tyzhu commited on
Commit
9080b47
1 Parent(s): ddc3ebf

Training in progress, epoch 16, checkpoint

Browse files
checkpoint-3187/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-3187/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:baaca81aa9fec701b9f9a29e2f12f5cfc60d32c9e34aa5f453b41c2d0304564e
3
+ size 143269386
checkpoint-3187/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-3187/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bb6661450b64a13c82b95387c5110684e10b5fe69004cde4a7eacef1bee5270e
3
- size 224537202
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3581f3a0866f29bea7978633e65108a6f248b38d44738aa2fbb9899d8739a5a3
3
+ size 286585234
checkpoint-3187/rng_state_0.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d3fd78d021136b3e2b73b1885acb32c6f149b625baec9ce5ec8128d659f8b791
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f2631247fd2a2a6215d801d3bb4b34e464fc191bc4d7c6d132657b78e40ba0e8
3
  size 15024
checkpoint-3187/rng_state_1.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bfbeb4ba7019a5aec346bd0905fc9e73d74740c0e1b3520854f41c939740b1ad
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e4e5b8373c1b17cebbb3b241f03e813b828e5cfe4bfa7fa8c43fae7b97fc72b7
3
  size 15024
checkpoint-3187/rng_state_2.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e284d19297663bccdfa5c9a7c24d23ed3d70a6c2b00e3db35bdc25629c982f89
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61dfc74303d2d4219b871879b3a1c842198f1f57dc55a9ce0603283244e15932
3
  size 15024
checkpoint-3187/rng_state_3.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9d9862d62d6f91dac7ddeca3e6d30adca84c4ec459b8ac7e489da983d6619244
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e22159ec345a7662d7def4595d1ec0860fd629ad073d01bf9a9a23619971b8e2
3
  size 15024
checkpoint-3187/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-3187/tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
 
