 0.1111,
1041
+ "step": 1660
1042
+ },
1043
+ {
1044
+ "epoch": 4.18,
1045
+ "learning_rate": 5.448467966573816e-06,
1046
+ "loss": 0.1158,
1047
+ "step": 1670
1048
+ },
1049
+ {
1050
+ "epoch": 4.21,
1051
+ "learning_rate": 5.281337047353761e-06,
1052
+ "loss": 0.1257,
1053
+ "step": 1680
1054
+ },
1055
+ {
1056
+ "epoch": 4.23,
1057
+ "learning_rate": 5.114206128133705e-06,
1058
+ "loss": 0.1553,
1059
+ "step": 1690
1060
+ },
1061
+ {
1062
+ "epoch": 4.26,
1063
+ "learning_rate": 4.947075208913649e-06,
1064
+ "loss": 0.1289,
1065
+ "step": 1700
1066
+ },
1067
+ {
1068
+ "epoch": 4.28,
1069
+ "learning_rate": 4.7799442896935936e-06,
1070
+ "loss": 0.1269,
1071
+ "step": 1710
1072
+ },
1073
+ {
1074
+ "epoch": 4.31,
1075
+ "learning_rate": 4.612813370473538e-06,
1076
+ "loss": 0.1514,
1077
+ "step": 1720
1078
+ },
1079
+ {
1080
+ "epoch": 4.33,
1081
+ "learning_rate": 4.445682451253482e-06,
1082
+ "loss": 0.1262,
1083
+ "step": 1730
1084
+ },
1085
+ {
1086
+ "epoch": 4.36,
1087
+ "learning_rate": 4.278551532033426e-06,
1088
+ "loss": 0.1363,
1089
+ "step": 1740
1090
+ },
1091
+ {
1092
+ "epoch": 4.38,
1093
+ "learning_rate": 4.111420612813371e-06,
1094
+ "loss": 0.1131,
1095
+ "step": 1750
1096
+ },
1097
+ {
1098
+ "epoch": 4.41,
1099
+ "learning_rate": 3.944289693593315e-06,
1100
+ "loss": 0.116,
1101
+ "step": 1760
1102
+ },
1103
+ {
1104
+ "epoch": 4.43,
1105
+ "learning_rate": 3.7771587743732592e-06,
1106
+ "loss": 0.1216,
1107
+ "step": 1770
1108
+ },
1109
+ {
1110
+ "epoch": 4.46,
1111
+ "learning_rate": 3.6100278551532034e-06,
1112
+ "loss": 0.1246,
1113
+ "step": 1780
1114
+ },
1115
+ {
1116
+ "epoch": 4.48,
1117
+ "learning_rate": 3.4428969359331475e-06,
1118
+ "loss": 0.1313,
1119
+ "step": 1790
1120
+ },
1121
+ {
1122
+ "epoch": 4.51,
1123
+ "learning_rate": 3.275766016713092e-06,
1124
+ "loss": 0.136,
1125
+ "step": 1800
1126
+ },
1127
+ {
1128
+ "epoch": 4.53,
1129
+ "learning_rate": 3.108635097493036e-06,
1130
+ "loss": 0.1494,
1131
+ "step": 1810
1132
+ },
1133
+ {
1134
+ "epoch": 4.56,
1135
+ "learning_rate": 2.9415041782729803e-06,
1136
+ "loss": 0.1381,
1137
+ "step": 1820
1138
+ },
1139
+ {
1140
+ "epoch": 4.58,
1141
+ "learning_rate": 2.774373259052925e-06,
1142
+ "loss": 0.1621,
1143
+ "step": 1830
1144
+ },
1145
+ {
1146
+ "epoch": 4.61,
1147
+ "learning_rate": 2.607242339832869e-06,
1148
+ "loss": 0.1307,
1149
+ "step": 1840
1150
+ },
1151
+ {
1152
+ "epoch": 4.63,
1153
+ "learning_rate": 2.4401114206128136e-06,
1154
+ "loss": 0.1164,
1155
+ "step": 1850
1156
+ },
1157
+ {
1158
+ "epoch": 4.66,
