GaetanMichelet commited on
Commit
75bd8dd
1 Parent(s): 0c9dc1b

Model save

Browse files
README.md ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: meta-llama/Meta-Llama-3.1-8B-Instruct
3
+ library_name: peft
4
+ license: llama3.1
5
+ tags:
6
+ - trl
7
+ - sft
8
+ - generated_from_trainer
9
+ model-index:
10
+ - name: Llama-31-8B_task-1_60-samples_config-1_full_auto
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
+ # Llama-31-8B_task-1_60-samples_config-1_full_auto
18
+
19
+ This model is a fine-tuned version of [meta-llama/Meta-Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) on an unknown dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 1.2038
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
+ | 2.1899 | 0.8696 | 5 | 2.0616 |
57
+ | 1.9473 | 1.9130 | 11 | 1.7736 |
58
+ | 1.6049 | 2.9565 | 17 | 1.4599 |
59
+ | 1.1934 | 4.0 | 23 | 1.0350 |
60
+ | 0.8711 | 4.8696 | 28 | 0.9062 |
61
+ | 0.8035 | 5.9130 | 34 | 0.8637 |
62
+ | 0.7366 | 6.9565 | 40 | 0.8406 |
63
+ | 0.6995 | 8.0 | 46 | 0.8247 |
64
+ | 0.6613 | 8.8696 | 51 | 0.8259 |
65
+ | 0.5531 | 9.9130 | 57 | 0.8318 |
66
+ | 0.5061 | 10.9565 | 63 | 0.8598 |
67
+ | 0.3776 | 12.0 | 69 | 0.9167 |
68
+ | 0.264 | 12.8696 | 74 | 1.0158 |
69
+ | 0.1878 | 13.9130 | 80 | 1.1250 |
70
+ | 0.1417 | 14.9565 | 86 | 1.2038 |
71
+
72
+
73
+ ### Framework versions
74
+
75
+ - PEFT 0.12.0
76
+ - Transformers 4.44.0
77
+ - Pytorch 2.1.2+cu121
78
+ - Datasets 2.20.0
79
+ - Tokenizers 0.19.1
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:01beebc54d878d582a12b87392d210b313174259a41de615d0af080680872711
3
  size 167832240
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4f47d5d758228da60b5bd9506923186b288afb99c1c97c4844668019b832db77
3
  size 167832240
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 14.956521739130435,
3
+ "total_flos": 1.1330036965572608e+16,
4
+ "train_loss": 0.8706800944150187,
5
+ "train_runtime": 834.1495,
6
+ "train_samples": 46,
7
+ "train_samples_per_second": 2.757,
8
+ "train_steps_per_second": 0.3
9
+ }
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 14.956521739130435,
3
+ "total_flos": 1.1330036965572608e+16,
4
+ "train_loss": 0.8706800944150187,
5
+ "train_runtime": 834.1495,
6
+ "train_samples": 46,
7
+ "train_samples_per_second": 2.757,
8
+ "train_steps_per_second": 0.3
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,479 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.8246700763702393,
3
+ "best_model_checkpoint": "data/Llama-31-8B_task-1_60-samples_config-1_full_auto/checkpoint-46",
4
+ "epoch": 14.956521739130435,
5
+ "eval_steps": 500,
6
+ "global_step": 86,
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": 1.8825360536575317,
14
+ "learning_rate": 4.000000000000001e-06,
15
+ "loss": 2.134,
16
+ "step": 1
17
+ },
18
+ {
19
+ "epoch": 0.34782608695652173,
20
+ "grad_norm": 1.797432780265808,
21
