gsmyrnis commited on
Commit
39f4cfd
·
verified ·
1 Parent(s): 7e18a62

End of training

Browse files
README.md CHANGED
@@ -4,6 +4,7 @@ license: apache-2.0
4
  base_model: mistralai/Mistral-7B-v0.3
5
  tags:
6
  - llama-factory
 
7
  - generated_from_trainer
8
  model-index:
9
  - name: mistral_7b_0-3_oh-dcft-v3.1-claude-3-5-haiku-20241022
@@ -15,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
15
 
16
  # mistral_7b_0-3_oh-dcft-v3.1-claude-3-5-haiku-20241022
17
 
18
- This model is a fine-tuned version of [mistralai/Mistral-7B-v0.3](https://huggingface.co/mistralai/Mistral-7B-v0.3) on an unknown dataset.
19
  It achieves the following results on the evaluation set:
20
  - Loss: 0.3708
21
 
 
4
  base_model: mistralai/Mistral-7B-v0.3
5
  tags:
6
  - llama-factory
7
+ - full
8
  - generated_from_trainer
9
  model-index:
10
  - name: mistral_7b_0-3_oh-dcft-v3.1-claude-3-5-haiku-20241022
 
16
 
17
  # mistral_7b_0-3_oh-dcft-v3.1-claude-3-5-haiku-20241022
18
 
19
+ This model is a fine-tuned version of [mistralai/Mistral-7B-v0.3](https://huggingface.co/mistralai/Mistral-7B-v0.3) on the mlfoundations-dev/oh-dcft-v3.1-claude-3-5-haiku-20241022 dataset.
20
  It achieves the following results on the evaluation set:
21
  - Loss: 0.3708
22
 
all_results.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "eval_loss": 0.37075862288475037,
4
+ "eval_runtime": 53.1485,
5
+ "eval_samples_per_second": 256.639,
6
+ "eval_steps_per_second": 1.016,
7
+ "total_flos": 2547731650314240.0,
8
+ "train_loss": 0.2954053143131192,
9
+ "train_runtime": 9131.3778,
10
+ "train_samples_per_second": 85.143,
11
+ "train_steps_per_second": 0.167
12
+ }
eval_results.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "eval_loss": 0.37075862288475037,
4
+ "eval_runtime": 53.1485,
5
+ "eval_samples_per_second": 256.639,
6
+ "eval_steps_per_second": 1.016
7
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.0,
3
+ "total_flos": 2547731650314240.0,
4
+ "train_loss": 0.2954053143131192,
5
+ "train_runtime": 9131.3778,
6
+ "train_samples_per_second": 85.143,
7
+ "train_steps_per_second": 0.167
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,1130 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 3.0,
5
+ "eval_steps": 500,
6
+ "global_step": 1521,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01972386587771203,
13
+ "grad_norm": 8.260985845744274,
14
+ "learning_rate": 6.493506493506493e-07,
15
+ "loss": 0.7556,
16
+ "step": 10
17
+ },
18
+ {
19
+ "epoch": 0.03944773175542406,
20
+ "grad_norm": 3.285262290141762,
21
+ "learning_rate": 1.2987012987012986e-06,
22
+ "loss": 0.6243,
23
+ "step": 20
24
+ },
25
+ {
26
+ "epoch": 0.05917159763313609,
27
+ "grad_norm": 1.8240152669785012,
28
+ "learning_rate": 1.9480519480519483e-06,
29
+ "loss": 0.527,
30
+ "step": 30
31
+ },
32
+ {
33
+ "epoch": 0.07889546351084813,
34
+ "grad_norm": 2.484918347587673,
35
+ "learning_rate": 2.597402597402597e-06,
36
+ "loss": 0.4858,
37
+ "step": 40
38
+ },
39
+ {
40
+ "epoch": 0.09861932938856016,
41
+ "grad_norm": 1.6071106659008887,
42
+ "learning_rate": 3.246753246753247e-06,
43
+ "loss": 0.4616,
44
+ "step": 50
45
+ },
46
+ {
47
+ "epoch": 0.11834319526627218,
48
+ "grad_norm": 1.7753595146102916,
49
+ "learning_rate": 3.896103896103897e-06,
50
+ "loss": 0.4467,
51
+ "step": 60
52
+ },
53
+ {
54
+ "epoch": 0.13806706114398423,
55
+ "grad_norm": 1.9952784096044014,
56
+ "learning_rate": 4.5454545454545455e-06,
57
+ "loss": 0.4355,
58
+ "step": 70
59
+ },
60
+ {
61
+ "epoch": 0.15779092702169625,
62
+ "grad_norm": 2.0090088525329057,
63
+ "learning_rate": 4.999952075361122e-06,
64
+ "loss": 0.4303,
65
+ "step": 80
66
+ },
67
+ {
68
+ "epoch": 0.17751479289940827,
69
+ "grad_norm": 1.9429866993343683,
70
+ "learning_rate": 4.99910013857428e-06,
71
+ "loss": 0.4213,
72
+ "step": 90
73
+ },
74
+ {
75
