broadfield-dev commited on
Commit
5bb7dbb
·
verified ·
1 Parent(s): 3dcaf39

Save benchmark: stable-run-sft-step-100

Browse files
checkpoints/stable-run-sft-step-100/config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_num_labels": 1,
3
+ "activation_function": "gelu_new",
4
+ "architectures": [
5
+ "GPT2LMHeadModel"
6
+ ],
7
+ "attn_pdrop": 0.1,
8
+ "bos_token_id": 50256,
9
+ "dtype": "float32",
10
+ "embd_pdrop": 0.1,
11
+ "eos_token_id": 50256,
12
+ "id2label": {
13
+ "0": "LABEL_0"
14
+ },
15
+ "initializer_range": 0.02,
16
+ "label2id": {
17
+ "LABEL_0": 0
18
+ },
19
+ "layer_norm_epsilon": 1e-05,
20
+ "model_type": "gpt2",
21
+ "n_ctx": 1024,
22
+ "n_embd": 768,
23
+ "n_head": 12,
24
+ "n_inner": null,
25
+ "n_layer": 6,
26
+ "n_positions": 1024,
27
+ "reorder_and_upcast_attn": false,
28
+ "resid_pdrop": 0.1,
29
+ "scale_attn_by_inverse_layer_idx": false,
30
+ "scale_attn_weights": true,
31
+ "summary_activation": null,
32
+ "summary_first_dropout": 0.1,
33
+ "summary_proj_to_labels": true,
34
+ "summary_type": "cls_index",
35
+ "summary_use_proj": true,
36
+ "task_specific_params": {
37
+ "text-generation": {
38
+ "do_sample": true,
39
+ "max_length": 50
40
+ }
41
+ },
42
+ "transformers_version": "4.56.1",
43
+ "use_cache": true,
44
+ "vocab_size": 50257
45
+ }
checkpoints/stable-run-sft-step-100/generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 50256,
4
+ "eos_token_id": 50256,
5
+ "transformers_version": "4.56.1"
6
+ }
checkpoints/stable-run-sft-step-100/hyperparameters.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "repo_id": "broadfield-dev/savant_2_gsm8k_final",
3
+ "resume_from_benchmark": "",
4
+ "benchmark_name_prefix": "stable-run",
5
+ "benchmark_interval": 20,
6
+ "sft_epochs": 4,
7
+ "sft_lr": 2e-05,
8
+ "sft_warmup_proportion": 0.1,
9
+ "rl_lr": 1.5e-06,
10
+ "sft_slice": "0:2000",
11
+ "rl_slice": "4000:7473",
12
+ "context_length": 1024,
13
+ "base_model_name": "distilgpt2",
14
+ "use_scheduler": true,
15
+ "sft_batch_size": 1,
16
+ "grad_accum_steps": 16
17
+ }
checkpoints/stable-run-sft-step-100/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
checkpoints/stable-run-sft-step-100/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2cd5ff10a5598fa8851069746590cb2d6b648e3782c2874709b5c5336693430b
3
+ size 327657928
checkpoints/stable-run-sft-step-100/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7362fb0838e639c3c94970d495c8b1532bdecbd66c118c69beaa2a90130b6ee1
3
+ size 655362635
checkpoints/stable-run-sft-step-100/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6661d24316a6877d1b37b57cdb523567d557a15196cd385aec9d4dd8d06de384
3
+ size 1401
checkpoints/stable-run-sft-step-100/special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<|endoftext|>",
3
+ "eos_token": "<|endoftext|>",
4
+ "pad_token": "<|endoftext|>",
5
+ "unk_token": "<|endoftext|>"
6
+ }
