Hugging Face's logo

GaetanMichelet
/
Mistral-7B_task-3_120-samples_config-2

PEFT
TensorBoard
Safetensors
mistral
alignment-handbook
trl
sft
Generated from Trainer
4-bit precision
bitsandbytes
Model card Files Metrics Community
GaetanMichelet commited on
Commit
9adc805
•
1 Parent(s): 1dd97b9

Model save

Browse files
Files changed (5) hide show
  1. README.md +76 -0
  2. adapter_model.safetensors +1 -1
  3. all_results.json +9 -0
  4. train_results.json +9 -0
  5. trainer_state.json +385 -0
README.md ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: mistralai/Mistral-7B-Instruct-v0.3
3
+ library_name: peft
4
+ license: apache-2.0
5
+ tags:
6
+ - trl
7
+ - sft
8
+ - generated_from_trainer
9
+ model-index:
10
+ - name: Mistral-7B_task-3_120-samples_config-2
11
+ results: []
12
+ ---
13
+
14
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
15
+ should probably proofread and complete it, then remove this comment. -->
16
+
17
+ # Mistral-7B_task-3_120-samples_config-2
18
+
19
+ This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.3](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3) on the None dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 0.7857
22
+
23
+ ## Model description
24
+
25
+ More information needed
26
+
27
+ ## Intended uses & limitations
28
+
29
+ More information needed
30
+
31
+ ## Training and evaluation data
32
+
33
+ More information needed
34
+
35
+ ## Training procedure
36
+
37
+ ### Training hyperparameters
38
+
39
+ The following hyperparameters were used during training:
40
+ - learning_rate: 0.0001
41
+ - train_batch_size: 1
42
+ - eval_batch_size: 1
43
+ - seed: 42
44
+ - distributed_type: multi-GPU
45
+ - gradient_accumulation_steps: 16
46
+ - total_train_batch_size: 16
47
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
48
+ - lr_scheduler_type: cosine
49
+ - lr_scheduler_warmup_ratio: 0.1
50
+ - num_epochs: 50
51
+
52
+ ### Training results
53
+
54
+ | Training Loss | Epoch | Step | Validation Loss |
55
+ |:-------------:|:-------:|:----:|:---------------:|
56
+ | 3.4696 | 0.9091 | 5 | 2.7714 |
57
+ | 1.5154 | 2.0 | 11 | 0.9233 |
58
+ | 0.4441 | 2.9091 | 16 | 0.5486 |
59
+ | 0.4453 | 4.0 | 22 | 0.4419 |
60
+ | 0.2943 | 4.9091 | 27 | 0.3947 |
61
+ | 0.209 | 6.0 | 33 | 0.3950 |
62
+ | 0.167 | 6.9091 | 38 | 0.4694 |
63
+ | 0.0931 | 8.0 | 44 | 0.5510 |
64
+ | 0.0271 | 8.9091 | 49 | 0.6278 |
65
+ | 0.0121 | 10.0 | 55 | 0.7051 |
66
+ | 0.0101 | 10.9091 | 60 | 0.7721 |
67
+ | 0.0546 | 12.0 | 66 | 0.7857 |
68
+
69
+
70
+ ### Framework versions
71
+
72
+ - PEFT 0.12.0
73
+ - Transformers 4.44.0
74
+ - Pytorch 2.1.2+cu121
75
+ - Datasets 2.20.0
76
+ - Tokenizers 0.19.1
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ad3645a2270d1d35c52040c81e6d599a825d8008ce9be6569947c7a84d50847f
3
  size 167832240
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1f9c6df514d8e9fc7dcfc127cfeec90ce0ffcafdc408d257d96b12198020499
3
  size 167832240
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.0,
3
+ "total_flos": 1.0256775657396634e+17,
4
+ "train_loss": 0.603098966773938,
5
+ "train_runtime": 3818.1022,
6
+ "train_samples": 88,
7
+ "train_samples_per_second": 1.152,
8
+ "train_steps_per_second": 0.065
9
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.0,
3