+ "epoch": 0.19723865877712032,
76
+ "grad_norm": 2.2150406308730166,
77
+ "learning_rate": 4.997183673954895e-06,
78
+ "loss": 0.4205,
79
+ "step": 100
80
+ },
81
+ {
82
+ "epoch": 0.21696252465483234,
83
+ "grad_norm": 2.3280715715799105,
84
+ "learning_rate": 4.994203588590157e-06,
85
+ "loss": 0.4132,
86
+ "step": 110
87
+ },
88
+ {
89
+ "epoch": 0.23668639053254437,
90
+ "grad_norm": 2.0514718162160617,
91
+ "learning_rate": 4.9901612929925455e-06,
92
+ "loss": 0.4097,
93
+ "step": 120
94
+ },
95
+ {
96
+ "epoch": 0.2564102564102564,
97
+ "grad_norm": 2.248051724393392,
98
+ "learning_rate": 4.985058700432217e-06,
99
+ "loss": 0.4078,
100
+ "step": 130
101
+ },
102
+ {
103
+ "epoch": 0.27613412228796846,
104
+ "grad_norm": 2.4477065193392114,
105
+ "learning_rate": 4.978898226031426e-06,
106
+ "loss": 0.4035,
107
+ "step": 140
108
+ },
109
+ {
110
+ "epoch": 0.2958579881656805,
111
+ "grad_norm": 2.3530821376592317,
112
+ "learning_rate": 4.97168278562142e-06,
113
+ "loss": 0.3988,
114
+ "step": 150
115
+ },
116
+ {
117
+ "epoch": 0.3155818540433925,
118
+ "grad_norm": 2.0658208779463796,
119
+ "learning_rate": 4.9634157943623345e-06,
120
+ "loss": 0.4008,
121
+ "step": 160
122
+ },
123
+ {
124
+ "epoch": 0.33530571992110453,
125
+ "grad_norm": 1.6308701318103827,
126
+ "learning_rate": 4.954101165126764e-06,
127
+ "loss": 0.3955,
128
+ "step": 170
129
+ },
130
+ {
131
+ "epoch": 0.35502958579881655,
132
+ "grad_norm": 1.8767575875235638,
133
+ "learning_rate": 4.943743306647738e-06,
134
+ "loss": 0.3964,
135
+ "step": 180
136
+ },
137
+ {
138
+ "epoch": 0.3747534516765286,
139
+ "grad_norm": 2.158851334024998,
140
+ "learning_rate": 4.932347121432018e-06,
141
+ "loss": 0.3955,
142
+ "step": 190
143
+ },
144
+ {
145
+ "epoch": 0.39447731755424065,
146
+ "grad_norm": 2.2424601067528367,
147
+ "learning_rate": 4.919918003439677e-06,
148
+ "loss": 0.3929,
149
+ "step": 200
150
+ },
151
+ {
152
+ "epoch": 0.41420118343195267,
153
+ "grad_norm": 1.4704562127782181,
154
+ "learning_rate": 4.9064618355310694e-06,
155
+ "loss": 0.3951,
156
+ "step": 210
157
+ },
158
+ {
159
+ "epoch": 0.4339250493096647,
160
+ "grad_norm": 1.5325962055467024,
161
+ "learning_rate": 4.8919849866823955e-06,
162
+ "loss": 0.3936,
163
+ "step": 220
164
+ },
165
+ {
166
+ "epoch": 0.4536489151873767,
167
+ "grad_norm": 1.752553432251344,
168
+ "learning_rate": 4.8764943089711876e-06,
169
+ "loss": 0.3894,
170
+ "step": 230
171
+ },
172
+ {
173
+ "epoch": 0.47337278106508873,
174
+ "grad_norm": 3.582185649197669,
175
+ "learning_rate": 4.859997134333133e-06,
176
+ "loss": 0.39,
177
+ "step": 240
178
+ },
179
+ {
180
+ "epoch": 0.4930966469428008,
181
+ "grad_norm": 2.283623608488685,
182
+ "learning_rate": 4.842501271091773e-06,
183
+ "loss": 0.3845,
184
+ "step": 250
185
+ },
186
+ {
187
+ "epoch": 0.5128205128205128,
188
+ "grad_norm": 2.954635543178996,
189
+ "learning_rate": 4.8240150002627285e-06,
190
+ "loss": 0.3853,
191
+ "step": 260
192
+ },
193
+ {
194
+ "epoch": 0.5325443786982249,
195
+ "grad_norm": 2.621411991175976,
196
+ "learning_rate": 4.80454707163418e-06,
197
+ "loss": 0.3802,
198
+ "step": 270
199
+ },
200
+ {
201
+ "epoch": 0.5522682445759369,
202
+ "grad_norm": 3.0076538937186554,
203
+ "learning_rate": 4.784106699625493e-06,
204
+ "loss": 0.3778,
205
+ "step": 280
206
+ },
207
+ {
208
+ "epoch": 0.571992110453649,
209
+ "grad_norm": 2.620788244299813,
210
+ "learning_rate": 4.762703558925907e-06,
211
+ "loss": 0.381,
212
+ "step": 290
213
+ },
214
+ {
215
+ "epoch": 0.591715976331361,
216
+ "grad_norm": 2.600774288511616,
217
+ "learning_rate": 4.740347779915384e-06,
218
+ "loss": 0.3795,
219
+ "step": 300
220
+ },
221
+ {
222
+ "epoch": 0.611439842209073,
223
+ "grad_norm": 2.825934593188172,
224
+ "learning_rate": 4.717049943869774e-06,
225
+ "loss": 0.3754,
226
+ "step": 310