checkpoints/stable-run-sft-step-100/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoints/stable-run-sft-step-100/tokenizer_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "added_tokens_decoder": {
4
+ "50256": {
5
+ "content": "<|endoftext|>",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false,
10
+ "special": true
11
+ }
12
+ },
13
+ "bos_token": "<|endoftext|>",
14
+ "clean_up_tokenization_spaces": false,
15
+ "eos_token": "<|endoftext|>",
16
+ "extra_special_tokens": {},
17
+ "model_max_length": 1024,
18
+ "pad_token": "<|endoftext|>",
19
+ "tokenizer_class": "GPT2Tokenizer",
20
+ "unk_token": "<|endoftext|>"
21
+ }
checkpoints/stable-run-sft-step-100/training_state.json ADDED
@@ -0,0 +1,1007 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "phase": "SFT",
3
+ "total_optimizer_steps": 100,
4
+ "metrics": [
5
+ {
6
+ "Step": 1,
7
+ "Metric": "SFT Loss",
8
+ "Value": 10.806696891784668
9
+ },
10
+ {
11
+ "Step": 1,
12
+ "Metric": "Learning Rate",
13
+ "Value": 4.0000000000000003e-07
14
+ },
15
+ {
16
+ "Step": 2,
17
+ "Metric": "SFT Loss",
18
+ "Value": 20.197086334228516
19
+ },
20
+ {
21
+ "Step": 2,
22
+ "Metric": "Learning Rate",
23
+ "Value": 8.000000000000001e-07
24
+ },
25
+ {
26
+ "Step": 3,
27
+ "Metric": "SFT Loss",
28
+ "Value": 13.669203758239746
29
+ },
30
+ {
31
+ "Step": 3,
32
+ "Metric": "Learning Rate",
33
+ "Value": 1.2000000000000002e-06
34
+ },
35
+ {
36
+ "Step": 4,
37
+ "Metric": "SFT Loss",
38
+ "Value": 19.52617835998535
39
+ },
40
+ {
41
+ "Step": 4,
42
+ "Metric": "Learning Rate",
43
+ "Value": 1.6000000000000001e-06
44
+ },
45
+ {
46
+ "Step": 5,
47
+ "Metric": "SFT Loss",
48
+ "Value": 14.46614933013916
49
+ },
50
+ {
51
+ "Step": 5,
52
+ "Metric": "Learning Rate",
53
+ "Value": 2.0000000000000003e-06
54
+ },
55
+ {
56
+ "Step": 6,
57
+ "Metric": "SFT Loss",
58
+ "Value": 11.709370613098145
59
+ },
60
+ {
61
+ "Step": 6,
62
+ "Metric": "Learning Rate",
63
+ "Value": 2.4000000000000003e-06
64
+ },
65
+ {
66
+ "Step": 7,
67
+ "Metric": "SFT Loss",
68
+ "Value": 14.79263687133789
69
+ },
70
+ {
71
+ "Step": 7,
72
+ "Metric": "Learning Rate",
73
+ "Value": 2.8000000000000003e-06
74
+ },
75
+ {
76
+ "Step": 8,
77
+ "Metric": "SFT Loss",
78
+ "Value": 16.564573287963867
79
+ },
80
+ {
81
+ "Step": 8,
82
+ "Metric": "Learning Rate",
83
+ "Value": 3.2000000000000003e-06
84
+ },
85
+ {
86
+ "Step": 9,
87
+ "Metric": "SFT Loss",
88
+ "Value": 16.437776565551758
89
+ },
90
+ {
91
+ "Step": 9,
92
+ "Metric": "Learning Rate",
93
+ "Value": 3.6000000000000003e-06
94
+ },
95
+ {
96
+ "Step": 10,
97
+ "Metric": "SFT Loss",
98
+ "Value": 14.215919494628906
99
+ },
100
+ {
101
+ "Step": 10,
102
+ "Metric": "Learning Rate",
103
+ "Value": 4.000000000000001e-06
104
+ },
105
+ {
106
+ "Step": 11,
107
+ "Metric": "SFT Loss",
108
+ "Value": 13.728559494018555
109
+ },
110
+ {
111
+ "Step": 11,
112
+ "Metric": "Learning Rate",
113
+ "Value": 4.4e-06
114
+ },
115
+ {
116
+ "Step": 12,
117
+ "Metric": "SFT Loss",
118
+ "Value": 17.120494842529297
119
+ },
120
+ {
121
