tmnam20 commited on
Commit
0e9b733
1 Parent(s): bcfdb9b

Training in progress, step 27000, checkpoint

Browse files
checkpoint-27000/added_tokens.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "<bos>": 40478,
3
+ "<eos>": 40479,
4
+ "<pad>": 40480
5
+ }
checkpoint-27000/config.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai-gpt",
3
+ "afn": "gelu",
4
+ "architectures": [
5
+ "OpenAIGPTForSequenceClassification"
6
+ ],
7
+ "attn_pdrop": 0.1,
8
+ "embd_pdrop": 0.1,
9
+ "id2label": {
10
+ "0": "entailment",
11
+ "1": "neutral",
12
+ "2": "contradiction"
13
+ },
14
+ "initializer_range": 0.02,
15
+ "label2id": {
16
+ "contradiction": 2,
17
+ "entailment": 0,
18
+ "neutral": 1
19
+ },
20
+ "layer_norm_epsilon": 1e-05,
21
+ "model_type": "openai-gpt",
22
+ "n_ctx": 512,
23
+ "n_embd": 768,
24
+ "n_head": 6,
25
+ "n_layer": 12,
26
+ "n_positions": 512,
27
+ "n_special": 0,
28
+ "pad_token_id": 40480,
29
+ "predict_special_tokens": true,
30
+ "problem_type": "single_label_classification",
31
+ "resid_pdrop": 0.1,
32
+ "summary_activation": null,
33
+ "summary_first_dropout": 0.1,
34
+ "summary_proj_to_labels": true,
35
+ "summary_type": "cls_index",
36
+ "summary_use_proj": true,
37
+ "task_specific_params": {
38
+ "text-generation": {
39
+ "do_sample": true,
40
+ "max_length": 50
41
+ }
42
+ },
43
+ "torch_dtype": "float32",
44
+ "transformers_version": "4.33.0.dev0",
45
+ "vocab_size": 40481
46
+ }
checkpoint-27000/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-27000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87786f73370381346c468ed817ffe125edb1305388889d586eb368a08914ef2a
3
+ size 932401285
checkpoint-27000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e84285aebaf65b2a32c3ee026aaa6e13796f6f011251116cb993491829434728
3
+ size 466205145
checkpoint-27000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ed4fbc89f9039d3a6f7dd0f6eb180ed8f31178752bafb5e941305dc0443e5105
3
+ size 14575
checkpoint-27000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f587b625c7607960fc08e1d336d90d848a943eed14da9f8dbb8235406e35392
3
+ size 627
checkpoint-27000/special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<bos>",
3
+ "eos_token": "<eos>",
4
+ "pad_token": "<pad>",
5
+ "unk_token": "<unk>"
6
+ }
checkpoint-27000/tokenizer_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "clean_up_tokenization_spaces": true,
3
+ "model_max_length": 512,
4
+ "padding_side": "left",
5
+ "tokenizer_class": "OpenAIGPTTokenizer",
6
+ "unk_token": "<unk>"
7
+ }
checkpoint-27000/trainer_state.json ADDED
@@ -0,0 +1,430 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.0436949729919434,
3
+ "best_model_checkpoint": "/kaggle/output/checkpoint-27000",
4
+ "epoch": 1.1000651890482398,
5
+ "eval_steps": 1000,
6
+ "global_step": 27000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "learning_rate": 2.7777777777777777e-11,
14
+ "loss": 1.2076,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.04,
19
+ "learning_rate": 2.7750000000000004e-08,
20
+ "loss": 1.1301,
21
+ "step": 1000
22
+ },
23
+ {
24
+ "epoch": 0.04,
25
+ "eval_accuracy": 0.3321357285429142,
26
+ "eval_loss": 1.1047232151031494,
27
+ "eval_runtime": 50.8344,
28
+ "eval_samples_per_second": 98.555,
29
+ "eval_steps_per_second": 12.334,
30
+ "step": 1000
31
+ },
32
+ {
33
+ "epoch": 0.08,
34
+ "learning_rate": 5.5527777777777784e-08,
35
