tmnam20 commited on
Commit
a0a8e92
1 Parent(s): 1500c31

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": 24,
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 sha256:85a06b33dfa666fe86b65589cf081f391ed3fa1fdc3b3c45fbab87c1e05b637b
3
+ size 932401285
checkpoint-27000/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a2b7e965988575372530c3b1ffff1064bad515e877071e67359ac9a5a01f1b8
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:a4f80c915ca1ec597fa701ad2e1c95548b8d77968fcc3a8db60702f7ce7f5ed4
3
+ size 14575
checkpoint-27000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:687ec5a748af635eef0859dfebb5ff6602acdb49e705153778f30b556220b447
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": 1.059377908706665,
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": 1.1506,
15
+ "step": 1
16
+ },
17
+ {
18
+ "epoch": 0.04,
19
+ "learning_rate": 2.7750000000000004e-08,
20
+ "loss": 1.1326,
21
+ "step": 1000
22
+ },
23
+ {
24
+ "epoch": 0.04,
25
+ "eval_accuracy": 0.3331337325349301,
26
+ "eval_loss": 1.1057127714157104,
27
+ "eval_runtime": 61.3538,
28
+ "eval_samples_per_second": 81.657,
29
+ "eval_steps_per_second": 10.219,
30
+ "step": 1000
31
+ },
32
+ {
33
+ "epoch": 0.08,
34
+ "learning_rate": 5.5527777777777784e-08,
35
+ "loss": 1.1238,
36
+ "step": 2000
37
+ },
38
+ {
39
+ "epoch": 0.08,
40
+ "eval_accuracy": 0.33812375249501,
41
+ "eval_loss": 1.1044437885284424,
42
+ "eval_runtime": 61.256,
43
+ "eval_samples_per_second": 81.788,
44
+ "eval_steps_per_second": 10.236,
45
+ "step": 2000
46
+ },
47
+ {
48
+ "epoch": 0.12,
49
+ "learning_rate": 8.327777777777778e-08,
50
+ "loss": 1.1251,
51
+ "step": 3000
52
+ },
53
+ {
54
+ "epoch": 0.12,
55
+ "eval_accuracy": 0.3407185628742515,
56
+ "eval_loss": 1.1043975353240967,
57
+ "eval_runtime": 61.2433,
58
+ "eval_samples_per_second": 81.805,
59
+ "eval_steps_per_second": 10.238,
60
+ "step": 3000
61
+ },
62
+ {
63
+ "epoch": 0.16,
64
+ "learning_rate": 1.1105555555555557e-07,
65
+ "loss": 1.1205,
66
+ "step": 4000
67
+ },
68
+ {
69
+ "epoch": 0.16,
70
+ "eval_accuracy": 0.3387225548902196,
71
+ "eval_loss": 1.100860357284546,
72
+ "eval_runtime": 61.4457,
73
+ "eval_samples_per_second": 81.535,
74
+ "eval_steps_per_second": 10.204,
75
+ "step": 4000
76
+ },
77
+ {
78
+ "epoch": 0.2,
79
+ "learning_rate": 1.3880555555555558e-07,
80
+ "loss": 1.1192,
81
+ "step": 5000
82
+ },
83
+ {
84
+ "epoch": 0.2,
85
+ "eval_accuracy": 0.34151696606786425,
86
+ "eval_loss": 1.1005476713180542,
87
+ "eval_runtime": 61.2568,
88
+ "eval_samples_per_second": 81.787,
89
+ "eval_steps_per_second": 10.236,
90
+ "step": 5000
91
+ },
92
+ {
93
+ "epoch": 0.24,
94
+ "learning_rate": 1.6658333333333335e-07,
95
+ "loss": 1.1145,
96
+ "step": 6000
97
+ },
98
+ {
99
+ "epoch": 0.24,