checkpoint-3187/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-3187/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-3187/trainer_state.json CHANGED
@@ -9,565 +9,520 @@
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
- "epoch": 0.5333333333333333,
13
- "grad_norm": 0.275871217250824,
14
  "learning_rate": 3e-05,
15
- "loss": 1.8545,
16
  "step": 100
17
  },
18
  {
19
- "epoch": 0.9973333333333333,
20
- "eval_accuracy": 0.6037399103139014,
21
- "eval_loss": 1.7123007774353027,
22
- "eval_runtime": 6.2976,
23
- "eval_samples_per_second": 79.395,
24
- "eval_steps_per_second": 10.004,
25
  "step": 187
26
  },
27
  {
28
- "epoch": 0.9973333333333333,
29
- "eval_exact_match": 14.4,
30
- "eval_f1": 22.843015873015876,
31
- "eval_qa_bleu": 9.471299126615676,
32
- "eval_qa_exact_match": 0.112,
33
- "eval_recite_bleu": 12.267251092702137,
34
  "eval_recite_exact_match": 0.0,
35
  "step": 187
36
  },
37
  {
38
- "epoch": 1.0666666666666667,
39
- "grad_norm": 0.32765188813209534,
40
  "learning_rate": 3e-05,
41
- "loss": 1.7408,
42
  "step": 200
43
  },
44
  {
45
  "epoch": 1.6,
46
- "grad_norm": 0.3532409369945526,
47
  "learning_rate": 3e-05,
48
- "loss": 1.7063,
49
  "step": 300
50
  },
51
  {
52
  "epoch": 2.0,
53
- "eval_accuracy": 0.6058565022421525,
54
- "eval_loss": 1.69453763961792,
55
- "eval_runtime": 6.5928,
56
- "eval_samples_per_second": 75.841,
57
- "eval_steps_per_second": 9.556,
58
  "step": 375
59
  },
60
  {
61
  "epoch": 2.0,
62
- "eval_exact_match": 14.4,
63
- "eval_f1": 23.29731601731603,
64
- "eval_qa_bleu": 9.4408132575913,
65
- "eval_qa_exact_match": 0.11,
66
- "eval_recite_bleu": 13.027794116434837,
67
  "eval_recite_exact_match": 0.0,
68
  "step": 375
69
  },
70
  {
71
- "epoch": 2.1333333333333333,
72
- "grad_norm": 0.3777780532836914,
73
  "learning_rate": 3e-05,
74
- "loss": 1.7094,
75
  "step": 400
76
  },
77
  {
78
- "epoch": 2.6666666666666665,
79
- "grad_norm": 0.3971036374568939,
80
  "learning_rate": 3e-05,
81
- "loss": 1.6702,
82
  "step": 500
83
  },
84
  {
85
- "epoch": 2.997333333333333,
86
- "eval_accuracy": 0.6065919282511211,
87
- "eval_loss": 1.6851236820220947,
88
- "eval_runtime": 6.3104,
89
- "eval_samples_per_second": 79.234,
90
- "eval_steps_per_second": 9.984,
91
  "step": 562
92
  },
93
  {
94
- "epoch": 2.997333333333333,
95
- "eval_exact_match": 14.0,
96
- "eval_f1": 23.306825396825417,
97
- "eval_qa_bleu": 8.683248465366908,
98
- "eval_qa_exact_match": 0.11,
99
- "eval_recite_bleu": 12.39336052958371,
100
- "eval_recite_exact_match": 0.006,
101
  "step": 562
102
  },
103
  {
104
  "epoch": 3.2,
105
- "grad_norm": 0.4782978594303131,
106
  "learning_rate": 3e-05,
107
- "loss": 1.6574,
108
  "step": 600
109
  },
110
  {
111
- "epoch": 3.7333333333333334,
112
- "grad_norm": 0.52582186460495,
113
  "learning_rate": 3e-05,
114
- "loss": 1.6356,
115
  "step": 700
116
  },
117
  {
118
  "epoch": 4.0,
119
- "eval_accuracy": 0.6074977578475337,
120
- "eval_loss": 1.682567834854126,
121
- "eval_runtime": 6.65,
122
- "eval_samples_per_second": 75.188,
123
- "eval_steps_per_second": 9.474,
124
  "step": 750
125
  },
126
  {
127
  "epoch": 4.0,
128
- "eval_exact_match": 15.0,
129
- "eval_f1": 23.366825396825416,
130
- "eval_qa_bleu": 9.05994637010322,
131
- "eval_qa_exact_match": 0.114,
132
- "eval_recite_bleu": 13.060415058384661,
133
- "eval_recite_exact_match": 0.006,
134
  "step": 750
135
  },
136
  {
137
- "epoch": 4.266666666666667,
138
- "grad_norm": 0.6418241262435913,
139
  "learning_rate": 3e-05,
140
- "loss": 1.6005,
141
  "step": 800
142
  },
143
  {
144
  "epoch": 4.8,
145
- "grad_norm": 0.6805307269096375,
146
  "learning_rate": 3e-05,
147
- "loss": 1.5775,
148
  "step": 900
149
  },
150
  {
151
- "epoch": 4.997333333333334,
152
- "eval_accuracy": 0.6071031390134529,
153
- "eval_loss": 1.69106125831604,
154
- "eval_runtime": 5.6598,
155
- "eval_samples_per_second": 88.343,
156
- "eval_steps_per_second": 11.131,
157
  "step": 937
158
  },
159
  {
160
- "epoch": 4.997333333333334,
161
- "eval_exact_match": 15.2,
162
- "eval_f1": 23.38539682539684,
163
- "eval_qa_bleu": 11.854591678140883,
164
- "eval_qa_exact_match": 0.112,
165
- "eval_recite_bleu": 12.694062454678122,
166
- "eval_recite_exact_match": 0.004,
167
  "step": 937
168
  },
169
  {
170
- "epoch": 5.333333333333333,
171
- "grad_norm": 0.8362048864364624,
172
  "learning_rate": 3e-05,
173
- "loss": 1.5173,
174
  "step": 1000
175
  },
176
  {
177
- "epoch": 5.866666666666667,
178
- "grad_norm": 0.8521270155906677,
179
  "learning_rate": 3e-05,
180
- "loss": 1.529,
181
  "step": 1100
182
  },
183
  {
184
  "epoch": 6.0,
185
- "eval_accuracy": 0.6069417040358744,
186
- "eval_loss": 1.7066757678985596,
187
- "eval_runtime": 6.2922,
188
- "eval_samples_per_second": 79.464,
189
- "eval_steps_per_second": 10.012,
190
  "step": 1125
191
  },
192
  {
193
  "epoch": 6.0,
194
- "eval_exact_match": 15.2,
195
- "eval_f1": 24.04682539682541,
196