+ "learning_rate": 8.000000000000001e-06,
22
+ "loss": 2.2794,
23
+ "step": 2
24
+ },
25
+ {
26
+ "epoch": 0.6956521739130435,
27
+ "grad_norm": 1.3481346368789673,
28
+ "learning_rate": 1.6000000000000003e-05,
29
+ "loss": 2.1899,
30
+ "step": 4
31
+ },
32
+ {
33
+ "epoch": 0.8695652173913043,
34
+ "eval_loss": 2.0615599155426025,
35
+ "eval_runtime": 5.0274,
36
+ "eval_samples_per_second": 2.387,
37
+ "eval_steps_per_second": 2.387,
38
+ "step": 5
39
+ },
40
+ {
41
+ "epoch": 1.0434782608695652,
42
+ "grad_norm": 1.4031779766082764,
43
+ "learning_rate": 2.4e-05,
44
+ "loss": 2.1487,
45
+ "step": 6
46
+ },
47
+ {
48
+ "epoch": 1.391304347826087,
49
+ "grad_norm": 1.6137045621871948,
50
+ "learning_rate": 3.2000000000000005e-05,
51
+ "loss": 2.0567,
52
+ "step": 8
53
+ },
54
+ {
55
+ "epoch": 1.7391304347826086,
56
+ "grad_norm": 1.0302890539169312,
57
+ "learning_rate": 4e-05,
58
+ "loss": 1.9473,
59
+ "step": 10
60
+ },
61
+ {
62
+ "epoch": 1.9130434782608696,
63
+ "eval_loss": 1.7736454010009766,
64
+ "eval_runtime": 5.0277,
65
+ "eval_samples_per_second": 2.387,
66
+ "eval_steps_per_second": 2.387,
67
+ "step": 11
68
+ },
69
+ {
70
+ "epoch": 2.0869565217391304,
71
+ "grad_norm": 1.3873261213302612,
72
+ "learning_rate": 4.8e-05,
73
+ "loss": 1.9107,
74
+ "step": 12
75
+ },
76
+ {
77
+ "epoch": 2.4347826086956523,
78
+ "grad_norm": 0.8510915637016296,
79
+ "learning_rate": 5.6000000000000006e-05,
80
+ "loss": 1.7363,
81
+ "step": 14
82
+ },
83
+ {
84
+ "epoch": 2.782608695652174,
85
+ "grad_norm": 0.6782408356666565,
86
+ "learning_rate": 6.400000000000001e-05,
87
+ "loss": 1.6049,
88
+ "step": 16
89
+ },
90
+ {
91
+ "epoch": 2.9565217391304346,
92
+ "eval_loss": 1.4598970413208008,
93
+ "eval_runtime": 5.0449,
94
+ "eval_samples_per_second": 2.379,
95
+ "eval_steps_per_second": 2.379,
96
+ "step": 17
97
+ },
98
+ {
99
+ "epoch": 3.130434782608696,
100
+ "grad_norm": 0.7629269361495972,
101
+ "learning_rate": 7.2e-05,
102
+ "loss": 1.5614,
103
+ "step": 18
104
+ },
105
+ {
106
+ "epoch": 3.4782608695652173,
107
+ "grad_norm": 0.9661106467247009,
108
+ "learning_rate": 8e-05,
109
+ "loss": 1.3718,
110
+ "step": 20
111
+ },
112
+ {
113
+ "epoch": 3.8260869565217392,
114
+ "grad_norm": 0.9902026653289795,
115
+ "learning_rate": 8.800000000000001e-05,
116
+ "loss": 1.1934,
117
+ "step": 22
118
+ },
119
+ {
120
+ "epoch": 4.0,
121
+ "eval_loss": 1.0349758863449097,
122
+ "eval_runtime": 5.0242,
123
+ "eval_samples_per_second": 2.388,
124
+ "eval_steps_per_second": 2.388,
125
+ "step": 23
126
+ },
127
+ {
128
+ "epoch": 4.173913043478261,
129
+ "grad_norm": 0.8432597517967224,
130
+ "learning_rate": 9.6e-05,
131
+ "loss": 1.0786,
132
+ "step": 24
133
+ },
134
+ {
135
+ "epoch": 4.521739130434782,
136
+ "grad_norm": 0.5363503694534302,
137
+ "learning_rate": 9.999512620046522e-05,
138
+ "loss": 0.9666,
139
+ "step": 26
140