+ "total_flos": 1.0256775657396634e+17,
4
+ "train_loss": 0.603098966773938,
5
+ "train_runtime": 3818.1022,
6
+ "train_samples": 88,
7
+ "train_samples_per_second": 1.152,
8
+ "train_steps_per_second": 0.065
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,385 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.39467453956604004,
3
+ "best_model_checkpoint": "data/Mistral-7B_task-3_120-samples_config-2/checkpoint-27",
4
+ "epoch": 12.0,
5
+ "eval_steps": 500,
6
+ "global_step": 66,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.18181818181818182,
13
+ "grad_norm": 21.41504669189453,
14
+ "learning_rate": 4.000000000000001e-06,
15
+ "loss": 4.4397,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.36363636363636365,
20
+ "grad_norm": 17.722919464111328,
21
+ "learning_rate": 8.000000000000001e-06,
22
+ "loss": 3.8584,
23
+ "step": 2
24
+ },
25
+ {
26
+ "epoch": 0.7272727272727273,
27
+ "grad_norm": 15.595569610595703,
28
+ "learning_rate": 1.6000000000000003e-05,
29
+ "loss": 3.4696,
30
+ "step": 4
31
+ },
32
+ {
33
+ "epoch": 0.9090909090909091,
34
+ "eval_loss": 2.7714335918426514,
35
+ "eval_runtime": 25.7676,
36
+ "eval_samples_per_second": 0.931,
37
+ "eval_steps_per_second": 0.931,
38
+ "step": 5
39
+ },
40
+ {
41
+ "epoch": 1.0909090909090908,
42
+ "grad_norm": 12.627395629882812,
43
+ "learning_rate": 2.4e-05,
44
+ "loss": 2.8196,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 1.4545454545454546,
49
+ "grad_norm": 9.122146606445312,
50
+ "learning_rate": 3.2000000000000005e-05,
51
+ "loss": 2.1912,
52
+ "step": 8
53
+ },
54
+ {
55
+ "epoch": 1.8181818181818183,
56
+ "grad_norm": 5.698470115661621,
57
+ "learning_rate": 4e-05,
58
+ "loss": 1.5154,
59
+ "step": 10
60
+ },
61
+ {
62
+ "epoch": 2.0,
63
+ "eval_loss": 0.9233453869819641,
64
+ "eval_runtime": 25.7652,
65
+ "eval_samples_per_second": 0.931,
66
+ "eval_steps_per_second": 0.931,
67
+ "step": 11
68
+ },
69
+ {
70
+ "epoch": 2.1818181818181817,
71
+ "grad_norm": 3.769749879837036,
72
+ "learning_rate": 4.8e-05,
73
+ "loss": 0.911,
74
+ "step": 12
75
+ },
76
+ {
77
+ "epoch": 2.5454545454545454,
78
+ "grad_norm": 1.8143624067306519,
79
+ "learning_rate": 5.6000000000000006e-05,
80
+ "loss": 0.7367,
81
+ "step": 14
82
+ },
83
+ {
84
+ "epoch": 2.909090909090909,
85
+ "grad_norm": 1.2752869129180908,
86
+ "learning_rate": 6.400000000000001e-05,
87
+ "loss": 0.4441,
88
+ "step": 16
89
+ },
90
+ {
91
+ "epoch": 2.909090909090909,
92
+ "eval_loss": 0.5486097931861877,
93
+ "eval_runtime": 25.7657,
94
+ "eval_samples_per_second": 0.931,
95
+ "eval_steps_per_second": 0.931,
96
+ "step": 16
97
+ },
98
+ {
99
+ "epoch": 3.2727272727272725,
100
+ "grad_norm": 0.5312128663063049,
101
+ "learning_rate": 7.2e-05,
102
+ "loss": 0.3839,
103
+ "step": 18
104
+ },
105
+ {
106
+ "epoch": 3.6363636363636362,
107
+ "grad_norm": 1.1591869592666626,
108
+ "learning_rate": 8e-05,
109
+ "loss": 0.4282,
110
+ "step": 20
111
+ },
112
+ {
113
+ "epoch": 4.0,
114
+ "grad_norm": 0.9519180655479431,
115
+ "learning_rate": 8.800000000000001e-05,
116
+ "loss": 0.4453,
117
+ "step": 22
118
+ },
119
+ {
120
+ "epoch": 4.0,
121
+ "eval_loss": 0.4419020116329193,
122
+ "eval_runtime": 25.7589,
123
+ "eval_samples_per_second": 0.932,
124
+ "eval_steps_per_second": 0.932,
