matthieulel commited on
Commit
ae16f21
1 Parent(s): acce1f1

End of training

Browse files
Files changed (5) hide show
  1. README.md +8 -6
  2. all_results.json +15 -0
  3. eval_results.json +11 -0
  4. train_results.json +7 -0
  5. trainer_state.json +948 -0
README.md CHANGED
@@ -2,6 +2,8 @@
2
  license: apache-2.0
3
  base_model: google/vit-large-patch32-384
4
  tags:
 
 
5
  - generated_from_trainer
6
  metrics:
7
  - accuracy
@@ -18,13 +20,13 @@ should probably proofread and complete it, then remove this comment. -->
18
 
19
  # vit-large-patch32-384-finetuned-galaxy10-decals
20
 
21
- This model is a fine-tuned version of [google/vit-large-patch32-384](https://huggingface.co/google/vit-large-patch32-384) on an unknown dataset.
22
  It achieves the following results on the evaluation set:
23
- - Loss: 0.6787
24
- - Accuracy: 0.8343
25
- - Precision: 0.8336
26
- - Recall: 0.8343
27
- - F1: 0.8326
28
 
29
  ## Model description
30
 
 
2
  license: apache-2.0
3
  base_model: google/vit-large-patch32-384
4
  tags:
5
+ - image-classification
6
+ - vision
7
  - generated_from_trainer
8
  metrics:
9
  - accuracy
 
20
 
21
  # vit-large-patch32-384-finetuned-galaxy10-decals
22
 
23
+ This model is a fine-tuned version of [google/vit-large-patch32-384](https://huggingface.co/google/vit-large-patch32-384) on the matthieulel/galaxy10_decals dataset.
24
  It achieves the following results on the evaluation set:
25
+ - Loss: 0.6766
26
+ - Accuracy: 0.8371
27
+ - Precision: 0.8374
28
+ - Recall: 0.8371
29
+ - F1: 0.8357
30
 
31
  ## Model description
32
 
all_results.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 29.76,
3
+ "eval_accuracy": 0.8370913190529876,
4
+ "eval_f1": 0.8357334072148467,
5
+ "eval_loss": 0.6766165494918823,
6
+ "eval_precision": 0.8373589976039267,
7
+ "eval_recall": 0.8370913190529876,
8
+ "eval_runtime": 21.6445,
9
+ "eval_samples_per_second": 81.961,
10
+ "eval_steps_per_second": 0.647,
11
+ "train_loss": 0.3928444991829575,
12
+ "train_runtime": 10565.0301,
13
+ "train_samples_per_second": 45.325,
14
+ "train_steps_per_second": 0.088
15
+ }
eval_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 29.76,
3
+ "eval_accuracy": 0.8370913190529876,
4
+ "eval_f1": 0.8357334072148467,
5
+ "eval_loss": 0.6766165494918823,
6
+ "eval_precision": 0.8373589976039267,
7
+ "eval_recall": 0.8370913190529876,
8
+ "eval_runtime": 21.6445,
9
+ "eval_samples_per_second": 81.961,
10
+ "eval_steps_per_second": 0.647
11
+ }
train_results.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 29.76,
3
+ "train_loss": 0.3928444991829575,
4
+ "train_runtime": 10565.0301,
5
+ "train_samples_per_second": 45.325,
6
+ "train_steps_per_second": 0.088
7
+ }
trainer_state.json ADDED
@@ -0,0 +1,948 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.8370913190529876,
3
+ "best_model_checkpoint": "vit-large-patch32-384-finetuned-galaxy10-decals/checkpoint-843",
4
+ "epoch": 29.76,
5
+ "eval_steps": 500,
6
+ "global_step": 930,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.32,
13
+ "learning_rate": 1.0752688172043012e-05,
14
+ "loss": 2.2838,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 0.64,
19
+ "learning_rate": 2.1505376344086024e-05,
20
+ "loss": 1.8298,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 0.96,