- "eval_qa_bleu": 12.150293484705452,
197
- "eval_qa_exact_match": 0.106,
198
- "eval_recite_bleu": 12.732083161926754,
199
- "eval_recite_exact_match": 0.002,
200
  "step": 1125
201
  },
202
  {
203
  "epoch": 6.4,
204
- "grad_norm": 0.9372308254241943,
205
  "learning_rate": 3e-05,
206
- "loss": 1.4613,
207
  "step": 1200
208
  },
209
  {
210
- "epoch": 6.933333333333334,
211
- "grad_norm": 1.0236693620681763,
212
  "learning_rate": 3e-05,
213
- "loss": 1.457,
214
  "step": 1300
215
  },
216
  {
217
- "epoch": 6.997333333333334,
218
- "eval_accuracy": 0.6055515695067265,
219
- "eval_loss": 1.72792649269104,
220
- "eval_runtime": 6.3278,
221
- "eval_samples_per_second": 79.016,
222
- "eval_steps_per_second": 9.956,
223
  "step": 1312
224
  },
225
  {
226
- "epoch": 6.997333333333334,
227
- "eval_exact_match": 15.0,
228
- "eval_f1": 23.28492063492065,
229
- "eval_qa_bleu": 9.849658422779212,
230
- "eval_qa_exact_match": 0.114,
231
- "eval_recite_bleu": 12.3997433702838,
232
- "eval_recite_exact_match": 0.006,
233
  "step": 1312
234
  },
235
  {
236
- "epoch": 7.466666666666667,
237
- "grad_norm": 1.0773049592971802,
238
  "learning_rate": 3e-05,
239
- "loss": 1.3996,
240
  "step": 1400
241
  },
242
  {
243
  "epoch": 8.0,
244
- "grad_norm": 1.0705862045288086,
245
  "learning_rate": 3e-05,
246
- "loss": 1.3907,
247
  "step": 1500
248
  },
249
  {
250
  "epoch": 8.0,
251
- "eval_accuracy": 0.6046188340807175,
252
- "eval_loss": 1.7511730194091797,
253
- "eval_runtime": 5.5895,
254
- "eval_samples_per_second": 89.453,
255
- "eval_steps_per_second": 11.271,
256
  "step": 1500
257
  },
258
  {
259
  "epoch": 8.0,
260
- "eval_exact_match": 13.0,
261
- "eval_f1": 19.93238095238096,
262
- "eval_qa_bleu": 11.93942669425287,
263
- "eval_qa_exact_match": 0.098,
264
- "eval_recite_bleu": 11.949414001720045,
265
- "eval_recite_exact_match": 0.006,
266
  "step": 1500
267
  },
268
  {
269
- "epoch": 8.533333333333333,
270
- "grad_norm": 1.2424529790878296,
271
  "learning_rate": 3e-05,
272
- "loss": 1.3309,
273
  "step": 1600
274
  },
275
  {
276
- "epoch": 8.997333333333334,
277
- "eval_accuracy": 0.6025022421524664,
278
- "eval_loss": 1.7774019241333008,
279
- "eval_runtime": 6.4013,
280
- "eval_samples_per_second": 78.109,
281
- "eval_steps_per_second": 9.842,
282
  "step": 1687
283
  },
284
  {
285
- "epoch": 8.997333333333334,
286
- "eval_exact_match": 13.6,
287
- "eval_f1": 21.089523809523815,
288
- "eval_qa_bleu": 10.094627644241111,
289
- "eval_qa_exact_match": 0.094,
290
- "eval_recite_bleu": 12.197681191541987,
291
- "eval_recite_exact_match": 0.006,
292
  "step": 1687
293
  },
294
  {
295
- "epoch": 9.066666666666666,
296
- "grad_norm": 1.535229206085205,
297
  "learning_rate": 3e-05,
298
- "loss": 1.3336,
299
  "step": 1700
300
  },
301
  {
302
  "epoch": 9.6,
303
- "grad_norm": 1.4936225414276123,
304
  "learning_rate": 3e-05,
305
- "loss": 1.2841,
306
  "step": 1800
307
  },
308
  {
309
  "epoch": 10.0,
310
- "eval_accuracy": 0.6012645739910314,
311
- "eval_loss": 1.804250717163086,
312
- "eval_runtime": 6.385,
313
- "eval_samples_per_second": 78.309,
314
- "eval_steps_per_second": 9.867,
315
  "step": 1875
316
  },
317
  {
318
  "epoch": 10.0,
319
- "eval_exact_match": 13.2,
320
- "eval_f1": 21.172832722832737,
321
- "eval_qa_bleu": 9.841075001407239,
322
- "eval_qa_exact_match": 0.096,
323
- "eval_recite_bleu": 11.691398764835105,
324
- "eval_recite_exact_match": 0.004,
325
  "step": 1875
326
  },
327
  {
328
- "epoch": 10.133333333333333,
329
- "grad_norm": 1.6687729358673096,
330
  "learning_rate": 3e-05,
331
- "loss": 1.2562,
332
  "step": 1900
333
  },
334
  {
335
- "epoch": 10.666666666666666,
336
- "grad_norm": 1.437538504600525,
337
  "learning_rate": 3e-05,
338
- "loss": 1.2308,
339
  "step": 2000
340
  },
341
  {
342
- "epoch": 10.997333333333334,
343
- "eval_accuracy": 0.6000717488789238,
344
- "eval_loss": 1.8528188467025757,
345
- "eval_runtime": 6.6644,
346
- "eval_samples_per_second": 75.026,
347
- "eval_steps_per_second": 9.453,
348
  "step": 2062
349
  },
350
  {
351
- "epoch": 10.997333333333334,
352
- "eval_exact_match": 12.6,
353
- "eval_f1": 20.865213675213692,
354
- "eval_qa_bleu": 7.784433449171618,
355
- "eval_qa_exact_match": 0.108,
356
- "eval_recite_bleu": 12.305413068460787,
357
- "eval_recite_exact_match": 0.008,
358
  "step": 2062
359
  },
360
  {
361
  "epoch": 11.2,
362
- "grad_norm": 1.708260416984558,
363
  "learning_rate": 3e-05,
364
- "loss": 1.2011,
365
  "step": 2100
366
  },
367
  {
368
- "epoch": 11.733333333333333,
369
- "grad_norm": 1.7026830911636353,
370
  "learning_rate": 3e-05,
371
- "loss": 1.1722,
372
  "step": 2200
373
  },
374
  {
375
  "epoch": 12.0,
376
- "eval_accuracy": 0.5987892376681614,
377
- "eval_loss": 1.8851206302642822,
378
- "eval_runtime": 5.5764,
379
- "eval_samples_per_second": 89.663,
380
- "eval_steps_per_second": 11.298,
381
  "step": 2250
382
  },
383
  {
384
  "epoch": 12.0,
385
- "eval_exact_match": 13.0,
386
