christineyu commited on
Commit
ab5f51d
·
verified ·
1 Parent(s): cba3da7

Model save

Browse files
Files changed (4) hide show
  1. README.md +57 -0
  2. all_results.json +9 -0
  3. train_results.json +9 -0
  4. trainer_state.json +387 -0
README.md ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: mistralai/Mistral-7B-Instruct-v0.2
3
+ library_name: transformers
4
+ model_name: prometheus-7b-v1.5-beta-1-over1218-pref-cat-v0
5
+ tags:
6
+ - generated_from_trainer
7
+ - trl
8
+ - sft
9
+ licence: license
10
+ ---
11
+
12
+ # Model Card for prometheus-7b-v1.5-beta-1-over1218-pref-cat-v0
13
+
14
+ This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2).
15
+ It has been trained using [TRL](https://github.com/huggingface/trl).
16
+
17
+ ## Quick start
18
+
19
+ ```python
20
+ from transformers import pipeline
21
+
22
+ question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
23
+ generator = pipeline("text-generation", model="christineyu/prometheus-7b-v1.5-beta-1-over1218-pref-cat-v0", device="cuda")
24
+ output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
25
+ print(output["generated_text"])
26
+ ```
27
+
28
+ ## Training procedure
29
+
30
+ [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/myexp/huggingface/runs/afnugt5n)
31
+
32
+ This model was trained with SFT.
33
+
34
+ ### Framework versions
35
+
36
+ - TRL: 0.12.2
37
+ - Transformers: 4.46.3
38
+ - Pytorch: 2.3.0
39
+ - Datasets: 3.2.0
40
+ - Tokenizers: 0.20.3
41
+
42
+ ## Citations
43
+
44
+
45
+
46
+ Cite TRL as:
47
+
48
+ ```bibtex
49
+ @misc{vonwerra2022trl,
50
+ title = {{TRL: Transformer Reinforcement Learning}},
51
+ author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
52
+ year = 2020,
53
+ journal = {GitHub repository},
54
+ publisher = {GitHub},
55
+ howpublished = {\url{https://github.com/huggingface/trl}}
56
+ }
57
+ ```
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.0,
3
+ "total_flos": 4.147792507305984e+16,
4
+ "train_loss": 0.04424134408650191,
5
+ "train_runtime": 975.4658,
6
+ "train_samples": 990,
7
+ "train_samples_per_second": 0.236,
8
+ "train_steps_per_second": 0.236
9
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.0,
3
+ "total_flos": 4.147792507305984e+16,
4
+ "train_loss": 0.04424134408650191,
5
+ "train_runtime": 975.4658,
6
+ "train_samples": 990,
7
+ "train_samples_per_second": 0.236,
8
+ "train_steps_per_second": 0.236
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,387 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.0,
5
+ "eval_steps": 500,
6
+ "global_step": 230,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.008695652173913044,
13
+ "grad_norm": 8.01160717010498,
14
+ "learning_rate": 4.347826086956522e-07,
15
+ "loss": 0.5054,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.043478260869565216,
20
+ "grad_norm": 6.416710376739502,
21
+ "learning_rate": 2.173913043478261e-06,
22
+ "loss": 0.5116,
23
+ "step": 5
24
+ },
25
+ {
26
+ "epoch": 0.08695652173913043,
27
+ "grad_norm": 2.897590160369873,