25
+ "learning_rate": 3.2258064516129034e-05,
26
+ "loss": 1.3342,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 0.99,
31
+ "eval_accuracy": 0.6313416009019166,
32
+ "eval_f1": 0.6051741872345295,
33
+ "eval_loss": 1.0491372346878052,
34
+ "eval_precision": 0.6077378601373529,
35
+ "eval_recall": 0.6313416009019166,
36
+ "eval_runtime": 21.763,
37
+ "eval_samples_per_second": 81.515,
38
+ "eval_steps_per_second": 0.643,
39
+ "step": 31
40
+ },
41
+ {
42
+ "epoch": 1.28,
43
+ "learning_rate": 4.301075268817205e-05,
44
+ "loss": 1.0557,
45
+ "step": 40
46
+ },
47
+ {
48
+ "epoch": 1.6,
49
+ "learning_rate": 5.3763440860215054e-05,
50
+ "loss": 0.8471,
51
+ "step": 50
52
+ },
53
+ {
54
+ "epoch": 1.92,
55
+ "learning_rate": 6.451612903225807e-05,
56
+ "loss": 0.7979,
57
+ "step": 60
58
+ },
59
+ {
60
+ "epoch": 1.98,
61
+ "eval_accuracy": 0.7671927846674182,
62
+ "eval_f1": 0.7652430186918779,
63
+ "eval_loss": 0.6901012659072876,
64
+ "eval_precision": 0.7716921452381017,
65
+ "eval_recall": 0.7671927846674182,
66
+ "eval_runtime": 21.1847,
67
+ "eval_samples_per_second": 83.74,
68
+ "eval_steps_per_second": 0.661,
69
+ "step": 62
70
+ },
71
+ {
72
+ "epoch": 2.24,
73
+ "learning_rate": 7.526881720430108e-05,
74
+ "loss": 0.741,
75
+ "step": 70
76
+ },
77
+ {
78
+ "epoch": 2.56,
79
+ "learning_rate": 8.60215053763441e-05,
80
+ "loss": 0.6928,
81
+ "step": 80
82
+ },
83
+ {
84
+ "epoch": 2.88,
85
+ "learning_rate": 9.677419354838711e-05,
86
+ "loss": 0.7197,
87
+ "step": 90
88
+ },
89
+ {
90
+ "epoch": 2.98,
91
+ "eval_accuracy": 0.7784667418263811,
92
+ "eval_f1": 0.7704940588846539,
93
+ "eval_loss": 0.6200318336486816,
94
+ "eval_precision": 0.7715960207478135,
95
+ "eval_recall": 0.7784667418263811,
96
+ "eval_runtime": 21.4187,
97
+ "eval_samples_per_second": 82.825,
98
+ "eval_steps_per_second": 0.654,
99
+ "step": 93
100
+ },
101
+ {
102
+ "epoch": 3.2,
103
+ "learning_rate": 9.916367980884111e-05,
104
+ "loss": 0.691,
105
+ "step": 100
106
+ },
107
+ {
108
+ "epoch": 3.52,
109
+ "learning_rate": 9.79689366786141e-05,
110
+ "loss": 0.6486,
111
+ "step": 110
112
+ },
113
+ {
114
+ "epoch": 3.84,
115
+ "learning_rate": 9.677419354838711e-05,
116
+ "loss": 0.6321,
117
+ "step": 120
118
+ },
119
+ {
120
+ "epoch": 4.0,
121
+ "eval_accuracy": 0.8060879368658399,
122
+ "eval_f1": 0.7957042518679185,
123
+ "eval_loss": 0.5692886710166931,
124
+ "eval_precision": 0.8035246330156649,
125
+ "eval_recall": 0.8060879368658399,
126
+ "eval_runtime": 21.6457,
127
+ "eval_samples_per_second": 81.956,
128
+ "eval_steps_per_second": 0.647,
129
+ "step": 125
130
+ },
131
+ {
132
+ "epoch": 4.16,
133
+ "learning_rate": 9.55794504181601e-05,
134
+ "loss": 0.5927,
135
+ "step": 130
136
+ },
137
+ {
138
+ "epoch": 4.48,
139
+ "learning_rate": 9.438470728793309e-05,
140
+ "loss": 0.5869,
141
+ "step": 140
142
+ },
143
+ {
144
+ "epoch": 4.8,
145
+ "learning_rate": 9.31899641577061e-05,
146
+ "loss": 0.5768,
147
+ "step": 150
148
+ },
149
+ {
150
+ "epoch": 4.99,
151
+ "eval_accuracy": 0.8111612175873731,
152
+ "eval_f1": 0.8134180707239236,
153
+ "eval_loss": 0.5500875115394592,
154
+ "eval_precision": 0.8213061319477135,
155
+ "eval_recall": 0.8111612175873731,
156
+ "eval_runtime": 21.6943,
157