125
+ "step": 22
126
+ },
127
+ {
128
+ "epoch": 4.363636363636363,
129
+ "grad_norm": 0.7251240015029907,
130
+ "learning_rate": 9.6e-05,
131
+ "loss": 0.3325,
132
+ "step": 24
133
+ },
134
+ {
135
+ "epoch": 4.7272727272727275,
136
+ "grad_norm": 0.5680262446403503,
137
+ "learning_rate": 9.999512620046522e-05,
138
+ "loss": 0.2943,
139
+ "step": 26
140
+ },
141
+ {
142
+ "epoch": 4.909090909090909,
143
+ "eval_loss": 0.39467453956604004,
144
+ "eval_runtime": 25.7651,
145
+ "eval_samples_per_second": 0.931,
146
+ "eval_steps_per_second": 0.931,
147
+ "step": 27
148
+ },
149
+ {
150
+ "epoch": 5.090909090909091,
151
+ "grad_norm": 0.895659327507019,
152
+ "learning_rate": 9.995614150494293e-05,
153
+ "loss": 0.2955,
154
+ "step": 28
155
+ },
156
+ {
157
+ "epoch": 5.454545454545454,
158
+ "grad_norm": 1.0607842206954956,
159
+ "learning_rate": 9.987820251299122e-05,
160
+ "loss": 0.2129,
161
+ "step": 30
162
+ },
163
+ {
164
+ "epoch": 5.818181818181818,
165
+ "grad_norm": 0.5883296132087708,
166
+ "learning_rate": 9.976136999909156e-05,
167
+ "loss": 0.209,
168
+ "step": 32
169
+ },
170
+ {
171
+ "epoch": 6.0,
172
+ "eval_loss": 0.3949660360813141,
173
+ "eval_runtime": 25.76,
174
+ "eval_samples_per_second": 0.932,
175
+ "eval_steps_per_second": 0.932,
176
+ "step": 33
177
+ },
178
+ {
179
+ "epoch": 6.181818181818182,
180
+ "grad_norm": 0.47473806142807007,
181
+ "learning_rate": 9.96057350657239e-05,
182
+ "loss": 0.2097,
183
+ "step": 34
184
+ },
185
+ {
186
+ "epoch": 6.545454545454545,
187
+ "grad_norm": 0.6922392845153809,
188
+ "learning_rate": 9.941141907232765e-05,
189
+ "loss": 0.1934,
190
+ "step": 36
191
+ },
192
+ {
193
+ "epoch": 6.909090909090909,
194
+ "grad_norm": 0.7173542976379395,
195
+ "learning_rate": 9.917857354066931e-05,
196
+ "loss": 0.167,
197
+ "step": 38
198
+ },
199
+ {
200
+ "epoch": 6.909090909090909,
201
+ "eval_loss": 0.4693656265735626,
202
+ "eval_runtime": 25.7635,
203
+ "eval_samples_per_second": 0.932,
204
+ "eval_steps_per_second": 0.932,
205
+ "step": 38
206
+ },
207
+ {
208
+ "epoch": 7.2727272727272725,
209
+ "grad_norm": 0.7706111073493958,
210
+ "learning_rate": 9.890738003669029e-05,
211
+ "loss": 0.1081,
212
+ "step": 40
213
+ },
214
+ {
215
+ "epoch": 7.636363636363637,
216
+ "grad_norm": 0.8268452286720276,
217
+ "learning_rate": 9.859805002892732e-05,
218
+ "loss": 0.0544,
219
+ "step": 42
220
+ },
221
+ {
222
+ "epoch": 8.0,
223
+ "grad_norm": 0.9987695217132568,
224
+ "learning_rate": 9.825082472361557e-05,
225
+ "loss": 0.0931,
226
+ "step": 44
227
+ },
228
+ {
229
+ "epoch": 8.0,
230
+ "eval_loss": 0.5509907007217407,
231
+ "eval_runtime": 25.7644,
232
+ "eval_samples_per_second": 0.932,
233
+ "eval_steps_per_second": 0.932,
234
+ "step": 44
235
+ },
236
+ {
237
+ "epoch": 8.363636363636363,
238
+ "grad_norm": 0.49683877825737,
239
+ "learning_rate": 9.786597487660337e-05,
240
+ "loss": 0.0463,
241
+ "step": 46
242
+ },
243
+ {
244
+ "epoch": 8.727272727272727,
245
+ "grad_norm": 0.6137292385101318,
246
+ "learning_rate": 9.744380058222483e-05,
247
+ "loss": 0.0271,
248
+ "step": 48
249
+ },
250
+ {
251
+ "epoch": 8.909090909090908,
252
+ "eval_loss": 0.627763569355011,
253
+ "eval_runtime": 25.7665,
254
+ "eval_samples_per_second": 0.931,
255