- "eval_f1": 21.17809523809525,
387
- "eval_qa_bleu": 11.19937598521782,
388
- "eval_qa_exact_match": 0.096,
389
- "eval_recite_bleu": 12.500534642121808,
390
- "eval_recite_exact_match": 0.01,
391
  "step": 2250
392
  },
393
  {
394
- "epoch": 12.266666666666667,
395
- "grad_norm": 1.6281794309616089,
396
  "learning_rate": 3e-05,
397
- "loss": 1.151,
398
  "step": 2300
399
  },
400
  {
401
  "epoch": 12.8,
402
- "grad_norm": 1.686490535736084,
403
  "learning_rate": 3e-05,
404
- "loss": 1.1354,
405
  "step": 2400
406
  },
407
  {
408
- "epoch": 12.997333333333334,
409
- "eval_accuracy": 0.5979910313901345,
410
- "eval_loss": 1.9114351272583008,
411
- "eval_runtime": 5.5718,
412
- "eval_samples_per_second": 89.737,
413
- "eval_steps_per_second": 11.307,
414
  "step": 2437
415
  },
416
  {
417
- "epoch": 12.997333333333334,
418
- "eval_exact_match": 13.8,
419
- "eval_f1": 21.856666666666673,
420
- "eval_qa_bleu": 10.206087008658956,
421
- "eval_qa_exact_match": 0.116,
422
- "eval_recite_bleu": 12.669262328097592,
423
- "eval_recite_exact_match": 0.008,
424
  "step": 2437
425
  },
426
  {
427
- "epoch": 13.333333333333334,
428
- "grad_norm": 1.8666125535964966,
429
  "learning_rate": 3e-05,
430
- "loss": 1.097,
431
  "step": 2500
432
  },
433
  {
434
- "epoch": 13.866666666666667,
435
- "grad_norm": 1.9091932773590088,
436
  "learning_rate": 3e-05,
437
- "loss": 1.0793,
438
  "step": 2600
439
  },
440
  {
441
  "epoch": 14.0,
442
- "eval_accuracy": 0.5960986547085202,
443
- "eval_loss": 1.9584920406341553,
444
- "eval_runtime": 6.3056,
445
- "eval_samples_per_second": 79.294,
446
- "eval_steps_per_second": 9.991,
447
  "step": 2625
448
  },
449
  {
450
  "epoch": 14.0,
451
- "eval_exact_match": 10.6,
452
- "eval_f1": 18.697936507936518,
453
- "eval_qa_bleu": 6.4581555766931436,
454
- "eval_qa_exact_match": 0.082,
455
- "eval_recite_bleu": 12.843526576123885,
456
- "eval_recite_exact_match": 0.01,
457
  "step": 2625
458
  },
459
  {
460
  "epoch": 14.4,
461
- "grad_norm": 2.042712688446045,
462
  "learning_rate": 3e-05,
463
- "loss": 1.0376,
464
  "step": 2700
465
  },
466
  {
467
- "epoch": 14.933333333333334,
468
- "grad_norm": 1.8875784873962402,
469
  "learning_rate": 3e-05,
470
- "loss": 1.037,
471
  "step": 2800
472
  },
473
  {
474
- "epoch": 14.997333333333334,
475
- "eval_accuracy": 0.5948161434977578,
476
- "eval_loss": 1.9966908693313599,
477
- "eval_runtime": 6.0322,
478
- "eval_samples_per_second": 82.888,
479
- "eval_steps_per_second": 10.444,
480
  "step": 2812
481
  },
482
  {
483
- "epoch": 14.997333333333334,
484
- "eval_exact_match": 12.8,
485
- "eval_f1": 20.438917748917753,
486
- "eval_qa_bleu": 10.19364322169534,
487
- "eval_qa_exact_match": 0.104,
488
- "eval_recite_bleu": 12.69742056629334,
489
- "eval_recite_exact_match": 0.012,
490
  "step": 2812
491
  },
492
  {
493
- "epoch": 15.466666666666667,
494
- "grad_norm": 2.027426242828369,
495
  "learning_rate": 3e-05,
496
- "loss": 0.9844,
497
  "step": 2900
498
  },
499
  {
500
  "epoch": 16.0,
501
- "grad_norm": 2.0708699226379395,
502
  "learning_rate": 3e-05,
503
- "loss": 0.9901,
504
  "step": 3000
505
  },
506
  {
507
  "epoch": 16.0,
508
- "eval_accuracy": 0.5934080717488789,
509
- "eval_loss": 2.0336477756500244,
510
- "eval_runtime": 5.9811,
511
- "eval_samples_per_second": 83.597,
512
- "eval_steps_per_second": 10.533,
513
  "step": 3000
514
  },
515
  {
516
  "epoch": 16.0,
517
- "eval_exact_match": 12.4,
518
- "eval_f1": 20.427142857142865,
519
- "eval_qa_bleu": 9.13217752730269,
520
- "eval_qa_exact_match": 0.096,
521
- "eval_recite_bleu": 12.688250071816965,
522
- "eval_recite_exact_match": 0.014,
523
  "step": 3000
524
  },
525
  {
526
- "epoch": 16.533333333333335,
527
- "grad_norm": 1.995875358581543,
528
  "learning_rate": 3e-05,
529
- "loss": 0.9316,
530
  "step": 3100
531
  },
532
  {
533
- "epoch": 16.997333333333334,
534
- "eval_accuracy": 0.591372197309417,
535
- "eval_loss": 2.088000535964966,
536
- "eval_runtime": 5.9631,
537
- "eval_samples_per_second": 83.849,
538
- "eval_steps_per_second": 10.565,
539
  "step": 3187
540
  },
541
  {
542
- "epoch": 16.997333333333334,
543
- "eval_exact_match": 11.4,
544
- "eval_f1": 18.528571428571443,
545
- "eval_qa_bleu": 7.115393148941145,
546
- "eval_qa_exact_match": 0.084,
547
- "eval_recite_bleu": 12.219544702526637,
548
- "eval_recite_exact_match": 0.014,
549
  "step": 3187
550
  }
551
  ],
552
  "logging_steps": 100,
553
  "max_steps": 9350,
554
- "num_input_tokens_seen": 0,
555
  "num_train_epochs": 50,
556
  "save_steps": 500,
557
- "stateful_callbacks": {
558
- "TrainerControl": {
559
- "args": {
560
- "should_epoch_stop": false,
561
- "should_evaluate": false,
562
- "should_log": false,
563
- "should_save": true,
564
- "should_training_stop": false
565
- },
566
- "attributes": {}
567
- }
568
- },
569
- "total_flos": 3.5176710108767846e+17,
570
- "train_batch_size": 2,
571
  "trial_name": null,
572
  "trial_params": null
573
  }
 