+ "eval_samples_per_second": 81.773,
158
+ "eval_steps_per_second": 0.645,
159
+ "step": 156
160
+ },
161
+ {
162
+ "epoch": 5.12,
163
+ "learning_rate": 9.199522102747909e-05,
164
+ "loss": 0.5286,
165
+ "step": 160
166
+ },
167
+ {
168
+ "epoch": 5.44,
169
+ "learning_rate": 9.080047789725208e-05,
170
+ "loss": 0.5513,
171
+ "step": 170
172
+ },
173
+ {
174
+ "epoch": 5.76,
175
+ "learning_rate": 8.960573476702509e-05,
176
+ "loss": 0.5173,
177
+ "step": 180
178
+ },
179
+ {
180
+ "epoch": 5.98,
181
+ "eval_accuracy": 0.8213077790304397,
182
+ "eval_f1": 0.8201874264260287,
183
+ "eval_loss": 0.5164684057235718,
184
+ "eval_precision": 0.8305619852770596,
185
+ "eval_recall": 0.8213077790304397,
186
+ "eval_runtime": 21.6677,
187
+ "eval_samples_per_second": 81.873,
188
+ "eval_steps_per_second": 0.646,
189
+ "step": 187
190
+ },
191
+ {
192
+ "epoch": 6.08,
193
+ "learning_rate": 8.84109916367981e-05,
194
+ "loss": 0.5143,
195
+ "step": 190
196
+ },
197
+ {
198
+ "epoch": 6.4,
199
+ "learning_rate": 8.72162485065711e-05,
200
+ "loss": 0.498,
201
+ "step": 200
202
+ },
203
+ {
204
+ "epoch": 6.72,
205
+ "learning_rate": 8.60215053763441e-05,
206
+ "loss": 0.4781,
207
+ "step": 210
208
+ },
209
+ {
210
+ "epoch": 6.98,
211
+ "eval_accuracy": 0.8105975197294251,
212
+ "eval_f1": 0.8089843496723177,
213
+ "eval_loss": 0.5220049023628235,
214
+ "eval_precision": 0.8161270722410587,
215
+ "eval_recall": 0.8105975197294251,
216
+ "eval_runtime": 21.7247,
217
+ "eval_samples_per_second": 81.658,
218
+ "eval_steps_per_second": 0.644,
219
+ "step": 218
220
+ },
221
+ {
222
+ "epoch": 7.04,
223
+ "learning_rate": 8.482676224611709e-05,
224
+ "loss": 0.4999,
225
+ "step": 220
226
+ },
227
+ {
228
+ "epoch": 7.36,
229
+ "learning_rate": 8.363201911589009e-05,
230
+ "loss": 0.4604,
231
+ "step": 230
232
+ },
233
+ {
234
+ "epoch": 7.68,
235
+ "learning_rate": 8.243727598566309e-05,
236
+ "loss": 0.4807,
237
+ "step": 240
238
+ },
239
+ {
240
+ "epoch": 8.0,
241
+ "learning_rate": 8.124253285543608e-05,
242
+ "loss": 0.451,
243
+ "step": 250
244
+ },
245
+ {
246
+ "epoch": 8.0,
247
+ "eval_accuracy": 0.818489289740699,
248
+ "eval_f1": 0.8152695783103332,
249
+ "eval_loss": 0.5133233666419983,
250
+ "eval_precision": 0.8227391031387269,
251
+ "eval_recall": 0.818489289740699,
252
+ "eval_runtime": 21.9192,
253
+ "eval_samples_per_second": 80.934,
254
+ "eval_steps_per_second": 0.639,
255
+ "step": 250
256
+ },
257
+ {
258
+ "epoch": 8.32,
259
+ "learning_rate": 8.004778972520908e-05,
260
+ "loss": 0.4252,
261
+ "step": 260
262
+ },
263
+ {
264
+ "epoch": 8.64,
265
+ "learning_rate": 7.885304659498209e-05,
266
+ "loss": 0.4359,
267
+ "step": 270
268
+ },
269
+ {
270
+ "epoch": 8.96,
271
+ "learning_rate": 7.765830346475508e-05,
272
+ "loss": 0.4373,
273
+ "step": 280
274
+ },
275
+ {
276
+ "epoch": 8.99,
277
+ "eval_accuracy": 0.8303269447576099,
278
+ "eval_f1": 0.8288008582358363,
279
+ "eval_loss": 0.511769711971283,
280
+ "eval_precision": 0.8325406030306683,
281
+ "eval_recall": 0.8303269447576099,
282
+ "eval_runtime": 21.4401,
283
+ "eval_samples_per_second": 82.742,
284
+ "eval_steps_per_second": 0.653,
285
+ "step": 281
286
+ },
287
+ {
288
+ "epoch": 9.28,
289
+ "learning_rate": 7.646356033452809e-05,
290
