tmnam20 commited on
Commit
bb905e7
1 Parent(s): 2b32933

Training in progress, step 28000, checkpoint

Browse files
checkpoint-28000/added_tokens.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "<bos>": 40478,
3
+ "<eos>": 40479,
4
+ "<pad>": 40480
5
+ }
checkpoint-28000/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": 2,
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-28000/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-28000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c920ed4402200130271daf5983f4db2baefd7455f367c0ea1955abbfc02483cb
3
+ size 932401285
checkpoint-28000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c8946ea548d4e5355d7c773d1a572f667346f18f83f0203e8ffe0b1b86639b1
3
+ size 466205145
checkpoint-28000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a83ce32f7f9beaedd0a31b46a4972281f322bb0e3caa9258105252a1934c8ae
3
+ size 14575
checkpoint-28000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1088e65e63211f0e6196db0949a299c6a498f4b5feb2c6484e1cc5c911de161
3
+ size 627
checkpoint-28000/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-28000/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-28000/trainer_state.json ADDED
@@ -0,0 +1,445 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 1.038167953491211,
3
+ "best_model_checkpoint": "/kaggle/output/checkpoint-28000",
4
+ "epoch": 1.1408083441981747,
5
+ "eval_steps": 1000,
6
+ "global_step": 28000,
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": 1.0788,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.04,
19
+ "learning_rate": 2.7750000000000004e-08,
20
+ "loss": 1.2063,
21
+ "step": 1000
22
+ },
23
+ {
24
+ "epoch": 0.04,
25
+ "eval_accuracy": 0.32894211576846305,
26
+ "eval_loss": 1.109653115272522,
27
+ "eval_runtime": 50.6057,
28
+ "eval_samples_per_second": 99.001,
29
+ "eval_steps_per_second": 12.39,
30
+ "step": 1000
31
+ },
32
+ {
33
+ "epoch": 0.08,
34
+ "learning_rate": 5.5527777777777784e-08,
35
+ "loss": 1.1293,
36
+ "step": 2000
37
+ },
38
+ {
39
+ "epoch": 0.08,
40
+ "eval_accuracy": 0.32994011976047904,
41
+ "eval_loss": 1.1056402921676636,
42
+ "eval_runtime": 50.7169,
43
+ "eval_samples_per_second": 98.784,
44
+ "eval_steps_per_second": 12.363,
45
+ "step": 2000
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "learning_rate": 8.327777777777778e-08,
50
+ "loss": 1.1263,
51
+ "step": 3000
52
+ },
53
+ {
54
+ "epoch": 0.12,
55
+ "eval_accuracy": 0.3347305389221557,
56
+ "eval_loss": 1.1063188314437866,
57
+ "eval_runtime": 50.8081,
58
+ "eval_samples_per_second": 98.606,
59
+ "eval_steps_per_second": 12.341,
60
+ "step": 3000
61
+ },
62
+ {
63
+ "epoch": 0.16,
64
+ "learning_rate": 1.1105555555555557e-07,
65
+ "loss": 1.1211,
66
+ "step": 4000
67
+ },
68
+ {
69
+ "epoch": 0.16,
70
+ "eval_accuracy": 0.34291417165668664,
71
+ "eval_loss": 1.1010773181915283,
72
+ "eval_runtime": 50.6469,
73
+ "eval_samples_per_second": 98.92,
74
+ "eval_steps_per_second": 12.38,
75
+ "step": 4000
76
+ },
77
+ {
78
+ "epoch": 0.2,
79
+ "learning_rate": 1.3883333333333335e-07,
80
+ "loss": 1.1176,
81
+ "step": 5000
82
+ },
83