+ "loss": 1.1286,
36
+ "step": 2000
37
+ },
38
+ {
39
+ "epoch": 0.08,
40
+ "eval_accuracy": 0.3377245508982036,
41
+ "eval_loss": 1.1038333177566528,
42
+ "eval_runtime": 50.8114,
43
+ "eval_samples_per_second": 98.6,
44
+ "eval_steps_per_second": 12.34,
45
+ "step": 2000
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "learning_rate": 8.327777777777778e-08,
50
+ "loss": 1.1214,
51
+ "step": 3000
52
+ },
53
+ {
54
+ "epoch": 0.12,
55
+ "eval_accuracy": 0.3409181636726547,
56
+ "eval_loss": 1.103529930114746,
57
+ "eval_runtime": 50.8258,
58
+ "eval_samples_per_second": 98.572,
59
+ "eval_steps_per_second": 12.336,
60
+ "step": 3000
61
+ },
62
+ {
63
+ "epoch": 0.16,
64
+ "learning_rate": 1.1105555555555557e-07,
65
+ "loss": 1.1249,
66
+ "step": 4000
67
+ },
68
+ {
69
+ "epoch": 0.16,
70
+ "eval_accuracy": 0.3469061876247505,
71
+ "eval_loss": 1.1000434160232544,
72
+ "eval_runtime": 50.6414,
73
+ "eval_samples_per_second": 98.931,
74
+ "eval_steps_per_second": 12.381,
75
+ "step": 4000
76
+ },
77
+ {
78
+ "epoch": 0.2,
79
+ "learning_rate": 1.3880555555555558e-07,
80
+ "loss": 1.1206,
81
+ "step": 5000
82
+ },
83
+ {
84
+ "epoch": 0.2,
85
+ "eval_accuracy": 0.34171656686626745,
86
+ "eval_loss": 1.0996097326278687,
87
+ "eval_runtime": 50.6956,
88
+ "eval_samples_per_second": 98.825,
89
+ "eval_steps_per_second": 12.368,
90
+ "step": 5000
91
+ },
92
+ {
93
+ "epoch": 0.24,
94
+ "learning_rate": 1.6658333333333335e-07,
95
+ "loss": 1.1161,
96
+ "step": 6000
97
+ },
98
+ {
99
+ "epoch": 0.24,
100
+ "eval_accuracy": 0.3590818363273453,
101
+ "eval_loss": 1.096977710723877,
102
+ "eval_runtime": 50.5826,
103
+ "eval_samples_per_second": 99.046,
104
+ "eval_steps_per_second": 12.396,
105
+ "step": 6000
106
+ },
107
+ {
108
+ "epoch": 0.29,
109
+ "learning_rate": 1.9433333333333334e-07,
110
+ "loss": 1.1138,
111
+ "step": 7000
112
+ },
113
+ {
114
+ "epoch": 0.29,
115
+ "eval_accuracy": 0.3532934131736527,
116
+ "eval_loss": 1.0949815511703491,
117
+ "eval_runtime": 50.5604,
118
+ "eval_samples_per_second": 99.089,
119
+ "eval_steps_per_second": 12.401,
120
+ "step": 7000
121
+ },
122
+ {
123
+ "epoch": 0.33,
124
+ "learning_rate": 2.2211111111111114e-07,
125
+ "loss": 1.1135,
126
+ "step": 8000
127
+ },
128
+ {
129
+ "epoch": 0.33,
130
+ "eval_accuracy": 0.36327345309381237,
131
+ "eval_loss": 1.093558430671692,
132
+ "eval_runtime": 50.608,
133
+ "eval_samples_per_second": 98.996,
134
+ "eval_steps_per_second": 12.389,
135
+ "step": 8000
136
+ },
137
+ {
138
+ "epoch": 0.37,
139
+ "learning_rate": 2.4986111111111113e-07,
140
+ "loss": 1.1063,
141
+ "step": 9000
142
+ },
143
+ {
144
+ "epoch": 0.37,
145
+ "eval_accuracy": 0.35548902195608784,
146
+ "eval_loss": 1.0992107391357422,
147
+ "eval_runtime": 50.5434,
148
+ "eval_samples_per_second": 99.123,
149
+ "eval_steps_per_second": 12.405,
150
+ "step": 9000
151
+ },
152
+ {
153
+ "epoch": 0.41,
154
+ "learning_rate": 2.776388888888889e-07,
155
+ "loss": 1.1083,
156
+ "step": 10000
157
+ },
158
+ {
159
+ "epoch": 0.41,
160
+ "eval_accuracy": 0.3836327345309381,
161
+ "eval_loss": 1.0920206308364868,
162
+ "eval_runtime": 50.5559,
163
+ "eval_samples_per_second": 99.098,
164
+ "eval_steps_per_second": 12.402,
165
+ "step": 10000
166
+ },
167
+ {
168
+ "epoch": 0.45,
169
+ "learning_rate": 3.053888888888889e-07,
170
+ "loss": 1.1044,
171
+ "step": 11000
172