+ "loss": 0.4023,
291
+ "step": 290
292
+ },
293
+ {
294
+ "epoch": 9.6,
295
+ "learning_rate": 7.526881720430108e-05,
296
+ "loss": 0.396,
297
+ "step": 300
298
+ },
299
+ {
300
+ "epoch": 9.92,
301
+ "learning_rate": 7.407407407407407e-05,
302
+ "loss": 0.3826,
303
+ "step": 310
304
+ },
305
+ {
306
+ "epoch": 9.98,
307
+ "eval_accuracy": 0.8258173618940248,
308
+ "eval_f1": 0.8243103846150182,
309
+ "eval_loss": 0.5279704928398132,
310
+ "eval_precision": 0.8268898344898333,
311
+ "eval_recall": 0.8258173618940248,
312
+ "eval_runtime": 21.4879,
313
+ "eval_samples_per_second": 82.558,
314
+ "eval_steps_per_second": 0.652,
315
+ "step": 312
316
+ },
317
+ {
318
+ "epoch": 10.24,
319
+ "learning_rate": 7.287933094384708e-05,
320
+ "loss": 0.363,
321
+ "step": 320
322
+ },
323
+ {
324
+ "epoch": 10.56,
325
+ "learning_rate": 7.168458781362007e-05,
326
+ "loss": 0.376,
327
+ "step": 330
328
+ },
329
+ {
330
+ "epoch": 10.88,
331
+ "learning_rate": 7.048984468339306e-05,
332
+ "loss": 0.378,
333
+ "step": 340
334
+ },
335
+ {
336
+ "epoch": 10.98,
337
+ "eval_accuracy": 0.8173618940248027,
338
+ "eval_f1": 0.8141777832297877,
339
+ "eval_loss": 0.5476882457733154,
340
+ "eval_precision": 0.8156045012406116,
341
+ "eval_recall": 0.8173618940248027,
342
+ "eval_runtime": 21.5324,
343
+ "eval_samples_per_second": 82.387,
344
+ "eval_steps_per_second": 0.65,
345
+ "step": 343
346
+ },
347
+ {
348
+ "epoch": 11.2,
349
+ "learning_rate": 6.929510155316607e-05,
350
+ "loss": 0.3416,
351
+ "step": 350
352
+ },
353
+ {
354
+ "epoch": 11.52,
355
+ "learning_rate": 6.810035842293908e-05,
356
+ "loss": 0.337,
357
+ "step": 360
358
+ },
359
+ {
360
+ "epoch": 11.84,
361
+ "learning_rate": 6.690561529271207e-05,
362
+ "loss": 0.3509,
363
+ "step": 370
364
+ },
365
+ {
366
+ "epoch": 12.0,
367
+ "eval_accuracy": 0.8280721533258174,
368
+ "eval_f1": 0.8244469418718917,
369
+ "eval_loss": 0.5437383055686951,
370
+ "eval_precision": 0.8291545990959505,
371
+ "eval_recall": 0.8280721533258174,
372
+ "eval_runtime": 21.4443,
373
+ "eval_samples_per_second": 82.726,
374
+ "eval_steps_per_second": 0.653,
375
+ "step": 375
376
+ },
377
+ {
378
+ "epoch": 12.16,
379
+ "learning_rate": 6.571087216248507e-05,
380
+ "loss": 0.3258,
381
+ "step": 380
382
+ },
383
+ {
384
+ "epoch": 12.48,
385
+ "learning_rate": 6.451612903225807e-05,
386
+ "loss": 0.3173,
387
+ "step": 390
388
+ },
389
+ {
390
+ "epoch": 12.8,
391
+ "learning_rate": 6.332138590203107e-05,
392
+ "loss": 0.3358,
393
+ "step": 400
394
+ },
395
+ {
396
+ "epoch": 12.99,
397
+ "eval_accuracy": 0.8258173618940248,
398
+ "eval_f1": 0.8240960752396663,
399
+ "eval_loss": 0.5626731514930725,
400
+ "eval_precision": 0.8267626478179836,
401
+ "eval_recall": 0.8258173618940248,
402
+ "eval_runtime": 21.2051,
403
+ "eval_samples_per_second": 83.659,
404
+ "eval_steps_per_second": 0.66,
405
+ "step": 406
406
+ },
407
+ {
408
+ "epoch": 13.12,
409
+ "learning_rate": 6.212664277180407e-05,
410
+ "loss": 0.3247,
411
+ "step": 410
412
+ },
413
+ {
414
+ "epoch": 13.44,
415
+ "learning_rate": 6.093189964157706e-05,
416
+ "loss": 0.3109,
417
+ "step": 420
418
+ },
419
+ {
420
+ "epoch": 13.76,
421
+ "learning_rate": 5.9737156511350064e-05,
422
+ "loss": 0.3027,
423
+ "step": 430
424
+ },
425
+ {