227
+ },
228
+ {
229
+ "epoch": 0.631163708086785,
230
+ "grad_norm": 1.9636455738063043,
231
+ "learning_rate": 4.692821077952556e-06,
232
+ "loss": 0.3709,
233
+ "step": 320
234
+ },
235
+ {
236
+ "epoch": 0.650887573964497,
237
+ "grad_norm": 1.465934093555826,
238
+ "learning_rate": 4.667672649995539e-06,
239
+ "loss": 0.3686,
240
+ "step": 330
241
+ },
242
+ {
243
+ "epoch": 0.6706114398422091,
244
+ "grad_norm": 1.6730733158146738,
245
+ "learning_rate": 4.641616563071003e-06,
246
+ "loss": 0.374,
247
+ "step": 340
248
+ },
249
+ {
250
+ "epoch": 0.6903353057199211,
251
+ "grad_norm": 1.6420981338152472,
252
+ "learning_rate": 4.6146651498578095e-06,
253
+ "loss": 0.3725,
254
+ "step": 350
255
+ },
256
+ {
257
+ "epoch": 0.7100591715976331,
258
+ "grad_norm": 1.7081311753490396,
259
+ "learning_rate": 4.586831166804191e-06,
260
+ "loss": 0.3723,
261
+ "step": 360
262
+ },
263
+ {
264
+ "epoch": 0.7297830374753451,
265
+ "grad_norm": 1.7004420091082517,
266
+ "learning_rate": 4.558127788089966e-06,
267
+ "loss": 0.3685,
268
+ "step": 370
269
+ },
270
+ {
271
+ "epoch": 0.7495069033530573,
272
+ "grad_norm": 1.813129623101683,
273
+ "learning_rate": 4.5285685993910246e-06,
274
+ "loss": 0.3693,
275
+ "step": 380
276
+ },
277
+ {
278
+ "epoch": 0.7692307692307693,
279
+ "grad_norm": 1.441392302489358,
280
+ "learning_rate": 4.49816759144906e-06,
281
+ "loss": 0.3672,
282
+ "step": 390
283
+ },
284
+ {
285
+ "epoch": 0.7889546351084813,
286
+ "grad_norm": 1.743528342139816,
287
+ "learning_rate": 4.466939153449565e-06,
288
+ "loss": 0.3629,
289
+ "step": 400
290
+ },
291
+ {
292
+ "epoch": 0.8086785009861933,
293
+ "grad_norm": 1.5505480061250534,
294
+ "learning_rate": 4.434898066211255e-06,
295
+ "loss": 0.3647,
296
+ "step": 410
297
+ },
298
+ {
299
+ "epoch": 0.8284023668639053,
300
+ "grad_norm": 1.748134152515452,
301
+ "learning_rate": 4.402059495190112e-06,
302
+ "loss": 0.3687,
303
+ "step": 420
304
+ },
305
+ {
306
+ "epoch": 0.8481262327416174,
307
+ "grad_norm": 1.888131474531523,
308
+ "learning_rate": 4.368438983301382e-06,
309
+ "loss": 0.368,
310
+ "step": 430
311
+ },
312
+ {
313
+ "epoch": 0.8678500986193294,
314
+ "grad_norm": 1.3077877777100417,
315
+ "learning_rate": 4.334052443562914e-06,
316
+ "loss": 0.364,
317
+ "step": 440
318
+ },
319
+ {
320
+ "epoch": 0.8875739644970414,
321
+ "grad_norm": 1.7143497390643974,
322
+ "learning_rate": 4.298916151563324e-06,
323
+ "loss": 0.3662,
324
+ "step": 450
325
+ },
326
+ {
327
+ "epoch": 0.9072978303747534,
328
+ "grad_norm": 1.2650560376490414,
329
+ "learning_rate": 4.263046737758557e-06,
330
+ "loss": 0.3634,
331
+ "step": 460
332
+ },
333
+ {
334
+ "epoch": 0.9270216962524654,
335
+ "grad_norm": 1.325272234023546,
336
+ "learning_rate": 4.226461179600474e-06,
337
+ "loss": 0.3647,
338
+ "step": 470
339
+ },
340
+ {
341
+ "epoch": 0.9467455621301775,
342
+ "grad_norm": 1.7799396783443953,
343
+ "learning_rate": 4.189176793501208e-06,
344
+ "loss": 0.3601,
345
+ "step": 480
346
+ },
347
+ {
348
+ "epoch": 0.9664694280078896,
349
+ "grad_norm": 1.6138030010077298,
350
+ "learning_rate": 4.151211226637083e-06,
351
+ "loss": 0.3639,
352
+ "step": 490
353
+ },
354
+ {
355
+ "epoch": 0.9861932938856016,
356
+ "grad_norm": 1.6475058606657829,
357
+ "learning_rate": 4.112582448595989e-06,
358
+ "loss": 0.3631,
359
+ "step": 500
360
+ },
361
+ {
362
+ "epoch": 1.0,
363
+ "eval_loss": 0.3610161542892456,
364
+ "eval_runtime": 46.5378,
365
+ "eval_samples_per_second": 293.095,
366
+ "eval_steps_per_second": 1.16,
367
+ "step": 507
368
+ },
369
+ {
370
+ "epoch": 1.0059171597633136,
371
+ "grad_norm": 2.318083617694004,
372
+ "learning_rate": 4.073308742872136e-06,
373
+ "loss": 0.339,
374
+ "step": 510
375
+ },
376
+ {
377
+ "epoch": 1.0256410256410255,
378
+ "grad_norm": 2.26507527796031,
379