+ },
173
+ {
174
+ "epoch": 0.45,
175
+ "eval_accuracy": 0.38762475049900197,
176
+ "eval_loss": 1.0904459953308105,
177
+ "eval_runtime": 50.9725,
178
+ "eval_samples_per_second": 98.288,
179
+ "eval_steps_per_second": 12.301,
180
+ "step": 11000
181
+ },
182
+ {
183
+ "epoch": 0.49,
184
+ "learning_rate": 3.331666666666667e-07,
185
+ "loss": 1.1019,
186
+ "step": 12000
187
+ },
188
+ {
189
+ "epoch": 0.49,
190
+ "eval_accuracy": 0.3912175648702595,
191
+ "eval_loss": 1.088543176651001,
192
+ "eval_runtime": 50.8297,
193
+ "eval_samples_per_second": 98.564,
194
+ "eval_steps_per_second": 12.335,
195
+ "step": 12000
196
+ },
197
+ {
198
+ "epoch": 0.53,
199
+ "learning_rate": 3.609166666666667e-07,
200
+ "loss": 1.1032,
201
+ "step": 13000
202
+ },
203
+ {
204
+ "epoch": 0.53,
205
+ "eval_accuracy": 0.3435129740518962,
206
+ "eval_loss": 1.0940682888031006,
207
+ "eval_runtime": 51.0031,
208
+ "eval_samples_per_second": 98.229,
209
+ "eval_steps_per_second": 12.293,
210
+ "step": 13000
211
+ },
212
+ {
213
+ "epoch": 0.57,
214
+ "learning_rate": 3.886944444444445e-07,
215
+ "loss": 1.1006,
216
+ "step": 14000
217
+ },
218
+ {
219
+ "epoch": 0.57,
220
+ "eval_accuracy": 0.36826347305389223,
221
+ "eval_loss": 1.0924910306930542,
222
+ "eval_runtime": 51.2753,
223
+ "eval_samples_per_second": 97.708,
224
+ "eval_steps_per_second": 12.228,
225
+ "step": 14000
226
+ },
227
+ {
228
+ "epoch": 0.61,
229
+ "learning_rate": 4.164444444444445e-07,
230
+ "loss": 1.1,
231
+ "step": 15000
232
+ },
233
+ {
234
+ "epoch": 0.61,
235
+ "eval_accuracy": 0.38323353293413176,
236
+ "eval_loss": 1.0879395008087158,
237
+ "eval_runtime": 51.3783,
238
+ "eval_samples_per_second": 97.512,
239
+ "eval_steps_per_second": 12.204,
240
+ "step": 15000
241
+ },
242
+ {
243
+ "epoch": 0.65,
244
+ "learning_rate": 4.442222222222223e-07,
245
+ "loss": 1.0982,
246
+ "step": 16000
247
+ },
248
+ {
249
+ "epoch": 0.65,
250
+ "eval_accuracy": 0.3964071856287425,
251
+ "eval_loss": 1.0822893381118774,
252
+ "eval_runtime": 51.2346,
253
+ "eval_samples_per_second": 97.786,
254
+ "eval_steps_per_second": 12.238,
255
+ "step": 16000
256
+ },
257
+ {
258
+ "epoch": 0.69,
259
+ "learning_rate": 4.7197222222222224e-07,
260
+ "loss": 1.0953,
261
+ "step": 17000
262
+ },
263
+ {
264
+ "epoch": 0.69,
265
+ "eval_accuracy": 0.4121756487025948,
266
+ "eval_loss": 1.0795228481292725,
267
+ "eval_runtime": 51.31,
268
+ "eval_samples_per_second": 97.642,
269
+ "eval_steps_per_second": 12.22,
270
+ "step": 17000
271
+ },
272
+ {
273
+ "epoch": 0.73,
274
+ "learning_rate": 4.997500000000001e-07,
275
+ "loss": 1.0943,
276
+ "step": 18000
277
+ },
278
+ {
279
+ "epoch": 0.73,
280
+ "eval_accuracy": 0.42015968063872255,
281
+ "eval_loss": 1.079458236694336,
282
+ "eval_runtime": 51.3255,
283
+ "eval_samples_per_second": 97.612,
284
+ "eval_steps_per_second": 12.216,
285
+ "step": 18000
286
+ },
287
+ {
288
+ "epoch": 0.77,
289
+ "learning_rate": 5.275e-07,
290
+ "loss": 1.0885,
291
+ "step": 19000
292
+ },
293
+ {
294
+ "epoch": 0.77,
295
+ "eval_accuracy": 0.437125748502994,
296
+ "eval_loss": 1.0667964220046997,
297
+ "eval_runtime": 51.1737,
298
+ "eval_samples_per_second": 97.902,
299
+ "eval_steps_per_second": 12.252,
300
+ "step": 19000
301
+ },
302
+ {
303
+ "epoch": 0.81,
304
+ "learning_rate": 5.552777777777778e-07,
305
+ "loss": 1.0792,
306
+ "step": 20000