426
+ "epoch": 13.98,
427
+ "eval_accuracy": 0.8325817361894025,
428
+ "eval_f1": 0.8309989322501777,
429
+ "eval_loss": 0.5557559728622437,
430
+ "eval_precision": 0.8341182995121916,
431
+ "eval_recall": 0.8325817361894025,
432
+ "eval_runtime": 21.7522,
433
+ "eval_samples_per_second": 81.555,
434
+ "eval_steps_per_second": 0.644,
435
+ "step": 437
436
+ },
437
+ {
438
+ "epoch": 14.08,
439
+ "learning_rate": 5.8542413381123063e-05,
440
+ "loss": 0.308,
441
+ "step": 440
442
+ },
443
+ {
444
+ "epoch": 14.4,
445
+ "learning_rate": 5.7347670250896056e-05,
446
+ "loss": 0.2783,
447
+ "step": 450
448
+ },
449
+ {
450
+ "epoch": 14.72,
451
+ "learning_rate": 5.615292712066906e-05,
452
+ "loss": 0.3027,
453
+ "step": 460
454
+ },
455
+ {
456
+ "epoch": 14.98,
457
+ "eval_accuracy": 0.8325817361894025,
458
+ "eval_f1": 0.8294913013624391,
459
+ "eval_loss": 0.5702593922615051,
460
+ "eval_precision": 0.835774810800051,
461
+ "eval_recall": 0.8325817361894025,
462
+ "eval_runtime": 21.7146,
463
+ "eval_samples_per_second": 81.696,
464
+ "eval_steps_per_second": 0.645,
465
+ "step": 468
466
+ },
467
+ {
468
+ "epoch": 15.04,
469
+ "learning_rate": 5.495818399044206e-05,
470
+ "loss": 0.2816,
471
+ "step": 470
472
+ },
473
+ {
474
+ "epoch": 15.36,
475
+ "learning_rate": 5.3763440860215054e-05,
476
+ "loss": 0.2662,
477
+ "step": 480
478
+ },
479
+ {
480
+ "epoch": 15.68,
481
+ "learning_rate": 5.256869772998806e-05,
482
+ "loss": 0.2708,
483
+ "step": 490
484
+ },
485
+ {
486
+ "epoch": 16.0,
487
+ "learning_rate": 5.137395459976105e-05,
488
+ "loss": 0.2786,
489
+ "step": 500
490
+ },
491
+ {
492
+ "epoch": 16.0,
493
+ "eval_accuracy": 0.8280721533258174,
494
+ "eval_f1": 0.8248780707496759,
495
+ "eval_loss": 0.5790648460388184,
496
+ "eval_precision": 0.8268104951659752,
497
+ "eval_recall": 0.8280721533258174,
498
+ "eval_runtime": 21.3703,
499
+ "eval_samples_per_second": 83.012,
500
+ "eval_steps_per_second": 0.655,
501
+ "step": 500
502
+ },
503
+ {
504
+ "epoch": 16.32,
505
+ "learning_rate": 5.017921146953405e-05,
506
+ "loss": 0.2498,
507
+ "step": 510
508
+ },
509
+ {
510
+ "epoch": 16.64,
511
+ "learning_rate": 4.898446833930705e-05,
512
+ "loss": 0.2711,
513
+ "step": 520
514
+ },
515
+ {
516
+ "epoch": 16.96,
517
+ "learning_rate": 4.778972520908005e-05,
518
+ "loss": 0.2379,
519
+ "step": 530
520
+ },
521
+ {
522
+ "epoch": 16.99,
523
+ "eval_accuracy": 0.8275084554678692,
524
+ "eval_f1": 0.825106664588863,
525
+ "eval_loss": 0.5864401459693909,
526
+ "eval_precision": 0.8264428043886305,
527
+ "eval_recall": 0.8275084554678692,
528
+ "eval_runtime": 21.5135,
529
+ "eval_samples_per_second": 82.46,
530
+ "eval_steps_per_second": 0.651,
531
+ "step": 531
532
+ },
533
+ {
534
+ "epoch": 17.28,
535
+ "learning_rate": 4.659498207885305e-05,
536
+ "loss": 0.2566,
537
+ "step": 540
538
+ },
539
+ {
540
+ "epoch": 17.6,
541
+ "learning_rate": 4.540023894862604e-05,
542
+ "loss": 0.2602,
543
+ "step": 550
544
+ },
545
+ {
546
+ "epoch": 17.92,
547
+ "learning_rate": 4.420549581839905e-05,
548
+ "loss": 0.2426,
549
+ "step": 560
550
+ },
551
+ {
552
+ "epoch": 17.98,
553
+ "eval_accuracy": 0.8320180383314544,
554
+ "eval_f1": 0.8305311136335276,
555
+ "eval_loss": 0.5984331369400024,
556
