tyzhu commited on
Commit
5f1cb01
1 Parent(s): beb1f41

Training in progress, epoch 24, checkpoint

Browse files
checkpoint-8537/adapter_config.json CHANGED
@@ -1,6 +1,6 @@
1
  {
2
  "auto_mapping": null,
3
- "base_model_name_or_path": "Qwen/Qwen1.5-4B",
4
  "bias": "none",
5
  "fan_in_fan_out": false,
6
  "inference_mode": true,
 
1
  {
2
  "auto_mapping": null,
3
+ "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
4
  "bias": "none",
5
  "fan_in_fan_out": false,
6
  "inference_mode": true,
checkpoint-8537/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a11b0b7a9fef8c82cdad7fa66064601d659142a8a8eac855a5cfee220f40e8fc
3
+ size 143269386
checkpoint-8537/added_tokens.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "<|endoftext|>": 151643,
3
- "<|im_end|>": 151645,
4
- "<|im_start|>": 151644
5
  }
 
1
  {
2
+ "</s>": 2,
3
+ "<s>": 1,
4
+ "<unk>": 0
5
  }
checkpoint-8537/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:def5b681dd5f8a42adb5c6373946f15c525e5d6f2f4f5e3c0115aacf588bc973
3
- size 224537202
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:284ac1ba02f9714b309fdc1a6a4595b9d2104735178ce1595d04c48f807ad78b
3
+ size 286585234
checkpoint-8537/rng_state_0.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ee03f7b3b32f33b12987a4d89fd618722e7c3311669daef7aeac9941b059f21f
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1969e0c4fe9621317b447e592abf0a2fd050e9c56958ec92a06ee6bec6c5e00d
3
  size 15024
checkpoint-8537/rng_state_1.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:489c43a75fc56a6166b7a105a1967b066e434f98c8cbee2ee3a6082061aff2fb
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac2bc771b593bb08ad3445e9b60edb130385fcc799d3ef7e670960a11535f28e
3
  size 15024
checkpoint-8537/rng_state_2.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:89d2d4084a4f3291fc25d2fd66c6d7a36fd4c520869f329b04c01afc749ea090
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10e301a70ddb6bb715f1db99e28a3647218ccd5bdc2be02f18cab990a16feb18
3
  size 15024
checkpoint-8537/rng_state_3.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fd102655098c835802375cc3c22a8a9df79b4b929e9d4c795c5c4200e96a8ff5
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77b46c5414547edb4726f0886d175eccbab527e280109235e0b15dc1f9268dbd
3
  size 15024
checkpoint-8537/special_tokens_map.json CHANGED
@@ -1,14 +1,6 @@
1
  {
2
- "additional_special_tokens": [
3
- "<|im_start|>",
4
- "<|im_end|>"
5
- ],
6
- "eos_token": {
7
- "content": "<|endoftext|>",
8
- "lstrip": false,
9
- "normalized": false,
10
- "rstrip": false,
11
- "single_word": false
12
- },
13
- "pad_token": "<|endoftext|>"
14
  }
 
1
  {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": "</s>",
5
+ "unk_token": "<unk>"
 
 
 
 
 
 
 
 
6
  }
checkpoint-8537/tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
 
checkpoint-8537/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-8537/tokenizer_config.json CHANGED
@@ -1,24 +1,23 @@
1
  {
2
- "add_prefix_space": false,
3
  "added_tokens_decoder": {
4
- "151643": {
5
- "content": "<|endoftext|>",
6
  "lstrip": false,
7
  "normalized": false,
8
  "rstrip": false,
9
  "single_word": false,
10
  "special": true
11
  },
12
- "151644": {
13
- "content": "<|im_start|>",
14
  "lstrip": false,
15
  "normalized": false,
16
  "rstrip": false,
17
  "single_word": false,
18
  "special": true
19
  },
20
- "151645": {
21
- "content": "<|im_end|>",
22
  "lstrip": false,
23
  "normalized": false,
24
  "rstrip": false,
@@ -26,18 +25,16 @@
26
  "special": true
27
  }
28
  },
29
- "additional_special_tokens": [
30
- "<|im_start|>",
31
- "<|im_end|>"
32
- ],
33
- "bos_token": null,
34
- "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
35
  "clean_up_tokenization_spaces": false,
36
- "eos_token": "<|endoftext|>",
37
- "errors": "replace",
38
- "model_max_length": 32768,
39
- "pad_token": "<|endoftext|>",
40
- "split_special_tokens": false,
41
- "tokenizer_class": "Qwen2Tokenizer",
42
- "unk_token": null
 
 
43
  }
 
1
  {
 
2
  "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "<unk>",
5
  "lstrip": false,
6
  "normalized": false,
7
  "rstrip": false,
8
  "single_word": false,
9
  "special": true
10
  },
11
+ "1": {
12
+ "content": "<s>",
13
  "lstrip": false,
14
  "normalized": false,
15
  "rstrip": false,
16
  "single_word": false,
17
  "special": true
18
  },
19
+ "2": {
20
+ "content": "</s>",
21
  "lstrip": false,
22
  "normalized": false,
23
  "rstrip": false,
 
25
  "special": true
26
  }
27
  },
28
+ "additional_special_tokens": [],
29
+ "bos_token": "<s>",
 
 
 
