Hugging Face's logo

GaetanMichelet
/
Mistral-7B_task-2_60-samples_config-1

PEFT
TensorBoard
Safetensors
mistral
alignment-handbook
trl
sft
Generated from Trainer
4-bit precision
bitsandbytes
Model card Files Metrics Community
GaetanMichelet commited on
Commit
e9be3c8
•
1 Parent(s): 976847a

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 +392 -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-2_60-samples_config-1
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-2_60-samples_config-1
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 an unknown dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 0.9425
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: 8
46
+ - total_train_batch_size: 8
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
+ | 0.9616 | 0.8696 | 5 | 0.8768 |
57
+ | 0.7463 | 1.9130 | 11 | 0.6409 |
58
+ | 0.6311 | 2.9565 | 17 | 0.5873 |
59
+ | 0.5306 | 4.0 | 23 | 0.5467 |
60
+ | 0.4454 | 4.8696 | 28 | 0.5375 |
61
+ | 0.3475 | 5.9130 | 34 | 0.5507 |
62
+ | 0.2291 | 6.9565 | 40 | 0.6329 |
63
+ | 0.1245 | 8.0 | 46 | 0.7403 |
64
+ | 0.0548 | 8.8696 | 51 | 0.8630 |
65
+ | 0.0302 | 9.9130 | 57 | 0.8626 |
66
+ | 0.014 | 10.9565 | 63 | 0.9664 |
67
+ | 0.0152 | 12.0 | 69 | 0.9425 |
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 sha256:81df90f144719872ddd04e40e0cef1e6af8200ffc51a9c2007b9304b268927cf
3
  size 167832240
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd1c995f407fe21ef60d5deaab0fe6f5bdcca389a1d39c4d0048d233b37daf79
3
  size 167832240
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.0,
3
+ "total_flos": 6.60947324633088e+16,
4
+ "train_loss": 0.3448392077740552,
5
+ "train_runtime": 2412.1178,
6
+ "train_samples": 46,
7
+ "train_samples_per_second": 0.954,
8
+ "train_steps_per_second": 0.104
9
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 12.0,
3
+ "total_flos": 6.60947324633088e+16,
4
+ "train_loss": 0.3448392077740552,
5
+ "train_runtime": 2412.1178,
6
+ "train_samples": 46,
7
+ "train_samples_per_second": 0.954,
8
+ "train_steps_per_second": 0.104
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,392 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.5375192165374756,
3
+ "best_model_checkpoint": "data/Mistral-7B_task-2_60-samples_config-1/checkpoint-28",
4
+ "epoch": 12.0,
5
+ "eval_steps": 500,
6
+ "global_step": 69,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.17391304347826086,
13
+ "grad_norm": 4.75992488861084,
14
+ "learning_rate": 4.000000000000001e-06,
15
+ "loss": 0.986,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.34782608695652173,
20
+ "grad_norm": 2.840470552444458,
21
+ "learning_rate": 8.000000000000001e-06,
22
+ "loss": 1.0599,
23
+ "step": 2
24
+ },
25
+ {
26
+ "epoch": 0.6956521739130435,
27
+ "grad_norm": 2.509486436843872,
28
+ "learning_rate": 1.6000000000000003e-05,
29
+ "loss": 0.9616,
30
+ "step": 4
31
+ },
32
+ {
33
+ "epoch": 0.8695652173913043,
34
+ "eval_loss": 0.8767836689949036,
35
+ "eval_runtime": 16.1173,