+ "eval_precision": 0.8319552266233834,
557
+ "eval_recall": 0.8320180383314544,
558
+ "eval_runtime": 21.366,
559
+ "eval_samples_per_second": 83.029,
560
+ "eval_steps_per_second": 0.655,
561
+ "step": 562
562
+ },
563
+ {
564
+ "epoch": 18.24,
565
+ "learning_rate": 4.301075268817205e-05,
566
+ "loss": 0.241,
567
+ "step": 570
568
+ },
569
+ {
570
+ "epoch": 18.56,
571
+ "learning_rate": 4.1816009557945046e-05,
572
+ "loss": 0.2346,
573
+ "step": 580
574
+ },
575
+ {
576
+ "epoch": 18.88,
577
+ "learning_rate": 4.062126642771804e-05,
578
+ "loss": 0.2325,
579
+ "step": 590
580
+ },
581
+ {
582
+ "epoch": 18.98,
583
+ "eval_accuracy": 0.826381059751973,
584
+ "eval_f1": 0.8251810645598993,
585
+ "eval_loss": 0.6216564774513245,
586
+ "eval_precision": 0.8280960122554618,
587
+ "eval_recall": 0.826381059751973,
588
+ "eval_runtime": 21.6911,
589
+ "eval_samples_per_second": 81.785,
590
+ "eval_steps_per_second": 0.645,
591
+ "step": 593
592
+ },
593
+ {
594
+ "epoch": 19.2,
595
+ "learning_rate": 3.9426523297491045e-05,
596
+ "loss": 0.2296,
597
+ "step": 600
598
+ },
599
+ {
600
+ "epoch": 19.52,
601
+ "learning_rate": 3.8231780167264044e-05,
602
+ "loss": 0.2296,
603
+ "step": 610
604
+ },
605
+ {
606
+ "epoch": 19.84,
607
+ "learning_rate": 3.7037037037037037e-05,
608
+ "loss": 0.2208,
609
+ "step": 620
610
+ },
611
+ {
612
+ "epoch": 20.0,
613
+ "eval_accuracy": 0.8258173618940248,
614
+ "eval_f1": 0.8235775028778175,
615
+ "eval_loss": 0.6165610551834106,
616
+ "eval_precision": 0.8229737232245232,
617
+ "eval_recall": 0.8258173618940248,
618
+ "eval_runtime": 22.0389,
619
+ "eval_samples_per_second": 80.494,
620
+ "eval_steps_per_second": 0.635,
621
+ "step": 625
622
+ },
623
+ {
624
+ "epoch": 20.16,
625
+ "learning_rate": 3.5842293906810036e-05,
626
+ "loss": 0.2187,
627
+ "step": 630
628
+ },
629
+ {
630
+ "epoch": 20.48,
631
+ "learning_rate": 3.4647550776583035e-05,
632
+ "loss": 0.2067,
633
+ "step": 640
634
+ },
635
+ {
636
+ "epoch": 20.8,
637
+ "learning_rate": 3.3452807646356034e-05,
638
+ "loss": 0.2196,
639
+ "step": 650
640
+ },
641
+ {
642
+ "epoch": 20.99,
643
+ "eval_accuracy": 0.8286358511837655,
644
+ "eval_f1": 0.825876975856749,
645
+ "eval_loss": 0.6308055520057678,
646
+ "eval_precision": 0.8279920279498516,
647
+ "eval_recall": 0.8286358511837655,
648
+ "eval_runtime": 21.4064,
649
+ "eval_samples_per_second": 82.872,
650
+ "eval_steps_per_second": 0.654,
651
+ "step": 656
652
+ },
653
+ {
654
+ "epoch": 21.12,
655
+ "learning_rate": 3.2258064516129034e-05,
656
+ "loss": 0.2034,
657
+ "step": 660
658
+ },
659
+ {
660
+ "epoch": 21.44,
661
+ "learning_rate": 3.106332138590203e-05,
662
+ "loss": 0.2098,
663
+ "step": 670
664
+ },
665
+ {
666
+ "epoch": 21.76,
667
+ "learning_rate": 2.9868578255675032e-05,
668
+ "loss": 0.2077,
669
+ "step": 680
670
+ },
671
+ {
672
+ "epoch": 21.98,
673
+ "eval_accuracy": 0.8325817361894025,
674
+ "eval_f1": 0.8305408181674179,
675
+ "eval_loss": 0.624207079410553,
676
+ "eval_precision": 0.8307265402722898,
677
+ "eval_recall": 0.8325817361894025,
678
+ "eval_runtime": 21.4301,
679
+ "eval_samples_per_second": 82.781,
680
+ "eval_steps_per_second": 0.653,
681
+ "step": 687
682
+ },
683
+ {
684
+ "epoch": 22.08,
685
+ "learning_rate": 2.8673835125448028e-05,
686