 
30
  "clean_up_tokenization_spaces": false,
31
+ "eos_token": "</s>",
32
+ "legacy": false,
33
+ "model_max_length": 1000000000000000019884624838656,
34
+ "pad_token": "</s>",
35
+ "padding_side": "left",
36
+ "sp_model_kwargs": {},
37
+ "tokenizer_class": "LlamaTokenizer",
38
+ "unk_token": "<unk>",
39
+ "use_default_system_prompt": true
40
  }
checkpoint-8537/trainer_state.json CHANGED
@@ -9,1020 +9,921 @@
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
- "epoch": 0.29282576866764276,
13
- "grad_norm": 0.26115021109580994,
14
  "learning_rate": 3e-05,
15
- "loss": 2.0719,
16
  "step": 100
17
  },
18
  {
19
- "epoch": 0.5856515373352855,
20
- "grad_norm": 0.2679530680179596,
21
  "learning_rate": 3e-05,
22
- "loss": 1.969,
23
  "step": 200
24
  },
25
  {
26
- "epoch": 0.8784773060029283,
27
- "grad_norm": 0.2916224002838135,
28
  "learning_rate": 3e-05,
29
- "loss": 1.9686,
30
  "step": 300
31
  },
32
  {
33
- "epoch": 0.9985358711566618,
34
- "eval_accuracy": 0.43743589743589745,
35
- "eval_loss": 3.6918885707855225,
36
- "eval_runtime": 5.2974,
37
- "eval_samples_per_second": 94.385,
38
- "eval_steps_per_second": 11.893,
39
  "step": 341
40
  },
41
  {
42
- "epoch": 0.9985358711566618,
43
- "eval_bleu": 0.1527908371131025,
44
- "eval_exact_match": 0.012,
45
- "eval_prefix_exact_match": 0.026,
46
  "step": 341
47
  },
48
  {
49
- "epoch": 1.171303074670571,
50
- "grad_norm": 0.3534347712993622,
51
  "learning_rate": 3e-05,
52
- "loss": 1.9603,
53
  "step": 400
54
  },
55
  {
56
- "epoch": 1.4641288433382138,
57
- "grad_norm": 0.3338424563407898,
58
  "learning_rate": 3e-05,
59
- "loss": 1.9235,
60
  "step": 500
61
  },
62
  {
63
- "epoch": 1.7569546120058566,
64
- "grad_norm": 0.4033390283584595,
65
  "learning_rate": 3e-05,
66
- "loss": 1.9337,
67
  "step": 600
68
  },
69
  {
70
  "epoch": 2.0,
71
- "eval_accuracy": 0.44764102564102565,
72
- "eval_loss": 3.7484843730926514,
73
- "eval_runtime": 5.0649,
74
- "eval_samples_per_second": 98.718,
75
- "eval_steps_per_second": 12.438,
76
  "step": 683
77
  },
78
  {
79
  "epoch": 2.0,
80
- "eval_bleu": 0.21311912714932216,
81
- "eval_exact_match": 0.02,
82
- "eval_prefix_exact_match": 0.032,
83
  "step": 683
84
  },
85
  {
86
- "epoch": 2.049780380673499,
87
- "grad_norm": 0.38882532715797424,
88
  "learning_rate": 3e-05,
89
- "loss": 1.9267,
90
  "step": 700
91
  },
92
  {
93
- "epoch": 2.342606149341142,
94
- "grad_norm": 0.44310709834098816,
95
  "learning_rate": 3e-05,
96
- "loss": 1.8823,
97
  "step": 800
98
  },
99
  {
100
- "epoch": 2.6354319180087846,
101
- "grad_norm": 0.5196680426597595,
102
  "learning_rate": 3e-05,
103
- "loss": 1.901,
104
  "step": 900
105
  },
106
  {
107
- "epoch": 2.9282576866764276,
108
- "grad_norm": 0.5293437242507935,
109
  "learning_rate": 3e-05,
110
- "loss": 1.9033,
111
  "step": 1000
112
  },
113
  {
114
- "epoch": 2.998535871156662,
115
- "eval_accuracy": 0.4495897435897436,
116
- "eval_loss": 3.882551670074463,
117
- "eval_runtime": 5.1314,
118
- "eval_samples_per_second": 97.439,
119
- "eval_steps_per_second": 12.277,
120
  "step": 1024
121
  },
122
  {
123
- "epoch": 2.998535871156662,
124
- "eval_bleu": 0.24307488023362148,
125
- "eval_exact_match": 0.016,
126
- "eval_prefix_exact_match": 0.032,
127
  "step": 1024
128
  },
129
  {
130
- "epoch": 3.22108345534407,
131
- "grad_norm": 0.552943229675293,
132
  "learning_rate": 3e-05,
133
- "loss": 1.8529,
134
  "step": 1100
135
  },
136
  {
137
- "epoch": 3.513909224011713,
138
- "grad_norm": 0.6421984434127808,
139
  "learning_rate": 3e-05,
140
- "loss": 1.8342,
141
  "step": 1200
142
  },
143
  {
144
- "epoch": 3.8067349926793557,
145
- "grad_norm": 0.7155198454856873,
146
  "learning_rate": 3e-05,
147
- "loss": 1.857,
148
  "step": 1300
149
  },
150
  {
151
  "epoch": 4.0,
152
- "eval_accuracy": 0.4481025641025641,
153
- "eval_loss": 3.970107316970825,
154
- "eval_runtime": 5.1563,
155
- "eval_samples_per_second": 96.969,
156
- "eval_steps_per_second": 12.218,
157
  "step": 1366
158
  },
159
  {
160
  "epoch": 4.0,
161
- "eval_bleu": 0.25309406645138827,
162
- "eval_exact_match": 0.016,
163
- "eval_prefix_exact_match": 0.044,
164
  "step": 1366
165
  },
166
  {
167
- "epoch": 4.099560761346998,
168
- "grad_norm": 0.7837833166122437,
169
  "learning_rate": 3e-05,
170
- "loss": 1.8284,
171
  "step": 1400
172
  },
173
  {
174
- "epoch": 4.392386530014641,
175
- "grad_norm": 0.784609854221344,
176
  "learning_rate": 3e-05,
177
- "loss": 1.7884,
178
  "step": 1500
179
  },
180
  {
181
- "epoch": 4.685212298682284,
182
- "grad_norm": 0.834195613861084,
183
  "learning_rate": 3e-05,
184
- "loss": 1.7841,
185
  "step": 1600
186
  },
187
  {
188
- "epoch": 4.978038067349927,
189
- "grad_norm": 0.8902062177658081,
190
  "learning_rate": 3e-05,
191
- "loss": 1.8042,
192
  "step": 1700
193
  },
194
  {
195
- "epoch": 4.998535871156662,
196
- "eval_accuracy": 0.4472820512820513,
197
- "eval_loss": 4.117065906524658,
198
- "eval_runtime": 5.2376,
199
- "eval_samples_per_second": 95.464,
200
- "eval_steps_per_second": 12.028,
201
  "step": 1707
202
  },
203
  {
204
- "epoch": 4.998535871156662,
205
- "eval_bleu": 0.2531049955279983,
206
- "eval_exact_match": 0.014,
207
- "eval_prefix_exact_match": 0.042,
208
  "step": 1707
209
  },
210
  {
211
- "epoch": 5.270863836017569,
212
- "grad_norm": 0.9495383501052856,
213
  "learning_rate": 3e-05,
214
- "loss": 1.7493,
215
  "step": 1800
216
  },
217
  {
218
- "epoch": 5.563689604685212,
219
- "grad_norm": 0.9647482633590698,
220
  "learning_rate": 3e-05,
221
- "loss": 1.7334,
222
  "step": 1900
223
  },
224
  {
225
- "epoch": 5.856515373352855,
226
- "grad_norm": 0.9825767874717712,
227
  "learning_rate": 3e-05,
228
- "loss": 1.7443,
229
  "step": 2000
230
  },
231
  {
232
  "epoch": 6.0,
233
- "eval_accuracy": 0.446974358974359,
234
- "eval_loss": 4.183700084686279,
235
- "eval_runtime": 5.2674,
236
- "eval_samples_per_second": 94.923,
237
- "eval_steps_per_second": 11.96,
238
  "step": 2049
239
  },
240
  {
241
  "epoch": 6.0,
242
- "eval_bleu": 0.23587359955369594,
243
- "eval_exact_match": 0.012,
244
- "eval_prefix_exact_match": 0.04,
245
  "step": 2049
246
  },
247
  {
248
- "epoch": 6.149341142020498,
249
- "grad_norm": 1.0430296659469604,
250
  "learning_rate": 3e-05,
251
- "loss": 1.6959,
252
  "step": 2100
253
  },
254
  {
255
- "epoch": 6.44216691068814,
256
- "grad_norm": 1.0777732133865356,
257
  "learning_rate": 3e-05,
258
- "loss": 1.6795,
259
  "step": 2200
260
  },
261
  {
262
- "epoch": 6.734992679355783,
263
- "grad_norm": 1.0959677696228027,
264
  "learning_rate": 3e-05,
265
- "loss": 1.7019,
266
  "step": 2300
267
  },
268
  {
269
- "epoch": 6.998535871156662,
270
- "eval_accuracy": 0.4461538461538462,
271
- "eval_loss": 4.260408401489258,
272
- "eval_runtime": 5.178,
273
- "eval_samples_per_second": 96.563,
274
- "eval_steps_per_second": 12.167,
275
  "step": 2390
276
  },
277
  {
278
- "epoch": 6.998535871156662,
279
- "eval_bleu": 0.2008158466316808,
280
- "eval_exact_match": 0.006,
281
- "eval_prefix_exact_match": 0.042,
282
  "step": 2390
283
  },
284
  {
285
- "epoch": 7.027818448023426,
286
- "grad_norm": 1.0649549961090088,
287
  "learning_rate": 3e-05,
288
- "loss": 1.6812,
289
  "step": 2400
290
  },
291
  {
292
- "epoch": 7.320644216691068,
293
- "grad_norm": 1.294122576713562,
294
  "learning_rate": 3e-05,
295
- "loss": 1.6226,
296
  "step": 2500
297
  },
298
  {
299
- "epoch": 7.613469985358711,
300
- "grad_norm": 1.2657020092010498,
301
  "learning_rate": 3e-05,
302
- "loss": 1.6404,
303
  "step": 2600
304
  },
305
  {
306
- "epoch": 7.906295754026354,
307
- "grad_norm": 1.1963133811950684,
308
  "learning_rate": 3e-05,
309
- "loss": 1.6305,
310
  "step": 2700
311
  },
312
  {
313
  "epoch": 8.0,
314
- "eval_accuracy": 0.44148717948717947,
315
- "eval_loss": 4.406544208526611,
316
- "eval_runtime": 5.3355,
317
- "eval_samples_per_second": 93.711,
318
- "eval_steps_per_second": 11.808,
319
  "step": 2732
320
  },
321
  {
322
  "epoch": 8.0,
323
- "eval_bleu": 0.24435500637461924,
324
- "eval_exact_match": 0.008,
325
- "eval_prefix_exact_match": 0.038,
326
  "step": 2732
327
  },
328
  {
329
- "epoch": 8.199121522693996,
330
- "grad_norm": 1.3801360130310059,
331
  "learning_rate": 3e-05,
332
- "loss": 1.5734,
333
  "step": 2800
334
  },
335
  {
336
- "epoch": 8.49194729136164,
337
- "grad_norm": 1.3306772708892822,
338
  "learning_rate": 3e-05,
339
- "loss": 1.5651,
340
  "step": 2900
341
  },
342
  {
343