307
+ },
308
+ {
309
+ "epoch": 0.81,
310
+ "eval_accuracy": 0.4315369261477046,
311
+ "eval_loss": 1.0628777742385864,
312
+ "eval_runtime": 51.4162,
313
+ "eval_samples_per_second": 97.44,
314
+ "eval_steps_per_second": 12.195,
315
+ "step": 20000
316
+ },
317
+ {
318
+ "epoch": 0.86,
319
+ "learning_rate": 5.830277777777779e-07,
320
+ "loss": 1.0755,
321
+ "step": 21000
322
+ },
323
+ {
324
+ "epoch": 0.86,
325
+ "eval_accuracy": 0.4343313373253493,
326
+ "eval_loss": 1.0604455471038818,
327
+ "eval_runtime": 51.2629,
328
+ "eval_samples_per_second": 97.731,
329
+ "eval_steps_per_second": 12.231,
330
+ "step": 21000
331
+ },
332
+ {
333
+ "epoch": 0.9,
334
+ "learning_rate": 6.108055555555556e-07,
335
+ "loss": 1.073,
336
+ "step": 22000
337
+ },
338
+ {
339
+ "epoch": 0.9,
340
+ "eval_accuracy": 0.43353293413173655,
341
+ "eval_loss": 1.0552711486816406,
342
+ "eval_runtime": 51.2402,
343
+ "eval_samples_per_second": 97.775,
344
+ "eval_steps_per_second": 12.236,
345
+ "step": 22000
346
+ },
347
+ {
348
+ "epoch": 0.94,
349
+ "learning_rate": 6.385555555555556e-07,
350
+ "loss": 1.0699,
351
+ "step": 23000
352
+ },
353
+ {
354
+ "epoch": 0.94,
355
+ "eval_accuracy": 0.4313373253493014,
356
+ "eval_loss": 1.052398920059204,
357
+ "eval_runtime": 51.1758,
358
+ "eval_samples_per_second": 97.898,
359
+ "eval_steps_per_second": 12.252,
360
+ "step": 23000
361
+ },
362
+ {
363
+ "epoch": 0.98,
364
+ "learning_rate": 6.663333333333334e-07,
365
+ "loss": 1.0695,
366
+ "step": 24000
367
+ },
368
+ {
369
+ "epoch": 0.98,
370
+ "eval_accuracy": 0.4407185628742515,
371
+ "eval_loss": 1.047132968902588,
372
+ "eval_runtime": 51.3549,
373
+ "eval_samples_per_second": 97.556,
374
+ "eval_steps_per_second": 12.209,
375
+ "step": 24000
376
+ },
377
+ {
378
+ "epoch": 1.02,
379
+ "learning_rate": 6.940833333333334e-07,
380
+ "loss": 1.0656,
381
+ "step": 25000
382
+ },
383
+ {
384
+ "epoch": 1.02,
385
+ "eval_accuracy": 0.4439121756487026,
386
+ "eval_loss": 1.045853853225708,
387
+ "eval_runtime": 51.1548,
388
+ "eval_samples_per_second": 97.938,
389
+ "eval_steps_per_second": 12.257,
390
+ "step": 25000
391
+ },
392
+ {
393
+ "epoch": 1.06,
394
+ "learning_rate": 7.218611111111112e-07,
395
+ "loss": 1.0645,
396
+ "step": 26000
397
+ },
398
+ {
399
+ "epoch": 1.06,
400
+ "eval_accuracy": 0.43273453093812375,
401
+ "eval_loss": 1.0467982292175293,
402
+ "eval_runtime": 51.1774,
403
+ "eval_samples_per_second": 97.895,
404
+ "eval_steps_per_second": 12.252,
405
+ "step": 26000
406
+ },
407
+ {
408
+ "epoch": 1.1,
409
+ "learning_rate": 7.496111111111112e-07,
410
+ "loss": 1.0639,
411
+ "step": 27000
412
+ },
413
+ {
414
+ "epoch": 1.1,
415
+ "eval_accuracy": 0.43313373253493015,
416
+ "eval_loss": 1.0436949729919434,
417
+ "eval_runtime": 51.1313,
418
+ "eval_samples_per_second": 97.983,
419
+ "eval_steps_per_second": 12.263,
420
+ "step": 27000
421
+ }
422
+ ],
423
+ "logging_steps": 1000,
424
+ "max_steps": 10000000,
425
+ "num_train_epochs": 408,
426
+ "save_steps": 1000,
427
+ "total_flos": 5.64393272255447e+16,
428
+ "trial_name": null,
429
+ "trial_params": null
430
+ }
checkpoint-27000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:933e5c21c45262af343f5d708824cb21f8d339a9e2662221f8b52a4157df35fa
3
+ size 4091
checkpoint-27000/vocab.json ADDED
The diff for this file is too large to render. See raw diff