28
+ "learning_rate": 4.347826086956522e-06,
29
+ "loss": 0.4135,
30
+ "step": 10
31
+ },
32
+ {
33
+ "epoch": 0.13043478260869565,
34
+ "grad_norm": 1.8552789688110352,
35
+ "learning_rate": 6.521739130434783e-06,
36
+ "loss": 0.3047,
37
+ "step": 15
38
+ },
39
+ {
40
+ "epoch": 0.17391304347826086,
41
+ "grad_norm": 1.323161244392395,
42
+ "learning_rate": 8.695652173913044e-06,
43
+ "loss": 0.2426,
44
+ "step": 20
45
+ },
46
+ {
47
+ "epoch": 0.21739130434782608,
48
+ "grad_norm": 1.178303599357605,
49
+ "learning_rate": 9.997696831512027e-06,
50
+ "loss": 0.1795,
51
+ "step": 25
52
+ },
53
+ {
54
+ "epoch": 0.2608695652173913,
55
+ "grad_norm": 1.5411367416381836,
56
+ "learning_rate": 9.971810547786794e-06,
57
+ "loss": 0.109,
58
+ "step": 30
59
+ },
60
+ {
61
+ "epoch": 0.30434782608695654,
62
+ "grad_norm": 1.1152806282043457,
63
+ "learning_rate": 9.917308508168712e-06,
64
+ "loss": 0.0577,
65
+ "step": 35
66
+ },
67
+ {
68
+ "epoch": 0.34782608695652173,
69
+ "grad_norm": 0.8210731744766235,
70
+ "learning_rate": 9.834504404631032e-06,
71
+ "loss": 0.0242,
72
+ "step": 40
73
+ },
74
+ {
75
+ "epoch": 0.391304347826087,
76
+ "grad_norm": 0.30621984601020813,
77
+ "learning_rate": 9.72387482452377e-06,
78
+ "loss": 0.0156,
79
+ "step": 45
80
+ },
81
+ {
82
+ "epoch": 0.43478260869565216,
83
+ "grad_norm": 0.30758821964263916,
84
+ "learning_rate": 9.586056507527266e-06,
85
+ "loss": 0.0135,
86
+ "step": 50
87
+ },
88
+ {
89
+ "epoch": 0.4782608695652174,
90
+ "grad_norm": 0.9909317493438721,
91
+ "learning_rate": 9.421842680832862e-06,
92
+ "loss": 0.0135,
93
+ "step": 55
94
+ },
95
+ {
96
+ "epoch": 0.5217391304347826,
97
+ "grad_norm": 0.37661653757095337,
98
+ "learning_rate": 9.232178493644006e-06,
99
+ "loss": 0.0088,
100
+ "step": 60
101
+ },
102
+ {
103
+ "epoch": 0.5652173913043478,
104
+ "grad_norm": 0.8875786066055298,
105
+ "learning_rate": 9.018155577274891e-06,
106
+ "loss": 0.0113,
107
+ "step": 65
108
+ },
109
+ {
110
+ "epoch": 0.6086956521739131,
111
+ "grad_norm": 0.6631242036819458,
112
+ "learning_rate": 8.781005762156593e-06,
113
+ "loss": 0.0082,
114
+ "step": 70
115
+ },
116
+ {
117
+ "epoch": 0.6521739130434783,
118
+ "grad_norm": 0.27046504616737366,
119
+ "learning_rate": 8.522093987913063e-06,
120
+ "loss": 0.0066,
121
+ "step": 75
122
+ },
123
+ {
124
+ "epoch": 0.6956521739130435,
125
+ "grad_norm": 0.4215288758277893,
126
+ "learning_rate": 8.24291044731378e-06,
127
+ "loss": 0.0078,
128
+ "step": 80
129
+ },
130
+ {
131
+ "epoch": 0.7391304347826086,
132
+ "grad_norm": 0.39369380474090576,
133
+ "learning_rate": 7.94506200931932e-06,
134
+ "loss": 0.0061,
135
+ "step": 85
136
+ },
137
+ {
138
+ "epoch": 0.782608695652174,
139
+ "grad_norm": 0.39114001393318176,
140
+ "learning_rate": 7.630262970585355e-06,
141
+ "loss": 0.0058,
142
+ "step": 90
143
+ },
144
+ {
145
+ "epoch": 0.8260869565217391,
146
+ "grad_norm": 0.32843726873397827,
147
+ "learning_rate": 7.300325188655762e-06,
148