+ "Step": 12,
122
+ "Metric": "Learning Rate",
123
+ "Value": 4.800000000000001e-06
124
+ },
125
+ {
126
+ "Step": 13,
127
+ "Metric": "SFT Loss",
128
+ "Value": 12.443873405456543
129
+ },
130
+ {
131
+ "Step": 13,
132
+ "Metric": "Learning Rate",
133
+ "Value": 5.2e-06
134
+ },
135
+ {
136
+ "Step": 14,
137
+ "Metric": "SFT Loss",
138
+ "Value": 10.322935104370117
139
+ },
140
+ {
141
+ "Step": 14,
142
+ "Metric": "Learning Rate",
143
+ "Value": 5.600000000000001e-06
144
+ },
145
+ {
146
+ "Step": 15,
147
+ "Metric": "SFT Loss",
148
+ "Value": 10.010440826416016
149
+ },
150
+ {
151
+ "Step": 15,
152
+ "Metric": "Learning Rate",
153
+ "Value": 6e-06
154
+ },
155
+ {
156
+ "Step": 16,
157
+ "Metric": "SFT Loss",
158
+ "Value": 9.959354400634766
159
+ },
160
+ {
161
+ "Step": 16,
162
+ "Metric": "Learning Rate",
163
+ "Value": 6.4000000000000006e-06
164
+ },
165
+ {
166
+ "Step": 17,
167
+ "Metric": "SFT Loss",
168
+ "Value": 9.42132568359375
169
+ },
170
+ {
171
+ "Step": 17,
172
+ "Metric": "Learning Rate",
173
+ "Value": 6.800000000000001e-06
174
+ },
175
+ {
176
+ "Step": 18,
177
+ "Metric": "SFT Loss",
178
+ "Value": 10.809995651245117
179
+ },
180
+ {
181
+ "Step": 18,
182
+ "Metric": "Learning Rate",
183
+ "Value": 7.2000000000000005e-06
184
+ },
185
+ {
186
+ "Step": 19,
187
+ "Metric": "SFT Loss",
188
+ "Value": 11.641499519348145
189
+ },
190
+ {
191
+ "Step": 19,
192
+ "Metric": "Learning Rate",
193
+ "Value": 7.600000000000001e-06
194
+ },
195
+ {
196
+ "Step": 20,
197
+ "Metric": "SFT Loss",
198
+ "Value": 8.137388229370117
199
+ },
200
+ {
201
+ "Step": 20,
202
+ "Metric": "Learning Rate",
203
+ "Value": 8.000000000000001e-06
204
+ },
205
+ {
206
+ "Step": 21,
207
+ "Metric": "SFT Loss",
208
+ "Value": 11.187762260437012
209
+ },
210
+ {
211
+ "Step": 21,
212
+ "Metric": "Learning Rate",
213
+ "Value": 8.400000000000001e-06
214
+ },
215
+ {
216
+ "Step": 22,
217
+ "Metric": "SFT Loss",
218
+ "Value": 6.696204662322998
219
+ },
220
+ {
221
+ "Step": 22,
222
+ "Metric": "Learning Rate",
223
+ "Value": 8.8e-06
224
+ },
225
+ {
226
+ "Step": 23,
227
+ "Metric": "SFT Loss",
228
+ "Value": 9.350967407226562
229
+ },
230
+ {
231
+ "Step": 23,
232
+ "Metric": "Learning Rate",
233
+ "Value": 9.200000000000002e-06
234
+ },
235
+ {
236
+ "Step": 24,
237
+ "Metric": "SFT Loss",
238
+ "Value": 7.601217269897461
239
+ },
240
+ {
241
+ "Step": 24,
242
+ "Metric": "Learning Rate",
243
+ "Value": 9.600000000000001e-06
244
+ },
245
+ {
246
+ "Step": 25,
247
+ "Metric": "SFT Loss",
248
+ "Value": 7.671919345855713
249
+ },
250
+ {
251
+ "Step": 25,
252
+ "Metric": "Learning Rate",
253
+ "Value": 1e-05
254
+ },
255
+ {
256
+ "Step": 26,
257
+ "Metric": "SFT Loss",
258
+ "Value": 5.797372341156006
259
+ },
260
+ {
261
+ "Step": 26,
262
+ "Metric": "Learning Rate",
263
+ "Value": 1.04e-05
264
+ },
265
+ {
266
+ "Step": 27,
267
+ "Metric": "SFT Loss",
268
+ "Value": 7.414616107940674
269
+ },
270
+ {
271
+ "Step": 27,
272