+ "loss": 0.2177,
687
+ "step": 690
688
+ },
689
+ {
690
+ "epoch": 22.4,
691
+ "learning_rate": 2.747909199522103e-05,
692
+ "loss": 0.2023,
693
+ "step": 700
694
+ },
695
+ {
696
+ "epoch": 22.72,
697
+ "learning_rate": 2.628434886499403e-05,
698
+ "loss": 0.2048,
699
+ "step": 710
700
+ },
701
+ {
702
+ "epoch": 22.98,
703
+ "eval_accuracy": 0.8275084554678692,
704
+ "eval_f1": 0.8263134782090512,
705
+ "eval_loss": 0.6801280379295349,
706
+ "eval_precision": 0.8303435791241508,
707
+ "eval_recall": 0.8275084554678692,
708
+ "eval_runtime": 21.9211,
709
+ "eval_samples_per_second": 80.927,
710
+ "eval_steps_per_second": 0.639,
711
+ "step": 718
712
+ },
713
+ {
714
+ "epoch": 23.04,
715
+ "learning_rate": 2.5089605734767026e-05,
716
+ "loss": 0.2072,
717
+ "step": 720
718
+ },
719
+ {
720
+ "epoch": 23.36,
721
+ "learning_rate": 2.3894862604540025e-05,
722
+ "loss": 0.2009,
723
+ "step": 730
724
+ },
725
+ {
726
+ "epoch": 23.68,
727
+ "learning_rate": 2.270011947431302e-05,
728
+ "loss": 0.2083,
729
+ "step": 740
730
+ },
731
+ {
732
+ "epoch": 24.0,
733
+ "learning_rate": 2.1505376344086024e-05,
734
+ "loss": 0.1886,
735
+ "step": 750
736
+ },
737
+ {
738
+ "epoch": 24.0,
739
+ "eval_accuracy": 0.826381059751973,
740
+ "eval_f1": 0.8255542142311617,
741
+ "eval_loss": 0.6615379452705383,
742
+ "eval_precision": 0.8279561568090393,
743
+ "eval_recall": 0.826381059751973,
744
+ "eval_runtime": 21.4625,
745
+ "eval_samples_per_second": 82.656,
746
+ "eval_steps_per_second": 0.652,
747
+ "step": 750
748
+ },
749
+ {
750
+ "epoch": 24.32,
751
+ "learning_rate": 2.031063321385902e-05,
752
+ "loss": 0.2118,
753
+ "step": 760
754
+ },
755
+ {
756
+ "epoch": 24.64,
757
+ "learning_rate": 1.9115890083632022e-05,
758
+ "loss": 0.1872,
759
+ "step": 770
760
+ },
761
+ {
762
+ "epoch": 24.96,
763
+ "learning_rate": 1.7921146953405018e-05,
764
+ "loss": 0.2007,
765
+ "step": 780
766
+ },
767
+ {
768
+ "epoch": 24.99,
769
+ "eval_accuracy": 0.8275084554678692,
770
+ "eval_f1": 0.8267436914020984,
771
+ "eval_loss": 0.6847040057182312,
772
+ "eval_precision": 0.8279552479088658,
773
+ "eval_recall": 0.8275084554678692,
774
+ "eval_runtime": 21.4978,
775
+ "eval_samples_per_second": 82.52,
776
+ "eval_steps_per_second": 0.651,
777
+ "step": 781
778
+ },
779
+ {
780
+ "epoch": 25.28,
781
+ "learning_rate": 1.6726403823178017e-05,
782
+ "loss": 0.1844,
783
+ "step": 790
784
+ },
785
+ {
786
+ "epoch": 25.6,
787
+ "learning_rate": 1.5531660692951016e-05,
788
+ "loss": 0.1852,
789
+ "step": 800
790
+ },
791
+ {
792
+ "epoch": 25.92,
793
+ "learning_rate": 1.4336917562724014e-05,
794
+ "loss": 0.1815,
795
+ "step": 810
796
+ },
797
+ {
798
+ "epoch": 25.98,
799
+ "eval_accuracy": 0.8325817361894025,
800
+ "eval_f1": 0.830537122348923,
801
+ "eval_loss": 0.6668907999992371,
802
+ "eval_precision": 0.8311252959744954,
803
+ "eval_recall": 0.8325817361894025,
804
+ "eval_runtime": 21.8156,
805
+ "eval_samples_per_second": 81.318,
806
+ "eval_steps_per_second": 0.642,
807
+ "step": 812
808
+ },
809
+ {
810
+ "epoch": 26.24,
811
+ "learning_rate": 1.3142174432497015e-05,
812
+ "loss": 0.197,
813
+ "step": 820
814
+ },
815
+ {
816
+ "epoch": 26.56,
817
+ "learning_rate": 1.1947431302270013e-05,
818
+ "loss": 0.1913,
819
+ "step": 830