+ "learning_rate": 4.033408698212244e-06,
380
+ "loss": 0.2904,
381
+ "step": 520
382
+ },
383
+ {
384
+ "epoch": 1.0453648915187377,
385
+ "grad_norm": 2.129210352759771,
386
+ "learning_rate": 3.99290119981726e-06,
387
+ "loss": 0.2845,
388
+ "step": 530
389
+ },
390
+ {
391
+ "epoch": 1.0650887573964498,
392
+ "grad_norm": 2.0458511034566897,
393
+ "learning_rate": 3.95180542040374e-06,
394
+ "loss": 0.2826,
395
+ "step": 540
396
+ },
397
+ {
398
+ "epoch": 1.0848126232741617,
399
+ "grad_norm": 2.34540520465628,
400
+ "learning_rate": 3.910140811129166e-06,
401
+ "loss": 0.2817,
402
+ "step": 550
403
+ },
404
+ {
405
+ "epoch": 1.1045364891518739,
406
+ "grad_norm": 1.5731137478504271,
407
+ "learning_rate": 3.8679270923854596e-06,
408
+ "loss": 0.2816,
409
+ "step": 560
410
+ },
411
+ {
412
+ "epoch": 1.1242603550295858,
413
+ "grad_norm": 1.9641564243584235,
414
+ "learning_rate": 3.825184244465071e-06,
415
+ "loss": 0.2833,
416
+ "step": 570
417
+ },
418
+ {
419
+ "epoch": 1.143984220907298,
420
+ "grad_norm": 1.5653763677552233,
421
+ "learning_rate": 3.7819324981040517e-06,
422
+ "loss": 0.2835,
423
+ "step": 580
424
+ },
425
+ {
426
+ "epoch": 1.1637080867850098,
427
+ "grad_norm": 1.4455902546137582,
428
+ "learning_rate": 3.7381923249065838e-06,
429
+ "loss": 0.2806,
430
+ "step": 590
431
+ },
432
+ {
433
+ "epoch": 1.183431952662722,
434
+ "grad_norm": 1.4589441051909717,
435
+ "learning_rate": 3.6939844276555146e-06,
436
+ "loss": 0.2842,
437
+ "step": 600
438
+ },
439
+ {
440
+ "epoch": 1.2031558185404339,
441
+ "grad_norm": 1.4737079619190827,
442
+ "learning_rate": 3.649329730513461e-06,
443
+ "loss": 0.2818,
444
+ "step": 610
445
+ },
446
+ {
447
+ "epoch": 1.222879684418146,
448
+ "grad_norm": 1.424470321783783,
449
+ "learning_rate": 3.6042493691191377e-06,
450
+ "loss": 0.2835,
451
+ "step": 620
452
+ },
453
+ {
454
+ "epoch": 1.242603550295858,
455
+ "grad_norm": 1.43822809638539,
456
+ "learning_rate": 3.558764680583589e-06,
457
+ "loss": 0.2829,
458
+ "step": 630
459
+ },
460
+ {
461
+ "epoch": 1.26232741617357,
462
+ "grad_norm": 1.4491877471048427,
463
+ "learning_rate": 3.51289719339106e-06,
464
+ "loss": 0.2823,
465
+ "step": 640
466
+ },
467
+ {
468
+ "epoch": 1.282051282051282,
469
+ "grad_norm": 1.4979353903583295,
470
+ "learning_rate": 3.4666686172092927e-06,
471
+ "loss": 0.2859,
472
+ "step": 650
473
+ },
474
+ {
475
+ "epoch": 1.301775147928994,
476
+ "grad_norm": 1.4793881592613725,
477
+ "learning_rate": 3.4201008326140596e-06,
478
+ "loss": 0.2849,
479
+ "step": 660
480
+ },
481
+ {
482
+ "epoch": 1.3214990138067062,
483
+ "grad_norm": 1.6343693105840815,
484
+ "learning_rate": 3.3732158807328116e-06,
485
+ "loss": 0.2875,
486
+ "step": 670
487
+ },
488
+ {
489
+ "epoch": 1.3412228796844181,
490
+ "grad_norm": 1.5638318327999918,
491
+ "learning_rate": 3.3260359528123266e-06,
492
+ "loss": 0.2877,
493
+ "step": 680
494
+ },
495
+ {
496
+ "epoch": 1.3609467455621302,
497
+ "grad_norm": 1.434550639059279,
498
+ "learning_rate": 3.2785833797153115e-06,
499
+ "loss": 0.2817,
500
+ "step": 690
501
+ },
502
+ {
503
+ "epoch": 1.3806706114398422,
504
+ "grad_norm": 1.3783604211664602,
505
+ "learning_rate": 3.2308806213509204e-06,
506
+ "loss": 0.2809,
507
+ "step": 700
508
+ },
509
+ {
510
+ "epoch": 1.4003944773175543,
511
+ "grad_norm": 1.7104337243982326,
512
+ "learning_rate": 3.182950256044188e-06,
513
+ "loss": 0.2825,
514
+ "step": 710
515
+ },
516
+ {
517
+ "epoch": 1.4201183431952662,
518
+ "grad_norm": 1.9527331404429782,
519
+ "learning_rate": 3.1348149698494233e-06,
520
+ "loss": 0.2827,
521
+ "step": 720
522
+ },
523
+ {
524
+ "epoch": 1.4398422090729783,
525
+ "grad_norm": 1.5082040480125063,
526
+ "learning_rate": 3.0864975458126158e-06,
527
+ "loss": 0.2857,
528
+ "step": 730
529
+ },
530
+ {
531
+ "epoch": 1.4595660749506902,
532