+ "Metric": "Learning Rate",
273
+ "Value": 1.0800000000000002e-05
274
+ },
275
+ {
276
+ "Step": 28,
277
+ "Metric": "SFT Loss",
278
+ "Value": 7.067437171936035
279
+ },
280
+ {
281
+ "Step": 28,
282
+ "Metric": "Learning Rate",
283
+ "Value": 1.1200000000000001e-05
284
+ },
285
+ {
286
+ "Step": 29,
287
+ "Metric": "SFT Loss",
288
+ "Value": 6.7167463302612305
289
+ },
290
+ {
291
+ "Step": 29,
292
+ "Metric": "Learning Rate",
293
+ "Value": 1.16e-05
294
+ },
295
+ {
296
+ "Step": 30,
297
+ "Metric": "SFT Loss",
298
+ "Value": 6.630039691925049
299
+ },
300
+ {
301
+ "Step": 30,
302
+ "Metric": "Learning Rate",
303
+ "Value": 1.2e-05
304
+ },
305
+ {
306
+ "Step": 31,
307
+ "Metric": "SFT Loss",
308
+ "Value": 6.276569366455078
309
+ },
310
+ {
311
+ "Step": 31,
312
+ "Metric": "Learning Rate",
313
+ "Value": 1.2400000000000002e-05
314
+ },
315
+ {
316
+ "Step": 32,
317
+ "Metric": "SFT Loss",
318
+ "Value": 7.236689567565918
319
+ },
320
+ {
321
+ "Step": 32,
322
+ "Metric": "Learning Rate",
323
+ "Value": 1.2800000000000001e-05
324
+ },
325
+ {
326
+ "Step": 33,
327
+ "Metric": "SFT Loss",
328
+ "Value": 6.829200267791748
329
+ },
330
+ {
331
+ "Step": 33,
332
+ "Metric": "Learning Rate",
333
+ "Value": 1.3200000000000002e-05
334
+ },
335
+ {
336
+ "Step": 34,
337
+ "Metric": "SFT Loss",
338
+ "Value": 6.348658561706543
339
+ },
340
+ {
341
+ "Step": 34,
342
+ "Metric": "Learning Rate",
343
+ "Value": 1.3600000000000002e-05
344
+ },
345
+ {
346
+ "Step": 35,
347
+ "Metric": "SFT Loss",
348
+ "Value": 6.114589691162109
349
+ },
350
+ {
351
+ "Step": 35,
352
+ "Metric": "Learning Rate",
353
+ "Value": 1.4e-05
354
+ },
355
+ {
356
+ "Step": 36,
357
+ "Metric": "SFT Loss",
358
+ "Value": 5.820864677429199
359
+ },
360
+ {
361
+ "Step": 36,
362
+ "Metric": "Learning Rate",
363
+ "Value": 1.4400000000000001e-05
364
+ },
365
+ {
366
+ "Step": 37,
367
+ "Metric": "SFT Loss",
368
+ "Value": 6.093341827392578
369
+ },
370
+ {
371
+ "Step": 37,
372
+ "Metric": "Learning Rate",
373
+ "Value": 1.48e-05
374
+ },
375
+ {
376
+ "Step": 38,
377
+ "Metric": "SFT Loss",
378
+ "Value": 6.195661544799805
379
+ },
380
+ {
381
+ "Step": 38,
382
+ "Metric": "Learning Rate",
383
+ "Value": 1.5200000000000002e-05
384
+ },
385
+ {
386
+ "Step": 39,
387
+ "Metric": "SFT Loss",
388
+ "Value": 7.604869365692139
389
+ },
390
+ {
391
+ "Step": 39,
392
+ "Metric": "Learning Rate",
393
+ "Value": 1.5600000000000003e-05
394
+ },
395
+ {
396
+ "Step": 40,
397
+ "Metric": "SFT Loss",
398
+ "Value": 5.295370101928711
399
+ },
400
+ {
401
+ "Step": 40,
402
+ "Metric": "Learning Rate",
403
+ "Value": 1.6000000000000003e-05
404
+ },
405
+ {
406
+ "Step": 41,
407
+ "Metric": "SFT Loss",
408
+ "Value": 6.263326644897461
409
+ },
410
+ {
411
+ "Step": 41,
412
+ "Metric": "Learning Rate",
413
+ "Value": 1.64e-05
414
+ },
415
+ {
416
+ "Step": 42,
417
+ "Metric": "SFT Loss",
418
+ "Value": 6.123704433441162
419
+ },
420
+ {
421
+ "Step": 42,
422
