Model save
Browse files- README.md +1 -1
- all_results.json +3 -3
- model-00001-of-00003.safetensors +1 -1
- model-00002-of-00003.safetensors +1 -1
- model-00003-of-00003.safetensors +1 -1
- runs/Jun30_16-42-14_n136-100-194/events.out.tfevents.1719737283.n136-100-194.302982.0 +2 -2
- train_results.json +3 -3
- trainer_state.json +453 -453
- training_args.bin +1 -1
README.md
CHANGED
@@ -32,7 +32,7 @@ More information needed
|
|
32 |
### Training hyperparameters
|
33 |
|
34 |
The following hyperparameters were used during training:
|
35 |
-
- learning_rate:
|
36 |
- train_batch_size: 4
|
37 |
- eval_batch_size: 4
|
38 |
- seed: 42
|
|
|
32 |
### Training hyperparameters
|
33 |
|
34 |
The following hyperparameters were used during training:
|
35 |
+
- learning_rate: 4e-07
|
36 |
- train_batch_size: 4
|
37 |
- eval_batch_size: 4
|
38 |
- seed: 42
|
all_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
-
"train_loss":
|
4 |
-
"train_runtime":
|
5 |
"train_samples": 52922,
|
6 |
-
"train_samples_per_second": 8.
|
7 |
"train_steps_per_second": 0.065
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
+
"train_loss": 3.8421780889894426,
|
4 |
+
"train_runtime": 6381.4933,
|
5 |
"train_samples": 52922,
|
6 |
+
"train_samples_per_second": 8.293,
|
7 |
"train_steps_per_second": 0.065
|
8 |
}
|
model-00001-of-00003.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4943162336
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4e88c6ac9b4127cd43c95069301c9c933118b1525797db81fc6d80f544cf7cec
|
3 |
size 4943162336
|
model-00002-of-00003.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4999819336
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bd64f1edcaa32df3d1b32cd35af92f119cd07ea2ba7ebf00b6828eed52da98f0
|
3 |
size 4999819336
|
model-00003-of-00003.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4540516344
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:87c91c34961d0019bf71d071d06cfd82742211fd4d3379b61e5bc9cad550096c
|
3 |
size 4540516344
|
runs/Jun30_16-42-14_n136-100-194/events.out.tfevents.1719737283.n136-100-194.302982.0
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ae7c4645315dc9db21552bd2b3b636781f6717efc0255ff69d2ebf3333897606
|
3 |
+
size 33985
|
train_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
-
"train_loss":
|
4 |
-
"train_runtime":
|
5 |
"train_samples": 52922,
|
6 |
-
"train_samples_per_second": 8.
|
7 |
"train_steps_per_second": 0.065
|
8 |
}
|
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
+
"train_loss": 3.8421780889894426,
|
4 |
+
"train_runtime": 6381.4933,
|
5 |
"train_samples": 52922,
|
6 |
+
"train_samples_per_second": 8.293,
|
7 |
"train_steps_per_second": 0.065
|
8 |
}
|
trainer_state.json
CHANGED
@@ -10,13 +10,13 @@
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.0,
|
13 |
-
"grad_norm":
|
14 |
-
"learning_rate":
|
15 |
"logits/chosen": -2.7005977630615234,
|
16 |
"logits/rejected": -2.6288318634033203,
|
17 |
"logps/chosen": -1.1158788204193115,
|
18 |
"logps/rejected": -1.1333446502685547,
|
19 |
-
"loss": 0.
|
20 |
"rewards/accuracies": 0.0,
|
21 |
"rewards/chosen": 0.0,
|
22 |
"rewards/margins": 0.0,
|
@@ -25,626 +25,626 @@
|
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.02,
|
28 |
-
"grad_norm":
|
29 |
-
"learning_rate":
|
30 |
-
"logits/chosen": -2.
|
31 |
-
"logits/rejected": -2.
|
32 |
-
"logps/chosen": -0.
|
33 |
-
"logps/rejected": -0.
|
34 |
-
"loss": 0.
|
35 |
-
"rewards/accuracies": 0.
|
36 |
-
"rewards/chosen": 0.
|
37 |
-
"rewards/margins": 0.
|
38 |
-
"rewards/rejected": -0.
|
39 |
"step": 10
|
40 |
},
|
41 |
{
|
42 |
"epoch": 0.05,
|
43 |
-
"grad_norm":
|
44 |
-
"learning_rate":
|
45 |
-
"logits/chosen": -2.
|
46 |
-
"logits/rejected": -2.
|
47 |
-
"logps/chosen": -0.
|
48 |
-
"logps/rejected": -1.
|
49 |
-
"loss":
|
50 |
-
"rewards/accuracies": 0.
|
51 |
-
"rewards/chosen": -0.
|
52 |
-
"rewards/margins": 0.
|
53 |
-
"rewards/rejected": -
|
54 |
"step": 20
|
55 |
},
|
56 |
{
|
57 |
"epoch": 0.07,
|
58 |
-
"grad_norm":
|
59 |
-
"learning_rate":
|
60 |
-
"logits/chosen": -2.
|
61 |
-
"logits/rejected": -2.
|
62 |
-
"logps/chosen": -0.
|
63 |
-
"logps/rejected": -0.
|
64 |
-
"loss":
|
65 |
-
"rewards/accuracies": 0.
|
66 |
-
"rewards/chosen": -
|
67 |
-
"rewards/margins": 0.
|
68 |
-
"rewards/rejected": -
|
69 |
"step": 30
|
70 |
},
|
71 |
{
|
72 |
"epoch": 0.1,
|
73 |
-
"grad_norm":
|
74 |
-
"learning_rate":
|
75 |
-
"logits/chosen": -2.
|
76 |
-
"logits/rejected": -2.
|
77 |
-
"logps/chosen": -0.
|
78 |
-
"logps/rejected": -0.
|
79 |
-
"loss":
|
80 |
-
"rewards/accuracies": 0.
|
81 |
-
"rewards/chosen": -0.
|
82 |
-
"rewards/margins":
|
83 |
-
"rewards/rejected": -
|
84 |
"step": 40
|
85 |
},
|
86 |
{
|
87 |
"epoch": 0.12,
|
88 |
-
"grad_norm":
|
89 |
-
"learning_rate":
|
90 |
-
"logits/chosen": -2.
|
91 |
-
"logits/rejected": -2.
|
92 |
-
"logps/chosen": -0.
|
93 |
-
"logps/rejected": -0.
|
94 |
-
"loss":
|
95 |
-
"rewards/accuracies": 0.
|
96 |
-
"rewards/chosen": -0.
|
97 |
-
"rewards/margins":
|
98 |
-
"rewards/rejected": -
|
99 |
"step": 50
|
100 |
},
|
101 |
{
|
102 |
"epoch": 0.15,
|
103 |
-
"grad_norm":
|
104 |
-
"learning_rate":
|
105 |
-
"logits/chosen": -2.
|
106 |
-
"logits/rejected": -2.
|
107 |
-
"logps/chosen": -0.
|
108 |
-
"logps/rejected": -0.
|
109 |
-
"loss":
|
110 |
-
"rewards/accuracies": 0.
|
111 |
-
"rewards/chosen":
|
112 |
-
"rewards/margins":
|
113 |
-
"rewards/rejected": -
|
114 |
"step": 60
|
115 |
},
|
116 |
{
|
117 |
"epoch": 0.17,
|
118 |
-
"grad_norm":
|
119 |
-
"learning_rate":
|
120 |
-
"logits/chosen": -2.
|
121 |
-
"logits/rejected": -2.
|
122 |
-
"logps/chosen": -0.
|
123 |
-
"logps/rejected": -0.
|
124 |
-
"loss":
|
125 |
-
"rewards/accuracies": 0.
|
126 |
-
"rewards/chosen":
|
127 |
-
"rewards/margins":
|
128 |
-
"rewards/rejected": -
|
129 |
"step": 70
|
130 |
},
|
131 |
{
|
132 |
"epoch": 0.19,
|
133 |
-
"grad_norm":
|
134 |
-
"learning_rate":
|
135 |
-
"logits/chosen": -2.