+ "grad_norm": 1.5939434329404958,
533
+ "learning_rate": 3.038020853187914e-06,
534
+ "loss": 0.2831,
535
+ "step": 740
536
+ },
537
+ {
538
+ "epoch": 1.4792899408284024,
539
+ "grad_norm": 1.425454732201556,
540
+ "learning_rate": 2.98940783661333e-06,
541
+ "loss": 0.2802,
542
+ "step": 750
543
+ },
544
+ {
545
+ "epoch": 1.4990138067061145,
546
+ "grad_norm": 1.4324944544127631,
547
+ "learning_rate": 2.940681505250742e-06,
548
+ "loss": 0.2848,
549
+ "step": 760
550
+ },
551
+ {
552
+ "epoch": 1.5187376725838264,
553
+ "grad_norm": 1.4082984304420074,
554
+ "learning_rate": 2.8918649218953624e-06,
555
+ "loss": 0.2801,
556
+ "step": 770
557
+ },
558
+ {
559
+ "epoch": 1.5384615384615383,
560
+ "grad_norm": 1.5895657718154816,
561
+ "learning_rate": 2.84298119205983e-06,
562
+ "loss": 0.2807,
563
+ "step": 780
564
+ },
565
+ {
566
+ "epoch": 1.5581854043392505,
567
+ "grad_norm": 1.6080440377232041,
568
+ "learning_rate": 2.7940534530380666e-06,
569
+ "loss": 0.2835,
570
+ "step": 790
571
+ },
572
+ {
573
+ "epoch": 1.5779092702169626,
574
+ "grad_norm": 1.404915797241871,
575
+ "learning_rate": 2.7451048629541045e-06,
576
+ "loss": 0.2808,
577
+ "step": 800
578
+ },
579
+ {
580
+ "epoch": 1.5976331360946747,
581
+ "grad_norm": 1.4879672080505235,
582
+ "learning_rate": 2.6961585898010523e-06,
583
+ "loss": 0.2806,
584
+ "step": 810
585
+ },
586
+ {
587
+ "epoch": 1.6173570019723866,
588
+ "grad_norm": 1.3888602093522253,
589
+ "learning_rate": 2.647237800475384e-06,
590
+ "loss": 0.2832,
591
+ "step": 820
592
+ },
593
+ {
594
+ "epoch": 1.6370808678500985,
595
+ "grad_norm": 1.3670120148082392,
596
+ "learning_rate": 2.5983656498117525e-06,
597
+ "loss": 0.2825,
598
+ "step": 830
599
+ },
600
+ {
601
+ "epoch": 1.6568047337278107,
602
+ "grad_norm": 1.2812642080517738,
603
+ "learning_rate": 2.54956526962351e-06,
604
+ "loss": 0.279,
605
+ "step": 840
606
+ },
607
+ {
608
+ "epoch": 1.6765285996055228,
609
+ "grad_norm": 1.252430854449729,
610
+ "learning_rate": 2.5008597577541288e-06,
611
+ "loss": 0.2814,
612
+ "step": 850
613
+ },
614
+ {
615
+ "epoch": 1.6962524654832347,
616
+ "grad_norm": 1.2750427994477165,
617
+ "learning_rate": 2.45227216714469e-06,
618
+ "loss": 0.2792,
619
+ "step": 860
620
+ },
621
+ {
622
+ "epoch": 1.7159763313609466,
623
+ "grad_norm": 1.354377403404739,
624
+ "learning_rate": 2.403825494922636e-06,
625
+ "loss": 0.282,
626
+ "step": 870
627
+ },
628
+ {
629
+ "epoch": 1.7357001972386588,
630
+ "grad_norm": 1.4267990848182481,
631
+ "learning_rate": 2.3555426715169396e-06,
632
+ "loss": 0.2791,
633
+ "step": 880
634
+ },
635
+ {
636
+ "epoch": 1.755424063116371,
637
+ "grad_norm": 1.252857555239978,
638
+ "learning_rate": 2.3074465498048303e-06,
639
+ "loss": 0.2826,
640
+ "step": 890
641
+ },
642
+ {
643
+ "epoch": 1.7751479289940828,
644
+ "grad_norm": 1.2876786054611615,
645
+ "learning_rate": 2.259559894295244e-06,
646
+ "loss": 0.2789,
647
+ "step": 900
648
+ },
649
+ {
650
+ "epoch": 1.7948717948717947,
651
+ "grad_norm": 1.2629901820145135,
652
+ "learning_rate": 2.2119053703540866e-06,
653
+ "loss": 0.2791,
654
+ "step": 910
655
+ },
656
+ {
657
+ "epoch": 1.8145956607495068,
658
+ "grad_norm": 1.3562733049556417,
659
+ "learning_rate": 2.1645055334764237e-06,
660
+ "loss": 0.2807,
661
+ "step": 920
662
+ },
663
+ {
664
+ "epoch": 1.834319526627219,
665
+ "grad_norm": 1.3132542320273741,
666
+ "learning_rate": 2.1173828186106828e-06,
667
+ "loss": 0.2782,
668
+ "step": 930
669
+ },
670
+ {
671
+ "epoch": 1.854043392504931,
672
+ "grad_norm": 1.372645351488049,
673
+ "learning_rate": 2.0705595295399e-06,
674
+ "loss": 0.28,
675
+ "step": 940
676
+ },
677
+ {
678
+ "epoch": 1.873767258382643,
679
+ "grad_norm": 1.286506818666612,
680
+ "learning_rate": 2.0240578283250596e-06,
681
+ "loss": 0.2788,
682
+ "step": 950
683
+ },