+ "loss": 0.0052,
149
+ "step": 95
150
+ },
151
+ {
152
+ "epoch": 0.8695652173913043,
153
+ "grad_norm": 0.17727030813694,
154
+ "learning_rate": 6.957147653634198e-06,
155
+ "loss": 0.0035,
156
+ "step": 100
157
+ },
158
+ {
159
+ "epoch": 0.9130434782608695,
160
+ "grad_norm": 0.4097553789615631,
161
+ "learning_rate": 6.6027055583554865e-06,
162
+ "loss": 0.0071,
163
+ "step": 105
164
+ },
165
+ {
166
+ "epoch": 0.9565217391304348,
167
+ "grad_norm": 0.34858328104019165,
168
+ "learning_rate": 6.2390389299645e-06,
169
+ "loss": 0.0064,
170
+ "step": 110
171
+ },
172
+ {
173
+ "epoch": 1.0,
174
+ "grad_norm": 0.37675490975379944,
175
+ "learning_rate": 5.8682408883346535e-06,
176
+ "loss": 0.0051,
177
+ "step": 115
178
+ },
179
+ {
180
+ "epoch": 1.0,
181
+ "eval_loss": 0.00741227762773633,
182
+ "eval_runtime": 1.2752,
183
+ "eval_samples_per_second": 0.784,
184
+ "eval_steps_per_second": 0.784,
185
+ "step": 115
186
+ },
187
+ {
188
+ "epoch": 1.0434782608695652,
189
+ "grad_norm": 0.171745166182518,
190
+ "learning_rate": 5.492445598905843e-06,
191
+ "loss": 0.0028,
192
+ "step": 120
193
+ },
194
+ {
195
+ "epoch": 1.0869565217391304,
196
+ "grad_norm": 0.19900259375572205,
197
+ "learning_rate": 5.113815989280528e-06,
198
+ "loss": 0.004,
199
+ "step": 125
200
+ },
201
+ {
202
+ "epoch": 1.1304347826086956,
203
+ "grad_norm": 0.3835945427417755,
204
+ "learning_rate": 4.7345313002762545e-06,
205
+ "loss": 0.0034,
206
+ "step": 130
207
+ },
208
+ {
209
+ "epoch": 1.1739130434782608,
210
+ "grad_norm": 0.6228604316711426,
211
+ "learning_rate": 4.356774543085845e-06,
212
+ "loss": 0.0039,
213
+ "step": 135
214
+ },
215
+ {
216
+ "epoch": 1.2173913043478262,
217
+ "grad_norm": 0.14324912428855896,
218
+ "learning_rate": 3.982719934736832e-06,
219
+ "loss": 0.0033,
220
+ "step": 140
221
+ },
222
+ {
223
+ "epoch": 1.2608695652173914,
224
+ "grad_norm": 0.2207939177751541,
225
+ "learning_rate": 3.6145203841665577e-06,
226
+ "loss": 0.0028,
227
+ "step": 145
228
+ },
229
+ {
230
+ "epoch": 1.3043478260869565,
231
+ "grad_norm": 0.19590801000595093,
232
+ "learning_rate": 3.2542951009381584e-06,
233
+ "loss": 0.0024,
234
+ "step": 150
235
+ },
236
+ {
237
+ "epoch": 1.3478260869565217,
238
+ "grad_norm": 0.1485009640455246,
239
+ "learning_rate": 2.9041173979166813e-06,
240
+ "loss": 0.0021,
241
+ "step": 155
242
+ },
243
+ {
244
+ "epoch": 1.391304347826087,
245
+ "grad_norm": 0.3463798463344574,
246
+ "learning_rate": 2.566002758108256e-06,
247
+ "loss": 0.0024,
248
+ "step": 160
249
+ },
250
+ {
251
+ "epoch": 1.434782608695652,
252
+ "grad_norm": 0.2143923044204712,
253
+ "learning_rate": 2.241897234344864e-06,
254
+ "loss": 0.0031,
255
+ "step": 165
256
+ },
257
+ {
258
+ "epoch": 1.4782608695652173,
259
+ "grad_norm": 0.21336747705936432,
260
+ "learning_rate": 1.933666248581418e-06,
261
+ "loss": 0.0024,
262
+ "step": 170
263
+ },
264
+ {
265
+ "epoch": 1.5217391304347827,
266
+ "grad_norm": 0.3059343099594116,
267