+ },
141
+ {
142
+ "epoch": 4.869565217391305,
143
+ "grad_norm": 0.5397949814796448,
144
+ "learning_rate": 9.995614150494293e-05,
145
+ "loss": 0.8711,
146
+ "step": 28
147
+ },
148
+ {
149
+ "epoch": 4.869565217391305,
150
+ "eval_loss": 0.906165361404419,
151
+ "eval_runtime": 5.0276,
152
+ "eval_samples_per_second": 2.387,
153
+ "eval_steps_per_second": 2.387,
154
+ "step": 28
155
+ },
156
+ {
157
+ "epoch": 5.217391304347826,
158
+ "grad_norm": 0.4831950068473816,
159
+ "learning_rate": 9.987820251299122e-05,
160
+ "loss": 0.8453,
161
+ "step": 30
162
+ },
163
+ {
164
+ "epoch": 5.565217391304348,
165
+ "grad_norm": 0.45016416907310486,
166
+ "learning_rate": 9.976136999909156e-05,
167
+ "loss": 0.8762,
168
+ "step": 32
169
+ },
170
+ {
171
+ "epoch": 5.913043478260869,
172
+ "grad_norm": 0.41508305072784424,
173
+ "learning_rate": 9.96057350657239e-05,
174
+ "loss": 0.8035,
175
+ "step": 34
176
+ },
177
+ {
178
+ "epoch": 5.913043478260869,
179
+ "eval_loss": 0.8637191653251648,
180
+ "eval_runtime": 5.0295,
181
+ "eval_samples_per_second": 2.386,
182
+ "eval_steps_per_second": 2.386,
183
+ "step": 34
184
+ },
185
+ {
186
+ "epoch": 6.260869565217392,
187
+ "grad_norm": 0.4248453378677368,
188
+ "learning_rate": 9.941141907232765e-05,
189
+ "loss": 0.7614,
190
+ "step": 36
191
+ },
192
+ {
193
+ "epoch": 6.608695652173913,
194
+ "grad_norm": 0.4545738995075226,
195
+ "learning_rate": 9.917857354066931e-05,
196
+ "loss": 0.7764,
197
+ "step": 38
198
+ },
199
+ {
200
+ "epoch": 6.956521739130435,
201
+ "grad_norm": 0.4439631700515747,
202
+ "learning_rate": 9.890738003669029e-05,
203
+ "loss": 0.7366,
204
+ "step": 40
205
+ },
206
+ {
207
+ "epoch": 6.956521739130435,
208
+ "eval_loss": 0.8405813574790955,
209
+ "eval_runtime": 5.0307,
210
+ "eval_samples_per_second": 2.385,
211
+ "eval_steps_per_second": 2.385,
212
+ "step": 40
213
+ },
214
+ {
215
+ "epoch": 7.304347826086957,
216
+ "grad_norm": 0.5056247115135193,
217
+ "learning_rate": 9.859805002892732e-05,
218
+ "loss": 0.7489,
219
+ "step": 42
220
+ },
221
+ {
222
+ "epoch": 7.6521739130434785,
223
+ "grad_norm": 0.44373199343681335,
224
+ "learning_rate": 9.825082472361557e-05,
225
+ "loss": 0.6803,
226
+ "step": 44
227
+ },
228
+ {
229
+ "epoch": 8.0,
230
+ "grad_norm": 0.5359618067741394,
231
+ "learning_rate": 9.786597487660337e-05,
232
+ "loss": 0.6995,
233
+ "step": 46
234
+ },
235
+ {
236
+ "epoch": 8.0,
237
+ "eval_loss": 0.8246700763702393,
238
+ "eval_runtime": 5.0242,
239
+ "eval_samples_per_second": 2.388,
240
+ "eval_steps_per_second": 2.388,
241
+ "step": 46
242
+ },
243
+ {
244
+ "epoch": 8.347826086956522,
245
+ "grad_norm": 0.5545627474784851,
246
+ "learning_rate": 9.744380058222483e-05,
247
+ "loss": 0.6248,
248
+ "step": 48
249
+ },
250
+ {
251
+ "epoch": 8.695652173913043,
252
+ "grad_norm": 0.6266648173332214,
253
+ "learning_rate": 9.698463103929542e-05,