684
+ {
685
+ "epoch": 1.893491124260355,
686
+ "grad_norm": 1.343985774681719,
687
+ "learning_rate": 1.9778997248155013e-06,
688
+ "loss": 0.2779,
689
+ "step": 960
690
+ },
691
+ {
692
+ "epoch": 1.913214990138067,
693
+ "grad_norm": 1.3873943864064089,
694
+ "learning_rate": 1.9321070662313824e-06,
695
+ "loss": 0.2768,
696
+ "step": 970
697
+ },
698
+ {
699
+ "epoch": 1.9329388560157792,
700
+ "grad_norm": 1.3822544572854645,
701
+ "learning_rate": 1.88670152682311e-06,
702
+ "loss": 0.2753,
703
+ "step": 980
704
+ },
705
+ {
706
+ "epoch": 1.952662721893491,
707
+ "grad_norm": 1.3724554338840655,
708
+ "learning_rate": 1.8417045976126347e-06,
709
+ "loss": 0.274,
710
+ "step": 990
711
+ },
712
+ {
713
+ "epoch": 1.972386587771203,
714
+ "grad_norm": 1.428387339598408,
715
+ "learning_rate": 1.797137576221482e-06,
716
+ "loss": 0.2775,
717
+ "step": 1000
718
+ },
719
+ {
720
+ "epoch": 1.9921104536489151,
721
+ "grad_norm": 1.2370547509299645,
722
+ "learning_rate": 1.753021556790314e-06,
723
+ "loss": 0.2746,
724
+ "step": 1010
725
+ },
726
+ {
727
+ "epoch": 2.0,
728
+ "eval_loss": 0.3482723832130432,
729
+ "eval_runtime": 46.4255,
730
+ "eval_samples_per_second": 293.804,
731
+ "eval_steps_per_second": 1.163,
732
+ "step": 1014
733
+ },
734
+ {
735
+ "epoch": 2.0118343195266273,
736
+ "grad_norm": 1.9502351693684774,
737
+ "learning_rate": 1.7093774199948004e-06,
738
+ "loss": 0.2309,
739
+ "step": 1020
740
+ },
741
+ {
742
+ "epoch": 2.0315581854043394,
743
+ "grad_norm": 1.5862323859503984,
744
+ "learning_rate": 1.6662258231625331e-06,
745
+ "loss": 0.2026,
746
+ "step": 1030
747
+ },
748
+ {
749
+ "epoch": 2.051282051282051,
750
+ "grad_norm": 1.3292614459089434,
751
+ "learning_rate": 1.6235871904956431e-06,
752
+ "loss": 0.2034,
753
+ "step": 1040
754
+ },
755
+ {
756
+ "epoch": 2.0710059171597632,
757
+ "grad_norm": 1.2370582334736997,
758
+ "learning_rate": 1.5814817034037715e-06,
759
+ "loss": 0.2008,
760
+ "step": 1050
761
+ },
762
+ {
763
+ "epoch": 2.0907297830374754,
764
+ "grad_norm": 1.325897622024457,
765
+ "learning_rate": 1.5399292909519422e-06,
766
+ "loss": 0.2042,
767
+ "step": 1060
768
+ },
769
+ {
770
+ "epoch": 2.1104536489151875,
771
+ "grad_norm": 1.4548395791353137,
772
+ "learning_rate": 1.4989496204278897e-06,
773
+ "loss": 0.2025,
774
+ "step": 1070
775
+ },
776
+ {
777
+ "epoch": 2.1301775147928996,
778
+ "grad_norm": 1.36179677292465,
779
+ "learning_rate": 1.458562088033273e-06,
780
+ "loss": 0.1978,
781
+ "step": 1080
782
+ },
783
+ {
784
+ "epoch": 2.1499013806706113,
785
+ "grad_norm": 1.4589926591648759,
786
+ "learning_rate": 1.4187858097032086e-06,
787
+ "loss": 0.2024,
788
+ "step": 1090
789
+ },
790
+ {
791
+ "epoch": 2.1696252465483234,
792
+ "grad_norm": 1.3095440667780154,
793
+ "learning_rate": 1.3796396120584576e-06,
794
+ "loss": 0.2032,
795
+ "step": 1100
796
+ },
797
+ {
798
+ "epoch": 2.1893491124260356,
799
+ "grad_norm": 1.3522834520399176,
800
+ "learning_rate": 1.341142023494537e-06,
801
+ "loss": 0.1992,
802
+ "step": 1110
803
+ },
804
+ {
805
+ "epoch": 2.2090729783037477,
806
+ "grad_norm": 1.3914925068585928,
807
+ "learning_rate": 1.3033112654120032e-06,
808
+ "loss": 0.2029,
809
+ "step": 1120
810
+ },
811
+ {
812
+ "epoch": 2.2287968441814594,
813
+ "grad_norm": 1.2392072409116117,
814
+ "learning_rate": 1.266165243592024e-06,
815
+ "loss": 0.2019,
816
+ "step": 1130
817
+ },
818
+ {
819
+ "epoch": 2.2485207100591715,
820
+ "grad_norm": 1.450828785906611,
821
+ "learning_rate": 1.2297215397213442e-06,
822
+ "loss": 0.2029,
823
+ "step": 1140
824
+ },
825
+ {
826
+ "epoch": 2.2682445759368837,
827
+ "grad_norm": 1.3539897715774756,
828
+ "learning_rate": 1.1939974030706499e-06,
829
+ "loss": 0.1989,
830
+ "step": 1150
831
+ },
832
+ {
833
+ "epoch": 2.287968441814596,
834
+ "grad_norm": 1.3124427663284721,