- "epoch": 8.784773060029282,
344
- "grad_norm": 1.5165746212005615,
345
  "learning_rate": 3e-05,
346
- "loss": 1.6056,
347
  "step": 3000
348
  },
349
  {
350
- "epoch": 8.998535871156662,
351
- "eval_accuracy": 0.4397948717948718,
352
- "eval_loss": 4.448728561401367,
353
- "eval_runtime": 5.187,
354
- "eval_samples_per_second": 96.395,
355
- "eval_steps_per_second": 12.146,
356
  "step": 3073
357
  },
358
  {
359
- "epoch": 8.998535871156662,
360
- "eval_bleu": 0.20120283162176952,
361
- "eval_exact_match": 0.008,
362
- "eval_prefix_exact_match": 0.036,
363
  "step": 3073
364
  },
365
  {
366
- "epoch": 9.077598828696924,
367
- "grad_norm": 1.4077060222625732,
368
  "learning_rate": 3e-05,
369
- "loss": 1.5612,
370
  "step": 3100
371
  },
372
  {
373
- "epoch": 9.370424597364568,
374
- "grad_norm": 1.475984811782837,
375
  "learning_rate": 3e-05,
376
- "loss": 1.5158,
377
  "step": 3200
378
  },
379
  {
380
- "epoch": 9.66325036603221,
381
- "grad_norm": 1.586300015449524,
382
  "learning_rate": 3e-05,
383
- "loss": 1.5145,
384
  "step": 3300
385
  },
386
  {
387
- "epoch": 9.956076134699854,
388
- "grad_norm": 1.510048270225525,
389
  "learning_rate": 3e-05,
390
- "loss": 1.5521,
391
  "step": 3400
392
  },
393
  {
394
  "epoch": 10.0,
395
- "eval_accuracy": 0.4388717948717949,
396
- "eval_loss": 4.5474138259887695,
397
- "eval_runtime": 5.169,
398
- "eval_samples_per_second": 96.731,
399
- "eval_steps_per_second": 12.188,
400
  "step": 3415
401
  },
402
  {
403
  "epoch": 10.0,
404
- "eval_bleu": 0.2258317707850195,
405
- "eval_exact_match": 0.004,
406
  "eval_prefix_exact_match": 0.032,
407
  "step": 3415
408
  },
409
  {
410
- "epoch": 10.248901903367496,
411
- "grad_norm": 1.6384536027908325,
412
  "learning_rate": 3e-05,
413
- "loss": 1.4631,
414
  "step": 3500
415
  },
416
  {
417
- "epoch": 10.541727672035138,
418
- "grad_norm": 1.7626007795333862,
419
  "learning_rate": 3e-05,
420
- "loss": 1.4623,
421
  "step": 3600
422
  },
423
  {
424
- "epoch": 10.834553440702782,
425
- "grad_norm": 1.6483694314956665,
426
  "learning_rate": 3e-05,
427
- "loss": 1.4934,
428
  "step": 3700
429
  },
430
  {
431
- "epoch": 10.998535871156662,
432
- "eval_accuracy": 0.4367179487179487,
433
- "eval_loss": 4.589761734008789,
434
- "eval_runtime": 5.853,
435
- "eval_samples_per_second": 85.426,
436
- "eval_steps_per_second": 10.764,
437
  "step": 3756
438
  },
439
  {
440
- "epoch": 10.998535871156662,
441
- "eval_bleu": 0.20792247275869374,
442
- "eval_exact_match": 0.004,
443
  "eval_prefix_exact_match": 0.026,
444
  "step": 3756
445
  },
446
  {
447
- "epoch": 11.127379209370424,
448
- "grad_norm": 1.8367844820022583,
449
  "learning_rate": 3e-05,
450
- "loss": 1.4521,
451
  "step": 3800
452
  },
453
  {
454
- "epoch": 11.420204978038067,
455
- "grad_norm": 1.8123779296875,
456
  "learning_rate": 3e-05,
457
- "loss": 1.4085,
458
  "step": 3900
459
  },
460
  {
461
- "epoch": 11.71303074670571,
462
- "grad_norm": 1.9437752962112427,
463
  "learning_rate": 3e-05,
464
- "loss": 1.4287,
465
  "step": 4000
466
  },
467
  {
468
  "epoch": 12.0,
469
- "eval_accuracy": 0.43553846153846154,
470
- "eval_loss": 4.691086769104004,
471
- "eval_runtime": 5.3738,
472
- "eval_samples_per_second": 93.043,
473
- "eval_steps_per_second": 11.723,
474
  "step": 4098
475
  },
476
  {
477
  "epoch": 12.0,
478
- "eval_bleu": 0.21672573731424186,
479
- "eval_exact_match": 0.006,
480
- "eval_prefix_exact_match": 0.028,
481
  "step": 4098
482
  },
483
  {
484
- "epoch": 12.005856515373353,
485
- "grad_norm": 1.5838289260864258,
486
  "learning_rate": 3e-05,
487
- "loss": 1.4361,
488
  "step": 4100
489
  },
490
  {
491
- "epoch": 12.298682284040996,
492
- "grad_norm": 1.8575290441513062,
493
  "learning_rate": 3e-05,
494
- "loss": 1.3433,
495
  "step": 4200
496
  },
497
  {
498
- "epoch": 12.591508052708638,
499
- "grad_norm": 1.9598041772842407,
500
  "learning_rate": 3e-05,
501
- "loss": 1.3661,
502
  "step": 4300
503
  },
504
  {
505
- "epoch": 12.88433382137628,
506
- "grad_norm": 1.972550868988037,
507
  "learning_rate": 3e-05,
508
- "loss": 1.3846,
509
  "step": 4400
510
  },
511
  {
512
- "epoch": 12.998535871156662,
513
- "eval_accuracy": 0.43553846153846154,
514
- "eval_loss": 4.762892246246338,
515
- "eval_runtime": 5.106,
516
- "eval_samples_per_second": 97.925,
517
- "eval_steps_per_second": 12.339,
518
  "step": 4439
519
  },
520
  {
521
- "epoch": 12.998535871156662,
522
- "eval_bleu": 0.2416046504676966,
523
- "eval_exact_match": 0.006,
524
- "eval_prefix_exact_match": 0.028,
525
  "step": 4439
526
  },
527
  {
528
- "epoch": 13.177159590043924,
529
- "grad_norm": 1.9469107389450073,
530
  "learning_rate": 3e-05,
531
- "loss": 1.3229,
532
  "step": 4500
533
  },
534
  {
535
- "epoch": 13.469985358711567,
536
- "grad_norm": 2.1680469512939453,
537
  "learning_rate": 3e-05,
538
- "loss": 1.3172,
539
  "step": 4600
540
  },
541
  {
542
- "epoch": 13.762811127379209,
543
- "grad_norm": 2.1960484981536865,
544
  "learning_rate": 3e-05,
545
- "loss": 1.3185,
546
  "step": 4700
547
  },
548
  {
549
  "epoch": 14.0,
550
- "eval_accuracy": 0.4327692307692308,
551
- "eval_loss": 4.758542537689209,
552
- "eval_runtime": 5.778,
553
- "eval_samples_per_second": 86.535,
554
- "eval_steps_per_second": 10.903,
555
  "step": 4781
556
  },
557
  {
558
  "epoch": 14.0,
559
- "eval_bleu": 0.19085834097154403,
560
- "eval_exact_match": 0.004,
561
- "eval_prefix_exact_match": 0.022,
562
  "step": 4781
563
  },
564
  {
565
- "epoch": 14.055636896046853,
566
- "grad_norm": 2.4733147621154785,
567
  "learning_rate": 3e-05,
568
- "loss": 1.3027,
569
  "step": 4800
570
  },
571
  {
572
- "epoch": 14.348462664714495,
573
- "grad_norm": 2.2826030254364014,
574
  "learning_rate": 3e-05,
575
- "loss": 1.2513,
576
  "step": 4900
577
  },
578
  {
579
- "epoch": 14.641288433382137,
580
- "grad_norm": 2.041346788406372,
581
  "learning_rate": 3e-05,
582
- "loss": 1.2637,
583
  "step": 5000
584
  },
585
  {
586
- "epoch": 14.93411420204978,
587
- "grad_norm": 2.3494467735290527,
588
  "learning_rate": 3e-05,
589
- "loss": 1.2667,
590
  "step": 5100
591
  },
592
  {
593
- "epoch": 14.998535871156662,
594
- "eval_accuracy": 0.4308717948717949,
595
- "eval_loss": 4.938868522644043,
596
- "eval_runtime": 5.9024,
597
- "eval_samples_per_second": 84.711,
598
- "eval_steps_per_second": 10.674,
599
  "step": 5122
600
  },
601
  {
602
- "epoch": 14.998535871156662,
603
- "eval_bleu": 0.18549760176612626,
604
- "eval_exact_match": 0.002,
605
- "eval_prefix_exact_match": 0.024,
606
  "step": 5122
607
  },
608
  {
609
- "epoch": 15.226939970717423,
610
- "grad_norm": 2.411022663116455,
611
  "learning_rate": 3e-05,
612
- "loss": 1.2158,
613
  "step": 5200
614
  },
615
  {
616
- "epoch": 15.519765739385067,
617
- "grad_norm": 2.3778491020202637,
618
  "learning_rate": 3e-05,
619
- "loss": 1.193,
620
  "step": 5300
621
  },
622
  {
623
- "epoch": 15.812591508052709,
624
- "grad_norm": 2.2003743648529053,
625
  "learning_rate": 3e-05,
626
- "loss": 1.2144,
627
  "step": 5400
628
  },
629
  {
630
  "epoch": 16.0,
631
- "eval_accuracy": 0.43025641025641026,
632
- "eval_loss": 4.898688316345215,
633
- "eval_runtime": 5.1834,
634
- "eval_samples_per_second": 96.463,
635
- "eval_steps_per_second": 12.154,
636
  "step": 5464
637
  },
638
  {
639
  "epoch": 16.0,
640
- "eval_bleu": 0.20104450766148693,
641
- "eval_exact_match": 0.004,
642
- "eval_prefix_exact_match": 0.026,
643
  "step": 5464
644
  },
645
  {
646
- "epoch": 16.105417276720353,
647
- "grad_norm": 2.1213016510009766,
648
  "learning_rate": 3e-05,
649
- "loss": 1.1842,
650
  "step": 5500
651
  },
652
  {
653
- "epoch": 16.398243045387993,
654
- "grad_norm": 2.4457600116729736,
655
  "learning_rate": 3e-05,
656
- "loss": 1.1527,
657
  "step": 5600
658
  },
659
  {
660
- "epoch": 16.691068814055637,
661
- "grad_norm": 2.5844104290008545,
662
  "learning_rate": 3e-05,
663
- "loss": 1.1536,
664
  "step": 5700
665
  },
666
  {
667
- "epoch": 16.98389458272328,
668
- "grad_norm": 2.7657299041748047,
669
  "learning_rate": 3e-05,
670
- "loss": 1.1708,
671
  "step": 5800
672
  },
673
  {
674
- "epoch": 16.998535871156662,
675
- "eval_accuracy": 0.4297435897435897,
676
- "eval_loss": 5.00166654586792,
677
- "eval_runtime": 5.4251,
678
- "eval_samples_per_second": 92.164,
679
- "eval_steps_per_second": 11.613,
680
  "step": 5805
681
  },
682
  {
683
- "epoch": 16.998535871156662,
684
- "eval_bleu": 0.2059810081312132,
685