|
136 |
-
"logits/rejected": -2.
|
137 |
-
"logps/chosen": -0.
|
138 |
-
"logps/rejected": -
|
139 |
-
"loss":
|
140 |
-
"rewards/accuracies": 0.
|
141 |
-
"rewards/chosen": -
|
142 |
-
"rewards/margins":
|
143 |
-
"rewards/rejected": -
|
144 |
"step": 80
|
145 |
},
|
146 |
{
|
147 |
"epoch": 0.22,
|
148 |
-
"grad_norm":
|
149 |
-
"learning_rate":
|
150 |
-
"logits/chosen": -2.
|
151 |
-
"logits/rejected": -2.
|
152 |
-
"logps/chosen": -1.
|
153 |
-
"logps/rejected": -1.
|
154 |
-
"loss":
|
155 |
-
"rewards/accuracies": 0.
|
156 |
-
"rewards/chosen":
|
157 |
-
"rewards/margins":
|
158 |
-
"rewards/rejected": -
|
159 |
"step": 90
|
160 |
},
|
161 |
{
|
162 |
"epoch": 0.24,
|
163 |
-
"grad_norm":
|
164 |
-
"learning_rate":
|
165 |
-
"logits/chosen": -2.
|
166 |
-
"logits/rejected": -2.
|
167 |
-
"logps/chosen": -0.
|
168 |
-
"logps/rejected": -0.
|
169 |
-
"loss":
|
170 |
"rewards/accuracies": 0.8062499761581421,
|
171 |
-
"rewards/chosen": -
|
172 |
-
"rewards/margins":
|
173 |
-
"rewards/rejected": -
|
174 |
"step": 100
|
175 |
},
|
176 |
{
|
177 |
"epoch": 0.27,
|
178 |
-
"grad_norm":
|
179 |
-
"learning_rate":
|
180 |
-
"logits/chosen": -2.
|
181 |
-
"logits/rejected": -2.
|
182 |
-
"logps/chosen": -0.
|
183 |
-
"logps/rejected": -0.
|
184 |
-
"loss":
|
185 |
-
"rewards/accuracies": 0.
|
186 |
-
"rewards/chosen":
|
187 |
-
"rewards/margins":
|
188 |
-
"rewards/rejected": -1.
|
189 |
"step": 110
|
190 |
},
|
191 |
{
|
192 |
"epoch": 0.29,
|
193 |
-
"grad_norm":
|
194 |
-
"learning_rate":
|
195 |
-
"logits/chosen": -2.
|
196 |
-
"logits/rejected": -2.
|
197 |
-
"logps/chosen": -0.
|
198 |
-
"logps/rejected": -
|
199 |
-
"loss":
|
200 |
-
"rewards/accuracies": 0.
|
201 |
-
"rewards/chosen": -
|
202 |
-
"rewards/margins":
|
203 |
-
"rewards/rejected": -
|
204 |
"step": 120
|
205 |
},
|
206 |
{
|
207 |
"epoch": 0.31,
|
208 |
-
"grad_norm":
|
209 |
-
"learning_rate":
|
210 |
-
"logits/chosen": -2.
|
211 |
-
"logits/rejected": -2.
|
212 |
-
"logps/chosen": -0.
|
213 |
-
"logps/rejected": -0.
|
214 |
-
"loss":
|
215 |
-
"rewards/accuracies": 0.
|
216 |
-
"rewards/chosen":
|
217 |
-
"rewards/margins":
|
218 |
-
"rewards/rejected": -
|
219 |
"step": 130
|
220 |
},
|
221 |
{
|
222 |
"epoch": 0.34,
|
223 |
-
"grad_norm":
|
224 |
-
"learning_rate":
|
225 |
-
"logits/chosen": -2.
|
226 |
-
"logits/rejected": -2.
|
227 |
-
"logps/chosen": -0.
|
228 |
-
"logps/rejected": -0.
|
229 |
-
"loss":
|
230 |
-
"rewards/accuracies": 0.
|
231 |
-
"rewards/chosen": -
|
232 |
-
"rewards/margins":
|
233 |
-
"rewards/rejected": -
|
234 |
"step": 140
|
235 |
},
|
236 |
{
|
237 |
"epoch": 0.36,
|
238 |
-
"grad_norm":
|
239 |
-
"learning_rate":
|
240 |
-
"logits/chosen": -2.
|
241 |
-
"logits/rejected": -2.
|
242 |
-
"logps/chosen": -0.
|
243 |
-
"logps/rejected": -0.
|
244 |
-
"loss":
|
245 |
-
"rewards/accuracies": 0.
|
246 |
-
"rewards/chosen":
|
247 |
-
"rewards/margins":
|
248 |
-
"rewards/rejected": -
|
249 |
"step": 150
|
250 |
},
|
251 |
{
|
252 |
"epoch": 0.39,
|
253 |
-
"grad_norm":
|
254 |
-
"learning_rate":
|
255 |
-
"logits/chosen": -2.
|
256 |
-
"logits/rejected": -2.
|
257 |
-
"logps/chosen": -0.
|
258 |
-
"logps/rejected": -0.
|
259 |
-
"loss":
|
260 |
-
"rewards/accuracies": 0.
|
261 |
-
"rewards/chosen": -0.
|
262 |
-
"rewards/margins":
|
263 |
-
"rewards/rejected": -
|
264 |
"step": 160
|
265 |
},
|
266 |
{
|
267 |
"epoch": 0.41,
|
268 |
-
"grad_norm":
|
269 |
-
"learning_rate":
|
270 |
-
"logits/chosen": -2.
|
271 |
-
"logits/rejected": -2.
|
272 |
-
"logps/chosen": -0.
|
273 |
-
"logps/rejected": -0.
|
274 |
-
"loss":
|
275 |
"rewards/accuracies": 0.800000011920929,
|
276 |
-
"rewards/chosen": -
|
277 |
-
"rewards/margins":
|
278 |
-
"rewards/rejected": -
|
279 |
"step": 170
|
280 |
},
|
281 |
{
|
282 |
"epoch": 0.44,
|
283 |
-
"grad_norm":
|
284 |
-
"learning_rate":
|
285 |
-
"logits/chosen": -2.
|
286 |
-
"logits/rejected": -2.
|
287 |
-
"logps/chosen": -0.
|
288 |
-
"logps/rejected": -0.
|
289 |
-
"loss":
|
290 |
-
"rewards/accuracies": 0.
|
291 |
-
"rewards/chosen":
|
292 |
-
"rewards/margins":
|
293 |
-
"rewards/rejected": -
|
294 |
"step": 180
|
295 |
},
|
296 |
{
|
297 |
"epoch": 0.46,
|
298 |
-
"grad_norm":
|
299 |
-
"learning_rate":
|
300 |
-
"logits/chosen": -2.
|
301 |
-
"logits/rejected": -2.
|
302 |
-
"logps/chosen": -0.
|
303 |
-
"logps/rejected": -1.
|
304 |
-
"loss":
|
305 |
-
"rewards/accuracies": 0.
|
306 |
-
"rewards/chosen": -
|
307 |
-
"rewards/margins":
|
308 |
-
"rewards/rejected": -
|
309 |
"step": 190
|
310 |
},
|
311 |
{
|
312 |
"epoch": 0.48,
|
313 |
-
"grad_norm":
|
314 |
-
"learning_rate":
|
315 |
-
"logits/chosen": -2.
|
316 |
-
"logits/rejected": -2.
|
317 |
-
"logps/chosen": -0.
|
318 |
-
"logps/rejected": -0.
|
319 |
-
"loss":
|
320 |
-
"rewards/accuracies": 0.
|
321 |
-
"rewards/chosen":
|
322 |
-
"rewards/margins":
|
323 |
-
"rewards/rejected": -
|
324 |
"step": 200
|
325 |
},
|
326 |
{
|
327 |
"epoch": 0.51,
|
328 |
-
"grad_norm":
|
329 |
-
"learning_rate":
|
330 |
-
"logits/chosen": -2.