9
  "is_world_process_zero": true,
10
  "log_history": [
11
  {
12
+ "epoch": 0.53,
 
13
  "learning_rate": 3e-05,
14
+ "loss": 1.4597,
15
  "step": 100
16
  },
17
  {
18
+ "epoch": 1.0,
19
+ "eval_accuracy": 0.6618980392156862,
20
+ "eval_loss": 1.2718806266784668,
21
+ "eval_runtime": 7.6194,
22
+ "eval_samples_per_second": 65.622,
23
+ "eval_steps_per_second": 8.268,
24
  "step": 187
25
  },
26
  {
27
+ "epoch": 1.0,
28
+ "eval_exact_match": 22.4,
29
+ "eval_f1": 30.941064425770318,
30
+ "eval_qa_bleu": 7.097403515221558,
31
+ "eval_qa_exact_match": 0.196,
32
+ "eval_recite_bleu": 13.258373740102225,
33
  "eval_recite_exact_match": 0.0,
34
  "step": 187
35
  },
36
  {
37
+ "epoch": 1.07,
 
38
  "learning_rate": 3e-05,
39
+ "loss": 1.2905,
40
  "step": 200
41
  },
42
  {
43
  "epoch": 1.6,
 
44
  "learning_rate": 3e-05,
45
+ "loss": 1.2157,
46
  "step": 300
47
  },
48
  {
49
  "epoch": 2.0,
50
+ "eval_accuracy": 0.6670039215686274,
51
+ "eval_loss": 1.206207036972046,
52
+ "eval_runtime": 7.6793,
53
+ "eval_samples_per_second": 65.11,
54
+ "eval_steps_per_second": 8.204,
55
  "step": 375
56
  },
57
  {
58
  "epoch": 2.0,
59
+ "eval_exact_match": 3.2,
60
+ "eval_f1": 3.96976236704394,
61
+ "eval_qa_bleu": 0.20484366294597936,
62
+ "eval_qa_exact_match": 0.026,
63
+ "eval_recite_bleu": 2.5778714769111824e-05,
64
  "eval_recite_exact_match": 0.0,
65
  "step": 375
66
  },
67
  {
68
+ "epoch": 2.13,
 
69
  "learning_rate": 3e-05,
70
+ "loss": 1.1852,
71
  "step": 400
72
  },
73
  {
74
+ "epoch": 2.67,
 
75
  "learning_rate": 3e-05,
76
+ "loss": 1.1861,
77
  "step": 500
78
  },
79
  {
80
+ "epoch": 3.0,
81
+ "eval_accuracy": 0.667513725490196,
82
+ "eval_loss": 1.19828462600708,
83
+ "eval_runtime": 7.1425,
84
+ "eval_samples_per_second": 70.004,
85
+ "eval_steps_per_second": 8.82,
86
  "step": 562
87
  },
88
  {
89
+ "epoch": 3.0,
90
+ "eval_exact_match": 4.4,
91
+ "eval_f1": 6.333660342551587,
92
+ "eval_qa_bleu": 0.522391008398391,
93
+ "eval_qa_exact_match": 0.028,
94
+ "eval_recite_bleu": 0.027569975851374308,
95
+ "eval_recite_exact_match": 0.0,
96
  "step": 562
97
  },
98
  {
99
  "epoch": 3.2,
 
100
  "learning_rate": 3e-05,
101
+ "loss": 1.1722,
102
  "step": 600
103
  },
104
  {
105
+ "epoch": 3.73,
 