- "eval_exact_match": 0.004,
686
- "eval_prefix_exact_match": 0.028,
687
  "step": 5805
688
  },
689
  {
690
- "epoch": 17.27672035139092,
691
- "grad_norm": 2.4050705432891846,
692
  "learning_rate": 3e-05,
693
- "loss": 1.083,
694
  "step": 5900
695
  },
696
  {
697
- "epoch": 17.569546120058565,
698
- "grad_norm": 2.584331750869751,
699
  "learning_rate": 3e-05,
700
- "loss": 1.0993,
701
  "step": 6000
702
  },
703
  {
704
- "epoch": 17.86237188872621,
705
- "grad_norm": 3.257659912109375,
706
  "learning_rate": 3e-05,
707
- "loss": 1.1146,
708
  "step": 6100
709
  },
710
  {
711
  "epoch": 18.0,
712
- "eval_accuracy": 0.43066666666666664,
713
- "eval_loss": 4.977816581726074,
714
- "eval_runtime": 5.2952,
715
- "eval_samples_per_second": 94.425,
716
- "eval_steps_per_second": 11.898,
717
  "step": 6147
718
  },
719
  {
720
  "epoch": 18.0,
721
- "eval_bleu": 0.1948234970665932,
722
- "eval_exact_match": 0.004,
723
- "eval_prefix_exact_match": 0.032,
724
  "step": 6147
725
  },
726
  {
727
- "epoch": 18.15519765739385,
728
- "grad_norm": 3.068880319595337,
729
  "learning_rate": 3e-05,
730
- "loss": 1.0673,
731
  "step": 6200
732
  },
733
  {
734
- "epoch": 18.448023426061493,
735
- "grad_norm": 2.737912893295288,
736
  "learning_rate": 3e-05,
737
- "loss": 1.0381,
738
  "step": 6300
739
  },
740
  {
741
- "epoch": 18.740849194729137,
742
- "grad_norm": 2.878462076187134,
743
  "learning_rate": 3e-05,
744
- "loss": 1.0531,
745
  "step": 6400
746
  },
747
  {
748
- "epoch": 18.998535871156662,
749
- "eval_accuracy": 0.42866666666666664,
750
- "eval_loss": 5.121572494506836,
751
- "eval_runtime": 5.3996,
752
- "eval_samples_per_second": 92.599,
753
- "eval_steps_per_second": 11.667,
754
  "step": 6488
755
  },
756
  {
757
- "epoch": 18.998535871156662,
758
- "eval_bleu": 0.19788734504730915,
759
  "eval_exact_match": 0.0,
760
  "eval_prefix_exact_match": 0.02,
761
  "step": 6488
762
  },
763
  {
764
- "epoch": 19.03367496339678,
765
- "grad_norm": 2.6934640407562256,
766
  "learning_rate": 3e-05,
767
- "loss": 1.0564,
768
  "step": 6500
769
  },
770
  {
771
- "epoch": 19.32650073206442,
772
- "grad_norm": 2.822169542312622,
773
  "learning_rate": 3e-05,
774
- "loss": 0.99,
775
  "step": 6600
776
  },
777
  {
778
- "epoch": 19.619326500732065,
779
- "grad_norm": 2.520814895629883,
780
  "learning_rate": 3e-05,
781
- "loss": 0.991,
782
  "step": 6700
783
  },
784
  {
785
- "epoch": 19.91215226939971,
786
- "grad_norm": 3.1566810607910156,
787
  "learning_rate": 3e-05,
788
- "loss": 1.0158,
789
  "step": 6800
790
  },
791
  {
792
  "epoch": 20.0,
793
- "eval_accuracy": 0.42728205128205127,
794
- "eval_loss": 5.120981216430664,
795
- "eval_runtime": 5.394,
796
- "eval_samples_per_second": 92.696,
797
- "eval_steps_per_second": 11.68,
798
  "step": 6830
799
  },
800
  {
801
  "epoch": 20.0,
802
- "eval_bleu": 0.15806516253752823,
803
- "eval_exact_match": 0.006,
804
- "eval_prefix_exact_match": 0.028,
805
  "step": 6830
806
  },
807
  {
808
- "epoch": 20.20497803806735,
809
- "grad_norm": 2.937910556793213,
810
  "learning_rate": 3e-05,
811
- "loss": 0.9615,
812
  "step": 6900
813
  },
814
  {
815
- "epoch": 20.497803806734993,
816
- "grad_norm": 3.2016685009002686,
817
  "learning_rate": 3e-05,
818
- "loss": 0.9404,
819
  "step": 7000
820
  },
821
  {
822
- "epoch": 20.790629575402637,
823
- "grad_norm": 3.2996864318847656,
824
  "learning_rate": 3e-05,
825
- "loss": 0.9555,
826
  "step": 7100
827
  },
828
  {
829
- "epoch": 20.998535871156662,
830
- "eval_accuracy": 0.42928205128205127,
831
- "eval_loss": 5.198832988739014,
832
- "eval_runtime": 5.8905,
833
- "eval_samples_per_second": 84.882,
834
- "eval_steps_per_second": 10.695,
835
  "step": 7171
836
  },
837
  {
838
- "epoch": 20.998535871156662,
839
- "eval_bleu": 0.18303907416923465,
840
- "eval_exact_match": 0.004,
841
- "eval_prefix_exact_match": 0.02,
842
  "step": 7171
843
  },
844
  {
845
- "epoch": 21.083455344070277,
846
- "grad_norm": 3.4803571701049805,
847
  "learning_rate": 3e-05,
848
- "loss": 0.9454,
849
  "step": 7200
850
  },
851
  {
852
- "epoch": 21.37628111273792,
853
- "grad_norm": 2.6315698623657227,
854
  "learning_rate": 3e-05,
855
- "loss": 0.8835,
856
  "step": 7300
857
  },
858
  {
859
- "epoch": 21.669106881405565,
860
- "grad_norm": 3.0926690101623535,
861
  "learning_rate": 3e-05,
862
- "loss": 0.9004,
863
  "step": 7400
864
  },
865
  {
866
- "epoch": 21.961932650073205,
867
- "grad_norm": 2.9990713596343994,
868
  "learning_rate": 3e-05,
869
- "loss": 0.9205,
870
  "step": 7500
871
  },
872
  {
873
  "epoch": 22.0,
874
- "eval_accuracy": 0.427025641025641,
875
- "eval_loss": 5.2239508628845215,
876
- "eval_runtime": 5.4025,
877
- "eval_samples_per_second": 92.55,
878
- "eval_steps_per_second": 11.661,
879
  "step": 7513
880
  },
881
  {
882
  "epoch": 22.0,
883
- "eval_bleu": 0.1854279974336752,
884
- "eval_exact_match": 0.004,
885
  "eval_prefix_exact_match": 0.02,
886
  "step": 7513
887
  },
888
  {
889
- "epoch": 22.25475841874085,
890
- "grad_norm": 3.2822601795196533,
891
  "learning_rate": 3e-05,
892
- "loss": 0.8461,
893
  "step": 7600
894
  },
895
  {
896
- "epoch": 22.547584187408493,
897
- "grad_norm": 3.4992525577545166,
898
  "learning_rate": 3e-05,
899
- "loss": 0.8485,
900
  "step": 7700
901
  },
902
  {
903
- "epoch": 22.840409956076133,
904
- "grad_norm": 3.7167255878448486,
905
  "learning_rate": 3e-05,
906
- "loss": 0.8711,
907
  "step": 7800
908
  },
909
  {
910
- "epoch": 22.998535871156662,
911
- "eval_accuracy": 0.42512820512820515,
912
- "eval_loss": 5.346656322479248,
913
- "eval_runtime": 5.4264,
914
- "eval_samples_per_second": 92.143,
915
- "eval_steps_per_second": 11.61,
916
  "step": 7854
917
  },
918
  {
919
- "epoch": 22.998535871156662,
920
- "eval_bleu": 0.18735744635859433,
921
- "eval_exact_match": 0.002,
922
- "eval_prefix_exact_match": 0.016,
923
  "step": 7854
924
  },
925
  {
926
- "epoch": 23.133235724743777,
927
- "grad_norm": 3.3037803173065186,
928
  "learning_rate": 3e-05,
929
- "loss": 0.8378,
930
  "step": 7900
931
  },
932
  {
933
- "epoch": 23.42606149341142,
934
- "grad_norm": 3.446135997772217,
935
  "learning_rate": 3e-05,
936
- "loss": 0.8102,
937
  "step": 8000
938
  },
939
  {
940
- "epoch": 23.718887262079065,
941
- "grad_norm": 3.0248115062713623,
942
  "learning_rate": 3e-05,
943
- "loss": 0.8082,
944
  "step": 8100
945
  },
946
  {
947
  "epoch": 24.0,
948
- "eval_accuracy": 0.42430769230769233,
949
- "eval_loss": 5.35550594329834,
950
- "eval_runtime": 5.1152,
951
- "eval_samples_per_second": 97.747,
952
- "eval_steps_per_second": 12.316,
953
  "step": 8196
954
  },
955
  {
956
  "epoch": 24.0,
957
- "eval_bleu": 0.1404162407909808,
958
- "eval_exact_match": 0.004,
959
- "eval_prefix_exact_match": 0.016,
960
  "step": 8196
961
  },
962
  {
963
- "epoch": 24.011713030746705,
964
- "grad_norm": 3.7103264331817627,
965
  "learning_rate": 3e-05,
966
- "loss": 0.8279,
967
  "step": 8200
968
  },
969
  {
970
- "epoch": 24.30453879941435,
971
- "grad_norm": 4.033411026000977,
972
  "learning_rate": 3e-05,
973
- "loss": 0.7449,
974
  "step": 8300
975
  },
976
  {
977
- "epoch": 24.597364568081993,
978
- "grad_norm": 3.5919322967529297,
979
  "learning_rate": 3e-05,
980
- "loss": 0.7654,
981
  "step": 8400
982
  },
983
  {
984
- "epoch": 24.890190336749633,
985
- "grad_norm": 3.9135005474090576,
986
  "learning_rate": 3e-05,
987
- "loss": 0.7854,
988
  "step": 8500
989
  },
990
  {
991
- "epoch": 24.998535871156662,
992
- "eval_accuracy": 0.4241025641025641,
993
- "eval_loss": 5.4628777503967285,
994
- "eval_runtime": 5.4188,
995
- "eval_samples_per_second": 92.272,
996
- "eval_steps_per_second": 11.626,
997
  "step": 8537
998
  },
999
  {
1000
- "epoch": 24.998535871156662,
1001
- "eval_bleu": 0.16704717168781627,
1002
- "eval_exact_match": 0.002,
1003
- "eval_prefix_exact_match": 0.018,
1004
  "step": 8537
1005
  }
1006
  ],
1007
  "logging_steps": 100,
1008
  "max_steps": 17050,
1009
- "num_input_tokens_seen": 0,
1010
  "num_train_epochs": 50,
1011
  "save_steps": 500,
1012
- "stateful_callbacks": {
1013
- "TrainerControl": {
1014
- "args": {
1015
- "should_epoch_stop": false,
1016
- "should_evaluate": false,
1017
- "should_log": false,
1018
- "should_save": true,
1019
- "should_training_stop": false
1020
- },
1021
- "attributes": {}
1022
- }
1023
- },
1024
- "total_flos": 7.331304629642199e+17,
1025
- "train_batch_size": 1,
1026
  "trial_name": null,
1027
  "trial_params": null
1028
  }
 