+ {
84
+ "epoch": 0.2,
85
+ "eval_accuracy": 0.3395209580838323,
86
+ "eval_loss": 1.099412441253662,
87
+ "eval_runtime": 50.7266,
88
+ "eval_samples_per_second": 98.765,
89
+ "eval_steps_per_second": 12.36,
90
+ "step": 5000
91
+ },
92
+ {
93
+ "epoch": 0.24,
94
+ "learning_rate": 1.6658333333333335e-07,
95
+ "loss": 1.1165,
96
+ "step": 6000
97
+ },
98
+ {
99
+ "epoch": 0.24,
100
+ "eval_accuracy": 0.3596806387225549,
101
+ "eval_loss": 1.0983413457870483,
102
+ "eval_runtime": 50.7493,
103
+ "eval_samples_per_second": 98.721,
104
+ "eval_steps_per_second": 12.355,
105
+ "step": 6000
106
+ },
107
+ {
108
+ "epoch": 0.29,
109
+ "learning_rate": 1.9436111111111112e-07,
110
+ "loss": 1.1127,
111
+ "step": 7000
112
+ },
113
+ {
114
+ "epoch": 0.29,
115
+ "eval_accuracy": 0.35768463073852297,
116
+ "eval_loss": 1.094804048538208,
117
+ "eval_runtime": 50.8322,
118
+ "eval_samples_per_second": 98.56,
119
+ "eval_steps_per_second": 12.335,
120
+ "step": 7000
121
+ },
122
+ {
123
+ "epoch": 0.33,
124
+ "learning_rate": 2.2213888888888891e-07,
125
+ "loss": 1.1095,
126
+ "step": 8000
127
+ },
128
+ {
129
+ "epoch": 0.33,
130
+ "eval_accuracy": 0.36407185628742517,
131
+ "eval_loss": 1.0937609672546387,
132
+ "eval_runtime": 50.8609,
133
+ "eval_samples_per_second": 98.504,
134
+ "eval_steps_per_second": 12.328,
135
+ "step": 8000
136
+ },
137
+ {
138
+ "epoch": 0.37,
139
+ "learning_rate": 2.499166666666667e-07,
140
+ "loss": 1.1074,
141
+ "step": 9000
142
+ },
143
+ {
144
+ "epoch": 0.37,
145
+ "eval_accuracy": 0.34770459081836325,
146
+ "eval_loss": 1.1025313138961792,
147
+ "eval_runtime": 50.8052,
148
+ "eval_samples_per_second": 98.612,
149
+ "eval_steps_per_second": 12.341,
150
+ "step": 9000
151
+ },
152
+ {
153
+ "epoch": 0.41,
154
+ "learning_rate": 2.776666666666667e-07,
155
+ "loss": 1.1065,
156
+ "step": 10000
157
+ },
158
+ {
159
+ "epoch": 0.41,
160
+ "eval_accuracy": 0.38163672654690617,
161
+ "eval_loss": 1.0920748710632324,
162
+ "eval_runtime": 50.8964,
163
+ "eval_samples_per_second": 98.435,
164
+ "eval_steps_per_second": 12.319,
165
+ "step": 10000
166
+ },
167
+ {
168
+ "epoch": 0.45,
169
+ "learning_rate": 3.054444444444444e-07,
170
+ "loss": 1.1026,
171
+ "step": 11000
172
+ },
173
+ {
174
+ "epoch": 0.45,
175
+ "eval_accuracy": 0.37944111776447104,
176
+ "eval_loss": 1.0899525880813599,
177
+ "eval_runtime": 50.6905,
178
+ "eval_samples_per_second": 98.835,
179
+ "eval_steps_per_second": 12.369,
180
+ "step": 11000
181
+ },
182
+ {
183
+ "epoch": 0.49,
184
+ "learning_rate": 3.3319444444444444e-07,
185
+ "loss": 1.1016,
186
+ "step": 12000
187
+ },
188
+ {
189
+ "epoch": 0.49,
190
+ "eval_accuracy": 0.4,
191
+ "eval_loss": 1.0867687463760376,
192
+ "eval_runtime": 50.8282,
193
+ "eval_samples_per_second": 98.567,
194
+ "eval_steps_per_second": 12.336,
195
+ "step": 12000
196
+ },
197
+ {
198
+ "epoch": 0.53,
199
+ "learning_rate": 3.6094444444444446e-07,
200
+ "loss": 1.0999,
201
+ "step": 13000
202
+ },
203
+ {
204
+ "epoch": 0.53,
205
+ "eval_accuracy": 0.3467065868263473,
206
+ "eval_loss": 1.0919947624206543,
207
+ "eval_runtime": 50.7953,