|
331 |
-
"logits/rejected": -2.
|
332 |
-
"logps/chosen": -1.
|
333 |
-
"logps/rejected": -0.
|
334 |
-
"loss":
|
335 |
-
"rewards/accuracies": 0.
|
336 |
-
"rewards/chosen":
|
337 |
-
"rewards/margins":
|
338 |
-
"rewards/rejected": -
|
339 |
"step": 210
|
340 |
},
|
341 |
{
|
342 |
"epoch": 0.53,
|
343 |
-
"grad_norm":
|
344 |
-
"learning_rate":
|
345 |
-
"logits/chosen": -2.
|
346 |
-
"logits/rejected": -2.
|
347 |
-
"logps/chosen": -0.
|
348 |
-
"logps/rejected": -0.
|
349 |
-
"loss":
|
350 |
-
"rewards/accuracies": 0.
|
351 |
-
"rewards/chosen":
|
352 |
-
"rewards/margins":
|
353 |
-
"rewards/rejected": -
|
354 |
"step": 220
|
355 |
},
|
356 |
{
|
357 |
"epoch": 0.56,
|
358 |
-
"grad_norm":
|
359 |
-
"learning_rate":
|
360 |
-
"logits/chosen": -2.
|
361 |
-
"logits/rejected": -2.
|
362 |
-
"logps/chosen": -0.
|
363 |
-
"logps/rejected": -0.
|
364 |
-
"loss":
|
365 |
-
"rewards/accuracies": 0.
|
366 |
-
"rewards/chosen": -
|
367 |
-
"rewards/margins":
|
368 |
-
"rewards/rejected": -
|
369 |
"step": 230
|
370 |
},
|
371 |
{
|
372 |
"epoch": 0.58,
|
373 |
-
"grad_norm":
|
374 |
-
"learning_rate":
|
375 |
-
"logits/chosen": -2.
|
376 |
-
"logits/rejected": -2.
|
377 |
-
"logps/chosen": -0.
|
378 |
-
"logps/rejected": -0.
|
379 |
-
"loss":
|
380 |
-
"rewards/accuracies": 0.
|
381 |
-
"rewards/chosen": 0.
|
382 |
-
"rewards/margins":
|
383 |
-
"rewards/rejected": -
|
384 |
"step": 240
|
385 |
},
|
386 |
{
|
387 |
"epoch": 0.6,
|
388 |
-
"grad_norm":
|
389 |
-
"learning_rate":
|
390 |
-
"logits/chosen": -2.
|
391 |
-
"logits/rejected": -2.
|
392 |
-
"logps/chosen": -0.
|
393 |
-
"logps/rejected": -0.
|
394 |
-
"loss":
|
395 |
-
"rewards/accuracies": 0.
|
396 |
-
"rewards/chosen":
|
397 |
-
"rewards/margins":
|
398 |
-
"rewards/rejected": -
|
399 |
"step": 250
|
400 |
},
|
401 |
{
|
402 |
"epoch": 0.63,
|
403 |
-
"grad_norm":
|
404 |
-
"learning_rate":
|
405 |
-
"logits/chosen": -2.
|
406 |
-
"logits/rejected": -2.
|
407 |
-
"logps/chosen": -1.
|
408 |
-
"logps/rejected": -0.
|
409 |
-
"loss":
|
410 |
-
"rewards/accuracies": 0.
|
411 |
-
"rewards/chosen":
|
412 |
-
"rewards/margins":
|
413 |
-
"rewards/rejected": -
|
414 |
"step": 260
|
415 |
},
|
416 |
{
|
417 |
"epoch": 0.65,
|
418 |
-
"grad_norm":
|
419 |
-
"learning_rate":
|
420 |
-
"logits/chosen": -2.
|
421 |
-
"logits/rejected": -2.
|
422 |
-
"logps/chosen": -0.
|
423 |
-
"logps/rejected": -0.
|
424 |
-
"loss":
|
425 |
-
"rewards/accuracies": 0.
|
426 |
-
"rewards/chosen": -0.
|
427 |
-
"rewards/margins":
|
428 |
-
"rewards/rejected": -
|
429 |
"step": 270
|
430 |
},
|
431 |
{
|
432 |
"epoch": 0.68,
|
433 |
-
"grad_norm":
|
434 |
-
"learning_rate":
|
435 |
-
"logits/chosen": -2.
|
436 |
-
"logits/rejected": -2.
|
437 |
-
"logps/chosen": -0.
|
438 |
-
"logps/rejected": -0.
|
439 |
-
"loss":
|
440 |
-
"rewards/accuracies": 0.
|
441 |
-
"rewards/chosen":
|
442 |
-
"rewards/margins":
|
443 |
-
"rewards/rejected": -
|
444 |
"step": 280
|
445 |
},
|
446 |
{
|
447 |
"epoch": 0.7,
|
448 |
-
"grad_norm":
|
449 |
-
"learning_rate":
|
450 |
-
"logits/chosen": -2.
|
451 |
-
"logits/rejected": -2.
|
452 |
-
"logps/chosen": -
|
453 |
-
"logps/rejected": -0.
|
454 |
-
"loss":
|
455 |
-
"rewards/accuracies": 0.
|
456 |
-
"rewards/chosen":
|
457 |
-
"rewards/margins":
|
458 |
-
"rewards/rejected": -
|
459 |
"step": 290
|
460 |
},
|
461 |
{
|
462 |
"epoch": 0.73,
|
463 |
-
"grad_norm":
|
464 |
-
"learning_rate":
|
465 |
-
"logits/chosen": -2.
|
466 |
-
"logits/rejected": -2.
|
467 |
-
"logps/chosen": -0.
|
468 |
-
"logps/rejected": -0.
|
469 |
-
"loss":
|
470 |
-
"rewards/accuracies": 0.
|
471 |
-
"rewards/chosen":
|
472 |
-
"rewards/margins":
|
473 |
-
"rewards/rejected": -
|
474 |
"step": 300
|
475 |
},
|
476 |
{
|
477 |
"epoch": 0.75,
|
478 |
-
"grad_norm":
|
479 |
-
"learning_rate":
|
480 |
-
"logits/chosen": -2.
|
481 |
-
"logits/rejected": -2.
|
482 |
-
"logps/chosen": -1.
|
483 |
-
"logps/rejected": -1.
|
484 |
-
"loss":
|
485 |
"rewards/accuracies": 0.8062499761581421,
|
486 |
-
"rewards/chosen": -
|
487 |
-
"rewards/margins":
|
488 |
-
"rewards/rejected": -
|
489 |
"step": 310
|
490 |
},
|
491 |
{
|
492 |
"epoch": 0.77,
|
493 |
-
"grad_norm":
|
494 |
-
"learning_rate":
|
495 |
-
"logits/chosen": -2.
|
496 |
-
"logits/rejected": -2.
|
497 |
-
"logps/chosen": -0.
|
498 |
-
"logps/rejected": -0.
|
499 |
-
"loss":
|
500 |
-
"rewards/accuracies": 0.
|
501 |
-
"rewards/chosen":
|
502 |
-
"rewards/margins":
|
503 |
-
"rewards/rejected": -
|
504 |
"step": 320
|
505 |
},
|
506 |
{
|
507 |
"epoch": 0.8,
|
508 |
-
"grad_norm":
|
509 |
-
"learning_rate":
|
510 |
-
"logits/chosen": -2.
|
511 |
-
"logits/rejected": -2.
|
512 |
-
"logps/chosen": -0.
|
513 |
-
"logps/rejected": -0.
|
514 |
-
"loss":
|
515 |
-
"rewards/accuracies": 0.
|
516 |
-
"rewards/chosen":
|
517 |
-
"rewards/margins":
|
518 |
-
"rewards/rejected": -
|
519 |
"step": 330
|
520 |
},
|
521 |
{
|
522 |
"epoch": 0.82,
|
523 |
-
"grad_norm":
|
524 |
-
"learning_rate":
|
525 |
-
"logits/chosen": -2.
|
526 |
-
"logits/rejected": -2.