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
+ "epoch": 0.29,
 
13
  "learning_rate": 3e-05,
14
+ "loss": 1.574,
15
  "step": 100
16
  },
17
  {
18
+ "epoch": 0.59,
 
19
  "learning_rate": 3e-05,
20
+ "loss": 1.474,
21
  "step": 200
22
  },
23
  {
24
+ "epoch": 0.88,
 
25
  "learning_rate": 3e-05,
26
+ "loss": 1.4104,
27
  "step": 300
28
  },
29
  {
30
+ "epoch": 1.0,
31
+ "eval_accuracy": 0.4536923076923077,
32
+ "eval_loss": 3.3574578762054443,
33
+ "eval_runtime": 4.4299,
34
+ "eval_samples_per_second": 112.869,
35
+ "eval_steps_per_second": 14.221,
36
  "step": 341
37
  },
38
  {
39
+ "epoch": 1.0,
40
+ "eval_bleu": 0.08840557310176213,
41
+ "eval_exact_match": 0.0,
42
+ "eval_prefix_exact_match": 0.058,
43
  "step": 341
44
  },
45
  {
46
+ "epoch": 1.17,
 
47
  "learning_rate": 3e-05,
48
+ "loss": 1.3889,
49
  "step": 400
50
  },
51
  {
52
+ "epoch": 1.46,
 
53
  "learning_rate": 3e-05,
54
+ "loss": 1.387,
55
  "step": 500
56
  },
57
  {
58
+ "epoch": 1.76,
 
59
  "learning_rate": 3e-05,
60
+ "loss": 1.389,
61
  "step": 600
62
  },
63
  {
64
  "epoch": 2.0,
65
+ "eval_accuracy": 0.4543589743589744,
66
+ "eval_loss": 3.4179794788360596,
67
+ "eval_runtime": 4.1997,
68
+ "eval_samples_per_second": 119.055,
69
+ "eval_steps_per_second": 15.001,
70
  "step": 683
71
  },
72
  {
73
  "epoch": 2.0,
74
+ "eval_bleu": 0.09820569565497561,
75
+ "eval_exact_match": 0.0,
76
+ "eval_prefix_exact_match": 0.056,
77
  "step": 683
78
  },
79
  {
80
+ "epoch": 2.05,
 
81
  "learning_rate": 3e-05,
82
+ "loss": 1.3725,
83
  "step": 700
84
  },
85
  {
86
+ "epoch": 2.34,
 
87
  "learning_rate": 3e-05,
88
+ "loss": 1.3588,
89
  "step": 800
90
  },
91
  {
92
+ "epoch": 2.64,
 
93
  "learning_rate": 3e-05,
94
+ "loss": 1.348,
95
  "step": 900
96
  },
97
  {
98
+ "epoch": 2.93,
 
99
  "learning_rate": 3e-05,
100
+ "loss": 1.3414,
101
  "step": 1000
102
  },
103
  {
104
+ "epoch": 3.0,
105
+ "eval_accuracy": 0.45476923076923076,
106
+ "eval_loss": 3.511923313140869,
107
+ "eval_runtime": 4.1985,
108
+ "eval_samples_per_second": 119.089,
109
+ "eval_steps_per_second": 15.005,
110
  "step": 1024
111
  },
112
  {
113
+ "epoch": 3.0,
114
+ "eval_bleu": 0.09889831624589397,
115
+ "eval_exact_match": 0.0,
116
+ "eval_prefix_exact_match": 0.056,
117
  "step": 1024
118
  },
119
  {
120
+ "epoch": 3.22,
 
121
  "learning_rate": 3e-05,
122
+ "loss": 1.3166,
123
  "step": 1100
124
  },
125
  {
126
+ "epoch": 3.51,
 