835
+ "learning_rate": 1.1590097423302681e-06,
836
+ "loss": 0.2013,
837
+ "step": 1160
838
+ },
839
+ {
840
+ "epoch": 2.3076923076923075,
841
+ "grad_norm": 1.2751387286158546,
842
+ "learning_rate": 1.1247751176070688e-06,
843
+ "loss": 0.2003,
844
+ "step": 1170
845
+ },
846
+ {
847
+ "epoch": 2.3274161735700196,
848
+ "grad_norm": 1.2826788452929796,
849
+ "learning_rate": 1.0913097325863526e-06,
850
+ "loss": 0.2013,
851
+ "step": 1180
852
+ },
853
+ {
854
+ "epoch": 2.3471400394477318,
855
+ "grad_norm": 1.3449233167779666,
856
+ "learning_rate": 1.0586294268624391e-06,
857
+ "loss": 0.2031,
858
+ "step": 1190
859
+ },
860
+ {
861
+ "epoch": 2.366863905325444,
862
+ "grad_norm": 1.3034368496811286,
863
+ "learning_rate": 1.026749668441587e-06,
864
+ "loss": 0.1994,
865
+ "step": 1200
866
+ },
867
+ {
868
+ "epoch": 2.386587771203156,
869
+ "grad_norm": 1.3565807097213252,
870
+ "learning_rate": 9.956855464207873e-07,
871
+ "loss": 0.2,
872
+ "step": 1210
873
+ },
874
+ {
875
+ "epoch": 2.4063116370808677,
876
+ "grad_norm": 1.451004027193357,
877
+ "learning_rate": 9.654517638459015e-07,
878
+ "loss": 0.1996,
879
+ "step": 1220
880
+ },
881
+ {
882
+ "epoch": 2.42603550295858,
883
+ "grad_norm": 1.3107553476519733,
884
+ "learning_rate": 9.360626307525231e-07,
885
+ "loss": 0.2004,
886
+ "step": 1230
887
+ },
888
+ {
889
+ "epoch": 2.445759368836292,
890
+ "grad_norm": 1.2866100592193557,
891
+ "learning_rate": 9.075320573928513e-07,
892
+ "loss": 0.2026,
893
+ "step": 1240
894
+ },
895
+ {
896
+ "epoch": 2.465483234714004,
897
+ "grad_norm": 1.3169876215045113,
898
+ "learning_rate": 8.798735476517964e-07,
899
+ "loss": 0.2027,
900
+ "step": 1250
901
+ },
902
+ {
903
+ "epoch": 2.485207100591716,
904
+ "grad_norm": 1.2821201625196061,
905
+ "learning_rate": 8.531001926554134e-07,
906
+ "loss": 0.2011,
907
+ "step": 1260
908
+ },
909
+ {
910
+ "epoch": 2.504930966469428,
911
+ "grad_norm": 1.315132765819279,
912
+ "learning_rate": 8.272246645747072e-07,
913
+ "loss": 0.199,
914
+ "step": 1270
915
+ },
916
+ {
917
+ "epoch": 2.52465483234714,
918
+ "grad_norm": 1.276154658164099,
919
+ "learning_rate": 8.022592106277332e-07,
920
+ "loss": 0.2008,
921
+ "step": 1280
922
+ },
923
+ {
924
+ "epoch": 2.544378698224852,
925
+ "grad_norm": 1.2274421062761773,
926
+ "learning_rate": 7.782156472828299e-07,
927
+ "loss": 0.1998,
928
+ "step": 1290
929
+ },
930
+ {
931
+ "epoch": 2.564102564102564,
932
+ "grad_norm": 1.2435720383981574,
933
+ "learning_rate": 7.551053546657356e-07,
934
+ "loss": 0.1995,
935
+ "step": 1300
936
+ },
937
+ {
938
+ "epoch": 2.583826429980276,
939
+ "grad_norm": 1.2327909078947592,
940
+ "learning_rate": 7.329392711732278e-07,
941
+ "loss": 0.2024,
942
+ "step": 1310
943
+ },
944
+ {
945
+ "epoch": 2.603550295857988,
946
+ "grad_norm": 1.1783489485507048,
947
+ "learning_rate": 7.117278882958421e-07,
948
+ "loss": 0.2003,
949
+ "step": 1320
950
+ },
951
+ {
952
+ "epoch": 2.6232741617357003,
953
+ "grad_norm": 1.2687230261577986,
954
+ "learning_rate": 6.914812456521138e-07,
955
+ "loss": 0.2006,
956
+ "step": 1330
957
+ },
958
+ {
959
+ "epoch": 2.6429980276134124,
960
+ "grad_norm": 1.2646158919927277,
961
+ "learning_rate": 6.722089262366993e-07,
962
+ "loss": 0.1982,
963
+ "step": 1340
964
+ },
965
+ {
966
+ "epoch": 2.662721893491124,
967
+ "grad_norm": 1.2236131305338422,
968
+ "learning_rate": 6.539200518846226e-07,
969
+ "loss": 0.2001,
970
+ "step": 1350
971
+ },
972
+ {
973
+ "epoch": 2.6824457593688362,
974
+ "grad_norm": 1.2428023457207789,
975
+ "learning_rate": 6.366232789537923e-07,
976
+ "loss": 0.2048,
977
+ "step": 1360
978
+ },
979
+ {
980
+ "epoch": 2.7021696252465484,
981
+ "grad_norm": 1.2559417256017682,
982
+ "learning_rate": 6.203267942278395e-07,
983
+ "loss": 0.2012,
984
+ "step": 1370
985
+ },
986
+ {
987