|
527 |
-
"logps/chosen": -0.
|
528 |
-
"logps/rejected": -0.
|
529 |
-
"loss":
|
530 |
-
"rewards/accuracies": 0.
|
531 |
-
"rewards/chosen":
|
532 |
-
"rewards/margins":
|
533 |
-
"rewards/rejected": -
|
534 |
"step": 340
|
535 |
},
|
536 |
{
|
537 |
"epoch": 0.85,
|
538 |
-
"grad_norm":
|
539 |
-
"learning_rate":
|
540 |
-
"logits/chosen": -2.
|
541 |
-
"logits/rejected": -2.
|
542 |
-
"logps/chosen": -0.
|
543 |
-
"logps/rejected": -0.
|
544 |
-
"loss":
|
545 |
-
"rewards/accuracies": 0.
|
546 |
-
"rewards/chosen": -
|
547 |
-
"rewards/margins":
|
548 |
-
"rewards/rejected": -
|
549 |
"step": 350
|
550 |
},
|
551 |
{
|
552 |
"epoch": 0.87,
|
553 |
-
"grad_norm":
|
554 |
-
"learning_rate":
|
555 |
-
"logits/chosen": -2.
|
556 |
-
"logits/rejected": -2.
|
557 |
-
"logps/chosen": -0.
|
558 |
-
"logps/rejected": -1.
|
559 |
-
"loss":
|
560 |
"rewards/accuracies": 0.7749999761581421,
|
561 |
-
"rewards/chosen":
|
562 |
-
"rewards/margins":
|
563 |
-
"rewards/rejected": -
|
564 |
"step": 360
|
565 |
},
|
566 |
{
|
567 |
"epoch": 0.89,
|
568 |
-
"grad_norm":
|
569 |
-
"learning_rate":
|
570 |
-
"logits/chosen": -2.
|
571 |
-
"logits/rejected": -2.
|
572 |
-
"logps/chosen": -0.
|
573 |
-
"logps/rejected": -0.
|
574 |
-
"loss":
|
575 |
-
"rewards/accuracies": 0.
|
576 |
-
"rewards/chosen": 0.
|
577 |
-
"rewards/margins":
|
578 |
-
"rewards/rejected": -
|
579 |
"step": 370
|
580 |
},
|
581 |
{
|
582 |
"epoch": 0.92,
|
583 |
-
"grad_norm":
|
584 |
-
"learning_rate":
|
585 |
-
"logits/chosen": -2.
|
586 |
-
"logits/rejected": -2.
|
587 |
-
"logps/chosen": -0.
|
588 |
-
"logps/rejected": -0.
|
589 |
-
"loss":
|
590 |
-
"rewards/accuracies": 0.
|
591 |
-
"rewards/chosen":
|
592 |
-
"rewards/margins":
|
593 |
-
"rewards/rejected": -
|
594 |
"step": 380
|
595 |
},
|
596 |
{
|
597 |
"epoch": 0.94,
|
598 |
-
"grad_norm":
|
599 |
-
"learning_rate":
|
600 |
-
"logits/chosen": -2.
|
601 |
-
"logits/rejected": -2.
|
602 |
-
"logps/chosen": -0.
|
603 |
-
"logps/rejected": -0.
|
604 |
-
"loss":
|
605 |
-
"rewards/accuracies": 0.
|
606 |
-
"rewards/chosen":
|
607 |
-
"rewards/margins":
|
608 |
-
"rewards/rejected": -
|
609 |
"step": 390
|
610 |
},
|
611 |
{
|
612 |
"epoch": 0.97,
|
613 |
-
"grad_norm":
|
614 |
-
"learning_rate":
|
615 |
-
"logits/chosen": -2.
|
616 |
-
"logits/rejected": -2.
|
617 |
-
"logps/chosen": -0.
|
618 |
-
"logps/rejected": -0.
|
619 |
-
"loss":
|
620 |
-
"rewards/accuracies": 0.
|
621 |
-
"rewards/chosen":
|
622 |
-
"rewards/margins":
|
623 |
-
"rewards/rejected": -
|
624 |
"step": 400
|
625 |
},
|
626 |
{
|
627 |
"epoch": 0.99,
|
628 |
-
"grad_norm":
|
629 |
-
"learning_rate":
|
630 |
-
"logits/chosen": -2.
|
631 |
-
"logits/rejected": -2.
|
632 |
-
"logps/chosen": -0.
|
633 |
-
"logps/rejected": -0.
|
634 |
-
"loss":
|
635 |
-
"rewards/accuracies": 0.
|
636 |
-
"rewards/chosen": -
|
637 |
-
"rewards/margins":
|
638 |
-
"rewards/rejected": -
|
639 |
"step": 410
|
640 |
},
|
641 |
{
|
642 |
"epoch": 1.0,
|
643 |
"step": 413,
|
644 |
"total_flos": 0.0,
|
645 |
-
"train_loss":
|
646 |
-
"train_runtime":
|
647 |
-
"train_samples_per_second": 8.
|
648 |
"train_steps_per_second": 0.065
|
649 |
}
|
650 |
],
|
|
|
10 |
"log_history": [
|
11 |
{
|
12 |
"epoch": 0.0,
|
13 |
+
"grad_norm": 6401.270603874373,
|
14 |
+
"learning_rate": 9.523809523809522e-09,
|
15 |
"logits/chosen": -2.7005977630615234,
|
16 |
"logits/rejected": -2.6288318634033203,
|
17 |
"logps/chosen": -1.1158788204193115,
|
18 |
"logps/rejected": -1.1333446502685547,
|
19 |
+
"loss": 0.7544,
|
20 |
"rewards/accuracies": 0.0,
|
21 |
"rewards/chosen": 0.0,
|
22 |
"rewards/margins": 0.0,
|
|
|
25 |
},
|
26 |
{
|
27 |
"epoch": 0.02,
|
28 |
+
"grad_norm": 9369.590990783972,
|
29 |
+
"learning_rate": 9.523809523809523e-08,
|
30 |
+
"logits/chosen": -2.76228666305542,
|
31 |
+
"logits/rejected": -2.6970374584198,
|
32 |
+
"logps/chosen": -0.837486743927002,
|
33 |
+
"logps/rejected": -0.8182350993156433,
|
34 |
+
"loss": 0.9695,
|
35 |
+
"rewards/accuracies": 0.4305555522441864,
|
36 |
+
"rewards/chosen": 0.06597563624382019,
|
37 |
+
"rewards/margins": 0.437710702419281,
|
38 |
+
"rewards/rejected": -0.3717350959777832,
|
39 |
"step": 10
|
40 |
},
|
41 |
{
|
42 |
"epoch": 0.05,
|
43 |
+
"grad_norm": 5966.657402243146,
|
44 |
+
"learning_rate": 1.9047619047619045e-07,
|
45 |
+
"logits/chosen": -2.6901049613952637,
|
46 |
+
"logits/rejected": -2.6502909660339355,
|
47 |
+
"logps/chosen": -0.9933319091796875,
|
48 |
+
"logps/rejected": -1.0394352674484253,
|
49 |
+
"loss": 1.0318,
|
50 |
+
"rewards/accuracies": 0.6499999761581421,
|
51 |
+
"rewards/chosen": -0.794396698474884,
|
52 |
+
"rewards/margins": 0.7471516728401184,
|
53 |
+
"rewards/rejected": -1.5415483713150024,
|
54 |
"step": 20
|
55 |
},
|
56 |
{
|
57 |
"epoch": 0.07,
|
58 |
+
"grad_norm": 8820.198504372876,
|
59 |
+
"learning_rate": 2.857142857142857e-07,
|
60 |
+
"logits/chosen": -2.7333264350891113,
|
61 |
+
"logits/rejected": -2.6793360710144043,
|
62 |
+
"logps/chosen": -0.9710652232170105,
|
63 |
+
"logps/rejected": -0.9799602627754211,
|
64 |
+
"loss": 1.3198,
|
65 |
+
"rewards/accuracies": 0.706250011920929,
|
66 |
+
"rewards/chosen": -2.275942325592041,
|
67 |
+
"rewards/margins": 0.9020620584487915,
|
68 |
+
"rewards/rejected": -3.178004264831543,
|
69 |
"step": 30
|
70 |
},
|
71 |
{
|
72 |
"epoch": 0.1,
|
73 |
+
"grad_norm": 8453.783513094899,
|
74 |
+
"learning_rate": 3.809523809523809e-07,
|
75 |
+
"logits/chosen": -2.6771621704101562,
|
76 |
+