127
  "learning_rate": 3e-05,
128
+ "loss": 1.3035,
129
  "step": 1200
130
  },
131
  {
132
+ "epoch": 3.81,
 
133
  "learning_rate": 3e-05,
134
+ "loss": 1.3002,
135
  "step": 1300
136
  },
137
  {
138
  "epoch": 4.0,
139
+ "eval_accuracy": 0.45543589743589746,
140
+ "eval_loss": 3.5288071632385254,
141
+ "eval_runtime": 4.4066,
142
+ "eval_samples_per_second": 113.466,
143
+ "eval_steps_per_second": 14.297,
144
  "step": 1366
145
  },
146
  {
147
  "epoch": 4.0,
148
+ "eval_bleu": 0.11209345982903633,
149
+ "eval_exact_match": 0.0,
150
+ "eval_prefix_exact_match": 0.054,
151
  "step": 1366
152
  },
153
  {
154
+ "epoch": 4.1,
 
155
  "learning_rate": 3e-05,
156
+ "loss": 1.2929,
157
  "step": 1400
158
  },
159
  {
160
+ "epoch": 4.39,
 
161
  "learning_rate": 3e-05,
162
+ "loss": 1.2693,
163
  "step": 1500
164
  },
165
  {
166
+ "epoch": 4.69,
 
167
  "learning_rate": 3e-05,
168
+ "loss": 1.2697,
169
  "step": 1600
170
  },
171
  {
172
+ "epoch": 4.98,
 
173
  "learning_rate": 3e-05,
174
+ "loss": 1.2574,
175
  "step": 1700
176
  },
177
  {
178
+ "epoch": 5.0,
179
+ "eval_accuracy": 0.45394871794871794,
180
+ "eval_loss": 3.6893365383148193,
181
+ "eval_runtime": 4.1122,
182
+ "eval_samples_per_second": 121.591,
183
+ "eval_steps_per_second": 15.32,
184
  "step": 1707
185
  },
186
  {
187
+ "epoch": 5.0,
188
+ "eval_bleu": 0.12076940663377221,
189
+ "eval_exact_match": 0.0,
190
+ "eval_prefix_exact_match": 0.056,
191
  "step": 1707
192
  },
193
  {
194
+ "epoch": 5.27,
 
195
  "learning_rate": 3e-05,
196
+ "loss": 1.2097,
197
  "step": 1800
198
  },
199
  {
200
+ "epoch": 5.56,
 
201
  "learning_rate": 3e-05,
202
+ "loss": 1.2188,
203
  "step": 1900
204
  },
205
  {
206
+ "epoch": 5.86,
 
207
  "learning_rate": 3e-05,
208
+ "loss": 1.2258,
209
  "step": 2000
210
  },
211
  {
212
  "epoch": 6.0,
213
+ "eval_accuracy": 0.45615384615384613,
214
+ "eval_loss": 3.725893497467041,
215
+ "eval_runtime": 4.0989,
216
+ "eval_samples_per_second": 121.984,
217
+ "eval_steps_per_second": 15.37,
218
  "step": 2049
219
  },
220
  {
221
  "epoch": 6.0,
222
+ "eval_bleu": 0.11884272898137646,
223
+ "eval_exact_match": 0.0,
224
+ "eval_prefix_exact_match": 0.058,
225
  "step": 2049
226
  },
227
  {
228
+ "epoch": 6.15,
 
229
  "learning_rate": 3e-05,
230
+ "loss": 1.205,
231
  "step": 2100
232
  },
233
  {
234
+ "epoch": 6.44,
 
235
  "learning_rate": 3e-05,
236
+ "loss": 1.1643,
237
  "step": 2200
238
  },
239
  {
240
+ "epoch": 6.73,
 
241
  "learning_rate": 3e-05,
242
+ "loss": 1.1844,
243
  "step": 2300
244
  },
245
  {
246
+ "epoch": 7.0,
247
+ "eval_accuracy": 0.45594871794871794,
248
+ "eval_loss": 3.724449396133423,
249
+ "eval_runtime": 4.8158,
250
+ "eval_samples_per_second": 103.825,
251
+ "eval_steps_per_second": 13.082,
252
  "step": 2390
253
  },
254
  {
255
+ "epoch": 7.0,
256
+ "eval_bleu": 0.12407896336995003,
257
+ "eval_exact_match": 0.0,
258
+ "eval_prefix_exact_match": 0.048,
259
  "step": 2390
260
  },
261
  {
262
+ "epoch": 7.03,
 
263
  "learning_rate": 3e-05,
264
+ "loss": 1.1771,
265
  "step": 2400
266
  },
267
  {
268
+ "epoch": 7.32,
 
269
  "learning_rate": 3e-05,
270
+ "loss": 1.1218,
271
  "step": 2500
272
  },
273
  {
274
+ "epoch": 7.61,
 
275
  "learning_rate": 3e-05,
276
+ "loss": 1.1491,
277
  "step": 2600
278
  },
279
  {
280
+ "epoch": 7.91,
 
281
  "learning_rate": 3e-05,
282
+ "loss": 1.1363,
283
  "step": 2700
284
  },
285
  {
286
  "epoch": 8.0,
287
+ "eval_accuracy": 0.4543589743589744,
288
+ "eval_loss": 3.8138701915740967,
289
+ "eval_runtime": 4.1963,
290
+ "eval_samples_per_second": 119.152,
291
+ "eval_steps_per_second": 15.013,
292
  "step": 2732
293
  },
294
  {
295
  "epoch": 8.0,
296
+ "eval_bleu": 0.11947629327387163,
297
+ "eval_exact_match": 0.0,
298
+ "eval_prefix_exact_match": 0.046,
299
  "step": 2732
300
  },
301
  {
302
+ "epoch": 8.2,
 
303
  "learning_rate": 3e-05,
304
+ "loss": 1.0858,
305
  "step": 2800
306
  },
307
  {
308
+ "epoch": 8.49,
 
309
  "learning_rate": 3e-05,
310
+ "loss": 1.0951,
311
  "step": 2900
312
  },
313
  {
314
+ "epoch": 8.78,
 
315
  "learning_rate": 3e-05,
316
+ "loss": 1.0903,
317
  "step": 3000
318
  },
319
  {
320
+ "epoch": 9.0,
321
+ "eval_accuracy": 0.4524102564102564,
322
+ "eval_loss": 3.9115548133850098,
323
+ "eval_runtime": 4.4011,
324
+ "eval_samples_per_second": 113.609,
325
+ "eval_steps_per_second": 14.315,
326
  "step": 3073
327
  },
328
  {
329
+ "epoch": 9.0,
330
+ "eval_bleu": 0.13426062809240774,
331
+ "eval_exact_match": 0.0,
332
+ "eval_prefix_exact_match": 0.034,
333
  "step": 3073
334
  },
335
  {
336
+ "epoch": 9.08,
 
337
  "learning_rate": 3e-05,
338
+ "loss": 1.0822,
339
  "step": 3100
340
  },
341
  {
342
+ "epoch": 9.37,
 
343
  "learning_rate": 3e-05,
344
+ "loss": 1.0414,
345
  "step": 3200
346
  },
347
  {
348
+ "epoch": 9.66,
 
349
  "learning_rate": 3e-05,
350
+ "loss": 1.0563,
351
  "step": 3300
352
  },
353
  {
354
+ "epoch": 9.96,
 
355
  "learning_rate": 3e-05,
356
+ "loss": 1.0538,
357
  "step": 3400
358
  },
359
  {
360
  "epoch": 10.0,
361
+ "eval_accuracy": 0.4515897435897436,
362
+ "eval_loss": 3.92203426361084,
363
+ "eval_runtime": 4.4105,
364
+ "eval_samples_per_second": 113.366,
365
+ "eval_steps_per_second": 14.284,
366
  "step": 3415
367
  },
368
  {
369
  "epoch": 10.0,
370
+ "eval_bleu": 0.11780869016319889,
371
+ "eval_exact_match": 0.0,
372
  "eval_prefix_exact_match": 0.032,
373
  "step": 3415
374
  },
375
  {
376
+ "epoch": 10.25,
 
377
  "learning_rate": 3e-05,
378
+ "loss": 0.991,
379
  "step": 3500
380
  },
381
  {
382
+ "epoch": 10.54,
 
383
  "learning_rate": 3e-05,
384
+ "loss": 1.0019,
385
  "step": 3600
386
  },
387
  {
388
+ "epoch": 10.83,
 