106
  "learning_rate": 3e-05,
107
+ "loss": 1.1445,
108
  "step": 700
109
  },
110
  {
111
  "epoch": 4.0,
112
+ "eval_accuracy": 0.6672,
113
+ "eval_loss": 1.197709560394287,
114
+ "eval_runtime": 7.7427,
115
+ "eval_samples_per_second": 64.577,
116
+ "eval_steps_per_second": 8.137,
117
  "step": 750
118
  },
119
  {
120
  "epoch": 4.0,
121
+ "eval_exact_match": 7.4,
122
+ "eval_f1": 10.950016966795278,
123
+ "eval_qa_bleu": 0.6700083846206139,
124
+ "eval_qa_exact_match": 0.052,
125
+ "eval_recite_bleu": 1.259776140465531,
126
+ "eval_recite_exact_match": 0.0,
127
  "step": 750
128
  },
129
  {
130
+ "epoch": 4.27,
 
131
  "learning_rate": 3e-05,
132
+ "loss": 1.1018,
133
  "step": 800
134
  },
135
  {
136
  "epoch": 4.8,
 
137
  "learning_rate": 3e-05,
138
+ "loss": 1.1,
139
  "step": 900
140
  },
141
  {
142
+ "epoch": 5.0,
143
+ "eval_accuracy": 0.6664941176470588,
144
+ "eval_loss": 1.2078713178634644,
145
+ "eval_runtime": 7.7721,
146
+ "eval_samples_per_second": 64.333,
147
+ "eval_steps_per_second": 8.106,
148
  "step": 937
149
  },
150
  {
151
+ "epoch": 5.0,
152
+ "eval_exact_match": 6.6,
153
+ "eval_f1": 10.12048028049545,
154
+ "eval_qa_bleu": 0.8314625442591895,
155
+ "eval_qa_exact_match": 0.05,
156
+ "eval_recite_bleu": 0.8030246920772521,
157
+ "eval_recite_exact_match": 0.0,
158
  "step": 937
159
  },
160
  {
161
+ "epoch": 5.33,
 
162
  "learning_rate": 3e-05,
163
+ "loss": 1.0666,
164
  "step": 1000
165
  },
166
  {
167
+ "epoch": 5.87,
 
168
  "learning_rate": 3e-05,
169
+ "loss": 1.0439,
170
  "step": 1100
171
  },
172
  {
173
  "epoch": 6.0,
174
+ "eval_accuracy": 0.6649725490196079,
175
+ "eval_loss": 1.22383713722229,
176
+ "eval_runtime": 7.6639,
177
+ "eval_samples_per_second": 65.241,
178
+ "eval_steps_per_second": 8.22,
179
  "step": 1125
180
  },
181
  {
182
  "epoch": 6.0,
183
+ "eval_exact_match": 9.0,
184
+ "eval_f1": 14.19791549030679,
185
+ "eval_qa_bleu": 2.7960558519683327,
186
+ "eval_qa_exact_match": 0.07,
187
+ "eval_recite_bleu": 2.082318255561628,
188
+ "eval_recite_exact_match": 0.0,
189
  "step": 1125
190
  },
191
  {
192
  "epoch": 6.4,
 
193
  "learning_rate": 3e-05,
194
+ "loss": 1.0029,
195
  "step": 1200
196
  },
197
  {
198
+ "epoch": 6.93,
 
199
  "learning_rate": 3e-05,
200
+ "loss": 0.9888,
201
  "step": 1300
202
  },
203
  {
204
+ "epoch": 7.0,
205
+ "eval_accuracy": 0.6638117647058823,
206
+ "eval_loss": 1.2457345724105835,
207
+ "eval_runtime": 6.7577,
208
+ "eval_samples_per_second": 73.99,
209
+ "eval_steps_per_second": 9.323,
210
  "step": 1312
211
  },
212
  {
213
+ "epoch": 7.0,
214
+ "eval_exact_match": 10.6,
215
+ "eval_f1": 15.30586844512479,
216
+ "eval_qa_bleu": 0.94556412718174,
217
+ "eval_qa_exact_match": 0.086,
218
+ "eval_recite_bleu": 3.1558565534273777,
219
+ "eval_recite_exact_match": 0.0,
220
  "step": 1312
221
  },
222
  {
223
+ "epoch": 7.47,
 
224
  "learning_rate": 3e-05,
225
+ "loss": 0.9453,
226
  "step": 1400
227
  },
228
  {
229
  "epoch": 8.0,
 
230
  "learning_rate": 3e-05,
231
+ "loss": 0.9364,
232
  "step": 1500
233
  },
234
  {
235
  "epoch": 8.0,
236
+ "eval_accuracy": 0.6615921568627451,
237
+ "eval_loss": 1.2816376686096191,
238
+ "eval_runtime": 7.5902,
239
+ "eval_samples_per_second": 65.875,
240
+ "eval_steps_per_second": 8.3,
241
  "step": 1500
242
  },
243
  {
244
  "epoch": 8.0,
245
+ "eval_exact_match": 9.2,
246
+ "eval_f1": 12.989213707726288,
247
+ "eval_qa_bleu": 1.9797940311992752,
248
+ "eval_qa_exact_match": 0.074,
249
+ "eval_recite_bleu": 2.551621610315755,
250
+ "eval_recite_exact_match": 0.0,
251
  "step": 1500
252
  },
253
  {
254
+ "epoch": 8.53,
 