36
+ "eval_samples_per_second": 0.745,
37
+ "eval_steps_per_second": 0.745,
38
+ "step": 5
39
+ },
40
+ {
41
+ "epoch": 1.0434782608695652,
42
+ "grad_norm": 2.077070474624634,
43
+ "learning_rate": 2.4e-05,
44
+ "loss": 0.8995,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 1.391304347826087,
49
+ "grad_norm": 1.53449547290802,
50
+ "learning_rate": 3.2000000000000005e-05,
51
+ "loss": 0.8161,
52
+ "step": 8
53
+ },
54
+ {
55
+ "epoch": 1.7391304347826086,
56
+ "grad_norm": 1.0011566877365112,
57
+ "learning_rate": 4e-05,
58
+ "loss": 0.7463,
59
+ "step": 10
60
+ },
61
+ {
62
+ "epoch": 1.9130434782608696,
63
+ "eval_loss": 0.6408722400665283,
64
+ "eval_runtime": 16.1143,
65
+ "eval_samples_per_second": 0.745,
66
+ "eval_steps_per_second": 0.745,
67
+ "step": 11
68
+ },
69
+ {
70
+ "epoch": 2.0869565217391304,
71
+ "grad_norm": 0.7578339576721191,
72
+ "learning_rate": 4.8e-05,
73
+ "loss": 0.6299,
74
+ "step": 12
75
+ },
76
+ {
77
+ "epoch": 2.4347826086956523,
78
+ "grad_norm": 0.6122817397117615,
79
+ "learning_rate": 5.6000000000000006e-05,
80
+ "loss": 0.5522,
81
+ "step": 14
82
+ },
83
+ {
84
+ "epoch": 2.782608695652174,
85
+ "grad_norm": 0.5560686588287354,
86
+ "learning_rate": 6.400000000000001e-05,
87
+ "loss": 0.6311,
88
+ "step": 16
89
+ },
90
+ {
91
+ "epoch": 2.9565217391304346,
92
+ "eval_loss": 0.5873411297798157,
93
+ "eval_runtime": 16.1188,
94
+ "eval_samples_per_second": 0.744,
95
+ "eval_steps_per_second": 0.744,
96
+ "step": 17
97
+ },
98
+ {
99
+ "epoch": 3.130434782608696,
100
+ "grad_norm": 0.5497525334358215,
101
+ "learning_rate": 7.2e-05,
102
+ "loss": 0.5468,
103
+ "step": 18
104
+ },
105
+ {
106
+ "epoch": 3.4782608695652173,
107
+ "grad_norm": 0.6230431199073792,
108
+ "learning_rate": 8e-05,
109
+ "loss": 0.5046,
110
+ "step": 20
111
+ },
112
+ {
113
+ "epoch": 3.8260869565217392,
114
+ "grad_norm": 0.45565032958984375,
115
+ "learning_rate": 8.800000000000001e-05,
116
+ "loss": 0.5306,
117
+ "step": 22
118
+ },
119
+ {
120
+ "epoch": 4.0,
121
+ "eval_loss": 0.5466979146003723,
122
+ "eval_runtime": 16.1126,
123
+ "eval_samples_per_second": 0.745,
124
+ "eval_steps_per_second": 0.745,
125
+ "step": 23
126
+ },
127
+ {
128
+ "epoch": 4.173913043478261,
129
+ "grad_norm": 0.42829808592796326,
130
+ "learning_rate": 9.6e-05,
131
+ "loss": 0.4993,
132
+ "step": 24
133
+ },
134
+ {
135
+ "epoch": 4.521739130434782,
136
+ "grad_norm": 0.4515722990036011,
137
+ "learning_rate": 9.999512620046522e-05,
138
+ "loss": 0.4521,
139
+ "step": 26
140
+ },
141
+ {
142
+ "epoch": 4.869565217391305,
143
+ "grad_norm": 0.5315717458724976,
144
+ "learning_rate": 9.995614150494293e-05,
145
+ "loss": 0.4454,
146
+ "step": 28
147
+ },
148
+ {
149
+ "epoch": 4.869565217391305,
150
+ "eval_loss": 0.5375192165374756,
151
+ "eval_runtime": 16.1187,
152
+ "eval_samples_per_second": 0.744,
153
+ "eval_steps_per_second": 0.744,
154
+ "step": 28
155
+ },
156
+ {
157
+ "epoch": 5.217391304347826,