389
  "learning_rate": 3e-05,
390
+ "loss": 0.9971,
391
  "step": 3700
392
  },
393
  {
394
+ "epoch": 11.0,
395
+ "eval_accuracy": 0.45143589743589746,
396
+ "eval_loss": 3.967252492904663,
397
+ "eval_runtime": 4.1003,
398
+ "eval_samples_per_second": 121.943,
399
+ "eval_steps_per_second": 15.365,
400
  "step": 3756
401
  },
402
  {
403
+ "epoch": 11.0,
404
+ "eval_bleu": 0.12407049349871409,
405
+ "eval_exact_match": 0.0,
406
  "eval_prefix_exact_match": 0.026,
407
  "step": 3756
408
  },
409
  {
410
+ "epoch": 11.13,
 
411
  "learning_rate": 3e-05,
412
+ "loss": 0.9861,
413
  "step": 3800
414
  },
415
  {
416
+ "epoch": 11.42,
 
417
  "learning_rate": 3e-05,
418
+ "loss": 0.9436,
419
  "step": 3900
420
  },
421
  {
422
+ "epoch": 11.71,
 
423
  "learning_rate": 3e-05,
424
+ "loss": 0.9699,
425
  "step": 4000
426
  },
427
  {
428
  "epoch": 12.0,
429
+ "eval_accuracy": 0.45076923076923076,
430
+ "eval_loss": 4.033609867095947,
431
+ "eval_runtime": 4.8103,
432
+ "eval_samples_per_second": 103.943,
433
+ "eval_steps_per_second": 13.097,
434
  "step": 4098
435
  },
436
  {
437
  "epoch": 12.0,
438
+ "eval_bleu": 0.12190601380876163,
439
+ "eval_exact_match": 0.0,
440
+ "eval_prefix_exact_match": 0.026,
441
  "step": 4098
442
  },
443
  {
444
+ "epoch": 12.01,
 
445
  "learning_rate": 3e-05,
446
+ "loss": 0.9783,
447
  "step": 4100
448
  },
449
  {
450
+ "epoch": 12.3,
 
451
  "learning_rate": 3e-05,
452
+ "loss": 0.9064,
453
  "step": 4200
454
  },
455
  {
456
+ "epoch": 12.59,
 
457
  "learning_rate": 3e-05,
458
+ "loss": 0.9178,
459
  "step": 4300
460
  },
461
  {
462
+ "epoch": 12.88,
 
463
  "learning_rate": 3e-05,
464
+ "loss": 0.9235,
465
  "step": 4400
466
  },
467
  {
468
+ "epoch": 13.0,
469
+ "eval_accuracy": 0.4492820512820513,
470
+ "eval_loss": 4.002023696899414,
471
+ "eval_runtime": 4.8204,
472
+ "eval_samples_per_second": 103.726,
473
+ "eval_steps_per_second": 13.069,
474
  "step": 4439
475
  },
476
  {
477
+ "epoch": 13.0,
478
+ "eval_bleu": 0.13115986163477472,
479
+ "eval_exact_match": 0.0,
480
+ "eval_prefix_exact_match": 0.026,
481
  "step": 4439
482
  },
483
  {
484
+ "epoch": 13.18,
 
485
  "learning_rate": 3e-05,
486
+ "loss": 0.8902,
487
  "step": 4500
488
  },
489
  {
490
+ "epoch": 13.47,
 
491
  "learning_rate": 3e-05,
492
+ "loss": 0.8703,
493
  "step": 4600
494
  },
495
  {
496
+ "epoch": 13.76,
 
497
  "learning_rate": 3e-05,
498
+ "loss": 0.891,
499
  "step": 4700
500
  },
501
  {
502
  "epoch": 14.0,
503
+ "eval_accuracy": 0.44774358974358974,
504
+ "eval_loss": 4.071566104888916,
505
+ "eval_runtime": 4.8316,
506
+ "eval_samples_per_second": 103.485,
507
+ "eval_steps_per_second": 13.039,
508
  "step": 4781
509
  },
510
  {
511
  "epoch": 14.0,
512
+ "eval_bleu": 0.12164300660650715,
513
+ "eval_exact_match": 0.0,
514
+ "eval_prefix_exact_match": 0.024,
515
  "step": 4781
516
  },
517
  {
518
+ "epoch": 14.06,
 
519
  "learning_rate": 3e-05,
520
+ "loss": 0.8698,
521
  "step": 4800
522
  },
523
  {
524
+ "epoch": 14.35,
 
525
  "learning_rate": 3e-05,
526
+ "loss": 0.8338,
527
  "step": 4900
528
  },
529
  {
530
+ "epoch": 14.64,
 
531
  "learning_rate": 3e-05,
532
+ "loss": 0.8362,
533
  "step": 5000
534
  },
535
  {
536
+ "epoch": 14.93,
 
537
  "learning_rate": 3e-05,
538
+ "loss": 0.845,
539
  "step": 5100
540
  },
541
  {
542
+ "epoch": 15.0,
543
+ "eval_accuracy": 0.44774358974358974,
544
+ "eval_loss": 4.09920597076416,
545
+ "eval_runtime": 4.4174,
546
+ "eval_samples_per_second": 113.19,
547
+ "eval_steps_per_second": 14.262,
548
  "step": 5122
549
  },
550
  {
551
+ "epoch": 15.0,
552
+ "eval_bleu": 0.1224977607294519,
553
+ "eval_exact_match": 0.0,
554
+ "eval_prefix_exact_match": 0.02,
555
  "step": 5122
556
  },
557
  {
558
+ "epoch": 15.23,
 
559
  "learning_rate": 3e-05,
560
+ "loss": 0.7981,
561
  "step": 5200
562
  },
563
  {
564
+ "epoch": 15.52,
 
565
  "learning_rate": 3e-05,
566
+ "loss": 0.7978,
567
  "step": 5300
568
  },
569
  {
570
+ "epoch": 15.81,
 
571
  "learning_rate": 3e-05,
572
+ "loss": 0.8009,
573
  "step": 5400
574
  },
575
  {
576
  "epoch": 16.0,
577
+ "eval_accuracy": 0.4464102564102564,
578
+ "eval_loss": 4.093270301818848,
579
+ "eval_runtime": 4.102,
580
+ "eval_samples_per_second": 121.892,
581
+ "eval_steps_per_second": 15.358,
582
  "step": 5464
583
  },
584
  {
585
  "epoch": 16.0,
586
+ "eval_bleu": 0.12201237729153987,
587
+ "eval_exact_match": 0.0,
588
+ "eval_prefix_exact_match": 0.022,
589
  "step": 5464
590
  },
591
  {
592
+ "epoch": 16.11,
 
593
  "learning_rate": 3e-05,
594
+ "loss": 0.7948,
595
  "step": 5500
596
  },
597
  {
598
+ "epoch": 16.4,
 
599
  "learning_rate": 3e-05,
600
+ "loss": 0.746,
601
  "step": 5600
602
  },
603
  {
604
+ "epoch": 16.69,
 
605
  "learning_rate": 3e-05,
606
+ "loss": 0.7627,
607
  "step": 5700
608
  },
609
  {
610
+ "epoch": 16.98,
 
611
  "learning_rate": 3e-05,
612
+ "loss": 0.782,
613
  "step": 5800
614
  },
615
  {
616
+ "epoch": 17.0,
617
+ "eval_accuracy": 0.44671794871794873,
618
+ "eval_loss": 4.1283488273620605,
619
+ "eval_runtime": 4.84,
620
+ "eval_samples_per_second": 103.305,
621
+ "eval_steps_per_second": 13.016,
622
  "step": 5805
623
  },
624
  {
625
+ "epoch": 17.0,
626
+ "eval_bleu": 0.1263852710208009,
627
+ "eval_exact_match": 0.0,
628
+ "eval_prefix_exact_match": 0.02,
629
  "step": 5805
630
  },
631
  {
632
+ "epoch": 17.28,
 
633
  "learning_rate": 3e-05,
634
+ "loss": 0.7142,
635
  "step": 5900
636
  },
637
  {
638
+ "epoch": 17.57,
 
639
  "learning_rate": 3e-05,
640
+ "loss": 0.7253,
641
  "step": 6000
642
  },
643
  {
644
+ "epoch": 17.86,
 
645
  "learning_rate": 3e-05,
646
+ "loss": 0.7294,
647
  "step": 6100
648
  },
649
  {
650
  "epoch": 18.0,
651
+ "eval_accuracy": 0.44564102564102565,
652
+ "eval_loss": 4.164257049560547,
653
+ "eval_runtime": 4.1037,
654
+ "eval_samples_per_second": 121.84,
655
+ "eval_steps_per_second": 15.352,
656
  "step": 6147
657
  },
658
  {
659
  "epoch": 18.0,
660
+ "eval_bleu": 0.10715019125675011,
661
+ "eval_exact_match": 0.0,
662
+ "eval_prefix_exact_match": 0.022,
663
  "step": 6147
664
  },
665
  {
666
+ "epoch": 18.16,
 