255
  "learning_rate": 3e-05,
256
+ "loss": 0.8889,
257
  "step": 1600
258
  },
259
  {
260
+ "epoch": 9.0,
261
+ "eval_accuracy": 0.6606196078431372,
262
+ "eval_loss": 1.3036144971847534,
263
+ "eval_runtime": 7.4845,
264
+ "eval_samples_per_second": 66.805,
265
+ "eval_steps_per_second": 8.417,
266
  "step": 1687
267
  },
268
  {
269
+ "epoch": 9.0,
270
+ "eval_exact_match": 14.2,
271
+ "eval_f1": 19.66897045942666,
272
+ "eval_qa_bleu": 3.3530752528407004,
273
+ "eval_qa_exact_match": 0.112,
274
+ "eval_recite_bleu": 5.3293362575452425,
275
+ "eval_recite_exact_match": 0.002,
276
  "step": 1687
277
  },
278
  {
279
+ "epoch": 9.07,
 
280
  "learning_rate": 3e-05,
281
+ "loss": 0.8853,
282
  "step": 1700
283
  },
284
  {
285
  "epoch": 9.6,
 
286
  "learning_rate": 3e-05,
287
+ "loss": 0.8373,
288
  "step": 1800
289
  },
290
  {
291
  "epoch": 10.0,
292
+ "eval_accuracy": 0.6587450980392157,
293
+ "eval_loss": 1.3337457180023193,
294
+ "eval_runtime": 6.6361,
295
+ "eval_samples_per_second": 75.345,
296
+ "eval_steps_per_second": 9.493,
297
  "step": 1875
298
  },
299
  {
300
  "epoch": 10.0,
301
+ "eval_exact_match": 14.4,
302
+ "eval_f1": 21.20886352105531,
303
+ "eval_qa_bleu": 2.50339041197013,
304
+ "eval_qa_exact_match": 0.106,
305
+ "eval_recite_bleu": 6.648771889251858,
306
+ "eval_recite_exact_match": 0.0,
307
  "step": 1875
308
  },
309
  {
310
+ "epoch": 10.13,
 
311
  "learning_rate": 3e-05,
312
+ "loss": 0.8353,
313
  "step": 1900
314
  },
315
  {
316
+ "epoch": 10.67,
 
317
  "learning_rate": 3e-05,
318
+ "loss": 0.7949,
319
  "step": 2000
320
  },
321
  {
322
+ "epoch": 11.0,
323
+ "eval_accuracy": 0.6575058823529412,
324
+ "eval_loss": 1.367775797843933,
325
+ "eval_runtime": 7.4976,
326
+ "eval_samples_per_second": 66.688,
327
+ "eval_steps_per_second": 8.403,
328
  "step": 2062
329
  },
330
  {
331
+ "epoch": 11.0,
332
+ "eval_exact_match": 15.4,
333
+ "eval_f1": 21.670913900913916,
334
+ "eval_qa_bleu": 4.131625289761948,
335
+ "eval_qa_exact_match": 0.104,
336
+ "eval_recite_bleu": 8.111027123751208,
337
+ "eval_recite_exact_match": 0.0,
338
  "step": 2062
339
  },
340
  {
341
  "epoch": 11.2,
 
342
  "learning_rate": 3e-05,
343
+ "loss": 0.7758,
344
  "step": 2100
345
  },
346
  {
347
+ "epoch": 11.73,
 
348
  "learning_rate": 3e-05,
349
+ "loss": 0.7539,
350
  "step": 2200
351
  },
352
  {
353
  "epoch": 12.0,
354
+ "eval_accuracy": 0.6553725490196078,
355
+ "eval_loss": 1.3984336853027344,
356
+ "eval_runtime": 7.6892,
357
+ "eval_samples_per_second": 65.026,
358
+ "eval_steps_per_second": 8.193,
359
  "step": 2250
360
  },
361
  {
362
  "epoch": 12.0,
363
+ "eval_exact_match": 16.2,
364
+ "eval_f1": 23.074922854922868,
365
+ "eval_qa_bleu": 10.80394366395148,
366
+ "eval_qa_exact_match": 0.132,
367
+ "eval_recite_bleu": 9.158207459098465,
368
+ "eval_recite_exact_match": 0.002,
369
  "step": 2250
370
  },
371
  {
372
+ "epoch": 12.27,
 
373
  "learning_rate": 3e-05,
374
+ "loss": 0.7415,
375
  "step": 2300
376
  },
377
  {
378
  "epoch": 12.8,
 
379
  "learning_rate": 3e-05,
380
+ "loss": 0.7133,
381
  "step": 2400
382
  },
383
  {
384
+ "epoch": 13.0,
385
+ "eval_accuracy": 0.6538196078431373,
386
+ "eval_loss": 1.4471032619476318,
387
+ "eval_runtime": 6.7177,
388
+ "eval_samples_per_second": 74.431,
389
+ "eval_steps_per_second": 9.378,
390
  "step": 2437
391
  },
392
  {
393
+ "epoch": 13.0,
394
+ "eval_exact_match": 16.4,
395
+ "eval_f1": 23.343965796845062,
396
+ "eval_qa_bleu": 4.473837470175788,
397
+ "eval_qa_exact_match": 0.136,
398
+ "eval_recite_bleu": 10.58390244263022,
399
+ "eval_recite_exact_match": 0.002,
400
  "step": 2437
401
  },
402
  {
403
+ "epoch": 13.33,
 