208
+ "eval_samples_per_second": 98.631,
209
+ "eval_steps_per_second": 12.344,
210
+ "step": 13000
211
+ },
212
+ {
213
+ "epoch": 0.57,
214
+ "learning_rate": 3.8872222222222223e-07,
215
+ "loss": 1.0976,
216
+ "step": 14000
217
+ },
218
+ {
219
+ "epoch": 0.57,
220
+ "eval_accuracy": 0.38003992015968063,
221
+ "eval_loss": 1.0882444381713867,
222
+ "eval_runtime": 50.8866,
223
+ "eval_samples_per_second": 98.454,
224
+ "eval_steps_per_second": 12.322,
225
+ "step": 14000
226
+ },
227
+ {
228
+ "epoch": 0.61,
229
+ "learning_rate": 4.1650000000000006e-07,
230
+ "loss": 1.0951,
231
+ "step": 15000
232
+ },
233
+ {
234
+ "epoch": 0.61,
235
+ "eval_accuracy": 0.4229540918163673,
236
+ "eval_loss": 1.0766451358795166,
237
+ "eval_runtime": 50.7597,
238
+ "eval_samples_per_second": 98.7,
239
+ "eval_steps_per_second": 12.352,
240
+ "step": 15000
241
+ },
242
+ {
243
+ "epoch": 0.65,
244
+ "learning_rate": 4.4425e-07,
245
+ "loss": 1.087,
246
+ "step": 16000
247
+ },
248
+ {
249
+ "epoch": 0.65,
250
+ "eval_accuracy": 0.4339321357285429,
251
+ "eval_loss": 1.0621711015701294,
252
+ "eval_runtime": 50.7559,
253
+ "eval_samples_per_second": 98.708,
254
+ "eval_steps_per_second": 12.353,
255
+ "step": 16000
256
+ },
257
+ {
258
+ "epoch": 0.69,
259
+ "learning_rate": 4.7202777777777785e-07,
260
+ "loss": 1.0769,
261
+ "step": 17000
262
+ },
263
+ {
264
+ "epoch": 0.69,
265
+ "eval_accuracy": 0.4357285429141717,
266
+ "eval_loss": 1.0624364614486694,
267
+ "eval_runtime": 50.8608,
268
+ "eval_samples_per_second": 98.504,
269
+ "eval_steps_per_second": 12.328,
270
+ "step": 17000
271
+ },
272
+ {
273
+ "epoch": 0.73,
274
+ "learning_rate": 4.998055555555556e-07,
275
+ "loss": 1.0798,
276
+ "step": 18000
277
+ },
278
+ {
279
+ "epoch": 0.73,
280
+ "eval_accuracy": 0.4311377245508982,
281
+ "eval_loss": 1.0611094236373901,
282
+ "eval_runtime": 50.7255,
283
+ "eval_samples_per_second": 98.767,
284
+ "eval_steps_per_second": 12.361,
285
+ "step": 18000
286
+ },
287
+ {
288
+ "epoch": 0.77,
289
+ "learning_rate": 5.275555555555556e-07,
290
+ "loss": 1.0717,
291
+ "step": 19000
292
+ },
293
+ {
294
+ "epoch": 0.77,
295
+ "eval_accuracy": 0.43852295409181635,
296
+ "eval_loss": 1.056412935256958,
297
+ "eval_runtime": 50.8955,
298
+ "eval_samples_per_second": 98.437,
299
+ "eval_steps_per_second": 12.319,
300
+ "step": 19000
301
+ },
302
+ {
303
+ "epoch": 0.81,
304
+ "learning_rate": 5.553333333333334e-07,
305
+ "loss": 1.0682,
306
+ "step": 20000
307
+ },
308
+ {
309
+ "epoch": 0.81,
310
+ "eval_accuracy": 0.43313373253493015,
311
+ "eval_loss": 1.0576339960098267,
312
+ "eval_runtime": 50.6675,
313
+ "eval_samples_per_second": 98.88,
314
+ "eval_steps_per_second": 12.375,
315
+ "step": 20000
316
+ },
317
+ {
318
+ "epoch": 0.86,
319
+ "learning_rate": 5.830833333333334e-07,
320
+ "loss": 1.0674,
321
+ "step": 21000
322
+ },
323
+ {
324
+ "epoch": 0.86,
325
+ "eval_accuracy": 0.4305389221556886,
326
+ "eval_loss": 1.057605504989624,
327
+ "eval_runtime": 50.8082,
328
+ "eval_samples_per_second": 98.606,
329
+ "eval_steps_per_second": 12.341,
330
+ "step": 21000