+ "eval_steps_per_second": 0.931,
256
+ "step": 49
257
+ },
258
+ {
259
+ "epoch": 9.090909090909092,
260
+ "grad_norm": 0.9003904461860657,
261
+ "learning_rate": 9.698463103929542e-05,
262
+ "loss": 0.0372,
263
+ "step": 50
264
+ },
265
+ {
266
+ "epoch": 9.454545454545455,
267
+ "grad_norm": 1.11908757686615,
268
+ "learning_rate": 9.648882429441257e-05,
269
+ "loss": 0.0182,
270
+ "step": 52
271
+ },
272
+ {
273
+ "epoch": 9.818181818181818,
274
+ "grad_norm": 0.6121003031730652,
275
+ "learning_rate": 9.595676696276172e-05,
276
+ "loss": 0.0121,
277
+ "step": 54
278
+ },
279
+ {
280
+ "epoch": 10.0,
281
+ "eval_loss": 0.7050547003746033,
282
+ "eval_runtime": 25.7565,
283
+ "eval_samples_per_second": 0.932,
284
+ "eval_steps_per_second": 0.932,
285
+ "step": 55
286
+ },
287
+ {
288
+ "epoch": 10.181818181818182,
289
+ "grad_norm": 0.17396920919418335,
290
+ "learning_rate": 9.538887392664544e-05,
291
+ "loss": 0.0075,
292
+ "step": 56
293
+ },
294
+ {
295
+ "epoch": 10.545454545454545,
296
+ "grad_norm": 0.3935445249080658,
297
+ "learning_rate": 9.478558801197065e-05,
298
+ "loss": 0.0059,
299
+ "step": 58
300
+ },
301
+ {
302
+ "epoch": 10.909090909090908,
303
+ "grad_norm": 0.3881361186504364,
304
+ "learning_rate": 9.414737964294636e-05,
305
+ "loss": 0.0101,
306
+ "step": 60
307
+ },
308
+ {
309
+ "epoch": 10.909090909090908,
310
+ "eval_loss": 0.7721331715583801,
311
+ "eval_runtime": 25.7619,
312
+ "eval_samples_per_second": 0.932,
313
+ "eval_steps_per_second": 0.932,
314
+ "step": 60
315
+ },
316
+ {
317
+ "epoch": 11.272727272727273,
318
+ "grad_norm": 0.46985000371932983,
319
+ "learning_rate": 9.347474647526095e-05,
320
+ "loss": 0.0089,
321
+ "step": 62
322
+ },
323
+ {
324
+ "epoch": 11.636363636363637,
325
+ "grad_norm": 0.5655374526977539,
326
+ "learning_rate": 9.276821300802534e-05,
327
+ "loss": 0.0102,
328
+ "step": 64
329
+ },
330
+ {
331
+ "epoch": 12.0,
332
+ "grad_norm": 38.224891662597656,
333
+ "learning_rate": 9.202833017478422e-05,
334
+ "loss": 0.0546,
335
+ "step": 66
336
+ },
337
+ {
338
+ "epoch": 12.0,
339
+ "eval_loss": 0.7857134342193604,
340
+ "eval_runtime": 25.7658,
341
+ "eval_samples_per_second": 0.931,
342
+ "eval_steps_per_second": 0.931,
343
+ "step": 66
344
+ },
345
+ {
346
+ "epoch": 12.0,
347
+ "step": 66,
348
+ "total_flos": 1.0256775657396634e+17,
349
+ "train_loss": 0.603098966773938,
350
+ "train_runtime": 3818.1022,
351
+ "train_samples_per_second": 1.152,
352
+ "train_steps_per_second": 0.065
353
+ }
354
+ ],
355
+ "logging_steps": 2,
356
+ "max_steps": 250,
357
+ "num_input_tokens_seen": 0,
358
+ "num_train_epochs": 50,
359
+ "save_steps": 25,
360
+ "stateful_callbacks": {
361
+ "EarlyStoppingCallback": {
362
+ "args": {
363
+ "early_stopping_patience": 7,
364
+ "early_stopping_threshold": 0.0
365
+ },
366
+ "attributes": {
367
+ "early_stopping_patience_counter": 0
368
+ }
369
+ },
370
+ "TrainerControl": {
371
+ "args": {
372
+ "should_epoch_stop": false,
373
+ "should_evaluate": false,
374
+ "should_log": false,
375
+ "should_save": true,
376
+ "should_training_stop": true
377
+ },
378
+ "attributes": {}
379
+ }
380
+ },
381
+ "total_flos": 1.0256775657396634e+17,
382
+ "train_batch_size": 1,
383
+ "trial_name": null,
384
+ "trial_params": null
385
+ }