404
  "learning_rate": 3e-05,
405
+ "loss": 0.682,
406
  "step": 2500
407
  },
408
  {
409
+ "epoch": 13.87,
 
410
  "learning_rate": 3e-05,
411
+ "loss": 0.6704,
412
  "step": 2600
413
  },
414
  {
415
  "epoch": 14.0,
416
+ "eval_accuracy": 0.6524862745098039,
417
+ "eval_loss": 1.4829775094985962,
418
+ "eval_runtime": 6.6339,
419
+ "eval_samples_per_second": 75.37,
420
+ "eval_steps_per_second": 9.497,
421
  "step": 2625
422
  },
423
  {
424
  "epoch": 14.0,
425
+ "eval_exact_match": 15.6,
426
+ "eval_f1": 23.615572621393074,
427
+ "eval_qa_bleu": 7.563435997390864,
428
+ "eval_qa_exact_match": 0.13,
429
+ "eval_recite_bleu": 11.120038652898813,
430
+ "eval_recite_exact_match": 0.008,
431
  "step": 2625
432
  },
433
  {
434
  "epoch": 14.4,
 
435
  "learning_rate": 3e-05,
436
+ "loss": 0.6412,
437
  "step": 2700
438
  },
439
  {
440
+ "epoch": 14.93,
 
441
  "learning_rate": 3e-05,
442
+ "loss": 0.6436,
443
  "step": 2800
444
  },
445
  {
446
+ "epoch": 15.0,
447
+ "eval_accuracy": 0.6507529411764705,
448
+ "eval_loss": 1.524308681488037,
449
+ "eval_runtime": 6.6481,
450
+ "eval_samples_per_second": 75.209,
451
+ "eval_steps_per_second": 9.476,
452
  "step": 2812
453
  },
454
  {
455
+ "epoch": 15.0,
456
+ "eval_exact_match": 17.8,
457
+ "eval_f1": 27.00021164021165,
458
+ "eval_qa_bleu": 7.90349748901623,
459
+ "eval_qa_exact_match": 0.15,
460
+ "eval_recite_bleu": 12.277483381668418,
461
+ "eval_recite_exact_match": 0.008,
462
  "step": 2812
463
  },
464
  {
465
+ "epoch": 15.47,
 
466
  "learning_rate": 3e-05,
467
+ "loss": 0.5969,
468
  "step": 2900
469
  },
470
  {
471
  "epoch": 16.0,
 
472
  "learning_rate": 3e-05,
473
+ "loss": 0.6073,
474
  "step": 3000
475
  },
476
  {
477
  "epoch": 16.0,
478
+ "eval_accuracy": 0.6499764705882353,
479
+ "eval_loss": 1.5640995502471924,
480
+ "eval_runtime": 7.7119,
481
+ "eval_samples_per_second": 64.835,
482
+ "eval_steps_per_second": 8.169,
483
  "step": 3000
484
  },
485
  {
486
  "epoch": 16.0,
487
+ "eval_exact_match": 19.2,
488
+ "eval_f1": 27.747719298245627,
489
+ "eval_qa_bleu": 9.793788404463436,
490
+ "eval_qa_exact_match": 0.14,
491
+ "eval_recite_bleu": 12.835117221544726,
492
+ "eval_recite_exact_match": 0.01,
493
  "step": 3000
494
  },
495
  {
496
+ "epoch": 16.53,
 
497
  "learning_rate": 3e-05,
498
+ "loss": 0.5567,
499
  "step": 3100
500
  },
501
  {
502
+ "epoch": 17.0,
503
+ "eval_accuracy": 0.6486588235294117,
504
+ "eval_loss": 1.6058622598648071,
505
+ "eval_runtime": 7.7727,
506
+ "eval_samples_per_second": 64.328,
507
+ "eval_steps_per_second": 8.105,
508
  "step": 3187
509
  },
510
  {
511
+ "epoch": 17.0,
512
+ "eval_exact_match": 19.6,
513
+ "eval_f1": 26.446776949066113,
514
+ "eval_qa_bleu": 10.294098070146385,
515
+ "eval_qa_exact_match": 0.156,
516
+ "eval_recite_bleu": 12.34815708077889,
517
+ "eval_recite_exact_match": 0.004,
518
  "step": 3187
519
  }
520
  ],
521
  "logging_steps": 100,
522
  "max_steps": 9350,
 
523
  "num_train_epochs": 50,
524
  "save_steps": 500,
525
+ "total_flos": 7.387434823901512e+17,
 
 
 
 
 
 
 
 
 
 
 
 
 
526
  "trial_name": null,
527
  "trial_params": null
528
  }
checkpoint-3187/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9e9c21c8367f69c99b21b20aa3dbf722814d1b222850e23dacbe471b235b2dd4
3
- size 5304
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9347580f7d2a7b3957c29c0f67ee2342d50aa99cfc14c2f59eba6127909f6932
3
+ size 4728