820
+ },
821
+ {
822
+ "epoch": 26.88,
823
+ "learning_rate": 1.0752688172043012e-05,
824
+ "loss": 0.1958,
825
+ "step": 840
826
+ },
827
+ {
828
+ "epoch": 26.98,
829
+ "eval_accuracy": 0.8370913190529876,
830
+ "eval_f1": 0.8357334072148467,
831
+ "eval_loss": 0.6766165494918823,
832
+ "eval_precision": 0.8373589976039267,
833
+ "eval_recall": 0.8370913190529876,
834
+ "eval_runtime": 21.0261,
835
+ "eval_samples_per_second": 84.371,
836
+ "eval_steps_per_second": 0.666,
837
+ "step": 843
838
+ },
839
+ {
840
+ "epoch": 27.2,
841
+ "learning_rate": 9.557945041816011e-06,
842
+ "loss": 0.1884,
843
+ "step": 850
844
+ },
845
+ {
846
+ "epoch": 27.52,
847
+ "learning_rate": 8.363201911589009e-06,
848
+ "loss": 0.1969,
849
+ "step": 860
850
+ },
851
+ {
852
+ "epoch": 27.84,
853
+ "learning_rate": 7.168458781362007e-06,
854
+ "loss": 0.1806,
855
+ "step": 870
856
+ },
857
+ {
858
+ "epoch": 28.0,
859
+ "eval_accuracy": 0.8359639233370914,
860
+ "eval_f1": 0.8341906674178488,
861
+ "eval_loss": 0.6678570508956909,
862
+ "eval_precision": 0.8353079467237093,
863
+ "eval_recall": 0.8359639233370914,
864
+ "eval_runtime": 21.3886,
865
+ "eval_samples_per_second": 82.941,
866
+ "eval_steps_per_second": 0.655,
867
+ "step": 875
868
+ },
869
+ {
870
+ "epoch": 28.16,
871
+ "learning_rate": 5.973715651135006e-06,
872
+ "loss": 0.1958,
873
+ "step": 880
874
+ },
875
+ {
876
+ "epoch": 28.48,
877
+ "learning_rate": 4.7789725209080055e-06,
878
+ "loss": 0.1751,
879
+ "step": 890
880
+ },
881
+ {
882
+ "epoch": 28.8,
883
+ "learning_rate": 3.5842293906810035e-06,
884
+ "loss": 0.1835,
885
+ "step": 900
886
+ },
887
+ {
888
+ "epoch": 28.99,
889
+ "eval_accuracy": 0.8348365276211951,
890
+ "eval_f1": 0.8327921869627218,
891
+ "eval_loss": 0.6767047047615051,
892
+ "eval_precision": 0.8333572524880747,
893
+ "eval_recall": 0.8348365276211951,
894
+ "eval_runtime": 21.4548,
895
+ "eval_samples_per_second": 82.685,
896
+ "eval_steps_per_second": 0.653,
897
+ "step": 906
898
+ },
899
+ {
900
+ "epoch": 29.12,
901
+ "learning_rate": 2.3894862604540028e-06,
902
+ "loss": 0.1893,
903
+ "step": 910
904
+ },
905
+ {
906
+ "epoch": 29.44,
907
+ "learning_rate": 1.1947431302270014e-06,
908
+ "loss": 0.1696,
909
+ "step": 920
910
+ },
911
+ {
912
+ "epoch": 29.76,
913
+ "learning_rate": 0.0,
914
+ "loss": 0.1796,
915
+ "step": 930
916
+ },
917
+ {
918
+ "epoch": 29.76,
919
+ "eval_accuracy": 0.8342728297632469,
920
+ "eval_f1": 0.8325855632553308,
921
+ "eval_loss": 0.6786999702453613,
922
+ "eval_precision": 0.8335690065497708,
923
+ "eval_recall": 0.8342728297632469,
924
+ "eval_runtime": 21.4373,
925
+ "eval_samples_per_second": 82.753,
926
+ "eval_steps_per_second": 0.653,
927
+ "step": 930
928
+ },
929
+ {
930
+ "epoch": 29.76,
931
+ "step": 930,
932
+ "total_flos": 3.853544865721595e+20,
933
+ "train_loss": 0.3928444991829575,
934
+ "train_runtime": 10565.0301,
935
+ "train_samples_per_second": 45.325,
936
+ "train_steps_per_second": 0.088
937
+ }
938
+ ],
939
+ "logging_steps": 10,
940
+ "max_steps": 930,
941
+ "num_input_tokens_seen": 0,
942
+ "num_train_epochs": 30,
943
+ "save_steps": 500,
944
+ "total_flos": 3.853544865721595e+20,
945
+ "train_batch_size": 128,
946
+ "trial_name": null,
947
+ "trial_params": null
948
+ }