"logits/rejected": -2.6321842670440674,
|
77 |
+
"logps/chosen": -0.989823043346405,
|
78 |
+
"logps/rejected": -0.9216930270195007,
|
79 |
+
"loss": 2.0555,
|
80 |
+
"rewards/accuracies": 0.762499988079071,
|
81 |
+
"rewards/chosen": -0.6586966514587402,
|
82 |
+
"rewards/margins": 5.100310325622559,
|
83 |
+
"rewards/rejected": -5.759006500244141,
|
84 |
"step": 40
|
85 |
},
|
86 |
{
|
87 |
"epoch": 0.12,
|
88 |
+
"grad_norm": 4192.139232222726,
|
89 |
+
"learning_rate": 3.995412608484087e-07,
|
90 |
+
"logits/chosen": -2.743403911590576,
|
91 |
+
"logits/rejected": -2.6878693103790283,
|
92 |
+
"logps/chosen": -0.9671042561531067,
|
93 |
+
"logps/rejected": -0.917597770690918,
|
94 |
+
"loss": 2.6495,
|
95 |
+
"rewards/accuracies": 0.793749988079071,
|
96 |
+
"rewards/chosen": -0.16885781288146973,
|
97 |
+
"rewards/margins": 5.186079978942871,
|
98 |
+
"rewards/rejected": -5.35493803024292,
|
99 |
"step": 50
|
100 |
},
|
101 |
{
|
102 |
"epoch": 0.15,
|
103 |
+
"grad_norm": 5643.860863524967,
|
104 |
+
"learning_rate": 3.976812391485896e-07,
|
105 |
+
"logits/chosen": -2.7438769340515137,
|
106 |
+
"logits/rejected": -2.676765203475952,
|
107 |
+
"logps/chosen": -0.911353588104248,
|
108 |
+
"logps/rejected": -0.9122518301010132,
|
109 |
+
"loss": 3.8047,
|
110 |
+
"rewards/accuracies": 0.8062499761581421,
|
111 |
+
"rewards/chosen": 2.4976494312286377,
|
112 |
+
"rewards/margins": 5.426072120666504,
|
113 |
+
"rewards/rejected": -2.928422212600708,
|
114 |
"step": 60
|
115 |
},
|
116 |
{
|
117 |
"epoch": 0.17,
|
118 |
+
"grad_norm": 4497.230754903385,
|
119 |
+
"learning_rate": 3.9440458281608213e-07,
|
120 |
+
"logits/chosen": -2.740940570831299,
|
121 |
+
"logits/rejected": -2.7162723541259766,
|
122 |
+
"logps/chosen": -0.9154363870620728,
|
123 |
+
"logps/rejected": -0.868497371673584,
|
124 |
+
"loss": 3.6432,
|
125 |
+
"rewards/accuracies": 0.824999988079071,
|
126 |
+
"rewards/chosen": 5.681364059448242,
|
127 |
+
"rewards/margins": 8.000432968139648,
|
128 |
+
"rewards/rejected": -2.319068431854248,
|
129 |
"step": 70
|
130 |
},
|
131 |
{
|
132 |
"epoch": 0.19,
|
133 |
+
"grad_norm": 9843.974275847575,
|
134 |
+
"learning_rate": 3.897347732134074e-07,
|
135 |
+
"logits/chosen": -2.679215908050537,
|
136 |
+
"logits/rejected": -2.625516891479492,
|
137 |
+
"logps/chosen": -0.9146322011947632,
|
138 |
+
"logps/rejected": -1.0181081295013428,
|
139 |
+
"loss": 5.767,
|
140 |
+
"rewards/accuracies": 0.800000011920929,
|
141 |
+
"rewards/chosen": -10.08639907836914,
|
142 |
+
"rewards/margins": 6.582289695739746,
|
143 |
+
"rewards/rejected": -16.668689727783203,
|
144 |
"step": 80
|
145 |
},
|
146 |
{
|
147 |
"epoch": 0.22,
|
148 |
+
"grad_norm": 4773.013380320505,
|
149 |
+
"learning_rate": 3.8370527539794614e-07,
|
150 |
+
"logits/chosen": -2.6771388053894043,
|
151 |
+
"logits/rejected": -2.6291418075561523,
|
152 |
+
"logps/chosen": -1.003847360610962,
|
153 |
+
"logps/rejected": -1.0297266244888306,
|
154 |
+
"loss": 4.6354,
|
155 |
+
"rewards/accuracies": 0.737500011920929,
|
156 |
+
"rewards/chosen": 4.863407611846924,
|
157 |
+
"rewards/margins": 9.78220272064209,
|
158 |
+
"rewards/rejected": -4.918795585632324,
|
159 |
"step": 90
|
160 |
},
|
161 |
{
|
162 |
"epoch": 0.24,
|
163 |
+
"grad_norm": 3074.8663144850243,
|
164 |
+
"learning_rate": 3.763592983027255e-07,
|
165 |
+
"logits/chosen": -2.705735683441162,
|
166 |
+
"logits/rejected": -2.6605448722839355,
|
167 |
+
"logps/chosen": -0.9163268804550171,
|
168 |
+
"logps/rejected": -0.9396775960922241,
|
169 |
+
"loss": 5.8585,
|
170 |
"rewards/accuracies": 0.8062499761581421,
|
171 |
+
"rewards/chosen": -12.477940559387207,
|
172 |
+
"rewards/margins": 7.702305793762207,
|
173 |
+
"rewards/rejected": -20.180248260498047,
|
174 |
"step": 100
|
175 |
},
|
176 |
{
|
177 |
"epoch": 0.27,
|
178 |
+
"grad_norm": 4099.610429119441,
|
179 |
+
"learning_rate": 3.6774948509008527e-07,
|
180 |
+
"logits/chosen": -2.714970111846924,
|
181 |
+
"logits/rejected": -2.6705470085144043,
|
182 |
+
"logps/chosen": -0.9598251581192017,
|
183 |
+
"logps/rejected": -0.9319995641708374,
|
184 |
+
"loss": 5.1529,
|
185 |
+
"rewards/accuracies": 0.7875000238418579,
|
186 |
+
"rewards/chosen": 9.19798755645752,
|
187 |
+
"rewards/margins": 10.779365539550781,
|
188 |
+
"rewards/rejected": -1.5813770294189453,
|
189 |
"step": 110
|
190 |
},
|
191 |
{
|
192 |
"epoch": 0.29,
|
193 |
+
"grad_norm": 4599.711217449366,
|
194 |
+
"learning_rate": 3.579375358972288e-07,
|
195 |
+
"logits/chosen": -2.678779125213623,
|
196 |
+
"logits/rejected": -2.6315762996673584,
|
197 |
+
"logps/chosen": -0.9081487655639648,
|
198 |
+
"logps/rejected": -1.0060938596725464,
|
199 |
+
"loss": 4.0915,
|
200 |
+
"rewards/accuracies": 0.768750011920929,
|
201 |
+
"rewards/chosen": -13.663342475891113,
|
202 |
+
"rewards/margins": 9.755656242370605,
|
203 |
+
"rewards/rejected": -23.418994903564453,
|
204 |
"step": 120
|
205 |
},
|
206 |
{
|
207 |
"epoch": 0.31,
|
208 |
+
"grad_norm": 4010.334966061441,
|
209 |
+
"learning_rate": 3.4699376567716156e-07,
|
210 |
+
"logits/chosen": -2.7230353355407715,
|
211 |
+
"logits/rejected": -2.684389591217041,
|
212 |
+
"logps/chosen": -0.8652521967887878,
|
213 |
+
"logps/rejected": -0.8799147605895996,
|
214 |
+
"loss": 4.4027,
|
215 |
+
"rewards/accuracies": 0.762499988079071,
|
216 |
+
"rewards/chosen": 2.19469952583313,
|
217 |
+
"rewards/margins": 15.263641357421875,
|
218 |
+
"rewards/rejected": -13.068939208984375,
|
219 |