100
+ "eval_accuracy": 0.3508982035928144,
101
+ "eval_loss": 1.0979759693145752,
102
+ "eval_runtime": 61.2221,
103
+ "eval_samples_per_second": 81.833,
104
+ "eval_steps_per_second": 10.241,
105
+ "step": 6000
106
+ },
107
+ {
108
+ "epoch": 0.29,
109
+ "learning_rate": 1.9433333333333334e-07,
110
+ "loss": 1.1164,
111
+ "step": 7000
112
+ },
113
+ {
114
+ "epoch": 0.29,
115
+ "eval_accuracy": 0.35508982035928144,
116
+ "eval_loss": 1.0955135822296143,
117
+ "eval_runtime": 61.255,
118
+ "eval_samples_per_second": 81.789,
119
+ "eval_steps_per_second": 10.236,
120
+ "step": 7000
121
+ },
122
+ {
123
+ "epoch": 0.33,
124
+ "learning_rate": 2.2211111111111114e-07,
125
+ "loss": 1.1091,
126
+ "step": 8000
127
+ },
128
+ {
129
+ "epoch": 0.33,
130
+ "eval_accuracy": 0.36147704590818364,
131
+ "eval_loss": 1.0946425199508667,
132
+ "eval_runtime": 61.3011,
133
+ "eval_samples_per_second": 81.728,
134
+ "eval_steps_per_second": 10.228,
135
+ "step": 8000
136
+ },
137
+ {
138
+ "epoch": 0.37,
139
+ "learning_rate": 2.4986111111111113e-07,
140
+ "loss": 1.107,
141
+ "step": 9000
142
+ },
143
+ {
144
+ "epoch": 0.37,
145
+ "eval_accuracy": 0.3471057884231537,
146
+ "eval_loss": 1.10226309299469,
147
+ "eval_runtime": 61.3193,
148
+ "eval_samples_per_second": 81.703,
149
+ "eval_steps_per_second": 10.225,
150
+ "step": 9000
151
+ },
152
+ {
153
+ "epoch": 0.41,
154
+ "learning_rate": 2.776388888888889e-07,
155
+ "loss": 1.1055,
156
+ "step": 10000
157
+ },
158
+ {
159
+ "epoch": 0.41,
160
+ "eval_accuracy": 0.3774451097804391,
161
+ "eval_loss": 1.0922847986221313,
162
+ "eval_runtime": 61.1866,
163
+ "eval_samples_per_second": 81.881,
164
+ "eval_steps_per_second": 10.247,
165
+ "step": 10000
166
+ },
167
+ {
168
+ "epoch": 0.45,
169
+ "learning_rate": 3.0541666666666667e-07,
170
+ "loss": 1.1028,
171
+ "step": 11000
172
+ },
173
+ {
174
+ "epoch": 0.45,
175
+ "eval_accuracy": 0.37425149700598803,
176
+ "eval_loss": 1.0914534330368042,
177
+ "eval_runtime": 61.341,
178
+ "eval_samples_per_second": 81.675,
179
+ "eval_steps_per_second": 10.222,
180
+ "step": 11000
181
+ },
182
+ {
183
+ "epoch": 0.49,
184
+ "learning_rate": 3.3319444444444444e-07,
185
+ "loss": 1.1024,
186
+ "step": 12000
187
+ },
188
+ {
189
+ "epoch": 0.49,
190
+ "eval_accuracy": 0.3872255489021956,
191
+ "eval_loss": 1.0888582468032837,
192
+ "eval_runtime": 61.4245,
193
+ "eval_samples_per_second": 81.564,
194
+ "eval_steps_per_second": 10.208,
195
+ "step": 12000
196
+ },
197
+ {
198
+ "epoch": 0.53,
199
+ "learning_rate": 3.6094444444444446e-07,
200
+ "loss": 1.1015,
201
+ "step": 13000
202
+ },
203
+ {
204
+ "epoch": 0.53,
205
+ "eval_accuracy": 0.3397205588822355,
206
+ "eval_loss": 1.096097707748413,
207
+ "eval_runtime": 61.3257,
208