158
+ "grad_norm": 0.46274569630622864,
159
+ "learning_rate": 9.987820251299122e-05,
160
+ "loss": 0.3712,
161
+ "step": 30
162
+ },
163
+ {
164
+ "epoch": 5.565217391304348,
165
+ "grad_norm": 0.5427079796791077,
166
+ "learning_rate": 9.976136999909156e-05,
167
+ "loss": 0.369,
168
+ "step": 32
169
+ },
170
+ {
171
+ "epoch": 5.913043478260869,
172
+ "grad_norm": 0.5853497385978699,
173
+ "learning_rate": 9.96057350657239e-05,
174
+ "loss": 0.3475,
175
+ "step": 34
176
+ },
177
+ {
178
+ "epoch": 5.913043478260869,
179
+ "eval_loss": 0.5507214665412903,
180
+ "eval_runtime": 16.1147,
181
+ "eval_samples_per_second": 0.745,
182
+ "eval_steps_per_second": 0.745,
183
+ "step": 34
184
+ },
185
+ {
186
+ "epoch": 6.260869565217392,
187
+ "grad_norm": 0.5865023136138916,
188
+ "learning_rate": 9.941141907232765e-05,
189
+ "loss": 0.2732,
190
+ "step": 36
191
+ },
192
+ {
193
+ "epoch": 6.608695652173913,
194
+ "grad_norm": 0.7380216121673584,
195
+ "learning_rate": 9.917857354066931e-05,
196
+ "loss": 0.2582,
197
+ "step": 38
198
+ },
199
+ {
200
+ "epoch": 6.956521739130435,
201
+ "grad_norm": 1.2702546119689941,
202
+ "learning_rate": 9.890738003669029e-05,
203
+ "loss": 0.2291,
204
+ "step": 40
205
+ },
206
+ {
207
+ "epoch": 6.956521739130435,
208
+ "eval_loss": 0.6328728199005127,
209
+ "eval_runtime": 16.1187,
210
+ "eval_samples_per_second": 0.744,
211
+ "eval_steps_per_second": 0.744,
212
+ "step": 40
213
+ },
214
+ {
215
+ "epoch": 7.304347826086957,
216
+ "grad_norm": 0.7579282522201538,
217
+ "learning_rate": 9.859805002892732e-05,
218
+ "loss": 0.183,
219
+ "step": 42
220
+ },
221
+ {
222
+ "epoch": 7.6521739130434785,
223
+ "grad_norm": 0.8423174023628235,
224
+ "learning_rate": 9.825082472361557e-05,
225
+ "loss": 0.1438,
226
+ "step": 44
227
+ },
228
+ {
229
+ "epoch": 8.0,
230
+ "grad_norm": 1.2522237300872803,
231
+ "learning_rate": 9.786597487660337e-05,
232
+ "loss": 0.1245,
233
+ "step": 46
234
+ },
235
+ {
236
+ "epoch": 8.0,
237
+ "eval_loss": 0.7402812838554382,
238
+ "eval_runtime": 16.1135,
239
+ "eval_samples_per_second": 0.745,
240
+ "eval_steps_per_second": 0.745,
241
+ "step": 46
242
+ },
243
+ {
244
+ "epoch": 8.347826086956522,
245
+ "grad_norm": 0.8192511200904846,
246
+ "learning_rate": 9.744380058222483e-05,
247
+ "loss": 0.077,
248
+ "step": 48
249
+ },
250
+ {
251
+ "epoch": 8.695652173913043,
252
+ "grad_norm": 1.1571916341781616,
253
+ "learning_rate": 9.698463103929542e-05,
254
+ "loss": 0.0548,
255
+ "step": 50
256
+ },
257
+ {
258
+ "epoch": 8.869565217391305,
259
+ "eval_loss": 0.8629714846611023,
260
+ "eval_runtime": 16.1114,
261
+ "eval_samples_per_second": 0.745,
262
+ "eval_steps_per_second": 0.745,
263
+ "step": 51
264
+ },
265
+ {
266
+ "epoch": 9.043478260869565,
267
+ "grad_norm": 1.533800721168518,
268
+ "learning_rate": 9.648882429441257e-05,
269
+ "loss": 0.0623,
270
+ "step": 52
271
+ },
272
+ {
273
+ "epoch": 9.391304347826088,
274
+ "grad_norm": 1.2077471017837524,