254
+ "loss": 0.6613,
255
+ "step": 50
256
+ },
257
+ {
258
+ "epoch": 8.869565217391305,
259
+ "eval_loss": 0.8258533477783203,
260
+ "eval_runtime": 5.0277,
261
+ "eval_samples_per_second": 2.387,
262
+ "eval_steps_per_second": 2.387,
263
+ "step": 51
264
+ },
265
+ {
266
+ "epoch": 9.043478260869565,
267
+ "grad_norm": 0.7139117121696472,
268
+ "learning_rate": 9.648882429441257e-05,
269
+ "loss": 0.6253,
270
+ "step": 52
271
+ },
272
+ {
273
+ "epoch": 9.391304347826088,
274
+ "grad_norm": 0.7528420686721802,
275
+ "learning_rate": 9.595676696276172e-05,
276
+ "loss": 0.5488,
277
+ "step": 54
278
+ },
279
+ {
280
+ "epoch": 9.73913043478261,
281
+ "grad_norm": 0.7976906895637512,
282
+ "learning_rate": 9.538887392664544e-05,
283
+ "loss": 0.5531,
284
+ "step": 56
285
+ },
286
+ {
287
+ "epoch": 9.91304347826087,
288
+ "eval_loss": 0.8317675590515137,
289
+ "eval_runtime": 5.0285,
290
+ "eval_samples_per_second": 2.386,
291
+ "eval_steps_per_second": 2.386,
292
+ "step": 57
293
+ },
294
+ {
295
+ "epoch": 10.08695652173913,
296
+ "grad_norm": 0.7176584005355835,
297
+ "learning_rate": 9.478558801197065e-05,
298
+ "loss": 0.5712,
299
+ "step": 58
300
+ },
301
+ {
302
+ "epoch": 10.434782608695652,
303
+ "grad_norm": 0.6586626768112183,
304
+ "learning_rate": 9.414737964294636e-05,
305
+ "loss": 0.4544,
306
+ "step": 60
307
+ },
308
+ {
309
+ "epoch": 10.782608695652174,
310
+ "grad_norm": 0.8539354205131531,
311
+ "learning_rate": 9.347474647526095e-05,
312
+ "loss": 0.5061,
313
+ "step": 62
314
+ },
315
+ {
316
+ "epoch": 10.956521739130435,
317
+ "eval_loss": 0.8597629070281982,
318
+ "eval_runtime": 5.029,
319
+ "eval_samples_per_second": 2.386,
320
+ "eval_steps_per_second": 2.386,
321
+ "step": 63
322
+ },
323
+ {
324
+ "epoch": 11.130434782608695,
325
+ "grad_norm": 0.6799076199531555,
326
+ "learning_rate": 9.276821300802534e-05,
327
+ "loss": 0.4456,
328
+ "step": 64
329
+ },
330
+ {
331
+ "epoch": 11.478260869565217,
332
+ "grad_norm": 0.6904007792472839,
333
+ "learning_rate": 9.202833017478422e-05,
334
+ "loss": 0.3692,
335
+ "step": 66
336
+ },
337
+ {
338
+ "epoch": 11.826086956521738,
339
+ "grad_norm": 0.926986813545227,
340
+ "learning_rate": 9.125567491391476e-05,
341
+ "loss": 0.3776,
342
+ "step": 68
343
+ },
344
+ {
345
+ "epoch": 12.0,
346
+ "eval_loss": 0.9167086482048035,
347
+ "eval_runtime": 5.0287,
348
+ "eval_samples_per_second": 2.386,
349
+ "eval_steps_per_second": 2.386,
350
+ "step": 69
351
+ },
352
+ {
353
+ "epoch": 12.173913043478262,
354
+ "grad_norm": 0.8241488337516785,
355
+ "learning_rate": 9.045084971874738e-05,
356
+ "loss": 0.3262,
357
+ "step": 70
358
+ },
359
+ {
360
+ "epoch": 12.521739130434783,
361
+ "grad_norm": 0.8996691703796387,
362
+ "learning_rate": 8.961448216775954e-05,
363
+ "loss": 0.3029,
364
+ "step": 72
365
+ },
366
+ {
367
+ "epoch": 12.869565217391305,