1159
+ "learning_rate": 2.2729805013927577e-06,
1160
+ "loss": 0.128,
1161
+ "step": 1860
1162
+ },
1163
+ {
1164
+ "epoch": 4.68,
1165
+ "learning_rate": 2.1058495821727023e-06,
1166
+ "loss": 0.1226,
1167
+ "step": 1870
1168
+ },
1169
+ {
1170
+ "epoch": 4.71,
1171
+ "learning_rate": 1.9387186629526464e-06,
1172
+ "loss": 0.1234,
1173
+ "step": 1880
1174
+ },
1175
+ {
1176
+ "epoch": 4.73,
1177
+ "learning_rate": 1.7715877437325906e-06,
1178
+ "loss": 0.1666,
1179
+ "step": 1890
1180
+ },
1181
+ {
1182
+ "epoch": 4.76,
1183
+ "learning_rate": 1.604456824512535e-06,
1184
+ "loss": 0.1072,
1185
+ "step": 1900
1186
+ },
1187
+ {
1188
+ "epoch": 4.78,
1189
+ "learning_rate": 1.437325905292479e-06,
1190
+ "loss": 0.1151,
1191
+ "step": 1910
1192
+ },
1193
+ {
1194
+ "epoch": 4.81,
1195
+ "learning_rate": 1.2701949860724234e-06,
1196
+ "loss": 0.1556,
1197
+ "step": 1920
1198
+ },
1199
+ {
1200
+ "epoch": 4.83,
1201
+ "learning_rate": 1.1030640668523677e-06,
1202
+ "loss": 0.1222,
1203
+ "step": 1930
1204
+ },
1205
+ {
1206
+ "epoch": 4.86,
1207
+ "learning_rate": 9.35933147632312e-07,
1208
+ "loss": 0.1251,
1209
+ "step": 1940
1210
+ },
1211
+ {
1212
+ "epoch": 4.88,
1213
+ "learning_rate": 7.688022284122563e-07,
1214
+ "loss": 0.1262,
1215
+ "step": 1950
1216
+ },
1217
+ {
1218
+ "epoch": 4.91,
1219
+ "learning_rate": 6.016713091922006e-07,
1220
+ "loss": 0.1096,
1221
+ "step": 1960
1222
+ },
1223
+ {
1224
+ "epoch": 4.93,
1225
+ "learning_rate": 4.3454038997214486e-07,
1226
+ "loss": 0.1115,
1227
+ "step": 1970
1228
+ },
1229
+ {
1230
+ "epoch": 4.96,
1231
+ "learning_rate": 2.6740947075208915e-07,
1232
+ "loss": 0.1065,
1233
+ "step": 1980
1234
+ },
1235
+ {
1236
+ "epoch": 4.98,
1237
+ "learning_rate": 1.0027855153203343e-07,
1238
+ "loss": 0.1307,
1239
+ "step": 1990
1240
+ },
1241
+ {
1242
+ "epoch": 5.0,
1243
+ "eval_accuracy": 0.9830832597822889,
1244
+ "eval_loss": 0.07888180762529373,
1245
+ "eval_runtime": 13.938,
1246
+ "eval_samples_per_second": 487.73,
1247
+ "eval_steps_per_second": 15.282,
1248
+ "step": 1995
1249
+ },
1250
+ {
1251
+ "epoch": 5.0,
1252
+ "step": 1995,
1253
+ "total_flos": 2.3180194781952e+18,
1254
+ "train_loss": 0.39961676239071037,
1255
+ "train_runtime": 1167.3183,
1256
+ "train_samples_per_second": 218.852,
1257
+ "train_steps_per_second": 1.709
1258
+ }
1259
+ ],
1260
+ "logging_steps": 10,
1261
+ "max_steps": 1995,
1262
+ "num_train_epochs": 5,
1263
+ "save_steps": 500,
1264
+ "total_flos": 2.3180194781952e+18,
1265
+ "trial_name": null,
1266
+ "trial_params": null
1267
+ }