"step": 130
|
220 |
},
|
221 |
{
|
222 |
"epoch": 0.34,
|
223 |
+
"grad_norm": 5239.11146834966,
|
224 |
+
"learning_rate": 3.349966003036421e-07,
|
225 |
+
"logits/chosen": -2.689558506011963,
|
226 |
+
"logits/rejected": -2.649766445159912,
|
227 |
+
"logps/chosen": -0.9352903366088867,
|
228 |
+
"logps/rejected": -0.9416161775588989,
|
229 |
+
"loss": 4.7953,
|
230 |
+
"rewards/accuracies": 0.7749999761581421,
|
231 |
+
"rewards/chosen": -4.734063148498535,
|
232 |
+
"rewards/margins": 8.841203689575195,
|
233 |
+
"rewards/rejected": -13.575268745422363,
|
234 |
"step": 140
|
235 |
},
|
236 |
{
|
237 |
"epoch": 0.36,
|
238 |
+
"grad_norm": 5394.35498681908,
|
239 |
+
"learning_rate": 3.220320145511884e-07,
|
240 |
+
"logits/chosen": -2.7070841789245605,
|
241 |
+
"logits/rejected": -2.647737979888916,
|
242 |
+
"logps/chosen": -0.9441506266593933,
|
243 |
+
"logps/rejected": -0.9885166883468628,
|
244 |
+
"loss": 4.2219,
|
245 |
+
"rewards/accuracies": 0.793749988079071,
|
246 |
+
"rewards/chosen": 5.9402689933776855,
|
247 |
+
"rewards/margins": 12.97706413269043,
|
248 |
+
"rewards/rejected": -7.036795139312744,
|
249 |
"step": 150
|
250 |
},
|
251 |
{
|
252 |
"epoch": 0.39,
|
253 |
+
"grad_norm": 5022.189692479379,
|
254 |
+
"learning_rate": 3.0819291597771795e-07,
|
255 |
+
"logits/chosen": -2.7051825523376465,
|
256 |
+
"logits/rejected": -2.667494297027588,
|
257 |
+
"logps/chosen": -0.911395251750946,
|
258 |
+
"logps/rejected": -0.939487099647522,
|
259 |
+
"loss": 4.7963,
|
260 |
+
"rewards/accuracies": 0.856249988079071,
|
261 |
+
"rewards/chosen": -0.6114660501480103,
|
262 |
+
"rewards/margins": 9.443866729736328,
|
263 |
+
"rewards/rejected": -10.055331230163574,
|
264 |
"step": 160
|
265 |
},
|
266 |
{
|
267 |
"epoch": 0.41,
|
268 |
+
"grad_norm": 5428.944545727042,
|
269 |
+
"learning_rate": 2.9357847912507786e-07,
|
270 |
+
"logits/chosen": -2.6787288188934326,
|
271 |
+
"logits/rejected": -2.609421968460083,
|
272 |
+
"logps/chosen": -0.8976411819458008,
|
273 |
+
"logps/rejected": -0.8857674598693848,
|
274 |
+
"loss": 4.6262,
|
275 |
"rewards/accuracies": 0.800000011920929,
|
276 |
+
"rewards/chosen": -2.82297945022583,
|
277 |
+
"rewards/margins": 10.640687942504883,
|
278 |
+
"rewards/rejected": -13.463666915893555,
|
279 |
"step": 170
|
280 |
},
|
281 |
{
|
282 |
"epoch": 0.44,
|
283 |
+
"grad_norm": 7317.882582449178,
|
284 |
+
"learning_rate": 2.7829343480875617e-07,
|
285 |
+
"logits/chosen": -2.6716930866241455,
|
286 |
+
"logits/rejected": -2.6018152236938477,
|
287 |
+
"logps/chosen": -0.9342878460884094,
|
288 |
+
"logps/rejected": -0.9536906480789185,
|
289 |
+
"loss": 4.5209,
|
290 |
+
"rewards/accuracies": 0.800000011920929,
|
291 |
+
"rewards/chosen": 4.173262596130371,
|
292 |
+
"rewards/margins": 8.933877944946289,
|
293 |
+
"rewards/rejected": -4.760615348815918,
|
294 |
"step": 180
|
295 |
},
|
296 |
{
|
297 |
"epoch": 0.46,
|
298 |
+
"grad_norm": 5046.2946182405685,
|
299 |
+
"learning_rate": 2.624473195899052e-07,
|
300 |
+
"logits/chosen": -2.737992763519287,
|
301 |
+
"logits/rejected": -2.7089955806732178,
|
302 |
+
"logps/chosen": -0.9629039764404297,
|
303 |
+
"logps/rejected": -1.039236307144165,
|
304 |
+
"loss": 4.5521,
|
305 |
+
"rewards/accuracies": 0.78125,
|
306 |
+
"rewards/chosen": -3.3569388389587402,
|
307 |
+
"rewards/margins": 13.995905876159668,
|
308 |
+
"rewards/rejected": -17.352848052978516,
|
309 |
"step": 190
|
310 |
},
|
311 |
{
|
312 |
"epoch": 0.48,
|
313 |
+
"grad_norm": 4268.8163809344915,
|
314 |
+
"learning_rate": 2.4615369080815547e-07,
|
315 |
+
"logits/chosen": -2.6982626914978027,
|
316 |
+
"logits/rejected": -2.6629488468170166,
|
317 |
+
"logps/chosen": -0.8523995280265808,
|
318 |
+
"logps/rejected": -0.9246847033500671,
|
319 |
+
"loss": 3.8184,
|
320 |
+
"rewards/accuracies": 0.78125,
|
321 |
+
"rewards/chosen": 2.3979854583740234,
|
322 |
+
"rewards/margins": 4.996596336364746,
|
323 |
+
"rewards/rejected": -2.5986106395721436,
|
324 |
"step": 200
|
325 |
},
|
326 |
{
|
327 |
"epoch": 0.51,
|
328 |
+
"grad_norm": 4211.242306423206,
|
329 |
+
"learning_rate": 2.2952931280049625e-07,
|
330 |
+
"logits/chosen": -2.7346115112304688,
|
331 |
+
"logits/rejected": -2.6734609603881836,
|
332 |
+
"logps/chosen": -1.0063531398773193,
|
333 |
+
"logps/rejected": -0.9570119976997375,
|
334 |
+
"loss": 4.9954,
|
335 |
+
"rewards/accuracies": 0.762499988079071,
|
336 |
+
"rewards/chosen": 7.86081600189209,
|
337 |
+
"rewards/margins": 13.075413703918457,
|
338 |
+
"rewards/rejected": -5.214597225189209,
|
339 |
"step": 210
|
340 |
},
|
341 |
{
|
342 |
"epoch": 0.53,
|
343 |
+
"grad_norm": 4906.2448320907815,
|
344 |
+
"learning_rate": 2.1269332013798747e-07,
|
345 |
+
"logits/chosen": -2.7431142330169678,
|
346 |
+
"logits/rejected": -2.7241249084472656,
|
347 |
+
"logps/chosen": -0.8835189938545227,
|
348 |
+
"logps/rejected": -0.8670462369918823,
|
349 |
+
"loss": 4.3795,
|
350 |
+
"rewards/accuracies": 0.78125,
|
351 |
+
"rewards/chosen": 4.582579135894775,
|
352 |
+
"rewards/margins": 8.150335311889648,
|
353 |
+
"rewards/rejected": -3.567755937576294,
|
354 |
"step": 220
|
355 |
},
|
356 |
{
|
357 |
"epoch": 0.56,
|
358 |
+
"grad_norm": 2995.5119741253625,
|
359 |
+
"learning_rate": 1.9576636387676436e-07,
|
360 |
+
"logits/chosen": -2.690732955932617,
|
361 |
+
"logits/rejected": -2.653067111968994,
|
362 |
+
"logps/chosen": -0.960831344127655,
|
363 |
+
"logps/rejected": -0.9556485414505005,
|
364 |
+