+ "Metric": "Learning Rate",
423
+ "Value": 1.6800000000000002e-05
424
+ },
425
+ {
426
+ "Step": 43,
427
+ "Metric": "SFT Loss",
428
+ "Value": 6.104567050933838
429
+ },
430
+ {
431
+ "Step": 43,
432
+ "Metric": "Learning Rate",
433
+ "Value": 1.72e-05
434
+ },
435
+ {
436
+ "Step": 44,
437
+ "Metric": "SFT Loss",
438
+ "Value": 6.757107257843018
439
+ },
440
+ {
441
+ "Step": 44,
442
+ "Metric": "Learning Rate",
443
+ "Value": 1.76e-05
444
+ },
445
+ {
446
+ "Step": 45,
447
+ "Metric": "SFT Loss",
448
+ "Value": 6.533822059631348
449
+ },
450
+ {
451
+ "Step": 45,
452
+ "Metric": "Learning Rate",
453
+ "Value": 1.8e-05
454
+ },
455
+ {
456
+ "Step": 46,
457
+ "Metric": "SFT Loss",
458
+ "Value": 6.507102012634277
459
+ },
460
+ {
461
+ "Step": 46,
462
+ "Metric": "Learning Rate",
463
+ "Value": 1.8400000000000003e-05
464
+ },
465
+ {
466
+ "Step": 47,
467
+ "Metric": "SFT Loss",
468
+ "Value": 6.377279758453369
469
+ },
470
+ {
471
+ "Step": 47,
472
+ "Metric": "Learning Rate",
473
+ "Value": 1.88e-05
474
+ },
475
+ {
476
+ "Step": 48,
477
+ "Metric": "SFT Loss",
478
+ "Value": 6.2640581130981445
479
+ },
480
+ {
481
+ "Step": 48,
482
+ "Metric": "Learning Rate",
483
+ "Value": 1.9200000000000003e-05
484
+ },
485
+ {
486
+ "Step": 49,
487
+ "Metric": "SFT Loss",
488
+ "Value": 5.53227424621582
489
+ },
490
+ {
491
+ "Step": 49,
492
+ "Metric": "Learning Rate",
493
+ "Value": 1.9600000000000002e-05
494
+ },
495
+ {
496
+ "Step": 50,
497
+ "Metric": "SFT Loss",
498
+ "Value": 6.743251800537109
499
+ },
500
+ {
501
+ "Step": 50,
502
+ "Metric": "Learning Rate",
503
+ "Value": 2e-05
504
+ },
505
+ {
506
+ "Step": 51,
507
+ "Metric": "SFT Loss",
508
+ "Value": 5.5736985206604
509
+ },
510
+ {
511
+ "Step": 51,
512
+ "Metric": "Learning Rate",
513
+ "Value": 1.9955555555555557e-05
514
+ },
515
+ {
516
+ "Step": 52,
517
+ "Metric": "SFT Loss",
518
+ "Value": 5.360650539398193
519
+ },
520
+ {
521
+ "Step": 52,
522
+ "Metric": "Learning Rate",
523
+ "Value": 1.9911111111111112e-05
524
+ },
525
+ {
526
+ "Step": 53,
527
+ "Metric": "SFT Loss",
528
+ "Value": 5.423285484313965
529
+ },
530
+ {
531
+ "Step": 53,
532
+ "Metric": "Learning Rate",
533
+ "Value": 1.9866666666666667e-05
534
+ },
535
+ {
536
+ "Step": 54,
537
+ "Metric": "SFT Loss",
538
+ "Value": 5.108491897583008
539
+ },
540
+ {
541
+ "Step": 54,
542
+ "Metric": "Learning Rate",
543
+ "Value": 1.9822222222222226e-05
544
+ },
545
+ {
546
+ "Step": 55,
547
+ "Metric": "SFT Loss",
548
+ "Value": 4.632509231567383
549
+ },
550
+ {
551
+ "Step": 55,
552
+ "Metric": "Learning Rate",
553
+ "Value": 1.977777777777778e-05
554
+ },
555
+ {
556
+ "Step": 56,
557
+ "Metric": "SFT Loss",
558
+ "Value": 4.9901933670043945
559
+ },
560
+ {
561
+ "Step": 56,
562
+ "Metric": "Learning Rate",
563
+ "Value": 1.9733333333333336e-05
564
+ },
565
+ {
566
+ "Step": 57,
567
+ "Metric": "SFT Loss",
568
+ "Value": 4.968444347381592
569
+ },
570
+ {
571
+ "Step": 57,
572