+ "epoch": 2.7218934911242605,
988
+ "grad_norm": 1.2572564112264348,
989
+ "learning_rate": 6.050383110412069e-07,
990
+ "loss": 0.1994,
991
+ "step": 1380
992
+ },
993
+ {
994
+ "epoch": 2.7416173570019726,
995
+ "grad_norm": 1.1764889460619852,
996
+ "learning_rate": 5.907650656283289e-07,
997
+ "loss": 0.2002,
998
+ "step": 1390
999
+ },
1000
+ {
1001
+ "epoch": 2.7613412228796843,
1002
+ "grad_norm": 1.2804661059833917,
1003
+ "learning_rate": 5.775138136986298e-07,
1004
+ "loss": 0.2002,
1005
+ "step": 1400
1006
+ },
1007
+ {
1008
+ "epoch": 2.7810650887573964,
1009
+ "grad_norm": 1.3077263435732718,
1010
+ "learning_rate": 5.652908272389604e-07,
1011
+ "loss": 0.1995,
1012
+ "step": 1410
1013
+ },
1014
+ {
1015
+ "epoch": 2.8007889546351086,
1016
+ "grad_norm": 1.231137370296971,
1017
+ "learning_rate": 5.541018915449863e-07,
1018
+ "loss": 0.1989,
1019
+ "step": 1420
1020
+ },
1021
+ {
1022
+ "epoch": 2.8205128205128203,
1023
+ "grad_norm": 1.3443797697665705,
1024
+ "learning_rate": 5.439523024829335e-07,
1025
+ "loss": 0.1983,
1026
+ "step": 1430
1027
+ },
1028
+ {
1029
+ "epoch": 2.8402366863905324,
1030
+ "grad_norm": 1.2092638219767884,
1031
+ "learning_rate": 5.348468639829871e-07,
1032
+ "loss": 0.2007,
1033
+ "step": 1440
1034
+ },
1035
+ {
1036
+ "epoch": 2.8599605522682445,
1037
+ "grad_norm": 1.2392545674361426,
1038
+ "learning_rate": 5.267898857655307e-07,
1039
+ "loss": 0.201,
1040
+ "step": 1450
1041
+ },
1042
+ {
1043
+ "epoch": 2.8796844181459567,
1044
+ "grad_norm": 1.255507262390408,
1045
+ "learning_rate": 5.19785181301299e-07,
1046
+ "loss": 0.2008,
1047
+ "step": 1460
1048
+ },
1049
+ {
1050
+ "epoch": 2.899408284023669,
1051
+ "grad_norm": 1.2545629120536586,
1052
+ "learning_rate": 5.138360660064146e-07,
1053
+ "loss": 0.1979,
1054
+ "step": 1470
1055
+ },
1056
+ {
1057
+ "epoch": 2.9191321499013805,
1058
+ "grad_norm": 1.2279624795193589,
1059
+ "learning_rate": 5.08945355673159e-07,
1060
+ "loss": 0.201,
1061
+ "step": 1480
1062
+ },
1063
+ {
1064
+ "epoch": 2.9388560157790926,
1065
+ "grad_norm": 1.2395946923655343,
1066
+ "learning_rate": 5.05115365137222e-07,
1067
+ "loss": 0.1999,
1068
+ "step": 1490
1069
+ },
1070
+ {
1071
+ "epoch": 2.9585798816568047,
1072
+ "grad_norm": 1.2212433583596156,
1073
+ "learning_rate": 5.023479071820607e-07,
1074
+ "loss": 0.1989,
1075
+ "step": 1500
1076
+ },
1077
+ {
1078
+ "epoch": 2.978303747534517,
1079
+ "grad_norm": 1.298954785136158,
1080
+ "learning_rate": 5.006442916808849e-07,
1081
+ "loss": 0.2019,
1082
+ "step": 1510
1083
+ },
1084
+ {
1085
+ "epoch": 2.998027613412229,
1086
+ "grad_norm": 1.3586461216494594,
1087
+ "learning_rate": 5.000053249766787e-07,
1088
+ "loss": 0.1999,
1089
+ "step": 1520
1090
+ },
1091
+ {
1092
+ "epoch": 3.0,
1093
+ "eval_loss": 0.37075862288475037,
1094
+ "eval_runtime": 53.9042,
1095
+ "eval_samples_per_second": 253.041,
1096
+ "eval_steps_per_second": 1.002,
1097
+ "step": 1521
1098
+ },
1099
+ {
1100
+ "epoch": 3.0,
1101
+ "step": 1521,
1102
+ "total_flos": 2547731650314240.0,
1103
+ "train_loss": 0.2954053143131192,
1104
+ "train_runtime": 9131.3778,
1105
+ "train_samples_per_second": 85.143,
1106
+ "train_steps_per_second": 0.167
1107
+ }
1108
+ ],
1109
+ "logging_steps": 10,
1110
+ "max_steps": 1521,
1111
+ "num_input_tokens_seen": 0,
1112
+ "num_train_epochs": 3,
1113
+ "save_steps": 500,
1114
+ "stateful_callbacks": {
1115
+ "TrainerControl": {
1116
+ "args": {
1117
+ "should_epoch_stop": false,
1118
+ "should_evaluate": false,
1119
+ "should_log": false,
1120
+ "should_save": true,
1121
+ "should_training_stop": true
1122
+ },
1123
+ "attributes": {}
1124
+ }
1125
+ },
1126
+ "total_flos": 2547731650314240.0,
1127
+ "train_batch_size": 16,
1128
+ "trial_name": null,
1129
+ "trial_params": null
1130
+ }
training_eval_loss.png ADDED
training_loss.png ADDED