"loss": 4.0487,
|
365 |
+
"rewards/accuracies": 0.7562500238418579,
|
366 |
+
"rewards/chosen": -8.449748039245605,
|
367 |
+
"rewards/margins": 10.095115661621094,
|
368 |
+
"rewards/rejected": -18.544864654541016,
|
369 |
"step": 230
|
370 |
},
|
371 |
{
|
372 |
"epoch": 0.58,
|
373 |
+
"grad_norm": 3504.414329050279,
|
374 |
+
"learning_rate": 1.7886974694151976e-07,
|
375 |
+
"logits/chosen": -2.7119805812835693,
|
376 |
+
"logits/rejected": -2.6879172325134277,
|
377 |
+
"logps/chosen": -0.990290641784668,
|
378 |
+
"logps/rejected": -0.9934972524642944,
|
379 |
+
"loss": 4.3644,
|
380 |
+
"rewards/accuracies": 0.793749988079071,
|
381 |
+
"rewards/chosen": -0.5965616106987,
|
382 |
+
"rewards/margins": 9.211602210998535,
|
383 |
+
"rewards/rejected": -9.808164596557617,
|
384 |
"step": 240
|
385 |
},
|
386 |
{
|
387 |
"epoch": 0.6,
|
388 |
+
"grad_norm": 3400.5848210057216,
|
389 |
+
"learning_rate": 1.6212455483752895e-07,
|
390 |
+
"logits/chosen": -2.756906270980835,
|
391 |
+
"logits/rejected": -2.6796135902404785,
|
392 |
+
"logps/chosen": -0.8838168978691101,
|
393 |
+
"logps/rejected": -0.9137406349182129,
|
394 |
+
"loss": 4.5034,
|
395 |
+
"rewards/accuracies": 0.7875000238418579,
|
396 |
+
"rewards/chosen": 6.5281982421875,
|
397 |
+
"rewards/margins": 8.64702033996582,
|
398 |
+
"rewards/rejected": -2.1188230514526367,
|
399 |
"step": 250
|
400 |
},
|
401 |
{
|
402 |
"epoch": 0.63,
|
403 |
+
"grad_norm": 6194.117841583386,
|
404 |
+
"learning_rate": 1.4565078792075733e-07,
|
405 |
+
"logits/chosen": -2.7132773399353027,
|
406 |
+
"logits/rejected": -2.6494650840759277,
|
407 |
+
"logps/chosen": -1.002362847328186,
|
408 |
+
"logps/rejected": -0.9982520341873169,
|
409 |
+
"loss": 4.8134,
|
410 |
+
"rewards/accuracies": 0.84375,
|
411 |
+
"rewards/chosen": 3.0224878787994385,
|
412 |
+
"rewards/margins": 16.206506729125977,
|
413 |
+
"rewards/rejected": -13.1840181350708,
|
414 |
"step": 260
|
415 |
},
|
416 |
{
|
417 |
"epoch": 0.65,
|
418 |
+
"grad_norm": 4565.495892627232,
|
419 |
+
"learning_rate": 1.295665014444281e-07,
|
420 |
+
"logits/chosen": -2.7381529808044434,
|
421 |
+
"logits/rejected": -2.6608738899230957,
|
422 |
+
"logps/chosen": -0.9501218795776367,
|
423 |
+
"logps/rejected": -0.9476363062858582,
|
424 |
+
"loss": 5.3754,
|
425 |
+
"rewards/accuracies": 0.8500000238418579,
|
426 |
+
"rewards/chosen": -0.570526123046875,
|
427 |
+
"rewards/margins": 12.367398262023926,
|
428 |
+
"rewards/rejected": -12.9379243850708,
|
429 |
"step": 270
|
430 |
},
|
431 |
{
|
432 |
"epoch": 0.68,
|
433 |
+
"grad_norm": 5337.153187944306,
|
434 |
+
"learning_rate": 1.1398695954469597e-07,
|
435 |
+
"logits/chosen": -2.6872425079345703,
|
436 |
+
"logits/rejected": -2.630267381668091,
|
437 |
+
"logps/chosen": -0.9056104421615601,
|
438 |
+
"logps/rejected": -0.8939152956008911,
|
439 |
+
"loss": 4.1053,
|
440 |
+
"rewards/accuracies": 0.793749988079071,
|
441 |
+
"rewards/chosen": 4.294297218322754,
|
442 |
+
"rewards/margins": 7.472552299499512,
|
443 |
+
"rewards/rejected": -3.1782548427581787,
|
444 |
"step": 280
|
445 |
},
|
446 |
{
|
447 |
"epoch": 0.7,
|
448 |
+
"grad_norm": 3582.07962645892,
|
449 |
+
"learning_rate": 9.902380922818425e-08,
|
450 |
+
"logits/chosen": -2.7334370613098145,
|
451 |
+
"logits/rejected": -2.6919913291931152,
|
452 |
+
"logps/chosen": -0.9840775728225708,
|
453 |
+
"logps/rejected": -0.9756690263748169,
|
454 |
+
"loss": 3.2759,
|
455 |
+
"rewards/accuracies": 0.762499988079071,
|
456 |
+
"rewards/chosen": 8.966680526733398,
|
457 |
+
"rewards/margins": 11.496904373168945,
|
458 |
+
"rewards/rejected": -2.5302233695983887,
|
459 |
"step": 290
|
460 |
},
|
461 |
{
|
462 |
"epoch": 0.73,
|
463 |
+
"grad_norm": 4767.591882910886,
|
464 |
+
"learning_rate": 8.478428028080398e-08,
|
465 |
+
"logits/chosen": -2.7305169105529785,
|
466 |
+
"logits/rejected": -2.6773815155029297,
|
467 |
+
"logps/chosen": -0.8988749384880066,
|
468 |
+
"logps/rejected": -0.9437707662582397,
|
469 |
+
"loss": 4.3175,
|
470 |
+
"rewards/accuracies": 0.7875000238418579,
|
471 |
+
"rewards/chosen": 0.8898951411247253,
|
472 |
+
"rewards/margins": 8.447718620300293,
|
473 |
+
"rewards/rejected": -7.55782413482666,
|
474 |
"step": 300
|
475 |
},
|
476 |
{
|
477 |
"epoch": 0.75,
|
478 |
+
"grad_norm": 4819.380329592898,
|
479 |
+
"learning_rate": 7.137041683151202e-08,
|
480 |
+
"logits/chosen": -2.7228643894195557,
|
481 |
+
"logits/rejected": -2.6581058502197266,
|
482 |
+
"logps/chosen": -1.0781683921813965,
|
483 |
+
"logps/rejected": -1.028840184211731,
|
484 |
+
"loss": 2.9744,
|
485 |
"rewards/accuracies": 0.8062499761581421,
|
486 |
+
"rewards/chosen": -2.4226202964782715,
|
487 |
+
"rewards/margins": 13.473236083984375,
|
488 |
+
"rewards/rejected": -15.895855903625488,
|
489 |
"step": 310
|
490 |
},
|
491 |
{
|
492 |
"epoch": 0.77,
|
493 |
+
"grad_norm": 7840.551721640683,
|
494 |
+
"learning_rate": 5.8878346077822135e-08,
|
495 |
+
"logits/chosen": -2.7280871868133545,
|
496 |
+
"logits/rejected": -2.649958848953247,
|
497 |
+
"logps/chosen": -0.9020591974258423,
|
498 |
+
"logps/rejected": -0.9361578822135925,
|
499 |
+
"loss": 2.7082,
|
500 |
+
"rewards/accuracies": 0.8812500238418579,
|
501 |
+
"rewards/chosen": 1.223115086555481,
|
502 |
+
"rewards/margins": 10.582406997680664,
|
503 |
+
"rewards/rejected": -9.35929012298584,
|
504 |
"step": 320
|
505 |
},
|
506 |
{
|
507 |
"epoch": 0.8,
|
508 |
+
"grad_norm": 4662.77535052248,
|
509 |
+