+ "Metric": "Learning Rate",
573
+ "Value": 1.968888888888889e-05
574
+ },
575
+ {
576
+ "Step": 58,
577
+ "Metric": "SFT Loss",
578
+ "Value": 5.028273582458496
579
+ },
580
+ {
581
+ "Step": 58,
582
+ "Metric": "Learning Rate",
583
+ "Value": 1.9644444444444447e-05
584
+ },
585
+ {
586
+ "Step": 59,
587
+ "Metric": "SFT Loss",
588
+ "Value": 5.093855381011963
589
+ },
590
+ {
591
+ "Step": 59,
592
+ "Metric": "Learning Rate",
593
+ "Value": 1.9600000000000002e-05
594
+ },
595
+ {
596
+ "Step": 60,
597
+ "Metric": "SFT Loss",
598
+ "Value": 4.720518112182617
599
+ },
600
+ {
601
+ "Step": 60,
602
+ "Metric": "Learning Rate",
603
+ "Value": 1.9555555555555557e-05
604
+ },
605
+ {
606
+ "Step": 61,
607
+ "Metric": "SFT Loss",
608
+ "Value": 4.872599124908447
609
+ },
610
+ {
611
+ "Step": 61,
612
+ "Metric": "Learning Rate",
613
+ "Value": 1.9511111111111113e-05
614
+ },
615
+ {
616
+ "Step": 62,
617
+ "Metric": "SFT Loss",
618
+ "Value": 4.398301124572754
619
+ },
620
+ {
621
+ "Step": 62,
622
+ "Metric": "Learning Rate",
623
+ "Value": 1.9466666666666668e-05
624
+ },
625
+ {
626
+ "Step": 63,
627
+ "Metric": "SFT Loss",
628
+ "Value": 5.259819030761719
629
+ },
630
+ {
631
+ "Step": 63,
632
+ "Metric": "Learning Rate",
633
+ "Value": 1.9422222222222223e-05
634
+ },
635
+ {
636
+ "Step": 64,
637
+ "Metric": "SFT Loss",
638
+ "Value": 5.536286354064941
639
+ },
640
+ {
641
+ "Step": 64,
642
+ "Metric": "Learning Rate",
643
+ "Value": 1.9377777777777778e-05
644
+ },
645
+ {
646
+ "Step": 65,
647
+ "Metric": "SFT Loss",
648
+ "Value": 4.998850345611572
649
+ },
650
+ {
651
+ "Step": 65,
652
+ "Metric": "Learning Rate",
653
+ "Value": 1.9333333333333333e-05
654
+ },
655
+ {
656
+ "Step": 66,
657
+ "Metric": "SFT Loss",
658
+ "Value": 4.723838806152344
659
+ },
660
+ {
661
+ "Step": 66,
662
+ "Metric": "Learning Rate",
663
+ "Value": 1.928888888888889e-05
664
+ },
665
+ {
666
+ "Step": 67,
667
+ "Metric": "SFT Loss",
668
+ "Value": 5.087211608886719
669
+ },
670
+ {
671
+ "Step": 67,
672
+ "Metric": "Learning Rate",
673
+ "Value": 1.9244444444444444e-05
674
+ },
675
+ {
676
+ "Step": 68,
677
+ "Metric": "SFT Loss",
678
+ "Value": 4.477473258972168
679
+ },
680
+ {
681
+ "Step": 68,
682
+ "Metric": "Learning Rate",
683
+ "Value": 1.9200000000000003e-05
684
+ },
685
+ {
686
+ "Step": 69,
687
+ "Metric": "SFT Loss",
688
+ "Value": 4.780407905578613
689
+ },
690
+ {
691
+ "Step": 69,
692
+ "Metric": "Learning Rate",
693
+ "Value": 1.9155555555555558e-05
694
+ },
695
+ {
696
+ "Step": 70,
697
+ "Metric": "SFT Loss",
698
+ "Value": 4.482248783111572
699
+ },
700
+ {
701
+ "Step": 70,
702
+ "Metric": "Learning Rate",
703
+ "Value": 1.9111111111111113e-05
704
+ },
705
+ {
706
+ "Step": 71,
707
+ "Metric": "SFT Loss",
708
+ "Value": 4.773695468902588
709
+ },
710
+ {
711
+ "Step": 71,
712
+ "Metric": "Learning Rate",
713
+ "Value": 1.9066666666666668e-05
714
+ },
715
+ {
716
+ "Step": 72,
717
+ "Metric": "SFT Loss",
718