+ "eval_samples_per_second": 81.695,
209
+ "eval_steps_per_second": 10.224,
210
+ "step": 13000
211
+ },
212
+ {
213
+ "epoch": 0.57,
214
+ "learning_rate": 3.8872222222222223e-07,
215
+ "loss": 1.1,
216
+ "step": 14000
217
+ },
218
+ {
219
+ "epoch": 0.57,
220
+ "eval_accuracy": 0.36387225548902197,
221
+ "eval_loss": 1.0940359830856323,
222
+ "eval_runtime": 61.2945,
223
+ "eval_samples_per_second": 81.737,
224
+ "eval_steps_per_second": 10.229,
225
+ "step": 14000
226
+ },
227
+ {
228
+ "epoch": 0.61,
229
+ "learning_rate": 4.1650000000000006e-07,
230
+ "loss": 1.0997,
231
+ "step": 15000
232
+ },
233
+ {
234
+ "epoch": 0.61,
235
+ "eval_accuracy": 0.38602794411177643,
236
+ "eval_loss": 1.0880299806594849,
237
+ "eval_runtime": 61.2546,
238
+ "eval_samples_per_second": 81.79,
239
+ "eval_steps_per_second": 10.236,
240
+ "step": 15000
241
+ },
242
+ {
243
+ "epoch": 0.65,
244
+ "learning_rate": 4.4425e-07,
245
+ "loss": 1.0987,
246
+ "step": 16000
247
+ },
248
+ {
249
+ "epoch": 0.65,
250
+ "eval_accuracy": 0.38962075848303396,
251
+ "eval_loss": 1.0837221145629883,
252
+ "eval_runtime": 61.1947,
253
+ "eval_samples_per_second": 81.87,
254
+ "eval_steps_per_second": 10.246,
255
+ "step": 16000
256
+ },
257
+ {
258
+ "epoch": 0.69,
259
+ "learning_rate": 4.7202777777777785e-07,
260
+ "loss": 1.0943,
261
+ "step": 17000
262
+ },
263
+ {
264
+ "epoch": 0.69,
265
+ "eval_accuracy": 0.4033932135728543,
266
+ "eval_loss": 1.081850290298462,
267
+ "eval_runtime": 61.2293,
268
+ "eval_samples_per_second": 81.824,
269
+ "eval_steps_per_second": 10.24,
270
+ "step": 17000
271
+ },
272
+ {
273
+ "epoch": 0.73,
274
+ "learning_rate": 4.997777777777779e-07,
275
+ "loss": 1.0966,
276
+ "step": 18000
277
+ },
278
+ {
279
+ "epoch": 0.73,
280
+ "eval_accuracy": 0.4031936127744511,
281
+ "eval_loss": 1.0839308500289917,
282
+ "eval_runtime": 61.2192,
283
+ "eval_samples_per_second": 81.837,
284
+ "eval_steps_per_second": 10.242,
285
+ "step": 18000
286
+ },
287
+ {
288
+ "epoch": 0.77,
289
+ "learning_rate": 5.275555555555556e-07,
290
+ "loss": 1.0935,
291
+ "step": 19000
292
+ },
293
+ {
294
+ "epoch": 0.77,
295
+ "eval_accuracy": 0.40099800399201596,
296
+ "eval_loss": 1.0814112424850464,
297
+ "eval_runtime": 61.1921,
298
+ "eval_samples_per_second": 81.873,
299
+ "eval_steps_per_second": 10.246,
300
+ "step": 19000
301
+ },
302
+ {
303
+ "epoch": 0.81,
304
+ "learning_rate": 5.553055555555556e-07,
305
+ "loss": 1.0939,
306
+ "step": 20000
307
+ },
308
+ {
309
+ "epoch": 0.81,
310
+ "eval_accuracy": 0.39241516966067863,
311
+ "eval_loss": 1.0840386152267456,
312
+ "eval_runtime": 61.2683,
313
+ "eval_samples_per_second": 81.772,
314
+ "eval_steps_per_second": 10.234,
315
+ "step": 20000
316
+ },
317
+ {
318
+ "epoch": 0.86,
319