"learning_rate": 4.73975894135696e-08,
|
510 |
+
"logits/chosen": -2.6770853996276855,
|
511 |
+
"logits/rejected": -2.6099040508270264,
|
512 |
+
"logps/chosen": -0.9263202548027039,
|
513 |
+
"logps/rejected": -0.9608638882637024,
|
514 |
+
"loss": 3.1985,
|
515 |
+
"rewards/accuracies": 0.8125,
|
516 |
+
"rewards/chosen": 0.2365754395723343,
|
517 |
+
"rewards/margins": 13.195585250854492,
|
518 |
+
"rewards/rejected": -12.959010124206543,
|
519 |
"step": 330
|
520 |
},
|
521 |
{
|
522 |
"epoch": 0.82,
|
523 |
+
"grad_norm": 4550.588002339864,
|
524 |
+
"learning_rate": 3.701042089556483e-08,
|
525 |
+
"logits/chosen": -2.756493330001831,
|
526 |
+
"logits/rejected": -2.687851667404175,
|
527 |
+
"logps/chosen": -0.8901381492614746,
|
528 |
+
"logps/rejected": -0.9301478266716003,
|
529 |
+
"loss": 3.841,
|
530 |
+
"rewards/accuracies": 0.78125,
|
531 |
+
"rewards/chosen": 0.11963929980993271,
|
532 |
+
"rewards/margins": 7.3289618492126465,
|
533 |
+
"rewards/rejected": -7.209322929382324,
|
534 |
"step": 340
|
535 |
},
|
536 |
{
|
537 |
"epoch": 0.85,
|
538 |
+
"grad_norm": 5464.471487236709,
|
539 |
+
"learning_rate": 2.779127764652889e-08,
|
540 |
+
"logits/chosen": -2.689107656478882,
|
541 |
+
"logits/rejected": -2.6330015659332275,
|
542 |
+
"logps/chosen": -0.9756801724433899,
|
543 |
+
"logps/rejected": -0.9646003842353821,
|
544 |
+
"loss": 3.6421,
|
545 |
+
"rewards/accuracies": 0.8062499761581421,
|
546 |
+
"rewards/chosen": -1.1974527835845947,
|
547 |
+
"rewards/margins": 9.013090133666992,
|
548 |
+
"rewards/rejected": -10.210542678833008,
|
549 |
"step": 350
|
550 |
},
|
551 |
{
|
552 |
"epoch": 0.87,
|
553 |
+
"grad_norm": 5949.708940984834,
|
554 |
+
"learning_rate": 1.9806226419516193e-08,
|
555 |
+
"logits/chosen": -2.704460620880127,
|
556 |
+
"logits/rejected": -2.656071186065674,
|
557 |
+
"logps/chosen": -0.9623576402664185,
|
558 |
+
"logps/rejected": -1.0082406997680664,
|
559 |
+
"loss": 3.5231,
|
560 |
"rewards/accuracies": 0.7749999761581421,
|
561 |
+
"rewards/chosen": 2.5273587703704834,
|
562 |
+
"rewards/margins": 11.88086223602295,
|
563 |
+
"rewards/rejected": -9.35350227355957,
|
564 |
"step": 360
|
565 |
},
|
566 |
{
|
567 |
"epoch": 0.89,
|
568 |
+
"grad_norm": 4320.933402478669,
|
569 |
+
"learning_rate": 1.3112490146559552e-08,
|
570 |
+
"logits/chosen": -2.7451281547546387,
|
571 |
+
"logits/rejected": -2.686728000640869,
|
572 |
+
"logps/chosen": -0.8951610326766968,
|
573 |
+
"logps/rejected": -0.89850914478302,
|
574 |
+
"loss": 3.0053,
|
575 |
+
"rewards/accuracies": 0.856249988079071,
|
576 |
+
"rewards/chosen": 0.6753175258636475,
|
577 |
+
"rewards/margins": 12.29626750946045,
|
578 |
+
"rewards/rejected": -11.620949745178223,
|
579 |
"step": 370
|
580 |
},
|
581 |
{
|
582 |
"epoch": 0.92,
|
583 |
+
"grad_norm": 2514.940389992379,
|
584 |
+
"learning_rate": 7.758037864413247e-09,
|
585 |
+
"logits/chosen": -2.7158432006835938,
|
586 |
+
"logits/rejected": -2.6906635761260986,
|
587 |
+
"logps/chosen": -0.9033122062683105,
|
588 |
+
"logps/rejected": -0.9709407091140747,
|
589 |
+
"loss": 2.8751,
|
590 |
+
"rewards/accuracies": 0.856249988079071,
|
591 |
+
"rewards/chosen": 0.9628832936286926,
|
592 |
+
"rewards/margins": 10.83133316040039,
|
593 |
+
"rewards/rejected": -9.868449211120605,
|
594 |
"step": 380
|
595 |
},
|
596 |
{
|
597 |
"epoch": 0.94,
|
598 |
+
"grad_norm": 3504.225752431698,
|
599 |
+
"learning_rate": 3.78124095609087e-09,
|
600 |
+
"logits/chosen": -2.6947999000549316,
|
601 |
+
"logits/rejected": -2.6553878784179688,
|
602 |
+
"logps/chosen": -0.9263744354248047,
|
603 |
+
"logps/rejected": -0.9935058355331421,
|
604 |
+
"loss": 3.019,
|
605 |
+
"rewards/accuracies": 0.862500011920929,
|
606 |
+
"rewards/chosen": 1.4072116613388062,
|
607 |
+
"rewards/margins": 10.741894721984863,
|
608 |
+
"rewards/rejected": -9.334683418273926,
|
609 |
"step": 390
|
610 |
},
|
611 |
{
|
612 |
"epoch": 0.97,
|
613 |
+
"grad_norm": 4987.634749508018,
|
614 |
+
"learning_rate": 1.2105981716597603e-09,
|
615 |
+
"logits/chosen": -2.7300946712493896,
|
616 |
+
"logits/rejected": -2.6389007568359375,
|
617 |
+
"logps/chosen": -0.9686774015426636,
|
618 |
+
"logps/rejected": -0.9328421354293823,
|
619 |
+
"loss": 3.7864,
|
620 |
+
"rewards/accuracies": 0.831250011920929,
|
621 |
+
"rewards/chosen": 0.6465551853179932,
|
622 |
+
"rewards/margins": 11.809611320495605,
|
623 |
+
"rewards/rejected": -11.163057327270508,
|
624 |
"step": 400
|
625 |
},
|
626 |
{
|
627 |
"epoch": 0.99,
|
628 |
+
"grad_norm": 5473.226219590305,
|
629 |
+
"learning_rate": 6.453139886395398e-11,
|
630 |
+
"logits/chosen": -2.7284317016601562,
|
631 |
+
"logits/rejected": -2.6886637210845947,
|
632 |
+
"logps/chosen": -0.9334842562675476,
|
633 |
+
"logps/rejected": -0.9600637555122375,
|
634 |
+
"loss": 3.6391,
|
635 |
+
"rewards/accuracies": 0.875,
|
636 |
+
"rewards/chosen": -2.157397747039795,
|
637 |
+
"rewards/margins": 13.2835054397583,
|
638 |
+
"rewards/rejected": -15.440902709960938,
|
639 |
"step": 410
|
640 |
},
|
641 |
{
|
642 |
"epoch": 1.0,
|
643 |
"step": 413,
|
644 |
"total_flos": 0.0,
|
645 |
+
"train_loss": 3.8421780889894426,
|
646 |
+
"train_runtime": 6381.4933,
|
647 |
+
"train_samples_per_second": 8.293,
|
648 |
"train_steps_per_second": 0.065
|
649 |
}
|
650 |
],
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 6264
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d13f3fc8759aca2ff57e3efc81099fc07a2315dc63120d75fb63db001a764cfe
|
3 |
size 6264
|