+ "Value": 4.399933815002441
719
+ },
720
+ {
721
+ "Step": 72,
722
+ "Metric": "Learning Rate",
723
+ "Value": 1.9022222222222223e-05
724
+ },
725
+ {
726
+ "Step": 73,
727
+ "Metric": "SFT Loss",
728
+ "Value": 4.540095329284668
729
+ },
730
+ {
731
+ "Step": 73,
732
+ "Metric": "Learning Rate",
733
+ "Value": 1.897777777777778e-05
734
+ },
735
+ {
736
+ "Step": 74,
737
+ "Metric": "SFT Loss",
738
+ "Value": 4.6496663093566895
739
+ },
740
+ {
741
+ "Step": 74,
742
+ "Metric": "Learning Rate",
743
+ "Value": 1.8933333333333334e-05
744
+ },
745
+ {
746
+ "Step": 75,
747
+ "Metric": "SFT Loss",
748
+ "Value": 5.276064872741699
749
+ },
750
+ {
751
+ "Step": 75,
752
+ "Metric": "Learning Rate",
753
+ "Value": 1.888888888888889e-05
754
+ },
755
+ {
756
+ "Step": 76,
757
+ "Metric": "SFT Loss",
758
+ "Value": 4.346099853515625
759
+ },
760
+ {
761
+ "Step": 76,
762
+ "Metric": "Learning Rate",
763
+ "Value": 1.8844444444444444e-05
764
+ },
765
+ {
766
+ "Step": 77,
767
+ "Metric": "SFT Loss",
768
+ "Value": 4.5329694747924805
769
+ },
770
+ {
771
+ "Step": 77,
772
+ "Metric": "Learning Rate",
773
+ "Value": 1.88e-05
774
+ },
775
+ {
776
+ "Step": 78,
777
+ "Metric": "SFT Loss",
778
+ "Value": 4.541317939758301
779
+ },
780
+ {
781
+ "Step": 78,
782
+ "Metric": "Learning Rate",
783
+ "Value": 1.8755555555555558e-05
784
+ },
785
+ {
786
+ "Step": 79,
787
+ "Metric": "SFT Loss",
788
+ "Value": 4.832979202270508
789
+ },
790
+ {
791
+ "Step": 79,
792
+ "Metric": "Learning Rate",
793
+ "Value": 1.8711111111111113e-05
794
+ },
795
+ {
796
+ "Step": 80,
797
+ "Metric": "SFT Loss",
798
+ "Value": 4.471445560455322
799
+ },
800
+ {
801
+ "Step": 80,
802
+ "Metric": "Learning Rate",
803
+ "Value": 1.866666666666667e-05
804
+ },
805
+ {
806
+ "Step": 81,
807
+ "Metric": "SFT Loss",
808
+ "Value": 5.19096040725708
809
+ },
810
+ {
811
+ "Step": 81,
812
+ "Metric": "Learning Rate",
813
+ "Value": 1.8622222222222224e-05
814
+ },
815
+ {
816
+ "Step": 82,
817
+ "Metric": "SFT Loss",
818
+ "Value": 4.298245906829834
819
+ },
820
+ {
821
+ "Step": 82,
822
+ "Metric": "Learning Rate",
823
+ "Value": 1.857777777777778e-05
824
+ },
825
+ {
826
+ "Step": 83,
827
+ "Metric": "SFT Loss",
828
+ "Value": 5.057285785675049
829
+ },
830
+ {
831
+ "Step": 83,
832
+ "Metric": "Learning Rate",
833
+ "Value": 1.8533333333333334e-05
834
+ },
835
+ {
836
+ "Step": 84,
837
+ "Metric": "SFT Loss",
838
+ "Value": 4.01115083694458
839
+ },
840
+ {
841
+ "Step": 84,
842
+ "Metric": "Learning Rate",
843
+ "Value": 1.848888888888889e-05
844
+ },
845
+ {
846
+ "Step": 85,
847
+ "Metric": "SFT Loss",
848
+ "Value": 4.00934362411499
849
+ },
850
+ {
851
+ "Step": 85,
852
+ "Metric": "Learning Rate",
853
+ "Value": 1.8444444444444448e-05
854
+ },
855
+ {
856
+ "Step": 86,
857
+ "Metric": "SFT Loss",
858
+ "Value": 4.560959339141846
859
+ },
860
+ {
861
+ "Step": 86,
862
+ "Metric": "Learning Rate",
863
+ "Value": 1.8400000000000003e-05
864
+ },
865
+ {
866