+ "learning_rate": 1.6430838552720168e-06,
268
+ "loss": 0.0034,
269
+ "step": 175
270
+ },
271
+ {
272
+ "epoch": 1.5652173913043477,
273
+ "grad_norm": 0.2732033431529999,
274
+ "learning_rate": 1.3718225306210049e-06,
275
+ "loss": 0.0026,
276
+ "step": 180
277
+ },
278
+ {
279
+ "epoch": 1.608695652173913,
280
+ "grad_norm": 0.25113445520401,
281
+ "learning_rate": 1.1214435464779006e-06,
282
+ "loss": 0.002,
283
+ "step": 185
284
+ },
285
+ {
286
+ "epoch": 1.6521739130434783,
287
+ "grad_norm": 0.29484084248542786,
288
+ "learning_rate": 8.933879842801558e-07,
289
+ "loss": 0.0026,
290
+ "step": 190
291
+ },
292
+ {
293
+ "epoch": 1.6956521739130435,
294
+ "grad_norm": 0.21308456361293793,
295
+ "learning_rate": 6.889684407639324e-07,
296
+ "loss": 0.0041,
297
+ "step": 195
298
+ },
299
+ {
300
+ "epoch": 1.7391304347826086,
301
+ "grad_norm": 0.1736830472946167,
302
+ "learning_rate": 5.0936147318152e-07,
303
+ "loss": 0.0035,
304
+ "step": 200
305
+ },
306
+ {
307
+ "epoch": 1.7826086956521738,
308
+ "grad_norm": 0.3409785032272339,
309
+ "learning_rate": 3.55600827507665e-07,
310
+ "loss": 0.0035,
311
+ "step": 205
312
+ },
313
+ {
314
+ "epoch": 1.8260869565217392,
315
+ "grad_norm": 0.286048948764801,
316
+ "learning_rate": 2.2857148861060552e-07,
317
+ "loss": 0.0029,
318
+ "step": 210
319
+ },
320
+ {
321
+ "epoch": 1.8695652173913042,
322
+ "grad_norm": 0.11789187043905258,
323
+ "learning_rate": 1.2900458663260506e-07,
324
+ "loss": 0.0023,
325
+ "step": 215
326
+ },
327
+ {
328
+ "epoch": 1.9130434782608696,
329
+ "grad_norm": 0.1927684098482132,
330
+ "learning_rate": 5.747318889684883e-08,
331
+ "loss": 0.0033,
332
+ "step": 220
333
+ },
334
+ {
335
+ "epoch": 1.9565217391304348,
336
+ "grad_norm": 0.14718472957611084,
337
+ "learning_rate": 1.4389001560803917e-08,
338
+ "loss": 0.0028,
339
+ "step": 225
340
+ },
341
+ {
342
+ "epoch": 2.0,
343
+ "grad_norm": 0.17989766597747803,
344
+ "learning_rate": 0.0,
345
+ "loss": 0.0033,
346
+ "step": 230
347
+ },
348
+ {
349
+ "epoch": 2.0,
350
+ "eval_loss": 0.00580478785559535,
351
+ "eval_runtime": 1.2826,
352
+ "eval_samples_per_second": 0.78,
353
+ "eval_steps_per_second": 0.78,
354
+ "step": 230
355
+ },
356
+ {
357
+ "epoch": 2.0,
358
+ "step": 230,
359
+ "total_flos": 4.147792507305984e+16,
360
+ "train_loss": 0.04424134408650191,
361
+ "train_runtime": 975.4658,
362
+ "train_samples_per_second": 0.236,
363
+ "train_steps_per_second": 0.236
364
+ }
365
+ ],
366
+ "logging_steps": 5,
367
+ "max_steps": 230,
368
+ "num_input_tokens_seen": 0,
369
+ "num_train_epochs": 2,
370
+ "save_steps": 100,
371
+ "stateful_callbacks": {
372
+ "TrainerControl": {
373
+ "args": {
374
+ "should_epoch_stop": false,
375
+ "should_evaluate": false,
376
+ "should_log": false,
377
+ "should_save": true,
378
+ "should_training_stop": true
379
+ },
380
+ "attributes": {}
381
+ }
382
+ },
383
+ "total_flos": 4.147792507305984e+16,
384
+ "train_batch_size": 1,
385
+ "trial_name": null,
386
+ "trial_params": null
387
+ }