275
+ "learning_rate": 9.595676696276172e-05,
276
+ "loss": 0.0261,
277
+ "step": 54
278
+ },
279
+ {
280
+ "epoch": 9.73913043478261,
281
+ "grad_norm": 0.8268634080886841,
282
+ "learning_rate": 9.538887392664544e-05,
283
+ "loss": 0.0302,
284
+ "step": 56
285
+ },
286
+ {
287
+ "epoch": 9.91304347826087,
288
+ "eval_loss": 0.8625869750976562,
289
+ "eval_runtime": 16.1131,
290
+ "eval_samples_per_second": 0.745,
291
+ "eval_steps_per_second": 0.745,
292
+ "step": 57
293
+ },
294
+ {
295
+ "epoch": 10.08695652173913,
296
+ "grad_norm": 0.9339629411697388,
297
+ "learning_rate": 9.478558801197065e-05,
298
+ "loss": 0.0289,
299
+ "step": 58
300
+ },
301
+ {
302
+ "epoch": 10.434782608695652,
303
+ "grad_norm": 0.7875646352767944,
304
+ "learning_rate": 9.414737964294636e-05,
305
+ "loss": 0.0126,
306
+ "step": 60
307
+ },
308
+ {
309
+ "epoch": 10.782608695652174,
310
+ "grad_norm": 0.8727169632911682,
311
+ "learning_rate": 9.347474647526095e-05,
312
+ "loss": 0.014,
313
+ "step": 62
314
+ },
315
+ {
316
+ "epoch": 10.956521739130435,
317
+ "eval_loss": 0.9664008021354675,
318
+ "eval_runtime": 16.1149,
319
+ "eval_samples_per_second": 0.745,
320
+ "eval_steps_per_second": 0.745,
321
+ "step": 63
322
+ },
323
+ {
324
+ "epoch": 11.130434782608695,
325
+ "grad_norm": 0.6830615997314453,
326
+ "learning_rate": 9.276821300802534e-05,
327
+ "loss": 0.0187,
328
+ "step": 64
329
+ },
330
+ {
331
+ "epoch": 11.478260869565217,
332
+ "grad_norm": 0.628609836101532,
333
+ "learning_rate": 9.202833017478422e-05,
334
+ "loss": 0.0121,
335
+ "step": 66
336
+ },
337
+ {
338
+ "epoch": 11.826086956521738,
339
+ "grad_norm": 0.7100380659103394,
340
+ "learning_rate": 9.125567491391476e-05,
341
+ "loss": 0.0152,
342
+ "step": 68
343
+ },
344
+ {
345
+ "epoch": 12.0,
346
+ "eval_loss": 0.9425068497657776,
347
+ "eval_runtime": 16.1161,
348
+ "eval_samples_per_second": 0.745,
349
+ "eval_steps_per_second": 0.745,
350
+ "step": 69
351
+ },
352
+ {
353
+ "epoch": 12.0,
354
+ "step": 69,
355
+ "total_flos": 6.60947324633088e+16,
356
+ "train_loss": 0.3448392077740552,
357
+ "train_runtime": 2412.1178,
358
+ "train_samples_per_second": 0.954,
359
+ "train_steps_per_second": 0.104
360
+ }
361
+ ],
362
+ "logging_steps": 2,
363
+ "max_steps": 250,
364
+ "num_input_tokens_seen": 0,
365
+ "num_train_epochs": 50,
366
+ "save_steps": 25,
367
+ "stateful_callbacks": {
368
+ "EarlyStoppingCallback": {
369
+ "args": {
370
+ "early_stopping_patience": 7,
371
+ "early_stopping_threshold": 0.0
372
+ },
373
+ "attributes": {
374
+ "early_stopping_patience_counter": 0
375
+ }
376
+ },
377
+ "TrainerControl": {
378
+ "args": {
379
+ "should_epoch_stop": false,
380
+ "should_evaluate": false,
381
+ "should_log": false,
382
+ "should_save": true,
383
+ "should_training_stop": true
384
+ },
385
+ "attributes": {}
386
+ }
387
+ },
388
+ "total_flos": 6.60947324633088e+16,
389
+ "train_batch_size": 1,
390
+ "trial_name": null,
391
+ "trial_params": null
392
+ }