368
+ "grad_norm": 1.148189663887024,
369
+ "learning_rate": 8.874722443520899e-05,
370
+ "loss": 0.264,
371
+ "step": 74
372
+ },
373
+ {
374
+ "epoch": 12.869565217391305,
375
+ "eval_loss": 1.0157562494277954,
376
+ "eval_runtime": 5.0275,
377
+ "eval_samples_per_second": 2.387,
378
+ "eval_steps_per_second": 2.387,
379
+ "step": 74
380
+ },
381
+ {
382
+ "epoch": 13.217391304347826,
383
+ "grad_norm": 0.9193633794784546,
384
+ "learning_rate": 8.784975278258783e-05,
385
+ "loss": 0.2267,
386
+ "step": 76
387
+ },
388
+ {
389
+ "epoch": 13.565217391304348,
390
+ "grad_norm": 1.1552317142486572,
391
+ "learning_rate": 8.692276703129421e-05,
392
+ "loss": 0.1833,
393
+ "step": 78
394
+ },
395
+ {
396
+ "epoch": 13.91304347826087,
397
+ "grad_norm": 0.9492861032485962,
398
+ "learning_rate": 8.596699001693255e-05,
399
+ "loss": 0.1878,
400
+ "step": 80
401
+ },
402
+ {
403
+ "epoch": 13.91304347826087,
404
+ "eval_loss": 1.1250226497650146,
405
+ "eval_runtime": 5.0348,
406
+ "eval_samples_per_second": 2.383,
407
+ "eval_steps_per_second": 2.383,
408
+ "step": 80
409
+ },
410
+ {
411
+ "epoch": 14.26086956521739,
412
+ "grad_norm": 1.2470932006835938,
413
+ "learning_rate": 8.498316702566828e-05,
414
+ "loss": 0.1453,
415
+ "step": 82
416
+ },
417
+ {
418
+ "epoch": 14.608695652173914,
419
+ "grad_norm": 1.0260671377182007,
420
+ "learning_rate": 8.397206521307584e-05,
421
+ "loss": 0.126,
422
+ "step": 84
423
+ },
424
+ {
425
+ "epoch": 14.956521739130435,
426
+ "grad_norm": 1.2034170627593994,
427
+ "learning_rate": 8.293447300593402e-05,
428
+ "loss": 0.1417,
429
+ "step": 86
430
+ },
431
+ {
432
+ "epoch": 14.956521739130435,
433
+ "eval_loss": 1.203776240348816,
434
+ "eval_runtime": 5.0321,
435
+ "eval_samples_per_second": 2.385,
436
+ "eval_steps_per_second": 2.385,
437
+ "step": 86
438
+ },
439
+ {
440
+ "epoch": 14.956521739130435,
441
+ "step": 86,
442
+ "total_flos": 1.1330036965572608e+16,
443
+ "train_loss": 0.8706800944150187,
444
+ "train_runtime": 834.1495,
445
+ "train_samples_per_second": 2.757,
446
+ "train_steps_per_second": 0.3
447
+ }
448
+ ],
449
+ "logging_steps": 2,
450
+ "max_steps": 250,
451
+ "num_input_tokens_seen": 0,
452
+ "num_train_epochs": 50,
453
+ "save_steps": 25,
454
+ "stateful_callbacks": {
455
+ "EarlyStoppingCallback": {
456
+ "args": {
457
+ "early_stopping_patience": 7,
458
+ "early_stopping_threshold": 0.0
459
+ },
460
+ "attributes": {
461
+ "early_stopping_patience_counter": 0
462
+ }
463
+ },
464
+ "TrainerControl": {
465
+ "args": {
466
+ "should_epoch_stop": false,
467
+ "should_evaluate": false,
468
+ "should_log": false,
469
+ "should_save": true,
470
+ "should_training_stop": true
471
+ },
472
+ "attributes": {}
473
+ }
474
+ },
475
+ "total_flos": 1.1330036965572608e+16,
476
+ "train_batch_size": 1,
477
+ "trial_name": null,
478
+ "trial_params": null
479
+ }