+ "Step": 87,
867
+ "Metric": "SFT Loss",
868
+ "Value": 3.884817361831665
869
+ },
870
+ {
871
+ "Step": 87,
872
+ "Metric": "Learning Rate",
873
+ "Value": 1.835555555555556e-05
874
+ },
875
+ {
876
+ "Step": 88,
877
+ "Metric": "SFT Loss",
878
+ "Value": 4.446163177490234
879
+ },
880
+ {
881
+ "Step": 88,
882
+ "Metric": "Learning Rate",
883
+ "Value": 1.8311111111111114e-05
884
+ },
885
+ {
886
+ "Step": 89,
887
+ "Metric": "SFT Loss",
888
+ "Value": 4.543856143951416
889
+ },
890
+ {
891
+ "Step": 89,
892
+ "Metric": "Learning Rate",
893
+ "Value": 1.826666666666667e-05
894
+ },
895
+ {
896
+ "Step": 90,
897
+ "Metric": "SFT Loss",
898
+ "Value": 4.007848262786865
899
+ },
900
+ {
901
+ "Step": 90,
902
+ "Metric": "Learning Rate",
903
+ "Value": 1.8222222222222224e-05
904
+ },
905
+ {
906
+ "Step": 91,
907
+ "Metric": "SFT Loss",
908
+ "Value": 4.391502857208252
909
+ },
910
+ {
911
+ "Step": 91,
912
+ "Metric": "Learning Rate",
913
+ "Value": 1.817777777777778e-05
914
+ },
915
+ {
916
+ "Step": 92,
917
+ "Metric": "SFT Loss",
918
+ "Value": 4.319259166717529
919
+ },
920
+ {
921
+ "Step": 92,
922
+ "Metric": "Learning Rate",
923
+ "Value": 1.8133333333333335e-05
924
+ },
925
+ {
926
+ "Step": 93,
927
+ "Metric": "SFT Loss",
928
+ "Value": 4.551959991455078
929
+ },
930
+ {
931
+ "Step": 93,
932
+ "Metric": "Learning Rate",
933
+ "Value": 1.808888888888889e-05
934
+ },
935
+ {
936
+ "Step": 94,
937
+ "Metric": "SFT Loss",
938
+ "Value": 4.233450412750244
939
+ },
940
+ {
941
+ "Step": 94,
942
+ "Metric": "Learning Rate",
943
+ "Value": 1.8044444444444445e-05
944
+ },
945
+ {
946
+ "Step": 95,
947
+ "Metric": "SFT Loss",
948
+ "Value": 4.298363208770752
949
+ },
950
+ {
951
+ "Step": 95,
952
+ "Metric": "Learning Rate",
953
+ "Value": 1.8e-05
954
+ },
955
+ {
956
+ "Step": 96,
957
+ "Metric": "SFT Loss",
958
+ "Value": 4.241702079772949
959
+ },
960
+ {
961
+ "Step": 96,
962
+ "Metric": "Learning Rate",
963
+ "Value": 1.7955555555555556e-05
964
+ },
965
+ {
966
+ "Step": 97,
967
+ "Metric": "SFT Loss",
968
+ "Value": 3.702441692352295
969
+ },
970
+ {
971
+ "Step": 97,
972
+ "Metric": "Learning Rate",
973
+ "Value": 1.791111111111111e-05
974
+ },
975
+ {
976
+ "Step": 98,
977
+ "Metric": "SFT Loss",
978
+ "Value": 4.456193923950195
979
+ },
980
+ {
981
+ "Step": 98,
982
+ "Metric": "Learning Rate",
983
+ "Value": 1.7866666666666666e-05
984
+ },
985
+ {
986
+ "Step": 99,
987
+ "Metric": "SFT Loss",
988
+ "Value": 3.991964817047119
989
+ },
990
+ {
991
+ "Step": 99,
992
+ "Metric": "Learning Rate",
993
+ "Value": 1.782222222222222e-05
994
+ },
995
+ {
996
+ "Step": 100,
997
+ "Metric": "SFT Loss",
998
+ "Value": 3.4140963554382324
999
+ },
1000
+ {
1001
+ "Step": 100,
1002
+ "Metric": "Learning Rate",
1003
+ "Value": 1.7777777777777777e-05
1004
+ }
1005
+ ],
1006
+ "sft_complete": false
1007
+ }
checkpoints/stable-run-sft-step-100/vocab.json ADDED
The diff for this file is too large to render. See raw diff