+ "learning_rate": 5.830833333333334e-07,
320
+ "loss": 1.0917,
321
+ "step": 21000
322
+ },
323
+ {
324
+ "epoch": 0.86,
325
+ "eval_accuracy": 0.38303393213572856,
326
+ "eval_loss": 1.083555817604065,
327
+ "eval_runtime": 61.2974,
328
+ "eval_samples_per_second": 81.733,
329
+ "eval_steps_per_second": 10.229,
330
+ "step": 21000
331
+ },
332
+ {
333
+ "epoch": 0.9,
334
+ "learning_rate": 6.108333333333333e-07,
335
+ "loss": 1.0926,
336
+ "step": 22000
337
+ },
338
+ {
339
+ "epoch": 0.9,
340
+ "eval_accuracy": 0.39481037924151696,
341
+ "eval_loss": 1.080513834953308,
342
+ "eval_runtime": 61.4078,
343
+ "eval_samples_per_second": 81.586,
344
+ "eval_steps_per_second": 10.21,
345
+ "step": 22000
346
+ },
347
+ {
348
+ "epoch": 0.94,
349
+ "learning_rate": 6.386111111111112e-07,
350
+ "loss": 1.09,
351
+ "step": 23000
352
+ },
353
+ {
354
+ "epoch": 0.94,
355
+ "eval_accuracy": 0.4115768463073852,
356
+ "eval_loss": 1.076084017753601,
357
+ "eval_runtime": 61.6647,
358
+ "eval_samples_per_second": 81.246,
359
+ "eval_steps_per_second": 10.168,
360
+ "step": 23000
361
+ },
362
+ {
363
+ "epoch": 0.98,
364
+ "learning_rate": 6.663888888888889e-07,
365
+ "loss": 1.0879,
366
+ "step": 24000
367
+ },
368
+ {
369
+ "epoch": 0.98,
370
+ "eval_accuracy": 0.42734530938123755,
371
+ "eval_loss": 1.0709619522094727,
372
+ "eval_runtime": 61.3291,
373
+ "eval_samples_per_second": 81.69,
374
+ "eval_steps_per_second": 10.224,
375
+ "step": 24000
376
+ },
377
+ {
378
+ "epoch": 1.02,
379
+ "learning_rate": 6.94138888888889e-07,
380
+ "loss": 1.0846,
381
+ "step": 25000
382
+ },
383
+ {
384
+ "epoch": 1.02,
385
+ "eval_accuracy": 0.417564870259481,
386
+ "eval_loss": 1.0698238611221313,
387
+ "eval_runtime": 61.3091,
388
+ "eval_samples_per_second": 81.717,
389
+ "eval_steps_per_second": 10.227,
390
+ "step": 25000
391
+ },
392
+ {
393
+ "epoch": 1.06,
394
+ "learning_rate": 7.219166666666666e-07,
395
+ "loss": 1.0821,
396
+ "step": 26000
397
+ },
398
+ {
399
+ "epoch": 1.06,
400
+ "eval_accuracy": 0.43313373253493015,
401
+ "eval_loss": 1.065280795097351,
402
+ "eval_runtime": 61.4486,
403
+ "eval_samples_per_second": 81.532,
404
+ "eval_steps_per_second": 10.204,
405
+ "step": 26000
406
+ },
407
+ {
408
+ "epoch": 1.1,
409
+ "learning_rate": 7.496666666666667e-07,
410
+ "loss": 1.078,
411
+ "step": 27000
412
+ },
413
+ {
414
+ "epoch": 1.1,
415
+ "eval_accuracy": 0.4367265469061876,
416
+ "eval_loss": 1.059377908706665,
417
+ "eval_runtime": 61.6257,
418
+ "eval_samples_per_second": 81.297,
419
+ "eval_steps_per_second": 10.174,
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:e31537d78ddfb65ec62a3bb3689996ba9eab16c9072c351cad313c78d3b7a608
3
+ size 4091
checkpoint-27000/vocab.json ADDED
The diff for this file is too large to render. See raw diff