331
+ },
332
+ {
333
+ "epoch": 0.9,
334
+ "learning_rate": 6.108611111111111e-07,
335
+ "loss": 1.0689,
336
+ "step": 22000
337
+ },
338
+ {
339
+ "epoch": 0.9,
340
+ "eval_accuracy": 0.4351297405189621,
341
+ "eval_loss": 1.0509696006774902,
342
+ "eval_runtime": 50.7595,
343
+ "eval_samples_per_second": 98.701,
344
+ "eval_steps_per_second": 12.352,
345
+ "step": 22000
346
+ },
347
+ {
348
+ "epoch": 0.94,
349
+ "learning_rate": 6.386388888888889e-07,
350
+ "loss": 1.065,
351
+ "step": 23000
352
+ },
353
+ {
354
+ "epoch": 0.94,
355
+ "eval_accuracy": 0.4363273453093812,
356
+ "eval_loss": 1.0513038635253906,
357
+ "eval_runtime": 50.685,
358
+ "eval_samples_per_second": 98.846,
359
+ "eval_steps_per_second": 12.371,
360
+ "step": 23000
361
+ },
362
+ {
363
+ "epoch": 0.98,
364
+ "learning_rate": 6.663888888888889e-07,
365
+ "loss": 1.0638,
366
+ "step": 24000
367
+ },
368
+ {
369
+ "epoch": 0.98,
370
+ "eval_accuracy": 0.4429141716566866,
371
+ "eval_loss": 1.04515540599823,
372
+ "eval_runtime": 50.9215,
373
+ "eval_samples_per_second": 98.387,
374
+ "eval_steps_per_second": 12.313,
375
+ "step": 24000
376
+ },
377
+ {
378
+ "epoch": 1.02,
379
+ "learning_rate": 6.941666666666667e-07,
380
+ "loss": 1.0607,
381
+ "step": 25000
382
+ },
383
+ {
384
+ "epoch": 1.02,
385
+ "eval_accuracy": 0.4471057884231537,
386
+ "eval_loss": 1.0432593822479248,
387
+ "eval_runtime": 50.8338,
388
+ "eval_samples_per_second": 98.556,
389
+ "eval_steps_per_second": 12.334,
390
+ "step": 25000
391
+ },
392
+ {
393
+ "epoch": 1.06,
394
+ "learning_rate": 7.219444444444444e-07,
395
+ "loss": 1.0578,
396
+ "step": 26000
397
+ },
398
+ {
399
+ "epoch": 1.06,
400
+ "eval_accuracy": 0.4407185628742515,
401
+ "eval_loss": 1.044198751449585,
402
+ "eval_runtime": 50.8791,
403
+ "eval_samples_per_second": 98.469,
404
+ "eval_steps_per_second": 12.323,
405
+ "step": 26000
406
+ },
407
+ {
408
+ "epoch": 1.1,
409
+ "learning_rate": 7.496944444444444e-07,
410
+ "loss": 1.0602,
411
+ "step": 27000
412
+ },
413
+ {
414
+ "epoch": 1.1,
415
+ "eval_accuracy": 0.4363273453093812,
416
+ "eval_loss": 1.045542597770691,
417
+ "eval_runtime": 50.8933,
418
+ "eval_samples_per_second": 98.441,
419
+ "eval_steps_per_second": 12.32,
420
+ "step": 27000
421
+ },
422
+ {
423
+ "epoch": 1.14,
424
+ "learning_rate": 7.774722222222223e-07,
425
+ "loss": 1.0566,
426
+ "step": 28000
427
+ },
428
+ {
429
+ "epoch": 1.14,
430
+ "eval_accuracy": 0.437125748502994,
431
+ "eval_loss": 1.038167953491211,
432
+ "eval_runtime": 50.8894,
433
+ "eval_samples_per_second": 98.449,
434
+ "eval_steps_per_second": 12.321,
435
+ "step": 28000
436
+ }
437
+ ],
438
+ "logging_steps": 1000,
439
+ "max_steps": 10000000,
440
+ "num_train_epochs": 408,
441
+ "save_steps": 1000,
442
+ "total_flos": 5.85296823559127e+16,
443
+ "trial_name": null,
444
+ "trial_params": null
445
+ }
checkpoint-28000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ba6621b829e7b7a24837f1c8da621351c891fe426b370399b04766c22090faa
3
+ size 4091
checkpoint-28000/vocab.json ADDED
The diff for this file is too large to render. See raw diff