667
  "learning_rate": 3e-05,
668
+ "loss": 0.7055,
669
  "step": 6200
670
  },
671
  {
672
+ "epoch": 18.45,
 
673
  "learning_rate": 3e-05,
674
+ "loss": 0.672,
675
  "step": 6300
676
  },
677
  {
678
+ "epoch": 18.74,
 
679
  "learning_rate": 3e-05,
680
+ "loss": 0.6792,
681
  "step": 6400
682
  },
683
  {
684
+ "epoch": 19.0,
685
+ "eval_accuracy": 0.44487179487179485,
686
+ "eval_loss": 4.185911655426025,
687
+ "eval_runtime": 4.823,
688
+ "eval_samples_per_second": 103.67,
689
+ "eval_steps_per_second": 13.062,
690
  "step": 6488
691
  },
692
  {
693
+ "epoch": 19.0,
694
+ "eval_bleu": 0.11314440439431668,
695
  "eval_exact_match": 0.0,
696
  "eval_prefix_exact_match": 0.02,
697
  "step": 6488
698
  },
699
  {
700
+ "epoch": 19.03,
 
701
  "learning_rate": 3e-05,
702
+ "loss": 0.7053,
703
  "step": 6500
704
  },
705
  {
706
+ "epoch": 19.33,
 
707
  "learning_rate": 3e-05,
708
+ "loss": 0.6448,
709
  "step": 6600
710
  },
711
  {
712
+ "epoch": 19.62,
 
713
  "learning_rate": 3e-05,
714
+ "loss": 0.6448,
715
  "step": 6700
716
  },
717
  {
718
+ "epoch": 19.91,
 
719
  "learning_rate": 3e-05,
720
+ "loss": 0.6672,
721
  "step": 6800
722
  },
723
  {
724
  "epoch": 20.0,
725
+ "eval_accuracy": 0.4436923076923077,
726
+ "eval_loss": 4.201004981994629,
727
+ "eval_runtime": 4.1985,
728
+ "eval_samples_per_second": 119.09,
729
+ "eval_steps_per_second": 15.005,
730
  "step": 6830
731
  },
732
  {
733
  "epoch": 20.0,
734
+ "eval_bleu": 0.11913021835274795,
735
+ "eval_exact_match": 0.0,
736
+ "eval_prefix_exact_match": 0.02,
737
  "step": 6830
738
  },
739
  {
740
+ "epoch": 20.2,
 
741
  "learning_rate": 3e-05,
742
+ "loss": 0.6146,
743
  "step": 6900
744
  },
745
  {
746
+ "epoch": 20.5,
 
747
  "learning_rate": 3e-05,
748
+ "loss": 0.6102,
749
  "step": 7000
750
  },
751
  {
752
+ "epoch": 20.79,
 
753
  "learning_rate": 3e-05,
754
+ "loss": 0.6258,
755
  "step": 7100
756
  },
757
  {
758
+ "epoch": 21.0,
759
+ "eval_accuracy": 0.4428717948717949,
760
+ "eval_loss": 4.230019569396973,
761
+ "eval_runtime": 4.4124,
762
+ "eval_samples_per_second": 113.318,
763
+ "eval_steps_per_second": 14.278,
764
  "step": 7171
765
  },
766
  {
767
+ "epoch": 21.0,
768
+ "eval_bleu": 0.12133891034082767,
769
+ "eval_exact_match": 0.0,
770
+ "eval_prefix_exact_match": 0.022,
771
  "step": 7171
772
  },
773
  {
774
+ "epoch": 21.08,
 
775
  "learning_rate": 3e-05,
776
+ "loss": 0.6149,
777
  "step": 7200
778
  },
779
  {
780
+ "epoch": 21.38,
 
781
  "learning_rate": 3e-05,
782
+ "loss": 0.5669,
783
  "step": 7300
784
  },
785
  {
786
+ "epoch": 21.67,
 
787
  "learning_rate": 3e-05,
788
+ "loss": 0.5923,
789
  "step": 7400
790
  },
791
  {
792
+ "epoch": 21.96,
 
793
  "learning_rate": 3e-05,
794
+ "loss": 0.599,
795
  "step": 7500
796
  },
797
  {
798
  "epoch": 22.0,
799
+ "eval_accuracy": 0.4418974358974359,
800
+ "eval_loss": 4.253176689147949,
801
+ "eval_runtime": 4.4121,
802
+ "eval_samples_per_second": 113.326,
803
+ "eval_steps_per_second": 14.279,
804
  "step": 7513
805
  },
806
  {
807
  "epoch": 22.0,
808
+ "eval_bleu": 0.1179919389999147,
809
+ "eval_exact_match": 0.0,
810
  "eval_prefix_exact_match": 0.02,
811
  "step": 7513
812
  },
813
  {
814
+ "epoch": 22.25,
 
815
  "learning_rate": 3e-05,
816
+ "loss": 0.536,
817
  "step": 7600
818
  },
819
  {
820
+ "epoch": 22.55,
 
821
  "learning_rate": 3e-05,
822
+ "loss": 0.5479,
823
  "step": 7700
824
  },
825
  {
826
+ "epoch": 22.84,
 
827
  "learning_rate": 3e-05,
828
+ "loss": 0.5625,
829
  "step": 7800
830
  },
831
  {
832
+ "epoch": 23.0,
833
+ "eval_accuracy": 0.443025641025641,
834
+ "eval_loss": 4.293737411499023,
835
+ "eval_runtime": 4.4185,
836
+ "eval_samples_per_second": 113.162,
837
+ "eval_steps_per_second": 14.258,
838
  "step": 7854
839
  },
840
  {
841
+ "epoch": 23.0,
842
+ "eval_bleu": 0.10699095482247514,
843
+ "eval_exact_match": 0.0,
844
+ "eval_prefix_exact_match": 0.022,
845
  "step": 7854
846
  },
847
  {
848
+ "epoch": 23.13,
 
849
  "learning_rate": 3e-05,
850
+ "loss": 0.5409,
851
  "step": 7900
852
  },
853
  {
854
+ "epoch": 23.43,
 
855
  "learning_rate": 3e-05,
856
+ "loss": 0.5067,
857
  "step": 8000
858
  },
859
  {
860
+ "epoch": 23.72,
 
861
  "learning_rate": 3e-05,
862
+ "loss": 0.5267,
863
  "step": 8100
864
  },
865
  {
866
  "epoch": 24.0,
867
+ "eval_accuracy": 0.44153846153846155,
868
+ "eval_loss": 4.254815578460693,
869
+ "eval_runtime": 4.2068,
870
+ "eval_samples_per_second": 118.854,
871
+ "eval_steps_per_second": 14.976,
872
  "step": 8196
873
  },
874
  {
875
  "epoch": 24.0,
876
+ "eval_bleu": 0.10975165733539657,
877
+ "eval_exact_match": 0.0,
878
+ "eval_prefix_exact_match": 0.024,
879
  "step": 8196
880
  },
881
  {
882
+ "epoch": 24.01,
 
883
  "learning_rate": 3e-05,
884
+ "loss": 0.5363,
885
  "step": 8200
886
  },
887
  {
888
+ "epoch": 24.3,
 
889
  "learning_rate": 3e-05,
890
+ "loss": 0.4806,
891
  "step": 8300
892
  },
893
  {
894
+ "epoch": 24.6,
 
895
  "learning_rate": 3e-05,
896
+ "loss": 0.49,
897
  "step": 8400
898
  },
899
  {
900
+ "epoch": 24.89,
 
901
  "learning_rate": 3e-05,
902
+ "loss": 0.5004,
903
  "step": 8500
904
  },
905
  {
906
+ "epoch": 25.0,
907
+ "eval_accuracy": 0.44035897435897436,
908
+ "eval_loss": 4.332499027252197,
909
+ "eval_runtime": 4.4052,
910
+ "eval_samples_per_second": 113.503,
911
+ "eval_steps_per_second": 14.301,
912
  "step": 8537
913
  },
914
  {
915
+ "epoch": 25.0,
916
+ "eval_bleu": 0.10401238605841524,
917
+ "eval_exact_match": 0.0,
918
+ "eval_prefix_exact_match": 0.024,
919
  "step": 8537
920
  }
921
  ],
922
  "logging_steps": 100,
923
  "max_steps": 17050,
 
924
  "num_train_epochs": 50,
925
  "save_steps": 500,
926
+ "total_flos": 1.6905304374467625e+18,
 
 
 
 
 
 
 
 
 
 
 
 
 
927
  "trial_name": null,
928
  "trial_params": null
929
  }
checkpoint-8537/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f9cef08c7f10bf396c269dfc26104fd5a94756f5431d6448312805ef8e298380
3
- size 5304
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28a33ee2cb1c54e667d59907eed3cc20ab64c3794314f21d17db553ff2af83d6
3
+ size 4728